75TexLRev989 2

DATE DOWNLOADED: Fri Sep 30 00:51:32 2022
SOURCE: Content Downloaded from HeinOnline
Citations:
Bluebook 21st ed.

Mark A. Lemley, Economics of Improvement in Intellectual Property Law, 75 TEX. L.
REV. 989 (1996-1997).
ALWD 7th ed.

Mark A. Lemley, Economics of Improvement in Intellectual Property Law, 75 Tex. L.
Rev. 989 (1996-1997).
APA 7th ed.

Lemley, M. A. (1996-1997). Economics of improvement in intellectual property law.
Texas Law Review, 75(5), 989-1084.
Chicago 17th ed.

Mark A. Lemley, "Economics of Improvement in Intellectual Property Law," Texas Law
Review 75, no. 5 (1996-1997): 989-1084
McGill Guide 9th ed.

Mark A. Lemley, "Economics of Improvement in Intellectual Property Law" (1996-1997)
75:5 Tex L Rev 989.
AGLC 4th ed.

Mark A. Lemley, 'Economics of Improvement in Intellectual Property Law' (1996-1997)
75(5) Texas Law Review 989
MLA 9th ed.

Lemley, Mark A. "Economics of Improvement in Intellectual Property Law." Texas Law
Review, vol. 75, no. 5, 1996-1997, pp. 989-1084. HeinOnline.
OSCOLA 4th ed.

Mark A. Lemley, 'Economics of Improvement in Intellectual Property Law' (1996-1997)
75 Tex L Rev 989
-- Your use of this HeinOnline PDF indicates your acceptance of HeinOnline's Terms and
Conditions of the license agreement available at
https://heinonline.org/HOL/License
-- The search text of this PDF is generated from uncorrected OCR text.
-- To obtain permission to use this article beyond the scope of your license, please use:
Copyright Information
The Economics of Improvement in Intellectual
Property Law
Mark A. Lemley*
I. The Economics of Invention and Improvement .......... 993
II. Treatment of Improvements in Patent Law. 1000

A. The Scope of Patent Claims ....... 1000
B. EvaluatingInfringement by Improvers 1005
C. Mechanismsfor ProtectingImprovers 1007
1. Minor Improvers ........... 1007
2. SignificantImprovers ......... 1008
3. Radical Improvers .......... 1010
I. Treatment of Improvements in Copyright Law .......... 1013

A. The Scope of Protection Under CopyrightLaw ....... 1013
B. EvaluatingInfringement by Improvers ............ 1015
C. Mechanismsfor ProtectingImprovers ............. 1019
1. Minor Improvers ........................ 1019
2. SignificantImprovers ..................... 1020
3. Radical Improvers ...................... 1023
IV. Is Differential Treatment of Improvement Justified? ....... 1029

A. Moral Rights Justifications ................... 1031
B. Different Works Justifications .................. 1034
C. Balance of PowerArguments .................. 1038
D. Market Power Arguments .................... 1041
V. Optimal Improvement Rules ..................... 1042

A. PropertyRights Theory ...................... 1044
* Assistant Professor, University of Texas School of Law; of counsel, Fish & Richardson P.C.,
Austin, Texas. I would like to thank Julie Cohen, Mark Gergen, Rose Hagan, Paul Heald, Doug
Laycock, Jessica Litman, Peter Menell, Robert Merges, Neil Netanel, Samuel Oddi, David Post, Chris
Wood, and the editors of the Texas LawReview for helpful comments on an earlier draft of this Article,
and Barbara Parvis and Nicole Stafford for research assistance. Copyright 1997 by Mark A. Lemley.
990 Texas Law Review [Vol. 75:989
B. Complicating the BargainingStructure ... ........ 1048

1. Identifying and OrganizingImprovers . ........ 1048
2. Licensing Improvers ............ ........ 1052
a. TransactionCosts ........... ........ 1053
b. Uncertainty ............... ........ 1055
c. Externalities .............. ........ 1056
d. Strategic Behavior .......... ........ 1058
e. Noneconomic Incentives ....... ........ 1059
f Implications of Imperfect Licensing ........ 1061
3. Market Power and Holdup Problems .. ........ 1065
C. Conclusions .................... ........ 1068
VI. Implications of the Improvements Model ..... ........ 1072

A. PatentLaw .................... ........ 1072
B. CopyrightLaw .................. ........ 1073
1. Derivative Works .............. ........ 1074
2. Fair Use ................... ........ 1077
VII. Conclusion ................................. 1083
I have a colleague in practice who claims that as a trade secrets

lawyer, his job is to "prosecute thieves and defend entrepreneurs." For a
lawyer in private practice, determining which is which may not be that dif-
ficult on a case-by-case basis: the answer may depend largely on who is
paying the bills. Courts hearing intellectual property cases are not so
fortunate, however. They must find a way to distinguish between
improvement-which is thought to be a noble task, a necessary part of
innovation, and generally to be encouraged-and imitation, which is
generally considered both illegal and even immoral.' This distinction is
not easy to make, but it is critical to achieving the proper balance of intel-
lectual property rights. Allow too much imitation, and you will stifle the
incentives for development and commercialization of new products.
Discourage improvements too strongly, and you will freeze development
at the first generation of products.
1. See generally Jeremy Waldron, From Authors to Copiers: IndividualRights andSocial Values
in IntellectualProperty, 68 CHI.-KENT L. REv. 841, 842 (1993) (arguing that a neutral treatment of
the morality of copying must incorporate the perspectives of both the author and the copier); Laurie
Steams, Comment, Copy Wrong: Plagiarism, Process, Property and the Law, 80 CAL. L. REV. 513
(1992) (distinguishing plagiarism, which emphasizes the creative process, from copyright infringement,
which emphasizes the creative result).
1997] Economics of Improvement
This problem is not new. Even before the United States adopted its
patent and copyright laws, Lord Mansfield said the following in an English
case involving a copyright claim against improved navigational charts:
The rule of decision in this case is a matter of great consequence to
the country. In deciding it we must take care to guard against two
extremes equally prejudicial; the one, that men of ability, who have
employed their time for the service of the community, may not be
deprived of their just merits, and the reward of their ingenuity and
labor; the other, that the world may not be deprived of
improvements, nor the progress of the arts be retarded.'
Unfortunately, there has not been much improvement in the law of
improvements since 1785.
A number of doctrines in modem copyright and patent law attempt to
strike some balance between the rights of original developers and the rights
of subsequent improvers. Both patents and copyrights are limited in dura-
tion and in scope. Each of these limitations provides some freedom of
action to subsequent improvers. Improvers are free to use material that is
in the public domain because the copyright or patent has expired. They are
free to skirt the edges of existing intellectual property rights, for example
by taking the ideas but not the expression from a copyrighted work,3 or
by "designing around" the claims of a patent. However, improvers cannot
always avoid the intellectual property rights of the basic work on which
they wish to improve. Some improvements fall within the scope of the
preexisting intellectual property right, either because of an expansive
definition of that right4 or because economic or technical necessity requires
that the improver hew closely to the work of the original creator in some
basic respect.5 Here, the improver is at the mercy of the original intel-
lectual property owner, unless there is some separate right that expressly
allows copying for the sake of improvement.
Patent and copyright law differ in how they treat improvements that
fall within the scope of the original intellectual property right. Patent
doctrines such as the rule of blocking patents and the reverse doctrine of
equivalents offer some protection to the developers of significant or radical
2. Sayre v. Moore, 102 Eng. Rep. 139, 140 (1785) (Lord Mansfield, C.J.), cited in Cary v.
Longman, 102 Eng. Rep. 138, 140 n.(b) (1801) (Lord Kenyon, C.J.). Lord Mansfield instructed the
jury to distinguish between a new chart, which corrected the errors in the plaintiff's work, and a "mere
servile imitation" of the plaintiff's chart. Only the latter work was actionable under copyright. The
jury found for the defendant. Id.
3. See 17 U.S.C. § 102(b) (1994); Baker v. Selden, 101 U.S. 99, 102-03 (1879).
4. See infra notes 53-69 and accompanying text (discussing how improvements can still fail within
the literal scope of a patent claim).
5. See infra notes 178-80 and accompanyingtext (discussing market factors that may require exact
duplication of an original work, particularly in the computer software or parody contexts).
Texas Law Review [Vol. 75:989
improvements. 6 Improvers are therefore in a position to bargain with

original patent owners to allocate the gains from their respective invention,
and in some circumstances may even be free to use that invention without
payment. By contrast, copyright doctrine currently offers little or no
protection to improvers. Rather, the exclusive rights granted to the copy-
right owner in section 106 extend to cover any "copy" or adaptation or
alteration of the original that is nonetheless "substantially similar" to the
original work.8 So improvers-even radical improvers-have no power
to bargain with copyright owners to divide the value of the improved work.
If the work is substantially similar to the underlying copyrighted work, the
original copyright owner has the right to exclude it from the market
entirely. Further, because the rule is that improvers cannot even copyright
their original contributions to what is on balance an infringing work, there
is no way for the improver to capture even part of the value of her
contribution. 9 To be sure, improvers may be protected under copyright's
"fair use" doctrine."° But application of that doctrine is fraught with
uncertainty, 1 and historically has emphasized market harm to the original
copyright owner. The result is a rule which in some respects gives broader
protection to copyright owners than that given to patentees, and arguably
provides insufficient incentive to improve upon copyrighted works.
In this Article, I argue that copyright rules regarding improvement
should look more like analogous patent doctrines. I suggest that there
should be a rule of "blocking copyrights" analogous to the blocking patents
doctrine that already exists, and that the "transformative use" doctrine
gaining currency among the courts may, if properly applied, protect radical
improvers from liability to original copyright owners even where the
improvement hurts the market for the original work. Part I sets out the
economic background of intellectual property and the general problem of
improvements. Part II examines the treatment of improvement under
patent law, with particular attention paid to the doctrine of blocking patents
and the reverse doctrine of equivalents. Part III examines the treatment of
similar improvements under copyright law, considering both the rules
governing derivative works and the fair use doctrine as it has historically
been applied. In Part IV, I consider whether the different treatment of
6. For a discussion of the meaning of these terms, see infra notes 85-98 and accompanying text.
7. See infra notes 80-106 and accompanying text.
8. 17 U.S.C. § 106 (1994).
9. 1 MELVILLE B. NIMMER & DAVID NIMMER, NIMMER ON COPYRIGHT § 3.06 (1996); see also
infra notes 154-62 and accompanying text.
10. 17 U.S.C. § 107 (1994).
11. See, e.g., Pierre N. Leval, Toward a Fair Use Standard, 103 HARV. L. REV. 1105, 1105
(1990) (criticizing courts for failing to develop a consistent set of principles or values to govern fair
use).
improvement under the patent and copyright laws is justifiable. For a

variety of reasons, I conclude that it is not. Part V then models the
optimal treatment of improvement in intellectual property law, concluding
that the patent rules regarding improvements make more sense than the
analogous copyright rules, largely because they reflect more realistic
assumptions about the economics of intellectual property licensing. Part
VI offers a way to interpret the fair use doctrine in copyright to reach a
result analogous to that in the patent cases, and provides examples of how
certain cases might be decided under that interpretation.
I. The Economics of Invention and Improvement

Intellectual property is fundamentally about incentives to invent and
create. While there are a number of noneconomic theories offered to
explain both copyright12 and patent law,13 both the United States
Constitution and judicial decisions seem to acknowledge the primacy of
incentive theory in justifying intellectual property. The Constitution
expressly conditions the grant of power in the Patent and Copyright Clause
on a particular end, namely "[tlo promote the Progress of Science and
useful Arts." 14 As the Supreme Court explained in the landmark case of
12. Noneconomic theories of copyright tend to center around a natural law or moral rights
approach, suggesting that authors have some sort of preexisting entitlement to control their works. For
discussions of these approaches, see, for example, Wendy J. Gordon, A Property Right in Self-
Expression:Equality and Individualism in the NaturalLaw of Intellectual Property, 102 YALE L.J.
1533, 1535 (1993) (arguing that courts have mistakenly interpreted the natural law theory of copyright
and afforded too much protection to authors at the expense of free speech interests); Alfred C. Yen,
Restoring the NaturalLaw: Copyright as LaborandPossession, 51 OHIO ST. L.J. 517,529-39 (1990)
(stressing that natural law concepts are inherent in copyright law); cf. Steven Cherensky, Comment,
A Pennyfor Their Thoughts: Employee-Inventors, PreinventionAssignmentAgreements, Property, and
Personhood, 81 CAL. L. REV. 595, 644-53 (1993) (applying Margaret Radin's theory of personal
investments of self in property to intellectual property). For what might be termed a "utilitarian but
noneconomic" view of copyright as designed to further the development of a democratic civil society,
see Neil Weinstock Netanel, Copyrightand a Democratic Civil Society, 106 YALE L.J. 283, 347-48
(1996).
13. For a discussion of reward-based and even natural law theories of scientific invention, see, for
example, A. Samuel Oddi, Un-Unified Economic Theories of Patents-The Not-Quite-Holy Grail, 71
NOTRE DAME L. REv. 267, 274-77 (1996); Kevin Rhodes, Comment, The Federal Circuit's Patent
NonobviousnessStandards:TheoreticalPerspectiveson RecentDoctrinalChanges, 85 Nw. U. L. REV.
1051, 1077-84 (1991); cf. Lawrence C. Beecker, Deserving to Own IntellectualProperty, 68 CHI.-KENT
L. REV. 609 (1993) (arguing that desert-based arguments for patent law are intuitively appealing, but
do not necessarily justify the scope of current patent doctrine). An alternative to classical incentive
theory is the prospect theory of patents, advanced by Edmund Kitch in an important article 20 years
ago. See Edmund W. Kitch, The Nature and Function of the Patent System, 20 J.L. & ECON. 265
(1977). Kitch offers a property-based vision of patents as entitlements to innovate within a particular
field, granted to those who have already started such innovation. For a refinement of Kitch's approach
that takes account of rent-seeking, see Mark F. Grady & Jay I. Alexander, Patent Law and Rent
Dissipation,78 VA. L. REV. 305 (1992).
14. U.S. CoNSr. art. I, § 8, cl. 8.
Mazer v. Stein,"5 "'[the copyright law, like the patent statutes, makes
reward to the owner a secondary consideration' .... The economic
philosophy behind the clause empowering Congress to grant patents and
copyrights is the conviction that it is the best way to advance public wel-
fare through
16
the talents of authors and inventors in 'Science and useful
gArts."
To understand why the Framers thought exclusive rights in inventions
and creations would promote the public welfare, consider what would
happen absent any sort of intellectual property protection. Invention and
creation require the investment of resources-the time of an author or
inventor, and often expenditures on facilities, prototypes, supplies, and the
like. In a private market economy, individuals will not invest in invention
or creation unless the expected return from doing so exceeds the cost of
doing so-that is, unless they can reasonably expect to make a profit from
the endeavor. To profit from a new idea or a work of authorship, the
creator must be able either to sell it to others for a price, or to put it to
some use which provides her with a comparative advantage in a market. 7
But ideas (and writings, for that matter) are notoriously hard to
control. Even if the idea is one that the creator can use herself, for
example to boost productivity in her business, she will reap a reward from
that idea only to the extent that her competitors do not find out about it.
A creator who depends on secrecy for value therefore lives in constant
peril of discovery and disclosure. Competitors may steal the idea, or learn
of it from an ex-employee. They may be able to figure it out by watching
the creator's production process, or by examining the products she sells.
Finally, they may come upon the idea on their own, or discover it in the
published literature. In all of these cases, the secrecy value of the idea will
be irretrievably lost. 8
The creator who wants to sell her idea is in an even more difficult
position. Selling information requires disclosing it to others. Once the
information has been disclosed outside a small group, however, it is
extremely difficult to control. Information has the characteristics of what
economists call a "public good"-it may be "consumed" by many people
without depletion, and it is difficult to identify those who will not pay and
15. 347 U.S. 201 (1954).

16. Id. at 219 (quoting United States v. Paramount Pictures, Inc., 334 U.S. 131, 158 (1948) and
U.S. CONST. art. I, § 8, cI. 8).
17. The latter may occur, for example, where an idea for a more efficient machine is used to
reduce the cost of producing goods, allowing the owner of the idea to compete more effectively in
selling those goods.
18. It is possible to address this problem by adopting rules preventing the misappropriation of trade
secrets, in effect adding legal support to a creator's efforts to keep a secret. See generally UNIF.
TRADE SECRETS ACT (amended 1985), 14 U.L.A. 138 (1990). Trade secrets laws fall within the
general category of intellectual property, so they do not disprove the rule stated in the text. Further,
there are problems with relying solely on secrecy in many cases, as noted below.
1997] Economics of Improvement 995
prevent them from using the information.19 To adapt an old parable, if

I give you a fish, I no longer have it, but if I teach you to fish, you or I
can teach a hundred others the same skill without appreciably reducing its
value.2 If we assume that it is nearly costless to distribute information
to others-an assumption that was once unrealistic but has become much
more reasonable with the development of the Internet-it will prove vir-
tually impossible to charge for information over the medium run. If the
author of a book charges more than the cost of distribution, hoping to
recover some of her expenditures in writing the work, competitors will
quickly jump in to offer the book at a lower price. Competition will drive
the price of the book towards its marginal cost-in this case the cost of
producing and distributing one additional copy. In this competitive market,
the author will be unable to recoup the fixed cost of writing the book.
More to the point, if this holds generally true authors may be expected to
leave the profession in droves, since they cannot make any money at it.
The result, according to economic theory, is an underproduction of books
and other works of invention and creation with similar public goods
characteristics."
Information is not the only example of a public good. Economists
generally offer lighthouses and national defense as examples of public
goods, since it is virtually impossible to provide the benefits of either one
solely to paying clients.' And indeed it would be inefficient to do so,
since consumption of these goods is "nonrivalrous" (meaning that everyone
can benefit from them once they are produced). In the case of national
defense (and most lighthouses),' we avoid the underproduction that
would result from leaving it to the market by having the government step
in and pay for the public good. For a variety of good reasons, we have
not gone that route with information.' Instead, government has created
19. See Kenneth J. Arrow, Economic Welfare and the Allocation of Resourcesfor Invention, in
THE RATE AND DIRECTION OF INvENTIvE ACTIvITY: ECONOMIC AND SOCIAL FACTORS 609, 614-16
(1962).
20. To some extent this oversimplifies the problem by ignoring possible second-order distorting
effects. In practice, if I taught several hundred million people to fish, the result might be depletion of
a physical resource (fish) that otherwise would not have occurred. Similarly, wide dissemination of
information-may have particular effects on secondary markets, depending on what the information is.
Nonetheless, the general point remains accurate.
21. See, e.g., F.M. SCHERER, INDUSTRIALMARKET STRUCTURE AND ECONOMIC PERFORMANCE
444 (2d ed. 1980) ("If pure and perfect competition in the strictest sense prevailed continuously...
incentives for invention and innovation would be fatally defective without a patent system or some
equivalent substitute."). Scherer goes on to note, however, that natural market imperfections may give
advantages to first movers, reducing the need for intellectual property protection. Id. at 444-45.
22. See, e.g., MICHAEL LAVER, THE POLITICS OF PRIVATE DEsIRES 30-31 (1981) (giving
examples of "collective consumption" goods).
23. But see Ronald H. Coase, TheLighthouse in Economics, 17 J.L. & ECON. 357 (1974) (casting
doubt on the assumption that lighthouses must always be publicly provided).
24. For example, government provision of all books or all ideas is probably inefficient if there is
a way to encourage competition between authors or inventors. Competition is a better spur to new
intellectual property rights in an effort to give authors and inventors control

over the use and distribution of their ideas, and therefore encourage
them to invest efficiently in the production of new ideas and works of
authorship.
Unfortunately, this approach comes at a cost. Granting authors and
inventors the right to exclude others from using their ideas necessarily
limits the diffusion of those ideas, and so prevents people from benefiting
from them.' In economic terms, intellectual property rights prevent com-
petition in the sale of the particular work or invention covered by the
intellectual property right, and therefore allow the intellectual property
owner to raise the price of that work above the marginal cost of repro-
ducing it. Indeed, intellectual property rights must permit prices to rise
above marginal cost in some cases if they are to have their intended effect
of providing an incentive to create. 6 This means that in many cases
fewer people will buy the work than if it were distributed on a competitive
basis, and they will pay more for the privilegeY Because intellectual
ideas than government mandate. Further, government control over new publications is fundamentally
inconsistent with the diversity of viewpoints and freedom of speech inherent in a democratic society.
See, e.g., Vincent Blasi, The Checking Value in FirstAmendment Theory, 1977 AM. B. FOUND. RES.
J. 521.
25. Glynn Lunney offers a second set of costs, which are not normally considered in economic
analyses of intellectual property. He suggests that incentives for creation distort allocative efficiency
by encouraging the inefficient shifting of resources from other sectors of the market into the production
of works of intellectual property. See Glynn S. Lunney, Jr., Reexamining Copyright's Incentives-
Access Paradigm, 49 VAND. L. REV. 483 (1996). While Lunney's critique of the copyright system,
if valid, is fundamental, a detailed analysis of his paper is beyond the scope of this Article.
26. This does not mean that intellectual property rights automatically confer market power or
create "monopolies" in an economic or antitrust sense, as some courts have erroneously presumed.
See, e.g., Jefferson Parish Hosp. Dist. No. 2 v. Hyde, 466 U.S. 2, 16 (1984) (assuming that a patent
on a product may confer market power on the seller); United States v. Loew's, Inc., 371 U.S. 38, 45
(1962) (finding the economic power necessary for a tying violation may be assumed where the tying
product is patented or copyrighted); Digidyne Corp. v. Data Gen. Corp., 734 F.2d 1336, 1344 (9th
Cir. 1984) (holding that a computer company's copyright created a presumption of economic power
for theproduct). While some intellectual property rights may in fact give their owner power in an eco-
nomically relevant product market, most do not; they merely prevent others from competing to sell
copies of a particular product, not from selling different products that compete with the original. See,
e.g., PHILLIP AREEDA & Louis KAPLOW, ANTITRUST ANALYSIS 441-42 (1988); HERBERT
HOVENKAMP, ECONOMICS AND FEDERAL ANTITRUST LAW § 8.3, at 219 (1985); Mark A. Lemley,
Comment, The Economic Irrationalityof the PatentMisuse Doctrine, 78 CAL. L. REV. 1599, 1626-28
(1990); Russell Lombardy, Comment, The Myth ofMarket Power: Why Market Power Should Not Be
PresumedWhen ApplyingAntitrust Principlesto the Analysis of lying Agreements Involving Intellectual
Property, 8 ST. THOMAS L. REv. 449, 453-55 (1996); William Montgomery, Note, The Presumption
of Economic Powerfor Patented and CopyrightedProducts in 7ying Arrangements, 85 COLum. L.
REV. 1140, 1156 (1985).
Whether the intellectual property owner can take advantage of his right to exclude to raise prices
on his product depends largely on two factors: whether perfect competition prevails in the industry and
whether his invention gives him a cost advantage over competitors. See SCHERER, supra note 21, at
444.
27. See SCHERER, supra note 21, at 450-51 (documenting instances of patent holders pricing in
excess of cost).
19971 Economics of Improvement
property rights impose costs on the public, the intellectual property laws
can be justified by the public goods argument only to the extent that they
do on balance encourage enough creation and dissemination of new works
to offset those costs. One of the reasons that intellectual property rights
are limited in scope, in duration, and in effect?' is precisely in order to
balance these costs and benefits.'
The situation becomes more complex when we consider not
just the market for a particular work, but the market for new works in
general. Creation does not occur in a vacuum. Rather, knowledge is
cumulative-authors and inventors must necessarily build on what came
before them.' Indeed, if they did not do so, the societal costs in terms
of reinvention would be enormous.3 1 Try to imagine building something
as complex as a car, without using any ideas from anyone who came before
you (including such things as the wheel, nuts and bolts, screws, glass, and
the combustion engine). As countless economists have demonstrated,
efficient creation of new works requires access to and use of old works.3'
And since "improvements" may in many cases dwarf the original work in
terms of their practical significance,33 dynamic market efficiency over dif-
ferent generations demands such access as well.
But intellectual property limits the public's ability to access and use
old works. The creators of old works can, if they choose, refuse to distri-
28. See infra notes 38-47 and accompanying text (briefly describing a few of the ways in which
intellectual property rights are limited).
29. For efforts to strike the optimal balance between these costs and benefits, see, for example,
WILLIAM D. NoRDHAus, INVENTION, GROWTH AND WELFARE: A THEORETICAL TREATMENT OF
TECHNOLOGICAL CHANGE (1969). On the indeterminacy of the whole endeavor, see Louis Kaplow,
The Patent-Antitrust Intersection: A Reappraisal, 97 HARV. L. REV. 1813 (1984). Cf. TED
O'D6NOGHUE ET AL., PATENT BREADTH, PATENT LIFE AND THE PACE OF TECHNOLOGICAL
IMPROVEMENT (Univ. of Cal. at Berkeley Dep't of Econs. Working Paper No. 95-242, Oct. 1995)
(asserting that effective patent life is a function not only of statutory term, but also of patent breadth,
which affects the period within which the patent will be superseded by a noninfringing invention).
30. The most famous formulation of this phenomenon is credited to Sir Isaac Newton, who
reportedly said, "If I have seen further it is by standing on the shoulders of giants." THE OXFORD
DICTIONARY OF QUOTATIONS 362 (3d ed. 1979). This is occasionally referred to in computer law as
the OTSOG ("On the Shoulders of Giants") principle.
31. Cy. Mark A. Lemley & David O'Brien, EncouragingSoftware Reuse, 49 STAN. L. REV.
(forthcoming Jan. 1997) (documenting costs of partial reinvention in the software industry).
32. See, e.g., PAUL DAVID, NEW TECHNOLOGY, DIFFUSION, PUBLIC POLICY, AND INDUSTRIAL
COMPETITIVENESS20 (Center for Econ. Policy Research, Publication Policy Paper No. 46, Apr. 1985);
MORTON I. KAMIEN & NANCY L. SCHWARTZ, MARKET STRUCTURE AND INNOVATION 3-9 (1982);
RICHARD R. NELSON & SIDNEY G. WINTER, AN EVOLUTIONARY THEORY OF ECONOMIC CHANGE 130
(1982); Robert P. Merges & Richard R. Nelson, On the Complex Economics of Patent Scope, 90
COLUM. L. REV. 839, 908 (1990); Nathan Rosenberg, FactorsAffecting the Dison of Technology,
10 EXPLORATIONS ECON. HIST. 3, 3 (1972); SuzanneScotchmer, Standingon the Shoulders of Giants:
Cumulative Research and the PatentLaw, J. ECON. PERSP., Winter 1991, at 29, 30-31.
33. See, e.g., John L. Enos, A Measure of the Rate of TechnologicalProgressin the Petroleum
Refining Industry, 6 J. INDUS. ECON. 180, 188-89 (1958); James Mak & Gary M. Walton, Steamboats
and the Great Productivity Surge in River Transportation, 32 J. ECON. HIST. 619, 625 (1972).
bute them to anyone at all, at any price, during the duration of intellectual
property protection.' While most creators will not prevent access alto-
gether, they can and do exercise control over who can use their creation,
the purposes for which they can use it, and the price they must pay. Intel-
lectual property owners may use these rights not only to obtain a return on
their investment in research and development, but also to exercise content
control over subsequent uses of their works or to prevent the development
of a competitive market for their products.35 Even if the intellectual
property owner is willing to license his rights on reasonable terms, or even
if the work turns out to be in the public domain, someone who wants to
use the work must expend time and resources identifying intellectual
property owners and negotiating with them. Improvers also bear the risk
of failing to identify every intellectual property owner and unknowingly
infringing on a prior intellectual property right. In each of these
circumstances, the existence of preexisting intellectual property rights
imposes a positive cost on improvers that they would not otherwise face.36
It is not enough to say that intellectual property law favors "creators"-for
here we have creators on both sides of the equation, and the law must
choose between them.37
Because original developers and improvers will not always find each
other and agree to license the original work ex ante, the intellectual
property rules governing improvements are important in understanding the
extent to which protection for first-generation innovation will impede
improvement in subsequent generations. The more absolute the property
right given to original authors and inventors, the more critical efficient
34. See 35 U.S.C. § 271(d)(4) (1994) (providing that patent owner may refuse to license patent);
Salinger v. Random House, Inc., 811 F.2d 90, 94 (2d Cir. 1987) (allowing J.D. Salinger to prevent
Random House from publishing his letters); Continental Paper Bag Co. v. Eastern Paper Bag Co., 210
U.S. 405, 424 (1908) (holding that a patentee is entitled to suppress a patent entirely). In some cases,
such as trademarks and trade secrets, the duration of the intellectual property right may be indefinite.
See 15 U.S.C. §§ 1058(a), 1059(a) (1994) (trademarks); Warner-Lambert Pharm. Co. v. John 3.
Reynolds, Inc., 178 F. Supp. 655, 661 (S.D.N.Y. 1959) (trade secrets), aff'd, 280 F.2d 197 (2d Cir.
1960).
35. The circumstances in which an intellectual property owner might refuse to license his idea are
explored in detail infra Part V.
36. See Merges & Nelson, supra note 32, at 868-70; Scotchmer, supra note 32, at 38; Stewart E.
Sterk, Rhetoric and Reality in Copyright Law, 94 MICH. L. REV. 1197, 1207-08 (1996); Brian R.
Landy, Comment, The Two Strands of the FairUse Web: A Theoryfor Resolving the Dilemma ofMusic
Parody, 54 OHIO ST. LJ. 227, 227 (1993).
37. For a discussion of the parameters of this choice, see Glynn S. Lunney, Jr., Lotus v. Borland:
CopyrightandComputer Programs,70 TUL. L. REv. 2397,2418-22 (1996). Paul Goldstein describes
a paradox wherein "every infringer of a derivative right is, by definition, itself the potential copyright
owner of a derivative work, with an equal claim on copyright's system of investment incentives." Paul
Goldstein, Derivative Rights and Derivative Works in Copyright, 30 J. COPYRIGHT Soc'y 209, 211
(1983). He is certainly correct that this fact "seriously complicates" the task of establishing optimal
infringement rules. Id.
licensing is to subsequent innovation, and the more sensitive the industry

is to market failures in licensing.
Sometimes improvers benefit from the limits placed on the scope or
duration of intellectual property rights. Patents, for example, last from the
time they are approved until twenty years from the date the patent applica-
tion was filed.3" After that time, the material claimed in the patent enters
the "public domain," and is free for anyone to use.39 Copyrights last
longer, anywhere from 50 to about 120 years depending on who wrote the
original work, whether they were paid to do so, whether they published it,
and how long they live.' Nonetheless, copyrights too expire eventually,
and the contents of such public domain works may be freely copied.41
The limited duration of patents and copyrights promotes improvement in
writings and inventions, by allowing subsequent authors and inventors to
build upon what came before them.4 2
Even during the term of a patent or copyright, the strength of the
owner's exclusive right is further limited by rules that delineate the scope
of that right. Thus, a copyright prevents copying, adaptation, distribution,
and performance, but does not prevent private use of a work.' Further,
the restriction on copying extends only to appropriation that is
"substantial."" Similarly, patent law grants the patentee the right to
make, use, and sell the claimed invention, but limits the scope of that
right to what is set forth in the patent claims and equivalents thereof.'
Improvers can avoid the reach of patents and copyrights still in force by
skirting the edges of these exclusive rights-for example, by taking only
ideas and not expression from a copyrighted work,' or by "designing
around" the claims of a patent.47
38. See 35 U.S.C. § 154(a)(2) (1994).

39. See Brulotte v. Thys Co., 379 U.S. 29, 32 (1964).
40. See generally 17 U.S.C. §§ 301-305 (1994) (setting out rules governing duration of
copyrights). A proposal pending in Congress would lengthen copyright terms by an additional 20
years, for both individual copyrights and works created "for hire." See Copyright Term Extension Act
of 1997, H.R. 604, 105th Cong. § 2.
41. Indeed, many important works of literature, such as the Bible and thewritings of Shakespeare,
are free for all to copy. On the importance of a vibrant public domain in copyright law, see David
Lange, Recognizing the PublicDomain,LAW & CONTEMP. PROBS., Autumn 1981, at 147, 147; Jessica
Litman, The Public Domain, 39 EMORY LJ. 965 (1990).
42. See supra note 32 and accompanying text; cf. Sterk, supra note 36, at 1223-24 (discussing
whether extension of the copyright period encourages authors to create enduring works).
43. See 17 U.S.C. § 106 (1994).
44. See Amy B. Cohen, Masking CopyrighsDecisionmaking:The Meaninglessnessof Substantial
Similarity, 20 U.C. DAVIS L. REv. 719 (1987).
45. See Graver Tank & Mfg. Co. v. Linde Air Prods. Co., 339 U.S. 605, 607-09 (1950).
46. See 17 U.S.C. § 102(b) (1994); Baker v. Selden, 101 U.S. 99 (1879); see also Justin Hughes,
The Philosophy of Intellectual Property, 77 GEO. L.J. 287, 319-20 (1988) (describing how "de-
propertizing" extraordinary ideas restores necessary tools to the common good).
47. "Designing around" a patent refers to using the written claims of a patent as a guide for
deciding what to avoid in producing a competing product. Recent Federal Circuit decisions have given
Robert Merges and Richard Nelson have modeled the scope of patent
rights in detailed economic terms.4 8 In this Article, I deal with a subset
of the broader question of the proper scope for intellectual property
rights-what happens in those cases in which an improvement falls squarely
within the scope of an intellectual property right still in force? In those
cases, the improver is at the mercy of the original intellectual property
owner, unless the intellectual property laws offer improvers relief from
infringement actions in certain circumstances. It is to that latter issue that
I turn in Parts II and III.
II. Treatment of Improvements in Patent Law

A. The Scope of Patent Claims
Patents in the United States are granted exclusively by the Patent and
Trademark Office (PTO). Inventors must apply to the PTO for patent pro-
tection, and obtain rights over the invention only when the PTO issues a
patent.49 To obtain a patent, the inventor must prove to the PTO that his
idea is novel, is not obvious, is useful, and is sufficiently described in
the application to allow others to use and benefit from it.' The patent
describes what the inventor has designed or built,5 1 and it concludes with
"claims" setting forth the legal boundaries of the invention. 2
The key to understanding the treatment of improvements in patent law
is recognizing that patents are legally defined by the language of the
patent's claims, not by what the patent owner has actually invented or
built.5 3 The claims establish the boundaries of the invention, and it is the
claims that are used in deciding whether a defendant's product infringes the
patent. If an accused product falls within the bounds of a valid claim (in
patent parlance, if it "literally infringes" the claim), it generally does not
more lenient treatment to accused infringers who designed around the patent in an effort to avoid
infringement. See Roton Barrier, Inc. v. Stanley Works, 79 F.3d 1112, 1127 (Fed. Cir. 1996); Hilton
Davis Chem. Co. v. Wamer-Jenkinson Co., 62 F.3d 1512, 1520 (Fed. Cir. 1995) (en bane) (both
holding that evidence of designing around a specific patent gives rise to an inference of no
infringement). However, the Supreme Court's recent decision in Warner-JenkinsonCo. v. HiltonDavis
Chemical Co., 65 U.S.L.W. 4162 (U.S. Mar. 3, 1997), appears to foreclose such favorable treatment.
48. Merges & Nelson, supra note 32.
49. By contrast, every other form of intellectual property vests at least partial rights in the owner
upon creation of the work or satisfaction of the requirements of protection.
50. See 35 U.S.C. §§ 101-103, 112 (1994).
51. An inventor can obtain a patent on a device without having ever built that device, so long as
it is described sufficiently that one skilled in the art could build it. See Hazeltine Corp. v. United
States, 820 F.2d 1190, 1196-97 (Fed. Cir. 1987).
52. 35 U.S.C. § 112 para. 2 (1994) ("The specification shall conclude with one or more claims
particularly pointing out and distinctly claiming the subject matter which the applicant regards as his
invention.").
53. See DatascopeCorp. v. SMEC, Inc., 879 F.2d 820, 824 (Fed. Cir. 1989) ("[I]t is claims, not
commercial embodiments, that are infringed." (emphasis in original)).
matter that the accused product is different from the inventor's product, or
even that the inventor could not possibly have built the accused product.
The defendant must avoid any infringement on the territory covered by the
claim.
For this reason, an inventor can expand the reach of his patent by
drafting claims as broadly as possible. For example, the inventor of a
pencil will be better off claiming an "instrument for writing" than a
"hexagonal cylinder constructed of wood, the center of which is hollow
and filled with graphite, and which is capable of being sharpened for use
in marking on paper." The former claim will capture more ground for the
patent owner. Not only will octagonal or round pencils fall within the
scope of the first claim, but so too might pens, mechanical pencils, and
signature machines.
There are two basic limits on the ability of inventors to draft broad
claims. First, because novelty and nonobviousness are tested with respect
to the language of the claims," a claim must not cover something already
invented, or which is merely an obvious extension of prior work. Thus,
the scope of a patentee's claims is limited by the "prior art" in the field.
The effect of the novelty rule is clear-you are not entitled to a patent on
something that someone else has already invented or described. If your
"invention" duplicates another's, then in theory you have not added any
social value to justify obtaining an exclusive right.55 The nonobviousness
rule requires still more from the patentee than an invention no one else has
come up with. What is required is an inventive leap-some quantum of
improvement over what has come before.56 In effect, the nonobviousness
requirement sets a minimum threshold social value the invention must con-
tribute in order to make it worth the trouble of issuing and enforcing a
patent. It also prevents patentees from overclaiming their invention. For
example, the inventor of the pencil cannot get away with claiming an
54. See RCA Corp. v. Applied Digital Data Sys., Inc., 730 F.2d 1440, 1444 (Fed. Cir. 1984).
55. Reality is not that simple. An inventor may add value in the way he describes the invention,
in making it known to a greater segment of the public, or in taking the steps necessary to commer-
cialize it. But these aspects of innovation are not normally protected by patent law if the invention
itself is already in the public domain. For a proposal to protect commercialization efforts directly, see
William Kingston, The Innovation Warrant, in DIRECT PROTECTION OF INNOVATION 59 (William
Kingston ed., 1987), and for criticism thereof, see, for example, Thomas Mandeville & Stuart
Macdonald, Innovation Protection Viewed from an Information Perspective, in DIRECT PROTECTION
OF INNOVATION, supra, at 157; Andrd Piatier, Innovation Patent, Invention Patent, or Both?, in
DIRECT PROTECTION OF INNOVATION, supra, at 125; and Gordon Tullock, Intellectual Property, in
DIRECT PROTECTION OF INNOVATION, supra, at 171.
56. See ROBERT PATRICK MEROES, PATENT LAW AND POLICY 379 (1992) ("Nonobviousness can
be thought of as a kind of 'nontriviality' requirement ... [that] attempts to measure an even more
abstract quality than novelty or utility: the technical accomplishment reflected in an invention. This
requirement asks in essence whether an invention is a big enough technical advance to merit the award
of a patent." (emphasis in original)).
"instrument for writing" if there are already other writing instruments in

the prior art.
Second, the patentee must be able to point to language in the patent
specification that enables those skilled in the art to make the claimed inven-
tion.57 Certainly, the patentee must describe what he has actually built or
designed. If he is to claim an entire class of products, though, he must
give some indication of how to produce other products in that class. Thus,
in a celebrated case, Thomas Edison defeated a prior patent on incandes-
cent light bulbs on the grounds that the patent owners had not enabled the
full scope of their claims.58 The patent owners, Sawyer and Man, had
built an incandescent bulb that did not work very well, largely because the
particular fiber they had chosen as the filament (carbonized paper) was not
durable. In their patent, Sawyer and Man claimed the use of any vegetable
fiber as a filament. When Edison discovered that a particular species of
bamboo worked much better than any other type of fiber, Sawyer and Man
sued him for patent infringement because Edison's bamboo was, after all,
vegetable fiber, and therefore within the scope of the claim. The Court
held their claim invalid on the grounds that they had not done sufficient
work to justify treating all vegetable fibers as interchangeable. In the
words of the Court:
Is the complainant entitled to a monopoly of all fibrous and
textile materials for incandescent conductors? If the patentees had
discovered in fibrous and textile substances a quality common to
them all, or to them generally, as distinguishing them from other
materials, such as minerals, etc., and such quality or characteristic
adapted them peculiarly to incandescent conductors, such claim might
not be too broad. If, for instance, minerals or porcelains had always
been used for a particular purpose, and a person should take out a
patent for a similar article of wood, and woods generally were
adapted to that purpose, the claim might not be too broad, though
defendant used wood of a different kind from that of the patentee.
But if woods generally were not adapted to the purpose, and yet the
patentee had discovered a wood possessing certain qualities, which
gave it a peculiar fitness for such purpose, it would not constitute an
infringement for another to discover and use a different kind of
wood, which was found to contain similar or superior qualities. The
present case is an apt illustration of this principle. Sawyer and Man
supposed they had discovered in carbonized paper the best material
for an incandescent conductor. Instead of confining themselves to
carbonized paper, as they might properly have done, and in fact did
in their third claim, they made a broad claim for every fibrous or
57. See 35 U.S.C. § 112 para. 1 (1994).

58. See The IncandescentLamp Patent, 159 U.S. 465 (1895).
textile material, when in fact an examination of over six thousand

vegetable growths showed that none of them possessed the peculiar
qualities that fitted them for that purpose. Was everybody then
precluded by this broad claim from making further investigation?
We think not.
The injustice of so holding is manifest in view of the
experiments made, and continued for several months, by Mr. Edison
and his assistants, among the different species of vegetable growth,
for the purpose of ascertaining the one best adapted to an
incandescent conductor.... The question really is whether the
imperfectly successful experiments of Sawyer and Man, with
carbonized paper and wood carbon, conceding all that is claimed for
them, authorize them to put under tribute the results of the brilliant
59
discoveries made by others.
This does not mean, however, that patent owners are limited only to
what they have actually discovered. Claims may reach beyond what the
patentee has in fact invented in three circumstances. First, where a number
of materials or devices are substitutable because they have similar charac-
teristics, the patentee may claim the generic class of materials, so long as
he describes the general class and its characteristics with sufficient
precision that others can identify and use them without "undue
experimentation."' Thus, in Atlas Powder Co. v. E.L du Pont De
Nemours & Co.,61 the court allowed a patent on explosive compounds
made from various salts, fuels, and emulsifiers, where the patentee had
listed the ingredients that might be used but had not given any indication
of which combinations would work. Even though du Pont had not tried all
of the possible combinations (there were thousands), and in practice forty
percent of the combinations tried were inert, the court held that du Pont
was entitled to claim the generic group of explosives.62 Thus, the enable-
ment requirement places some limits on broad claiming, but it certainly
does not limit claims to the precise embodiments tested by the patent
owner.
Second, the doctrine of equivalents provides a means for broadening
the scope of a patent beyond the literal language of the claims (and hence
beyond the invention originally made by the patent owner). The doctrine
of equivalents was originally intended to provide equitable relief to patent
owners against imitators who would avoid literal infringement of a nar-
rowly written claim by changing one insubstantial feature of the
59. Id. at 472-74.

60. See, e.g., In re Wands, 858 F.2d 731,737 (Fed. Cir. 1988); Utter v. Hiraga, 845 F.2d 993,
998 (Fed. Cir. 1988); In re Angstadt, 537 F.2d 498, 502-03 (C.C.P.A. 1976).
61. 750 F.2d 1569 (Fed. Cir. 1984).
62. Id. at 1576-77.
invention.' It has since been expanded by the courts into an integral part
of the infringement analysis of every patent,' though the scope and
application of the doctrine remain a matter of some dispute. As presently
conceived, the doctrine of equivalents provides that accused products or
processes that do not fall within the literal scope of the patent claims
nonetheless infringe the patent if they are only "insubstantially different"
from the patent claims.' The effect is to create a "penumbra" around the
literal scope of the claims, and therefore to expand the protection given to
patent owners.' Figure 1 diagrams the scope of patent protection in
simple terms.
Figure 1
invention
range of equivalents
non-infringing products
63. See Graver Tank & Mfg. Co. v. Linde Air Prods. Co., 339 U.S. 605, 607-08 (1950);
Westinghouse v. Boyden Power Brake Co., 170 U.S. 537, 569 (1898).
64. See Hilton Davis Chem.Co. v. Warner-Jenkinson Co., 62 F.3d 1512 (Fed. Cir. 1995) (en
banc) (per curiam) (holding that in every infringement case the trial judge must apply the doctrine of
equivalents), rev'd on other grounds, 65 U.S.L.W. 4162 (U.S. Mar. 3, 1997); Peter K. Schalestock,
Comment, Equityfor Whom? Defining the Reach of Non-Literal PatentInfringement, 19 SFAiTE U.
L. REV. 323 (1996) (criticizing this change).
65. Hilton Davis, 62 F.3d at 1521-22. The court in Hilton Davis provided significant guidance
as to what factors should be used in determining the substantiality of a difference. The court stated that
while the "function-way-result" test developed by the Supreme Court in Graver Tank will often be suf-
ficient to show the extent of differences between the patented product and the accused device, the ulti-
mate test of infringement under the doctrine of equivalents is whether the differences between the
claims and the accused device are "insubstantial." Id. The court also recognized the utility of other
factors, including evidence of copying and designing around. Id.
The Supreme Court's recent decision in Warner-JenkinsonCo. v. Hilton Davis Chemical Co.,
65 U.S.L.W. 4162 (U.S. Mar. 3, 1997), is largely consistent with this approach. The Court concluded
that "the particular linguistic framework used is less important than whether the test is probative of the
essential inquiry" of equivalence between each element. Id. at 4168. It did modify the Federal
Circuit's opinion in some respects, rejecting the use of copying and design-around and strengthening
the role of prosecution history estoppel. Id. at 4167-68.
66. HiltonDavis, 62 F.3d at 1534.
19971 Economics of Improvement 1005
Finally, patent claims may reach new and unanticipated inventions

made after the patent issues, but which fall within the literal language of
the claims. 7 An example is Hughes Aircraft Co. v. United States."' In
that case, Hughes held a patent on technology developed in the 1970s for
controlling the orientation of a communications satellite by sending control
signals from a ground control computer to the satellite. When advances
in computer technology allowed the necessary processing power to be
installed on the satellite itself, the government began to control its satellites
using on-board computers. In Hughes's patent suit against the government,
the Federal Circuit held that the government's method of on-board com-
puter control infringed the Hughes patent, even though that patent was
based on old technology that required communications from the ground.'
In the Hughes case, it is clear that the government's technology represented
a significant advance over the technology conceived by the patentee.
Nonetheless, the patentee was entitled to capture the benefits of these sub-
sequent technological improvements.
B. EvaluatingInfringement by Improvers
Changes in products and processes occur all the time. Trying some-
thing new in the hope of improving on an existing product or process is an
integral part of the competitive process. Sometimes these changes will
actually improve the existing product or make its production more efficient,
and therefore contribute to social value. Other changes may be value-
neutral (that is, the new product may be just as good or as efficient as the
existing product), or they may actually be inefficient (producing less useful
products, or products that cost more than the existing ones).
Where the existing product is patented, improvers must avoid the
literal scope of the patent claims and equivalents thereof in order to sell a
competing product. Not every change from the actual product invented by
the patent owner will avoid infringement, since some will fall within the
literal scope of the patent claims. In Figure 1, these changes will fall
within circle C-D, although they are not within the actual scope of what the
inventor created (A-B). Still other changes will fall outside the literal
scope of the patent claims, but within the somewhat amorphous range of
67. Cf. Diamond Rubber Co. v. Consolidated Rubber Tire Co., 220 U.S. 428, 435-36 (1911)
(holding that a patentee need not understand why or how his invention works, so long as it does);
Eames v. Andrews, 122 U.S. 40, 55 (1887) (holding that a patent is no less valid simply because the
inventor neither knows nor includes the scientific principle on which the invention is based).
68. 717 F.2d 1351 (Fed. Cir. 1983).
69. Id. at 1365; accord Laser Alignment, Inc. v. Woodruff& Sons, Inc., 491 F.2d 866 (7th Cir.
1974) (holding that the use of a laser to align pipe segments infringed a patent for using ordinary
beams of light to align pipe).
protection afforded the invention under the doctrine of equivalents. In

Figure 1, these changes are within circle E-F, but are outside circle C-D.
This determination is largely a factual one, and will depend on the
nature of the accused products as well as the scope of the claims. The
Federal Circuit has offered a number of ways to test infringement under
the doctrine of equivalents, for example.' Equivalence is tested on an
element-by-element basis,71 inquiring whether there are substantial dif-
ferences in the function each element performs, or in the way it performs
that function,' or alternatively whether the claimed device and the
accused device are reasonably interchangeable.73 It should be obvious
from the foregoing discussion that one cannot avoid patent infringement
merely by building something different than what the patentee has built (or
even described). Subsequent developers of products must be careful to
avoid treading on the literal language of the patent claims, whether or not
the patentee envisioned the particular device at issue. Further, subsequent
developers must attempt to guarantee that a jury will not find their product
to be insubstantially different from the patent claim language. Even
subsequently developed products that unquestionably improve on the work
of the original inventor may infringe the inventor's patent.74
What may be less obvious is that this infringement determination does
not take into account the value of the improvement made by the accused
infringer. Dramatic improvements may still fall within either the literal
scope of the claims or the range of equivalents afforded the original inven-
tion. Consider the improvement at issue in Hughes Aircraft, which was
held to infringe an old patent even though subsequent events had revolu-
tionized satellite technology. 5 Similar results have occurred when elec-
70. See Keith A. Robb, Hilton Davis and the DoctrineofEquivalents-AnInsubstantialDfference

4 TEx. IrrTE. PROP. L.J. 275 (1996).
71. See Warner-Jenkinson Co. v. Hilton Davis Chem. Co., 65 U.S.L.W. 4162,4165 (U.S. Mar.
3, 1997) (characterizing each element in a patent claim as "material" to analysis under the doctrine of
equivalents); Pennwalt Corp. v. Durand-Wayland, Inc., 833 F.2d 931, 935 (Fed. Cir. 1987) (en bane)
(determining infringement by comparing the patent claim to the accused device on an element-by-
element basis).
72. See Graver Tank & Mfg. Co. v. Linde Air Prods. Co., 339 U.S. 605, 607-10 (1950)
(adopting the "function-way-result" test for infringement under the doctrine of equivalents). But see
Hilton Davis Chem. Co. v. Wamer-JenkinsonCo., 62F.3d 1512, 1518 (Fed. Cir. 1995) (en banc) (per
curiam) (arguing that the function-way-resulttest is not the only way of determining equivalence), rev'd
on other grounds, 65 U.S.L.W. 4162 (U.S. Mar. 3, 1997).
73. See Graver Tank, 339 U.S. at 609; HiltonDavis, 62 F.3d at 1519.
74. See Marsh-McBirney, Inc. v. Montedoro-Whitney Corp., 882 F.2d 498, 504 (Fed. Cir. 1989);
Richardson v. Suzuki Motor Co., 868 F.2d 1226, 1240 (Fed. Cir. 1989); Water Techs. Corp. v.
Calco, Ltd., 850 F.2d 660, 669 (Fed. Cir. 1988); Atlas Powder Co. v. E.I. du Pont De Nemours &
Co., 750 F.2d 1569, 1580 (Fed. Cir. 1984); Bendix Corp. v. United States, 600 F.2d 1364, 1382 (Ct.
Cl. 1979).
75. Hughes Aircraft Co. v. United States, 717 F.2d 1351, 1352-54 (Fed. Cir. 1983).
trical patents based on analog technology have been asserted against new
(and in most cases dramatically improved) digital implementations.7 6 At
the other end of the spectrum, changes with little or no social value may
be sufficient to avoid infringement of a patent, especially if the claims are
narrowly drafted. For example, the narrow claim for a pencil described
above might not be infringed by a seven-sided cylinder, or a cylinder made
of plastic, even though changing the shape or composition of the pencil has
no impact on its performance. Indeed, in some cases competitors may be
encouraged to adopt inefficient solutions in order to avoid infringement.
The only clear way to avoid infringing on a patent is to use a product that
was in the prior artl--an approach which does not capture any of the
value of the patented invention.
C. Mechanismsfor ProtectingImprovers
Patent law does, however, provide two doctrines which encourage
improvements, even when those improvements infringe on the underlying
patent. It is useful to distinguish three general cases, that I will call here
the minor improver, the significant improver, and the radical improver.
1. Minor Improvers.-As noted above, inventors are entitled to a

patent on their invention only if they show some inventive leap over the
prior art. Generally speaking, this translates into a minimum social con-
tribution necessary to obtain a patent.7" For the sake of argument, let us
set this minimum value the patent owner has contributed to society at $200.
Subsequent developers may make changes to the patented technology in an
76. See Decca Ltd. v. United States, 544 F.2d 1070, 1081 (Ct. Cl. 1976).
77. See Wilson Sporting Goods Co. v. David Geoffrey & Assocs., 904 F.2d 677, 684 (Fed. Cir.
1990) ("[Slince prior art always limits what an inventor could have claimed, it limits the range of
permissible equivalents of a claim."). This rule seems an obvious and necessary extension of the gene-
ral principle that patent claims which cover, even in part, material that was in the prior art are invalid.
See, e.g., Malta v. Schulmerich Carillons, Inc., 952 F.2d 1320, 1342 (Fed. Cir. 1991); Intel Corp.
v. United States Int'l Trade Comm'n, 946 F.2d 821, 842 (Fed. Cir. 1991); see also Paul M. Janicke,
Heat of Passion: lfha Really Happened in Graver Tank, 24 AIPLA QJ. 1, 91 n.329 (1996) (citing
over 30 cases and commentators in agreement). Even this seemingly obvious rule has been called into
question, however. See Baxter Healthcare Corp. v. Spectramed, Inc., 49 F.3d 1575, 1583 (Fed. Cir.)
(holding that accused device may infringe patent even though accused device is obvious in view of the
prior art), cert. denied, 116 S. Ct. 272 (1995). The approach taken in Baxter seems both incorrect as
a matter of policy and flatly inconsistent with Wilson Sporting Goods.
78. This need not be the case. The requirement that a patented invention be "useful" is only laxly
enforced. See Tol-O-Matic, Inc. v. Proma Produkt-Und Marketing G.m.b.H., 945 F.2d 1546, 1553
(Fed. Cir. 1991); Moleculon Research Corp. v. CBS, Inc., 793 F.2d 1261, 1269 (Fed. Cir. 1986);
cf. United States Steel Corp. v. Phillips Petroleum Co., 865 F.2d 1247 (Fed. Cir. 1989) (holding that
a finding of infringement compels a finding of utility). It is therefore theoretically possible to obtain
a patent on a nonobvious but inefficient (relative to the prior art) way of doing things. Nonetheless,
this hypothetical case need not detain us, since infringement of such a patent would presumably be rare.
effort to avoid infringement. As noted above, those changes may result in

real improvements to the original patented process, but they need not do
so. Let us define all such changes with an additionalsocial value of less
than $200 as "minor improvements." This includes those whose changes
are inefficient, as well as those whose changes add some value, but less
than the minimum additional value required for an improvement patent
(that is, those developers whose improvements are obvious in view of the
original patent or the prior art).
The law offers no protection to such minor improvers, beyond the
possibility of avoiding infringement as described above. In other words,
minor improvers must take their chances with infringement. Further,
because their contribution does not rise to the minimum value necessary for
patent protection, minor improvers cannot prevent either the original patent
owner or other competitors from copying their improvement and using it
in competing products. While in some cases the minor improver may be
able to appropriate some of the value of her improvement through trade
secrecy or first mover advantages,79 there is no guarantee that she will
be able to do so. And if the minor improvement does turn out to be
infringing, only the original patent owner can use it during the term of his
patent. Thus, the patent owner captures the value of the improvement in
this situation.
2. Significant Improvers.-Subsequent developers whose improve-

ments exceed the minimum social value threshold for patentability fare
slightly better under patent doctrine. Let us define a "significant
improver" as a developer whose improvement has an additional value to
society of more than $200, but less than some higher number. For now,
we will set that higher number at $1000.1, Developers whose improve-
ment on a patented technology adds between $200 and $1000 of social
value still risk infringement of the underlying patent, either literally or
under the doctrine of equivalents. But these significant improvers have
another weapon in their arsenal-they can patent their improvements.81
79. On the value of first mover advantages, see Richard C. Levin et al., Appropriatingthe Returns
from Industrial Research and Development, 3 BROOKINGS PAPERS ON ECON. ACTIVITY 783, 816
(1987). On the limitations of such advantages, particularly in industries where copying is easy, see
William L. Baldwin & Gerald L. Childs, The FastSecond and Rivalry in Research and Development,
36 S. ECON. 1. 18, 21 (1969); Janusz A. Ordover, Economic Foundations'andConsiderationsin
ProtectingIndustrialand Intellectual Property, 53 ANTITRUST L.J. 503, 507 (1985).
80. I discuss factors relevant to setting the proper upper value for "significant" improvements in
Part VI.
81. By hypothesis, the improvement is nonobvious-it has contributed more than the minimum
social value required for patentability. Naturally, the minimum social value test is highly stylized.
Improvers must meet other requirements, such as enablement and utility, in order to qualify for a patent
on their improvement. Still, these factors should not concern us unduly. Enablement is generally
1997] Economics of'Improvement 1009
It may seem odd at first that an improvement which is nonobvious in

view of the prior art can still infringe on a patent which is part of that prior
art. Such a situation can arise in at least three different ways. First,
adding to an infringing machine does not relieve a defendant of liability for
infringement. Thus, if the improvement consists of additions to the basic
structure claimed in the original patent, that improvement will not avoid
infringement even though it would not have been obvious to the original
patentee. A clear example of this case is Marconi Wireless Telegraph Co.
v. De Forest Radio Telephone & Telegraph Co., I where the court held
that a triode (a container having three electrodes) infringed a prior patent
on a diode (a container having two electrodes), because the triode neces-
sarily contained two electrodes in a container. The second reason such a
situation might arise is that enablement is tested only as of the time the ori-
ginal inventor files for a patent. If the original inventor has at that time
enabled the use of an entire class of products, a claim covering that entire
class is warranted. But if the class subsequently expands to include other
species not conceived at the time of the first patent, the generic claim lan-
guage will allow the first inventor to capture those new species within the
scope of his claim.' Finally, a patent on a new product will extend to
new and unanticipated uses of that product. For example, in B. G. Corp.
v. Walter Kidde & Co.," the court held that the inventor of the spark
plug was entitled to control the use of spark plugs in airplanes, even
though he never foresaw that use at the time of his invention.
Patenting an improvement does not prevent it from infringing on the
original patent, and so it does not directly protect the significant improver
from suit.' But an improvement patent will affect what value the original
patent owner can capture in such a suit. The original patent owner is
entitled to damages for past infringement, and an injunction against future
use of the infringing improvement. In contrast to the case of the minor
improver, however, the original patent owner cannot capture the value of
the significant improvement because that improvement is itself protected by
a patent. Should the original patent owner try to use the patented
improvement, the significant improver can sue him for damages and
an injunction." This situation is known as the case of "blocking
within the control of a patent applicant, and an improvement to an existing patent with significant social
value will normally clear the utility hurdle with ease.
82. 236 F. 942 (S.D.N.Y. 1916), aftd, 243 F. 560 (2d Cir. 1917).
83. On this problem, see Robert P. Merges, Rent Control in the PatentDistrict: Observationson
the Grady-AlexanderThesis, 78 VA. L. REV. 359, 379 n.73 (1992).
84. 79 F.2d 20, 22 (2d Cir. 1935).
85. See supra note 74. But see Zygo Corp. v. Wyko Corp., 79 F.3d 1563 (Fed. Cir. 1996)
(suggesting that an improvement's patentability is evidence of noninfringement).
86. See 35 U.S.C. § 154 (1994) (providing only the right to exclude others from working the
invention, rather than an affirmative right to practice the invention). Cases involving blocking patents
patents."" The original patent owner can prevent the improver from
using his patented technology, but the improver can also prevent the
original patent owner from using the improvement. Unless the parties
bargain, no one gets the benefit of the improvement."
3. Radical Improvers.-In some sense, any invention, no matter how

pioneering, can be thought of as an improvement on prior work. 9 One
might conceive of the automobile as an improvement on the horse and
buggy, for example, or of computers as an improvement on calculators.'
Clearly, not all improvements escape infringement-even significant
improvements do not do so.91 But at some point-$1000 in our stylized
example-the patent law treats an improvement as sufficiently radical
to constitute a departure from all that came before it, even though the
improvement may fall within the literal claims of the original patent.
These radical improvements are protected under the "reverse doctrine of
equivalents. "'
include Cantrellv. Wallick, 117 U.S. 689, 694 (1886); Cochranev. Deener, 94 U.S. 780, 787 (1877);
Atlas Powder Co. v. E.L du PontDe Nemours & Co., 750 F.2d 1569, 1576 (Fed. Cir. 1984); Ziegler
v. Phillips Petroleum Co., 483 F.2d 858, 871-72 (5th Cir. 1973); and Bryan v. Sid W. Richardson,
Inc., 254 F.2d 191, 194 (5th Cir. 1958), as well as those cited above.
87. For general treatments of blocking patents in the literature, see, for example, Robert Merges,
Intellectual PropertyRights and BargainingBreakdown: The Case ofBlocking Patents, 62 TENN. L.
REV. 75 (1994); Merges & Nelson, supra note 32, at 860-62. For our purposes, patents on comple-
mentary goods owned by different parties might also be thought to be "blocking" in a limited sense.
If I own the patent on a hammer and you own the patent on an anvil, each of us is free to make our
own products, so the patents are not strictly "blocking." However, absent a license agreement neither
of us may be able to exploit our patents to maximum advantage, because of the synergy between the
two. See Michael A. Sanzo, Antitrust Law and PatentMisconduct in the ProprietaryDrugIndustry,
39 ViLL. L. REV. 1209, 1245-46 (1994); David S. Taylor, Note, The Sinking of the United States
Electronics Industry Within JapanesePatent Pools, 26 GEO. WASH. J.INT'L L. & ECON. 181, 201
(1992).
88. This is strictly true only if the patents are completely rather than partially blocking. Because
of the way a particular pair of patent claims is written, it is possible that the overlap between the two
claims will be only partial. For example, a claim on an improvement might prevent the original inven-
tor from using that improvement only in a certain way or in combination with some but not all other
products. Such situations may be thought of as involving "partially blocking patents."
89. See supra notes 30-33 and accompanying text.
90. This last example is not exactly hypothetical. Texas Instruments, for example, has asserted
its basic calculatorpatent against a variety of modem products containing integrated circuits. See, e.g.,
Texas Instruments, Inc. v. United States Int'l Trade Comm'n, 805 F.2d 1558 (Fed. Cir. 1986)
(asserting Texas Instruments' patent rights against importers of modem calculators).
91. See Phillips Petroleum Co. v. United States Steel Corp., 673 F. Supp. 1278 (D. Del. 1987),
aft'd, 865 F.2d 1247 (Fed. Cir. 1989) ("[Tlhe mere fact that the accused products are superior does
not, in and of itself, permit Defendants to escape liability under the reverse doctrine of equivalents.").
92. On the reverse doctrine of equivalents generally, see SR[ Internationalv. MatsushitaElectric
Corp., 775 F.2d 1107 (Fed. Cir. 1985); Laura A. Handley, Refining the Graver Tank Analysis with
Hypothetical Claims: A Biotechnology Exemplar, 5 HARV. J.L. & TECH. 31, 40-41 (1991); C.F.
Pigott, Jr., Equivalents in Reverse, 48 J. PAT. & TRADEMARK OFF. SOc'Y 291 (1966); Michael S.
Greenfield, Note, RecombinantDNA Technology. A Science Struggling with the PatentLaw, 44 STAN.
Despite its name and its origin,' the reverse doctrine of equivalents
is not merely the "flip side" of the doctrine of equivalents used in infringe-
ment analysis.' The reverse doctrine protects from infringement a
product "so far changed in principle from a patented article that it performs
the same or a similar function in a substantially different way, but
nevertheless falls within the literal words of the claim."' While the
reverse doctrine offers relief from infringement for radical improvers, its
application by the courts to excuse literal infringement is rare.' Unlike
the doctrine of equivalents, which can apply both to technology known at
the time of the patent application and to subsequently developed
technology,' the only application of the reverse doctrine is to protect
subsequent improvements from infringement if they are sufficiently
radical.98 An example of the reverse doctrine at work is Scripps Clinic
& Research Foundationv. Genentech, Inc." In that case, Scripps had a
patent for a blood clotting product, human Factor VIII:C, which it purified
from human blood. Genentech produced identical Factor VIII:C from
laboratory bacteria using recombinant DNA processes. Genentech argued
that its recombinant Factor VIII:C was "changed in principle" from
Scripps's purified Factor VIII:C.1 ° The Federal Circuit held that
L. REV. 1051, 1078-79 (1992); and Jonathan Geld, Back Upi-Using the Reverse Doctrine of
Equivalents to Halt the Advance of Functionally Claimed Software 3-24 (1996) (unpublished
manuscript, on file with the Texas Law Review).
93. Westinghouse v. Boyden Power Brake Co., 170 U.S. 537 (1898) is generally cited as the first
case establishing the reverse doctrine of equivalents. In Graver Tank, the Supreme Court referred to
the doctrine of equivalents as a unified phenomenon that included within its scope the reverse doctrine
of equivalents. See Graver Tank & Mfg. Co. v. Linde Air Prods. Co., 339 U.S. 605, 608-09 (1950).
Nonetheless, the development of the doctrines has not been entirely parallel.
94. See Mark D. Janis, Unmasking Structural Equivalency: The Intersection of § 112, 6
Equivalentsandthe Doctrine ofEquivalents, 4 ALE. L.J. Sa. & TECH. 205,213-15 (1994) (identifying
differences in the two analyses).
95. Graver Tank, 339 U.S. at 608-09; see also Handley, supra note 92, at 41 (comparing the
"different in principle" test to the function-way-resulttest traditionally applied in doctrine of equivalents
cases).
96. See Geld, supra note 92, at 9 (observing that only five reported decisions accept a reverse
doctrine of equivalents argument and the vast majority of courts reject the argument when it is pre-
sented to them). Merges and Nelson suggest that the doctrine must be rarely used in order to preserve
the certainty (and therefore the incentive effect) associated with the patent. Merges & Nelson, supra
note 32, at 867 n.120.
97. See, e.g., Pennwalt Corp. v. Durand-Wayland, Inc., 833 F.2d 931, 941-42 n.4 (Fed. Cir.
1987) (en banc) (Bennett, J., dissenting); Texas Instruments, Inc. v. United States Int'l Trade Cormm'n,
805 F.2d 1558, 1563 (Fed. Cir. 1986); Merges & Nelson, supra note 32, at 855.
98. It is also worth noting that the reverse doctrine of equivalents seems to involve evaluating
equivalence between the claimed invention and the accused product viewed as a whole, as opposed to
the element-by-element approach used in the doctrine of equivalents. CompareTexas Instruments, 805
F.2d at 1571 (discussing the reverse doctrine of equivalents), vith Pennwalt, 833 F.2d at 935-39
(discussing the doctrine of equivalents).
99. 927 F.2d 1565 (Fed. Cir. 1991).
100. Id. at 1581.
material issues of fact precluded summary judgment on this issue, and

reversed a district court finding of infringement." °' Michael Greenfield
offers another potential application of the reverse doctrine of equivalents
in the biotechnology field. He suggests that patents on human gene-
fragment sequences called "express sequence tags" (ESTs) should not
preclude the later use of those gene sequences in recombinant protein
production."° The case he poses' is instructive for our purposes. The
patented ESTs are primarily useful in gene mapping-a valuable use, but
not nearly as valuable as the use of the gene sequence to produce recom-
binant proteins would be. 1 Where the value of the improvement greatly
exceeds the value of the original invention, application of the reverse
doctrine of equivalents seems most likely."4
The existence of the doctrine provides protection for radical improvers
who nonetheless literally infringe the original patent-those whose improve-
ments change the original invention "in principle" or in a substantial way,
but who still fall within the literal language of the claims.' 5 The reverse
doctrine of equivalents therefore benefits radical improvers at the expense
101. Id.; see also Leesona Corp. v. United States, 530 F.2d 896, 906 (Ct. Cl. 1976); Precision
Metal Fabricators, Inc. v. Jetstream Sys. Co., 693 F. Supp. 814, 819 (N.D. Cal. 1988); Beloit Corp.
v. J.M. Voith, G.m.b.H., 626 F. Supp. 991, 1007-08 (E.D. Va.), aff'd, 802 F.2d 471 (Fed. Cir.
1986) (all finding no infringement based on reverse doctrine of equivalents). But see Kevin J.
McGough & Daniel P. Burke, A Casefor Expansive Patent Protectionof Biotechnology Inventions, 6
HARv. J.L. & TECH. 85, 97-98 (1992) (arguing that Scripps provides weak precedent for the reverse
doctrine of equivalents, since the court also noted possible differences between the products produced
by the two methods).
102. See Greenfield, supra note 92, at 1090-91. For a general discussion of the patentability of
ESTs, and a brief review of past efforts to patent them, see Rebecca S.Eisenberg & Robert P. Merges,
OpinionLetter as to the Patentabilityof CertainInventions Associatedwith the Identificationof Partial
eDNA Sequences, 23 AIPLA Q.J. 1 (1995).
103. Eisenberg and Merges raise this point obliquely when they suggest that patents on research
tools are more troubling than patents on end products because of the danger that research patents will
restrict further productive uses of those research tools. However, they do not treat the problem
explicitly under the reverse doctrine of equivalents. The debate over patents on research tools is played
out in a recent issue of the AIPLA Quarterly Journal. See Eisenberg & Merges, supra note 102, at
19; Scott A. Chambers, Comments on the Patentabili of Certain Inventions Associated with the
Identification of PartialcDNA Sequences, 23 AIPLA Q.J. 53, 53-54 (1995) (arguing that patents on
research tools are no more dangerous than other patents because they remove nothing from the public
domain); Rebecca S.Eisenberg& Robert P. Merges, Reply to Comments on the Patentabilityof Certain
InventionsAssociated with the Identification of PartialcDNA Sequences, 23 AIPLA Q.J. 61, 62 (1995)
(arguing that patents on research tools will restrict the dissemination of these tools to other researchers).
104. For a discussion of how to calibrate these values in practice, as well as the economic impli-
cations of the reverse doctrine of equivalents, see infra Part V.
105. Radical improvers who do not fall within the literal language of the claims do not, strictly
speaking, benefit from the reverse doctrine of equivalents, which applies only in cases of literal
infringement. See Merges & Nelson, supra note 32, at 867 n.120. In practical terms, however, the
effect of applying the reverse doctrine is the same as failing to find infringement under the doctrine of
equivalents-in either case, the question is whether there are substantial differences between the
patented method and the accused product. Thus, in some sense all findings of nonequivalencehave the
same basis.
of the original patentee, and so encourages radical improvements, just as

the blocking patents rule provides some lesser encouragement to significant
improvements." o°
III. Treatment of Improvements in Copyright Law

A. The Scope of Protection Under CopyrightLaw
While the basic incentive structure of copyright law is the same
as patent law-a limited grant of exclusive rights to creators in order
to encourage both more creation and the dissemination of existing
works-there are substantial differences between the two doctrines.
Copyright law does not require examination by the government or the
"issuance" of a copyright; instead, original works of authorship are
protected immediately upon their creation."rc While there are certain
requirements for protection-the work must fall within the scope of
copyright protection,"'8 must be original,"rc and must be fixed in a tan-
gible medium of expression"1°-they are nowhere near as stringent as the
requirements imposed by the patent law. And the duration of copyright
protection is much longer than the term of patent protection-usually
seventy-five years or more."'
Taken together, these facts about copyright-the automatic protec-
tion of a broad range of works for a long period of time, with minimal
requirements-suggest that it is a more powerful intellectual property
106. For an interesting suggestion that application of the reverse doctrine of equivalents may
actually benefit patentees in some circumstances by cementing the validityof the patent claims, see Karl
Bozicevic, The "Reverse Doctrine of Equivalents" in the World of Reverse Transcriptase,71 J. PAT.
& TRADEMARK OFF. Soc'Y 353, 354 (1989).
107. One caveat is in order here. Under U.S. copyright law, registration of the copyrighted work
is a prerequisite to filing suit for copyright infringement. See 17 U.S.C. § 411(a) (1994). In that
limited sense, a copyright in a work is only a potential right until it is registered.
108. Section 102(a) provides a nonexclusive list of the types of works eligible for copyright pro-
tection. Id. § 102(a).
109. See Feist Publications, Inc. v. Rural Tel. Serv. Co., 499 U.S. 340, 345 (1991) ("Original,
as the term is used in copyright, means only that the work was independently created by the author...
and that it possesses at least some minimal degree of creativity.").
110. See 17 U.S.C. § 102(a) (1994). The requirement is satisfied when the embodiment of the
work "issufficiently permanent or stable to permit it to be perceived, reproduced, or otherwise
communicated for a period of more than transitory duration." Id. § 101. The fixation requirement is
quite easily satisfied, at least under certain modem cases. See MAI Sys. Corp. v. Peak Computer,
Inc., 991 F.2d 511, 518 (9th Cir. 1993) (holding that the loading of copyrighted software into a
computer's RAM for diagnosis of computer problems creates a copy that is sufficiently permanent to
be fixed).
111. The term of a copyright depends on who creates it. In the case of individual authors, it
extends for the life of the author plus 50 years. See 17 U.S.C. § 302(a) (1994). In the case of works
made for hire, the term is 75 years from the date of publication, or 100 years from creation, whichever
is shorter. See id. § 302(c). Legislation currently pending before Congress would extend each of these
terms by 20 years. See Copyright Term Extension Act of 1997, H.R. 604, 105th Cong. § 2.
right than patent law. The power of copyrights is tempered in practice,

however, by two significant limitations on what activities infringe a
copyright." 2 First, copyright protection does not extend to the ideas,
facts, or functional elements of a work, but only to the author's original
expression of those ideas or elements. Thus, a copyright owner in a
database of facts cannot prevent a user from copying the facts themselves
from the database. Only the creative effort (if any) that has gone into the
selection or organization of materials is entitled to protection."'
Similarly, copyright owners cannot protect the basic ideas of their novels
or movies115 or the functional aspects of their lamps11 6 or computer
programs,117 though they are entitled to protect the particular implemen-
tations of those ideas or functions to the extent that the implementations are
original and creative. For works whose value resides primarily in ideas or
functional attributes, therefore, copyright is inferior to patent as a means
of protection. 1
The second limitation on the power of copyright owners relative to
patent owners lies in the nature of activities deemed infringing. Anyone
who makes, uses, or sells a product within the claims of a patent infringes
that patent, 1 whether or not they copied the idea from the patent owner,
and indeed whether or not they were even aware that the patented tech-
nology existed. By contrast, a copyright protects its owner only against
those who have actually taken the plaintiff's work. Independent develop-
ment of the work is a complete defense. 1" As a result, copyright in-
fringement cases frequently focus on the defendant's actions and intentions
in preparing the accused work. 21 Further, copyright law normally does
not prevent another from using the copyrighted work, as long as she does
not make a copy of the work or perform or display it publicly."
112. The fair use defense to infringement is a third significant limitation on copyright. Fair use
is discussed in more detail below. See infra notes 171-96 and accompanying text.
113. See 17 U.S.C. § 102(b) (1994).
114. See Feist, 499 U.S. at 347-51.
115. See, e.g., Nicholsv. Universal Pictures Corp., 45 F.2d 119, 121 (2d Cir. 1930).
116. See Mazer v. Stein, 347 U.S. 201, 212-13 (1954).
117. See Computer Assocs. Int'l, Inc. v. Altai, Inc., 982 F.2d 693, 714 (2d Cir. 1992).
118. See generally A. Samuel Oddi, An Uneasier Casefor Copyrightthanfor PatentProtection
of Computer Programs, 72 NEB. L. Rsv. 351 (1993); Pamela Samuelson et al., A Manifesto
Concerningthe Legal Protection of Computer Programs,94 COLUM. L. REV. 2308 (1994) (both dis-
cussing the limitations of copyright protection of computer programs).
119. Subject to the narrow limitation discussed supra notes 89-106 and accompanying text.
120. See 2 PAUL GOLDSTMIN COPYRIGHT § 7.2.2, at 7:19 (2d ed. 1996).
121. Because direct evidence of copying is typically unavailable, courts allow evidence that a
defendant had access to the copyrighted work and that the two works are substantially similar to serve
as a surrogate for proof of copying. See, e.g., Sid & Marty Krofft Television Prods., Inc. v.
McDonald's Corp., 562 F.2d 1157, 1162 (9th Cir. 1977); Arnstein v. Porter, 154 F.2d 464, 468 (2d
Cir. 1946).
122. See 17 U.S.C. § 106 (1994) (providing that a copyright gives the owner the exclusive right
to copy, adapt, distribute, and in some cases publicly perform or display the work). The exception to
B. EvaluatingInfringement by Improvers
Unlike patents, copyrights do not have claims. Infringement is tested
by comparing the accused work to the copyrighted work to determine
whether substantial copying has occurred." z The simplest case is one in
which the entire copyrighted work has been copied verbatim. This of
course constitutes infringement, 124 just as copying the patent owner's in-
vention infringes the patent. Beyond this point, however, the patent and
copyright infringement tests diverge. For one thing, it is possible to
infringe a copyright by copying only a small portion of the copyright
owner's work. Copying even a few hundred words from a book,"z or
a few seconds of music from a song,12 may be enough to infringe the
copyright in the entire work. By contrast, copying part but not all of a
claimed invention is not patent infringement, since every element of the
this rule is digitized information, where the courts have effectively enabled the copyright owner to
control private uses of the work by defining any use in a computer or computer network as involving
the creation of a copy. See, e.g., MAI Sys. Corp. v. Peak Computer, Inc., 991 F.2d 511, 519 (9th
Cir. 1993) (holdingthat the transfer of a computer program from a permanent storage device to a com-
puter RAM creates a copy for copyright purposes). For criticism of this approach, see, for example,
Fred H. Cate, Law in Cyberspace, 39 HOwARD L.J. 565, 575-77 (1996); Margaret Chon, New Wine
Burstingfrom Old Bottles: Collaborative InternetArt, Joint Works, and Entrepreneurship,75 OR. L.
REV. 257,260-63 (1996); Niva Elkin-Koren, Copyright Law andSocial Dialogue on the Information
Superhighway:The Case Against CopyrightLiability of Bulletin Board Operators, 13 CARDOZO ARTS
& ENT. L.J. 345, 381-83 (1995); Mark A. Lemley, Dealing with Overlapping Copyrights on the
Internet, 22 U. DAYTON L. REV. (forthcoming 1997) (manuscript at 5-6, on file with the Texas Law
Review); Jessica Litman, Revising Copyright Lawfor the Information Age, 75 OR. L. REv. 19, 21
(1996); Jessica Litman, The Exclusive Right to Read, 13 CARDozO ARTS & ENT. L.J. 29, 41-43
(1994); Charles R. McManis, Taking TRIPs on the InformationSuperhighway:InternationalIntellectual
Property Protection and Emerging Computer Technology, 41 VILL. L. REV. 207, 263-68 (1996);
Barbara Cohen, Note, A ProposedRegime for Copyright Protectionon the Internet,22 BROOKLYN J.
INT'L L. 401, 412 (1996); Michael E. Johnson, Note, The Uncertain Future of Computer Software
Users' Rights in the Aftermath of MAI Systems, 44 DuKE L.J. 327, 333-36 (1994); Katrine Levin,
Note, MAI v. Peak: Should Loading Operating System Software into RAM Constitute Copyright
Infringement?, 24 GOLDEN GATE U. L. REv. 649 (1994); James V. Mahon, Note, A Commentary on
ProposalsforCopyrightProtectionon the NationalInformationInfrastructure,22 RUTRmS COMPUTER
& TECH. L.J. 233, 240-45 (1996); Carol G. Stovsky, Note, MAI Systems Corp. v. Peak Computer,
Inc.: Using CopyrightLaw to ProhibitUnauthorizedUse of ComputerSoftware, 56 OHIO ST. L.J. 593,
600-09 (1995); and compare David Nimmer, Brains and Other Paraphernalia of the Digital Age, 10
HARV. J.L. & TECH. 1, 10-12 (1996) (endorsing the result of MAI under existing law, but proposing
statutory changes to ameliorate its effects).
123. While courts often confuse the issue of "substantial similarity" between two works as evi-
dence of copying with the issue of whether substantial protected expression has in fact been taken, the
two issues are analytically distinct. See Cohen, supra note 44, at 735-39 (criticizing courts for con-
flating the determination that a work was actually copied with the determination that such copying went
so far as to constitute improper appropriation).
124. The finding of infringement is subject to the fair use defense. See Sony Corp. v. Universal
City Studios, Inc., 464 U.S. 417, 433 (1984) (holding, in the context of videotaping television shows
for later viewing, that even verbatim copying of an entire work can be fair use in some circumstances).
125. See Harper & Row Publishers, Inc. v. Nation Enters., 471 U.S. 539, 548-49 (1985).
126. See Jarvis v. A & M Records, 827 F. Supp. 282, 288-93 (D.N.J. 1993); Grand Upright
Music Ltd. v. Warner Bros. Records, Inc., 780 F. Supp. 182, 183-85 (S.D.N.Y. 1991).
claimed invention (or an equivalent) must be present in the accused

device.127 Thus, improvers may be liable for copying even a relatively
small portion of the original work." s
Further, copyright infringement is not limited to cases of verbatim
copying. The courts long ago recognized that such a limitation would ren-
der copyright protection ineffective, allowing "plagiarist[s] [to] escape by
immaterial variations." 129 Defendants may infringe the copyright by
taking only the "nonliteral" elements of a work, such as the plot outline of
a movie, 13° the structure, sequence and organization of a computer pro-
gram,13 1 or even the "total concept and feel" of a song, a television
show, or a greeting card.13 1 While one might analogize this protection
against nonliteral infringement to patent law's doctrine of equivalents, the
animating principle is somewhat different. Nonliteral copyright infringe-
ment involves the copying of elements of a work which, while not literally
contained in the physical manifestation of the work, nonetheless constitute
part of the original expression of the work, as distinguished from its idea
or its functional attributes. 33 Thus, improvements in the text of a work
which keep the structure identical likely will not escape the web of non-
literal copyright infringement. On the other hand, the fact that two dif-
ferent plot devices for a mystery novel arefunctionally equivalentdoes not
mean that one will infringe the copyright of the other. By contrast, the test
for the doctrine of equivalents in patent law-insubstantial differences, as
measured by factors such as reasonable interchangeability--allows the law
to reach different works that operate on similar functional principles.
127. See Permwalt Corp. v. Durand-Wayland, Inc., 833 F.2d 931, 939 (Fed. Cir. 1987) (en
bane).
128. There are a number of defenses in the Copyright Act permitting various acts of copying for
certain public interest purposes, or allowing certain acts on the payment of a compulsory license fee.
See 17 U.S.C. §§ 107-120 (1994). One of these defenses, the fair use doctrine, is discussed in greater
detail infra notes 171-96 and accompanying text.
129. Nichols v. Universal Pictures Corp., 45 F.2d 119, 121 (2d Cir. 1930).
130. See, e.g., Sheldon v. Metro-Goldwyn Pictures Corp., 309 U.S. 390, 396-98, 397 (1940)
(confirming that the appropriation of a story line beyond the "mere use of [the] basic plot" could
amount to copyright infringement).
131. See Whelan Assocs., Inc. v. Jaslow Dental Lab., Inc., 797 F.2d 1222, 1248 (3d Cir. 1986);
Computer Assocs. Int'l, Inc. v. Altai, Inc., 982 F.2d 693, 702-12 (2d Cir. 1992).
132. See Roth Greeting Cards v. United Card Co., 429 F.2d 1106, 1110 (9th Cir. 1970); see also
Sid & Marty Krofft Television Prods., Inc. v. McDonald's Corp., 562 F.2d 1157, 1167 (9th Cir.
1977); Arnstein v. Porter, 154 F.2d 464, 469-73 (2d Cir. 1946) (both applying the "total concept and
feel" test).
133. See Sheldon, 309 U.S. at 409 (enjoining a successful movie based in part on a copyrighted
play, even though only a small part of the movie literally infringed the script); see also Abend v. MCA,
Inc., 863 F.2d 1465, 1478-80 (9th Cir. 1988) (noting the potential availability of damages on remand
for the continued use of the movie Rear Window, a derivative work based on plaintiff's original short
story, but refusing to make available an injunction against continued distribution because of a valid
license existing when the movie was made), aff'd on other grounds sub nom. Stewart v. Abend, 495
U.S. 207 (1990).
In addition to the broad definition of a copy subject to the constraints

of the Copyright Act, section 106(2) of the Act also gives the copyright
owner the right to control the making of adaptations, or "derivative
works."" It is not clear precisely how a derivative work differs from
a nonliteral copy, or what section 106(2) adds to the provisions of
106(1).3 Section 101 of the Copyright Act defines a derivative work
only in terms of examples: "A 'derivative work' is a work based upon one
or more preexisting works, such as a translation, musical arrangement,
dramatization, fictionalization, motion picture version, sound recording, art
reproduction, abridgement, condensation, or any other form in which a
work may be recast, transformed, or adapted. "136 Under this definition,
a derivative work can infringe despite having no literal text in common
with the original (for example, in the case of a Navajo translation of an
Hungarian work). Nonetheless, many of the same limitations that apply to
infringement by copying apply equally to infringement by derivation. In
Litchfield v. Spielberg,137 for example, Litchfield claimed that a screen-
play she had written and submitted to the defendants called Lokey from
Maldomar was infringed by defendants' enormously popular movie
E.T.3 The defendants denied copying the script, and the Ninth Circuit
concluded that it could not infer copying because the two works were not
"substantially similar." 13 9 The court rejected Litchfield's claim that E.T.
was a derivative work for the same reasons, holding that:
[A] work is not derivative unless it has been substantially copied
from the prior work.
We have stated that "[a] work will be considered a derivative
work only ifit would be consideredan infringingwork ifthe material
which it has derived from a prior work had been taken without the
consent of a copyright proprietor of such prior work." 140
134. 17 U.S.C. § 106 (1994).

135. See Goldstein, supra note 37, at 215-17 (lamenting that the case law has largely failed to
draw a principled line distinguishing derivative rights from reproduction rights). Goldstein suggests
that copies end-and derivativeworks begin-wherethe alteration of the original work causes it to com-
pete in a new market. See id. at 217. Thus, movie versions of books and French translations of
English works are both derivative works, since in both cases the new work competes in a different mar-
ket from the original. See 17 U.S.C. § 101 (offering examples of derivative works); see also Lone
Ranger Television, Inc. v. Program Radio Corp., 740 F.2d 718, 721 (9th Cir. 1984) (quoting Goldstein
and finding that the alterations required in producing taped drama episodes for radio rendered them
derivative products of the original copyright-protected scripts).
136. 17 U.S.C. § 101 (1994).
137. 736 F.2d 1352 (9th Cir. 1984).
138. Id. at 1354-55.
139. Id. at 1358.
140. Id. at 1357 (emphasis added) (citing United States v. Taxe, 540 F.2d 961, 965 n.2 (9th Cir.
1976)); accord Vault Corp. v. Quaid Software Ltd., 847 F.2d 255, 267-68 (5th Cir. 1988) (finding
no derivative work is created where only an insignificant portion of the plaintiff's work was copied).
In light of this, it is not evident what section 106(2) adds to the

rights already granted under section 106(1). 141 A few theories have been
advanced. The derivative works right may extend the reach of the protec-
tion against copying in certain limited factual circumstances-for example,
where the derivative work is not fixed in a tangible medium of expression
and therefore does not constitute a "copy."1 It may preclude owners
of a particular copy of a work from altering or transforming that copy not-
withstanding the first sale doctrine,'43 though such an application of the
adaptation right has been controversial." And in conjunction with sec-
tion 103 of the Act, it may allow copyright owners to protect a "chain" of
works, even though works at the end of the chain bear little or no resem-
blance to the original works in the chain. Thus, Disney might protect a
book, a script derived from that book, a movie made from the script, and
stuffed animals derived from the movie, even though the stuffed animals
have no resemblance to anything described in the original book.'
Finally, in some cases copyright law will extend to cover an
"improvement" which does not copy from the original work at all, but is
merely designed to interoperate with the copyrighted work. In two cases,
courts have held that a copyrighted talking teddy bear which contained an
audio tape controlling its voice was infringed by a competitor's audio tape
141. In his treatise, Nimmer refers to § 106(2) as "completely superfluous." 2 NIMMER &
NIMMER, supra note 9, § 8.09[A], at 8-123; see also Sterk, supra note 36, at 1215-16 (arguing that
in most cases derivative rights are pure economic rents to copyright owners).
142. Section 106(2) does not require that a derivative work be fixed in order to be infringing.
CompareMidway Mfg. Co. v. Artic Int'l, Inc., 704 F.2d 1009, 1014 (7th Cir. 1983) (holding that cir-
cuit boards that speed up copyrighted video games create unlawful derivative works), with Lewis
Galoob Toys, Inc. v. Nintendo of Am., Inc., 780 F. Supp. 1283, 1291 (N.D. Cal. 1991) (holding that
game devices that speed up copyrighted video games do not create unlawful derivative works), aft'd,
964 F.2d 965 (9th Cir. 1992). The Ninth Circuit's decision in Galoob acknowledged that an accused
work did not need to be "fixed" within the meaning of the Copyright Act in order to infringe, but did
hold that it must incorporate the copyrighted material in some concrete or permanent form. Galoob,
964 F.2d at 967-69. It is not clear how this incorporationwould differ from fixation, which has been
given an expansive interpretation by other decisions. See, e.g., MAI Sys. Corp. v. Peak Computer,
Inc., 991 F.2d 511,517-19 (9th Cir. 1993); see also Carol S. Curme, Note, Derivative Works of Video
Game Displays: Lewis Galoob Toys, Inc. v. Nintendo of Am. Inc., 61 U. CINN. L. REv. 999, 1022-27
(1993) (criticizing the Galoob court's treatment of the derivative works issue).
143. See Mirage Editions, Inc. v. Albuquerque A.R.T. Co., 856 F.2d 1341, 1344 (9th Cir. 1988);
Mufioz v. Albuquerque A.R.T. Co., 829 F. Supp. 309,314-15 (D. Alaska 1993); National Geographic
Soc'yv. Classified Geographic, Inc., 27 F. Supp. 655, 660 (D. Mass. 1939) (all permitting the copy-
right owner to restrict purchasers of copies from cutting up magazines or books and reassembling them
in different form). The first sale doctrine generally permits a purchaser to dispose of the purchased
copy at his discretion, but it does not apply to derivative works. See 17 U.S.C. § 109(a) (1994).
144. See Lee v. Deck the Walls, Inc., 925 F. Supp. 576, 579-80 (N.D. 1l1. 1996) (reaching the
opposite conclusion from the Mirage court); 2 GOLDsTEIN, supra note 120, § 5.3, at 5:81 (condemning
Mirage as an unwarranted extension of the Copyright Act).
145. Of course, § 103 (which provides that derivative works containing new original expression
are separately copyrightable) is arguably sufficient to take care of this by itself. It is not clear what
§ 106(2) adds to the "chain of copyright" theory.
containing different sounds, but which was designed to be played in the

same bear by people who already own the bear. 1" Other cases have
given more favorable treatment to independent works that interoperate with
an original copyrighted work, but they have generally done so under the
fair use doctrine after concluding that the production of such works
constituted infringement. 7
C. Mechanismsfor ProtectingImprovers
One thing that the derivative works right does do is make it clear that
copyright law is intended to reach improvers as well as counterfeiters. The
copyright owner's control over works in markets other than the one he
entered is no accident, but part of the scheme of copyright protection.
Nonetheless, there are a few legal rules that function to protect improvers
in copyright law. To compare these rules to analogous doctrines in patent
law, let us once again hypothesize three types of improvers: minor
improvers, significant improvers, and radical improvers.'"
1. Minor Improvers.-In Part II, I defined a minor patent improver

as one who advances social utility by adding to the basic invention, but
who does not contribute enough to justify receiving a patent in her own
right on the improvement. For consistency, I will adopt the same defini-
tion here. Because copyright protection is so much easier to obtain than
patent protection, however, the effect of this rule is to make the group of
"minor improvers" very small. Most improvements will be sufficiently
original that they would ordinarily qualify for copyright protection in their
own right, and will therefore fall into the category of "significant"
improvements. Nonetheless, certain improvements to an original work can
be classified as "minor" even under this restrictive definition. Editorial
corrections or revisions that do not themselves contribute copyrightable
material fall into this category, as do factual corrections to a database or
certain types of restoration efforts to works of art.
Minor improvers receive no special protection under copyright law.
They are subject to the same rules of copyright infringement as any other
copier, notwithstanding their improvements. Thus, it is possible that minor
146. See Worlds of Wonder, Inc. v. Veritel Learning Sys., Inc., 658 F. Supp. 351, 356 (N.D.
Tex. 1986); Worlds of Wonder, Inc. v. Vector Intercontinental, Inc., 653 F. Supp. 135 (N.D. Ohio
1986). For criticism of these decisions, see Christian H. Nadan, Comment, A Proposal to Recognize
Component Works: How a Teddy Bears on the Competing Ends of Copyright Law, 78 CAL. L. REV.
1633 (1990).
147. See, e.g., Bateman v. Mnemonics, Inc., 79 F.3d 1532, 1542 (11th Cir. 1996); Sega Enters.
Ltd. v. Accolade, Inc., 977 F.2d 1510, 1527-30 (9th Cir. 1992);Lewis Galoob Toys, 964 F.2d at 970-
72.
148. Glynn Lunney attempts a somewhat similar taxonomy. See Lunney, supra note 37, at 2428-
improvements will qualify as a fair use, particularly if they are done in the
context of copies made for a public benefit and do not harm the market for
the original work.149 But most copies that contain only minor improve-
ments will not qualify as fair uses. And while it is theoretically possible
that the changes made to the original work by the minor improver will
cause the improved work to be noninfringing, in practice the likelihood of
a change being both sufficiently minor as not to qualify for copyright pro-
tection and sufficiently major as to preclude a finding of substantial
copying (or infringement of the derivative works right) is minimal. Minor
improvers are therefore likely to be guilty of copyright infringement. As
with minor improvements that infringe patents,1 improvements that
infringe copyrights are effectively captured by the copyright owner, since
anyone copying the new work would similarly be guilty of infringing the
copyright on the original work. 1
2. Significant Improvers.-This class of improvers is quite broad,

encompassing any improvement to a copyrighted work that itself contains
original, creative expression, up to some arbitrary limit separating sig-
nificant from radical improvements." Virtually any work that qualifies
as a "derivative work" under the definition in section 101 will qualify as
at least a significant improvement, since the additional effort involved in
adapting the original work to a new market will almost always include
original expression contributed by the improver. Thus, translating an
original work into a new language, making a movie from a screenplay, and
even producing stuffed animals based on a movie involve the inclusion of
significant new creative elements. Further, nonliteral copies that compete
directly with the original will often qualify as significant improvements as
well. A novel may copy the detailed plot structure of an original work, for
example, but contribute entirely new text and characters. In all of these
cases, the improver is guilty of copying from the original work, but has
also contributed valuable material to the new work.
In the patent context, we referred to this situation as involving
"blocking patents."153 The improver could not legally use the material
149. See, e.g., Sony Corp. v. Universal City Studios, Inc., 464 U.S. 417, 454-56 (1984).
151. See Lipton v. Nature Co., 71 F.3d 464, 471 (2d Cir. 1995) (holding that innocent copying
of a work will not protect the copier from infringement if the copied work itself infringed another
work); 2 NIMMER& NIMMmI, supra note 9, § 8.01[C].
152. Because there is no equivalent in copyright to the "nonobviousness" and "utility" require-
ments in patent law, even changes with a negative social value might fall into the category of significant
improvements. See, e.g., Dallas Cowboys Cheerleaders, Inc. v. Pussycat Cinema, Ltd., 604 F.2d 200
(2d Cir. 1979) (discussing the copyright implications of an "improvement" consisting of a pornographic
movie portraying the plaintiffs).
covered by the original patent without permission, but the original patent
owner similarly could not use the new material contributed by the
improver. Perhaps surprisingly, there is no analogous doctrine of
"blocking copyrights" granting copyright infringers the rights to their own
improvements. 1" Instead, section 103 of the Act provides that original
works of authorship contributed as part of the creation of a derivative work
are copyrightable only by the original copyright owners or their
licensee.155 This means, for example, that if an improver makes a movie
which is later determined to have infringed the copyright on the book on
which it was based, the improver may not be entitled to any copyright
protection for the movie-even those portions of the movie which were
contributed by the improver. 56 Indeed, in one case the fact that a script-
writer used the "Rocky Balboa" character from a prior film in creating an
154. See, e.g., MARSHALL LEAFFER, UNDERSTANDING COPYRIGHT LAW 47 (2d ed. 1995) ("A
derivative work copyright can only be obtained when the author legally used the material on which the
derivative work was based."); Goldstein, supra note 37, at 244 (explaining the logical consistency of
linking protection of a derivative work to authorization for use of the underlying work, but criticizing
the sometimes harsh results); Wendy J.Gordon, Toward a Jurisprudenceof Benefits: The Norms of
Copyrightand the Problem of Private Censorship, 57 U. CHI. L. REV. 1009, 1019, 1018-20 (1990)
(reviewing PAUL GOLDSTEIN, COPYRIGHT: PRINCIPLES, LAW AND PRACTICE (1989)) (describing the
status of unauthorized derivative works and contrasting it to the blocking patent, which yields a "more
equitable result"); Robert P. Merges, Of PropertyRules, Coase,andIntellectualProperty, 94 COLUM.
L. REV. 2655, 2659 n.15 (1994) (suggesting that "copyright law's policy favoring the reputational
interests of authors" offers the best explanation for the lack of a "blocking-copyright" doctrine). The
rule against blocking copyrights is sometimes obscured by the line the Copyright Act draws between
those who create a derivative work with the original permission of the copyright owner and those who
do not. See 17 U.S.C. § 103(a) (1994); see also JBJ Fabrics, Inc. v. Brylane, Inc., 714 F. Supp. 107,
110 (S.D.N.Y. 1989) ("[flf plaintiff did indeed have informal authorization ... to use the painting in
its fabric design, such as a non-exclusive license, then it is not using the painting unlawfully and may
not face any limits on the scope of its [derivative] copyright protection."); Goldstein, supra note 37,
at 239-40 (distinguishing these situations). Those who prepare a derivative work with permission are
presumably entitled to copyright their original contribution in the derivative work, and thus might be
thought to have a right to "block" the original copyright owner from using their added expression in
violation of the terms of the license. See JAY DRATLER JR., INTELLECTUAL PROPERTY LAW:
COMMERCIAL, CREATIVE, AND INDUSTRIALPROPEPTY § 6.01[2][c][iii] (1993) (describing the potential
for "standoffs" between licensors and licensees who have created derivative works). Even then,
though, the derivative work creator-licensee may face unexpected limits on his ability to commercialize
his derivative work, as cases like G. Ricordi & Co. v. ParamountPictures,Inc., 189 F.2d 469 (2d Cir.
1951), make clear. See generally I NIMMER & NIMMER, supra note 9, § 3.07[A][1] (illustrating
Ricordi's repudiation of the theory that a derivative-right licensee gains a property right in the new
work that is independent of the underlying work); Peter Jaszi, When Works Collide:Derivative Motion
Pictures, UnderlyingRights, and the Public Interest, 28 UCLA L. REV. 715 (1981).
155. Section 103(a) provides that "[tihe subject matter of copyright as specified by section 102
includes compilations and derivative works, but protection for a work employing preexisting material
in which copyright subsists does not extend to any part of the work in which such material has been
used unlawfully.' 17 U.S.C. § 103(a) (1994).
156. See Gracen v. Bradford Exch., 698 F.2d 300, 302-03 (7th Cit. 1983); Pamfiloff v. Giant
Records, Inc., 794 F. Supp. 933, 938 (N.D. Cal. 1992); Dynamic Solutions, Inc. v. Planning &
Control, Inc., 646 F. Supp. 1329, 1340 (S.D.N.Y. 1986); Gallery House, Inc. v. Yi, 582 F. Supp.
1294, 1297 (N.D. Ill. 1984); 1 NIMMER & NIMMER, supra note 9, § 3.06.
unauthorized sequel to Rocky caused the court not only to find that the
scriptwriter was infringing the copyright in the "Rocky" character,15 7but to
invalidate the scriptwriter's derivative work copyright completely.
Because section 103 of the Act denies copyright only in the "part of
the work" that incorporates infringing material, some infringing derivative
works may nonetheless be copyrightable in part. This is particularly true
in the case of compilations, where the infringing material can be separated
easily from other parts of the work. Thus, the House Report accom-
panying the 1976 Act provides that "an unauthorized translation of a novel
could not be copyrighted at all, but the owner of copyright in an anthology
of poetry could sue someone who infringed the whole anthology, even
though the infringer proves that publication of one of the poems was
unauthorized."15 This limitation on section 103 may provide significant
improvers with solace in some cases, notably where the copied material is
easily separated from the improver's contribution. In other cases, how-
ever, where the material is inextricably intertwined, the improver loses the
entire value of her contribution. This is the case with regard to movies or
plays based on a book. It is also likely to be the case with many improve-
ments in the field of computer software. Finally, the promise of digital
information-that it can easily be altered, translated between media, and
transformed-now seems to be its peril as well. 59
The effect of this rule is to allow the original copyright owner to cap-
ture the value of even significant improvements made by others. While the
improvements themselves are nominally in the public domain, no one can
use those improvements in conjunction with the original work without
infringing the copyright on that work.1" Further, the existence of this
157. See Andersonv. Stallone, 11 U.S.P.Q.2d (BNA) 1161, 1167-69 (C.D. Cal. 1989).
158. H.R. REP. No. 94-1476, at 57-58 (1976), reprintedin 1976 U.S.C.C.A.N. 5659,5671. One
might analogize the latter situation to the case of "partially blocking patents," discussed supra note 87.
159. See Chon, supra note 122, at 266-72 (illustrating the inadequacy of current joint-author doc-
trine in copyright law when dealing with collaborative Internet art); Niva Elkin-Koren, Cyberlawand
Social Change:A DemocraticApproach to CopyrightLaw in Cyberspace, 14 CARDOZO ARTS & ENT.
LJ. 215, 278-81 (1996) (pointing to the fluidity of works in the digital medium as one of the principal
problems copyright law must face in cyberspace, and noting that a broad derivative-works right will
reduce the usefulness of the digital medium).
160. See Russell v. Price, 612 F.2d 1123, 1128 (9th Cir. 1979) (holding that the public screening
of a film in the public domain violates the performance rights of the original play on which the film
was based); Central Point Software, Inc. v. Nugent, 903 F. Supp. 1057, 1060 (E.D. Tex. 1995)
(stating that upgrades to copyrighted software made without permission were infringing derivative
works and fell within the scope of the original copyright grant); Grove Press, Inc. v. Greenleaf Publ'g
Co., 247 F. Supp. 518, 527 (E.D.N.Y. 1965) (holding that the copyright owner of an original French
novel could prevent the copying of English translations of his work, even though the English translation
had fallen into the public domain, since copying the public domain translation would infringe the copy-
right owner's right to prepare derivative works); see also Francis M. Nevins, The Doctrine of
CopyrightAmbush:Limitations on the Free Use of Public DomainDerivative Works, 25 ST. LOUIS U.
L.J. 58 (1981). A more familiar example to many people is the film It's a Wonderful Life, which
power of capture may influence license negotiations between copyright

owners and potential improvers, tipping the balance in favor of the creators
of original works at the expense of improvers."' Relative to patent law,
then, copyright law provides less encouragement for significant improve-
ments, and gives more power to original copyright owners." 2
3. Radical Improvers.-The final case involves improvers who have

made a major contribution to social value, for example a work in which the
new material predominates over infringing material. In this situation,
patent law offers the possibility of complete immunity from infringement
under the reverse doctrine of equivalents. Copyright law has no corres-
ponding doctrine. Radical improvers still infringe the original copyright
so long as they have copied any substantial portion of the original work.
Thus, the maker of a major motion picture relying on only a small portion
of a play," the publisher of a news story which includes three hundred
words quoted from a forthcoming biography,"' and the composer who
writes a song which is similar in only a few notes to a previously published
song" are all subject to suit for copyright infringement notwithstanding
entered the public domain in the 1970s because the copyright owner failed to renew it. The movie was
shown freely by numerous broadcasters until the 1990s, when Republic Films bought the rights to the
short story on which the film was based. Armed with the copyright in the short story, they were able
to "lock up" the right to show the film, even though it was nominally in the public domain. See, e.g.,
James Bates, A Not-So-Wonderful Copyright Issue, L.A. TIMES, Mar. 26, 1996, at D6; 'Life' to Air
Once This Season, PALM BEACH PosT, Dec. 4, 1994, at TV POST 2, available in LEXIS, Regnws
Library, Arcnws File.
161. For more discussion on this point, see infra Part V.
162. For criticism of this result, see Goldstein, supra note 37, at 244; Gordon, supra note 154,
at 1019; Nadan, supra note 146, at 1639-40, 1651-54.
The rule that derivative copyrights must be authorized to be valid creates other problems as well.
Consider the situation of a translator who translates a copyrighted work without permission. Under
the legislative history to § 103, there seems no question that the translator loses her copyright in the
translation because it was unauthorized. The creative effort that went into the translation falls into the
public domain. But suppose that aftershe createsthe translation,the translator obtains authorization
from the underlying copyright owner. One might argue that the translation is now lawful, entitling the
translator to retroactively claim copyright in it. But that would take material out of the public domain
by private agreement, which seems an odd result.
At least one decision offers protection to the creators of derivative works, though that protection
is razor thin. In Lewis Galoob Toys, Inc. v. Aintendo ofAmerica, Inc., 964 F.2d 965 (9th Cir. 1992),
the Ninth Circuit held that individual consumers of Nintendo game cartridges who used the plaintiff's
device to speed up or alter the parameters of their Nintendo games were not guilty of copyright
infringement, both because the temporarily altered games did not qualify as derivative works and
because the private home use of the Galoob device was a fair use. Id. at 965. However, this decision
cannot be read to offer solace to the creators of derivative works in more permanent form who wish
to market those works.
163. See Sheldon v. Metro-Goldwyn Pictures Corp., 309 U.S. 390 (1940).
164. See Harper& Row Publishers, Inc. v. Nation Enters., 471 U.S. 539 (1985).
165. See Arnstein v. Porter, 154 F.2d 464 (2d Cir. 1946). But see id. at 475 (Clark, J.,
dissenting) (arguing that the overall musical impression, rather than the comparative technical analysis
of the particular notes or patterns of the accused work, should determine a finding of plagiarism).
the value of their contributions." Infringement under the derivative

works right may even extend to complementary goods that do not directly
copy any of the underlying work, as in Addison-Wesley Publishing Co. v.
Brown, 67 where the court held that a copyrighted physics textbook which
included problems for students was infringed by the publication of a
manual containing the solutions to those problems.16 A particularly
dramatic example is Woods v. Universal City Studios, Inc.," 6 in which
the owner of a copyrighted drawing of a chair in a room successfully
enjoined Universal from continuing to show or distribute the movie 12
Monkeys in the middle of its theater run because the movie featured a scene
involving a very similar chair."7
To be sure, copyright's fair use doctrine does offer some protection
to improvers in certain circumstances. The fair use doctrine provides a
defense to some charges of infringement on a case-by-case basis, generally
where there is a public interest served by the copying or where the copying
is unlikely to have a significant market impact." While the fair use doc-
trine may serve to protect radical improvers in some cases, it is not
expressly designed to do so. Fair use may be invoked to protect pure
copiers-those who do not improve at all-as well as to prefer radical
improvers." n It does not play a role akin to the reverse doctrine of
equivalents in patent law, for several reasons.
First, a critical factor in fair use analysis is the market impact of the
use on the copyright owner. The Supreme Court has described this impact
166. See 1 GOLDSTMIN, supra note 120, § 5.3, at 5:79 (noting that derivative rights may extend
to derivations "that will sometimes contain only the faintest trace of the underlying work"). To be
sure, the value contributed by the improver may be a disputed issue in any given case. See Harper&
Row, 471 U.S. at562 (characterizing The Nation's review as relying for its value largely on the under-
lying work). But the value of the improvement does not affect the outcome of the case. See infra note
182.
167. 223 F. Supp. 219 (E.D.N.Y. 1963).
1,68. Id. at 220. Goldstein describes this case as residing at the "speculative outer edges of the
derivative right." 1 GOLDSTEIN, supra note 120, § 5.3, at 5:80.
169. 920 F. Supp. 62 (S.D.N.Y. 1996).
170. Id. at 63-65. To be fair, the chair at issue appears to have been extremely distinctive, and
its surroundings were copied as well. Id.
171. Section 107 provides a list of four nonexclusive factors to consider in determining whether
a particular use is fair. 17 U.S.C. § 107 (1994). Those factors are the purpose and character of the
use, the nature of the copyrighted work, the amount of the copyrighted work taken, and the potential
market impact on the copyright owner. It is clear that fair use protects the creators of derivative works
as well as those who make more traditional copies. See, e.g., Campbell v. Acuff-Rose Music, Inc.,
510 U.S. 569, 579-80 (1994) (holding that "parody, like other comment or criticism, may claim fair
use under § 107"); Lewis Galoob Toys, Inc. v. Nintendo of Am., Inc., 964 F.2d 965, 969-72 (9th Cir.
1992) (holding that the use by an individual Nintendo game consumer of a device that alters the charac-
teristics of the games was a fair use).
172. See Sony Corp. v. Universal City Studios, Inc., 464 U.S. 417, 447-55 (1984) (holding that
the creation of verbatim copies of movies for private use was a fair use, since the movies were broad-
cast for free over a television network, and the plaintiffs therefore suffered no economic harm).
as "undoubtedly the single most important element of fair use,"" 7 and

Melville Nimmer has gone so far as to say that it is determinative. 174
The Supreme Court has indicated that "[flair use, when properly applied,
is limited to copying by others which does not materially impair the mar-
ketability of the work which is copied." 7 5 And the courts have further
made it clear that this market impact includes not just the potential for lost
sales of the original work, but also potential losses in the markets for deri-
vative works' 76 and licensing fees.'" This emphasis on market harm
to the original copyright owner seems to preclude the possibility that a
radical improvement which competes with the original work, or even with
actual or potential licensed derivatives of that work, can nonetheless be a
fair use. For example, in Triad Systems Corp. v. Southeastern Express
Co.," 5 the Ninth Circuit held that the copyright in software components
of an auto parts inventory system could be used to prevent the defendants
from servicing that system, since servicing necessarily required the
defendants to make RAM copies of the software. The court rejected
Southeastern's claim of fair use, reasoning that "market harm" to the
copyright owner rightfully included loss of revenues it received for ser-
vicing the computer system.' 79 The fact that sales may be attributable to
173. Harper & Row Publishers, Inc. v. Nation Enters., 471 U.S. 539, 566 (1985); see also
Campbell, 510 U.S. at 590 (observing that a defendant claiming fair use "would have difficulty
carrying the burden of demonstrating fair use without favorable evidence about relevant markets");
Wendy J. Gordon, Fair Use as Market Failure:A Structuraland Economic Analysis of the Betamar
Case and Its Predecessors, 82 COLUM. L. REV. 1600, 1615 (1982) (explaining that "one of the
necessary preconditions for premising fair use on economic grounds is that market failure must be
present").
174. See I NIMMER& NIMMER, supra note 9, § 1.10[D], at 1-92.
175. Harper & Row, 471 U.S. at 566-67 (quoting 1 NIMMER & NIMMER, supra note 9,
§ 1.10[DI).
176. See Campbell, 510 U.S. at 592-93.
177. See American Geophysical Union v. Texaco Inc., 60 F.3d 913, 929-31 (2d Cir. 1994). The
Sixth Circuit had taken a contrary position in Princeton University Press v. Michigan Document
Services, Inc., 74 F.3d 1512(6th Cir. 1996), but the opinion was withdrawn for en bane consideration,
74 F.3d 1528 (6th Cir. 1996). The en banc court then issued an opinion that followed American
Geophysical. See Princeton Univ. Press v. Michigan DocumentServs., Inc., 99 F.3d 1381, 1387-88,
1391 (6th Cir. 1996) (en banc), petitionfor cert.filed, 65 U.S.L.W. 2324 (U.S. Jan. 31, 1997) (No.
96-1219).
178. 64 F.3d 1330, 1336 (9th Cir. 1995).
179. Id. at 1336-37. It is reasonableto disagreewith the court's conclusionthata computeropera-
ting system copyright extends to the service market for that system, allowing the copyright owner to
capture the entire market for such service. See generally David McGowan, Regulating Competition
in the Information Age: Computer Software as an Essential Facility Under the Sherman Act, 18
HASTINGS COMM. & ENT. L.J. 771, 808-16 (1996) (discussing the extension of copyright law as a bar-
rier to competition). Triad's conclusion is at odds with the Ninth Circuit's earlier holding in Sega
EnterprisesLtd. v. Accolade, Inc., 977 F.2d 1510, 1527-28 (9th Cir. 1992), which allowed a com-
petitor to copy Sega computer cartridge codes in order to design video game cartridges which were
compatible with Sega's video game consoles. The Triad decision may also be inconsistent with the
antitrust laws. See Eastman Kodak Co. v. Image Technical Servs., Inc., 504 U.S. 451, 469 (1992)
the improvement itself will not help the improver if those sales are taken
away from the original copyright owner, at least under the rationale of
cases like Triad."°
Second, the focus of the fair use analysis is on the material taken from
the original copyright owner, not on the material added by the improver.
Numerous courts have held that a defendant cannot "excuse the wrong by
showing how much of his work he did not pirate."' In fair use, the
focus of the third factor-amount copied-has primarily been on the portion
of the original copyrighted work taken, not on the percentage of the
infringing work that contains copyrighted material." The one notable
exception to this rule, the Supreme Court's decision in Harper & Row
Publishers, Inc. v. Nation Enterprises,"8 uses the percentage of the
defendant's work that consists of copies to defeat rather than to support a
fair use claim."8 4 Thus, the contributions made by an infringer are not
directly considered in this factor of the fair use analysis.
Finally, the traditional focus of fair use doctrine on the purpose of the
use is unlikely to benefit radical improvers, except perhaps incidentally.
Traditional applications of this first factor in fair use analysis have tried to
distinguish between uses that served some public (generally nonprofit)
interest, such as education, research, or news reporting, and those that
were commercially motivated."a The Supreme Court has even gone so
(finding an antitrust violation for tying the sale of equipment parts to an agreement not to purchase ser-
vices from a competitor). For an argument that RAM should almost always qualify as fair use, see
Chad G. Asarch, Note, Is Turn About Fair Play? Copyright Law and the Fair Use of Computer
Sofiware Loaded into RAM, 95 MIcH. L. REv. 654, 660-61 (1996).
180. This situation should be distinguished from cases where the effect of a defendant's work is
to suppress demand for the copyrighted work on its merits. For example, a critical book review may
reduce sales of the book it criticizes, but that is not the sort of market loss relevant to the fourth factor
of fair use doctrine. See Campbell, 510 U.S. at 592 (noting that there is no observable derivative mar-
ket for criticism); Fisher v. Dees, 794 F.2d 432, 438 (9th Cir. 1986) (distinguishing between biting
criticism which suppresses demand for the original and copyright infringement which replaces it).
181. Sheldon v. Metro-Goldwyn Pictures Corp., 81 F.2d 49, 56 (2d Cir. 1936); see also Harper
& Row Publishers, Inc. v. Nation Enters., 471 U.S. 539, 565 (1985).
182. See 17 U.S.C. § 107(3) (1994) (referring to the amount of the copyrighted work taken);
Harper& Row, 471 U.S. at 565 ("[A] taking may not be excused merely because it is insubstantial
with respect to the infringingwork." (emphasis in original)); see also Meeropol v. Nizer, 560 F.2d
1061, 1069 (2d Cir. 1977) (stating that "the amount and substantiality of the work used" is a factor
in determining whether fair use applies); Roy Export Co. v. CBS, Inc., 503 F. Supp. 1137, 1145
(S.D.N.Y. 1980) (finding that even where the portion of the work used is quantitatively small, it may
still be quantitatively substantial, thus militating against a finding of fair use), aff'd, 672 F.2d 1095 (2d
Cir. 1982); Woods v. Universal City Studios, Inc., 920 F. Supp. 62, 65 (S.D.N.Y. 1996) (determining
that the relevant inquiry is not whether an infringement is de minimis in the infringing work, but
whether it is de minimis in the infringed work).
183. 471 U.S. 539 (1985).
184. Id. at 565-66. In that case, the defendant's story on the Ford memoirs took only a tiny por-
tion (less than 1%) of the plaintiff's book, but the excerpts taken represented a more significant percen-
tage of the defendant's story (around 13 %). Id.
185. See, e.g., American Geophysical Union v. Texaco Inc., 60 F.3d 913,921-22 (2d Cir. 1994)
(characterizing the photocopying of copyrighted articles by Texaco research scientists as an
far as to create a presumption that commercially motivated uses are

unfair,"8 6 though it backed off from that presumption in its most recent
treatment of fair use."s Reliance on the commercial nature of a use
would likely doom many fair use claims by radical improvers, since
improvers are more likely to be motivated by commercial purposes than
many classes of users, such as teachers and reporters.
The first factor does offer some hope for radical improvers, however.
Courts and commentators have focused increasing attention in recent years
on the "transformative" character of a use in evaluating the first factor.' 8
The Supreme Court spoke effusively of transformative works in its most
recent fair use opinion:
The central purpose of this [fair use] investigation is to see, in Justice
Story's words, whether the new work merely "supersede[s] the
objects" of the original creation.., or instead adds something new,
with a further purpose or different character, altering the first with
new expression, meaning, or message; it asks, in other words,
whether and to what extent the new work is "transformative."...
Although such transformative use is not absolutely necessary for a
finding of fair use ...the goal of copyright, to promote science and
"intermediate use" indirectly benefitting Texaco's commercial performance); Sega Enters. Ltd. v.
Accolade, Inc., 977 F.2d 1510, 1522 (9th Cir. 1992) (deciding that copying for commercial purposes
favors the copyright owner in fair use analysis); Basic Books, Inc. v. Kinko's Graphics Corp., 758 F.
Supp. 1522 (S.D.N.Y. 1991) (holding that the reproduction of copyrighted materials by a "copy shop"
for sale to university students was commercially motivated and therefore unfair). This analysis of the
first factor finds support in § 107, which requires the courts to determine "the purpose and character
of the use, including whether such use is of a commercial nature or is for nonprofit educational
purposes.' 17 U.S.C. § 107(1) (1994).
186. See Sony Corp. v. Universal City Studios, Inc., 464 U.S. 417, 449 (1984) ("If the Betamax
were used to make copies for a commercial or profit-making purpose, such use would presumptively
be unfair.").
187. See Campbell v. Acuff-Rose Music, Inc., 510 U.S. 569, 584 (1994) (finding that the Court
of Appeals erred by "giving virtually dispositive weight to the commercial nature' of the use under
Sony). The Campbell Court held that "[t]he language of the statute makes clear that the commercial
or nonprofit educational purpose of a work is only one element of the first factor enquiry into its
purpose and character.' Id. For academic criticism of the Court's reliance on the commercial nature
of a use, see Robert A. Kreiss, Accessibility and Commercialization in Copyright Theory, 43 UCLA
L. REv. 1, 62-63 (1995) (theorizing that because the aim of copyright is to allow competitors to learn
and use the ideas of others, the commercial purpose of a competitor "would seem to be either a strike
in favor of fair use or irrelevant to the fair use analysis"); William F. Patty & Shira Perlmutter, Fair
Use Misconstrued:Profit, Presumptions,and Parody, 11 CARDOZo ARTS & ENT. L.J. 667, 667-708
(1993) (arguing that the artificial distinctions between commercial and noncommercial uses should be
replaced with a sensitive balancing of interests test).
188. The origin of the concept of transformative versus superseding uses is generally attributed
to Justice Story's statement in Folsom v. Marsh, 9 F. Cas. 342 (C.C.D. Mass. 1841) (No. 4901), that,
in approaching a question of fair use, one must "look to ... the degree in which the use may prejudice
the sale, or diminish the profits, or supersede the objects, of the original work." Id. at 348. See
generally Leval, supra note 11, at 1111-16 (elaborating on the concept of transformative use and its
importance in evaluating a secondary use's purpose and character).
the arts, is generally furthered by the creation of transformative

works. Such works thus lie at the heart of the fair use doctrine's
guarantee of breathing space within the confines of copyright...
and the more transformative the new work, the less will be the
significance of other factors,1 89like commercialism, that may weigh
against a finding of fair use.
Taken seriously, these words might spark the development of something
akin to the reverse doctrine of equivalents-protection for radical improvers
against any claim of copyright infringement."ro But the evolution of the
transformative use doctrine in this regard will depend critically on how the
transformative nature of the use is weighed against market harm to the
copyright owner. Expansive definitions of market harm, like those adopted
in Triad and American Geophysical Union, may render the transformative
use analysis a mere secondary consideration, unlikely to protect radical
improvers who use their improvements to compete with the original
copyright owner.19' This is particularly true if market harm is used to
trump a finding of transformative use, giving the fourth factor
primacy.192
189. Campbell, 510 U.S. at 579 (citations omitted) (quoting Folsom, 9 F. Cas. at 348).
190. But see Roxana Badin, Comment, An Appropriate(d) Place in Transformative Value:
AppropriationArt's Exclusionfrom Campbell v. Acuff-Rose Music, Inc., 60 BROOK. L. REV. 1653,
1653-54 (1995). Badin reads Campbell as having carved out only a narrow exception for commercial
parodies that will hurt the fair use case for other forms of appropriative art. Id. She goes on to sug-
gest that an expansive reading of Campbell is more desirable, however. Id. at 1654.
Commentators have debated the wisdom of the transformative use doctrine. Those endorsing
some form of transformative use analysis include Jay Dratler, Jr., Distilling the Witches' Brew of Fair
Use in Copyright Law, 43 U. MIAMi L. REV. 233, 291-93 (1988); William W. Fisher I,
Reconstructingthe Fair Use Doctrine, 101 HARV. L. REv. 1661, 1730 (1988); and Leval, supranote
11, at 1111. Those opposed to its use, or at least to certain aspects of the way it has been applied,
include Laura G. Lape, Transforming Fair Use: The Productive Use Factorin Fair Use Doctrine, 58
ALB. L. REV. 677,712-22 (1995); L. Ray Patterson, FreeSpeech, Copyright, andFairUse, 40 VAND.
L. REV. 1, 45-46 n.149 (1987); and Lloyd L. Weinreb, Fair'sFair: A Comment on the Fair Use
Doctrine, 103 HARV. L. REV. 1137, 1143-44 (1990).
191. Cf. Douglas J. Ellis, Comment, The Right ofPublicity andthe FirstAmendment: A Comment
on Why Celebrity Parodies Are Fair Game for Fair Use, 64 U. CIN. L. REv. 575, 611 (1996)
(suggesting that uncertainty regarding the theoretical basis for copyright means that "transformative
use will have a checkered and inconsistent application"). For discussion of the role of theoretical
approaches to copyright in treating radical improvers, see infra Part IV.
192. See Pamela Samuelson, Modifying Copyrighted Software: Adjusting Copyright Doctrineto
Accommodate a Technology, 28 JURIMETRICS J. 179, 195 (1988) (noting that fair use is unlikely to pro-
tect those who modify software, because of the Court's emphasis on harm to both original and potential
derivative markets); Stephen B. Thau, Copyright, Privacy, and FairUse, 24 HOFSTRA L. REv. 179,
189-90 (1995) ("While the Court often purports to use fair use 'to avoid rigid application of the
copyright statute when, on occasion, it would stifle the very creativity which that law is designed to
foster,' in actuality the analysis always turns on the economic harm suffered by the plaintiff." (quoting
Stewart v. Abend, 495 U.S. 207, 236-38 (1990) (quoting Iowa State Univ. Research Found., Inc. v.
American Broad. Cos., 621 F.2d 57, 60 (2d Cit. 1980)))); see also Lape, supra note 190, at 715-16
(arguing against giving primacy to the market harm factor).
The Supreme Court's statement in Harper& Row that "to negate fair
use one need only show that if the challenged use 'should become
widespread, it would adversely affect the potential market for the
copyrighted work'" 1" certainly suggests that market harm is already the
de facto test for fair use. Even Campbell itself may represent only an
illusory victory for the transformative use doctrine, since the Court in that
case indicated that the defendants bore the burden of proving the absence
of harm to the market for derivative works on remand, notwithstanding the
transformative nature of their use. 1" Further, the "commercial use"
prong of factor one is not dead, and at least one court has used it to negate
the impact of transformative use-in effect holding that transformative but
commercial uses cancel each other out, leaving the first factor neutral.' 95
In either case, fair use protection appears effectively to be foreclosed to
transformative works designed to compete with the original or its derivative
progeny."'
IV. Is Differential Treatment of Improvement Justified?

Comparing the treatment of improvers under patent and copyright law
leads to a rather surprising result: copyright law is significantly more hos-
tile to improvements than is patent law. What is surprising is not so much
that the rules differ," 9 but the way in which they differ. Copyright is
traditionally thought to afford weaker, not stronger, protection than patent
law, in part to compensate for the fact that copyrights are so much easier
to obtain than patents and last so much longer.19 But in the context of
improvements, the opposite result obtains. In this Part, I consider several
possible reasons that might be offered to explain this differential treatment.
193. Harper& Row Publishers, Inc. v. Nation Enters., 471 U.S. 539, 568 (1985) (emphasis added
by the Harper & Row Court) (quoting Sony Corp. v. Universal City Studios, Inc., 464 U.S. 417, 451
(1984)).
194. Campbell, 510 U.S. at 590-94. Justice Kennedy's concurring opinion would limit the deci-
sion even further, giving it virtually no application outside the parody context. Id. at 596-600
(Kennedy, J., concurring).
195. See College Entrance Examination Bd. v. Pataki, 889 F. Supp. 554, 568 (N.D.N.Y.),
modified, 893 F. Supp. 152 (N.D.N.Y. 1995). The neutrality of the first factor in that case allowed
the copyright owner to prevail by showing market harm. Id. at 575.
196. Of course, where a use both is transformative and does not harm the market for the original
work (or licensed derivatives thereof), it will likely be found to be fair. This was the court's
conclusion, for example, in New York Tunes Co. v. Roxbury Data Interface, Inc., 434 F. Supp. 217,
223-25 (D.N.J. 1977). But many of the most important transformative uses of a work will compete
with the work, or at least will compete in markets which the original copyright owner could foreseeably
enter. In these cases, fair use will be ineffective to protect such transformations.
197. But see John Shepard Wiley Jr., Copyrightatthe School ofPatent, 58 U. CHI. L. REV. 119,
180-81 (1991) (arguing that because patent law and copyright law share the same purposes, their doc-
trines and procedures should also be consistent).
Because I find those reasons ultimately unpersuasive, I proceed in Part V

to compare the two rules on their merits. 1 9
My goal in this Part is not tb argue that patent and copyright should
be coextensive in all respects. Obviously they are not. Rather, this section
focuses on the specific justifications that might be offered for treating
improvement differently in copyright law than in patent law.' One can
imagine several different types of arguments along these lines. First, one
might argue that copyright law serves a different purpose than patent law-
notably, the protection of the moral rights of authors. Refusing to allow
others to improve on a copyrighted work without the copyright owner's
permission grants the copyright owner a sort of moral right against altera-
tion of her work. I shall refer to this as the "moral rights" argument.
Second, one might explain the difference in treatment by reference to the
different sorts of works that copyrights and patents protect. Perhaps there
is simply more need for improvement in patent law, and hence more need
to encourage it. I shall call this the "different works" argument. Third,
one might argue that copyright's broad prohibitions against unlicensed
improvement are justified by the very fact that copyright provides a
narrower scope of protection than patent law-that the one makes up for
the other. I shall call this the "balance of power" argument. Finally, one
might make an argument based on the relative market power conferred by
patents and copyrights. Let us consider each argument in turn.
199. Thus, arguments that seek to justify the copyright rule, but would also justify the same rule
in patent cases, are treated in Part V rather than this Part, because they are not arguments in favor of
differential treatment.
200. At the outset, one can quite easily reject a potential argument from history-an argument that
says, in effect, that the differences are justified because they have always been there. This argument
fails because copyright's strong power over derivative works is in fact a relatively recent innovation.
Early cases not only did not give the original creator strong power over exploitation of derivative
works, but in fact held that alteration of a copyrighted work excused infringement. See, e.g., Stowe
v. Thomas, 23 F. Cas. 201 (C.C.E.D. Pa. 1853) (No. 13514); Folsom v. Marsh, 9 F. Cas. 342
(C.C.D. Mass. 1841) (No. 4901); Burnett v. Chetwood, 35 Eng. Rep. 1008 (Ch. 1817); Gyles v.
Wilcox, 26 Eng. Rep. 489 (Ch. 1740); Ralph S. Brown, The Widening Gyre: Are Derivative Works
Getting Out of Hand?, 3 CARDozo ARTS & ENT. LJ. 1, 2 (1984); Goldstein, supra note 37, at 211-
14; Gordon, supra note 154, at 1018. It is also worth noting the "historical kinship" between patent
and copyright law remarked upon by the Supreme Court in Sony Corp. v. UniversalCity Studios, Inc.,
464 U.S. 417, 439 (1984).
The history of the Semiconductor Chip Protection Act (SCPA), 17 U.S.C. §§ 901-914 (1994),
provides yet another reason to reject this interpretation. The SCPA, which is included in the copyright
statute and is modeled on the Copyright Act in significant respects, nonetheless has a very different rule
regarding improvements. Improvers are not held liable under the SCPA if they (1) engaged in reverse
engineering and (2) contributed their own innovation resulting in an "original" product. The effect of
this rule is to make it harder for an SCPA plaintiff to show infringement if the defendant has engaged
in any significant improvementto the original chip. See Brooktree Corp. v. Advanced Micro Devices,
Inc., 977 F.2d 1555 (Fed. Cir. 1992) (affirming a judgment in favor of a plaintiff claiming patent
infringement of semiconductor mask registrations despite the defendant's claims of reverse engineering
and innovation).
A. Moral Rights Justifications

The moral rights argument has some intuitive appeal. Copyright theo-
ry has a strong moral rights tradition, particularly outside the United
States, that is almost entirely lacking in patent theory. 1 Thus, a moral
rights justification would explain why copyright but not patent law was af-
fected. Further, copyright's restrictions on improvement might be thought
to further important moral rights. Signatories to the Berne Convention are
in theory required to recognize certain minimal moral rights protections for
authors, including the right to attribution of authorship and the right to
object to modifications to the work.' The failure of the United States
to create such moral rights protection despite its accession to Berne has
caused considerable controversy.' Precluding improvers from creating
derivative works without the permission of the original copyright owner
may have the effect of creating a "right to object to modifications" in the
original creator.' For some, this may be sufficient to justify the rule.
201. Even what Samuel Oddi calls the "reward theory" of patent law is essentially incentive-based:
inventors must be rewarded in order to (a) encourage more inventions, or (b) prevent "free-riding."
See Oddi, supra note 13, at 275-77 (discussing the reward theory). A rare example of a natural law
justification in patent law is the French patent statute of 1791, which was premised on the belief that
.every novel idea whose realization or development can become useful to society belongs primarily to
him who conceived it, and that it would be a violation of the rights of man in their very essence if an
industrial invention were not regarded as the property of its creator." STAFF OF SENATE SUBCOMM.
ON PATENTS, TRADEMARKS AND COPYRIGHTS, 85TH CONG., AN ECONOMIC REVIEW OF THE PATENT
SYSTEM, STUDY No. 15, at 22 (Comm. Print 1958) (Fritz Machlup, author) (quoting the French patent
statute). Even this example, while it suggests a natural law basis for patents, does not justify the
equivalent of the moral rights of attribution and integrity found in the copyright context.
202. Berne Convention for the Protection of Literary and Artistic Works, Sept. 9, 1886, art.
S. TREATY Doc. No. 99-27, at 5 (1986), 828 U.N.T.S. 221,235.
203. See, e.g., Edward J. Damich, The Right of Personality: A Common-Law Basis for the
Protection ofthe MoralRightsofAuthors, 23 GA. L. REv. 1 (1988) (arguing that while U.S. law does
not protect moral rights to the extent required by the Berne Convention, expansion of existing common-
law precepts could achieve such protection); Roberta Rosenthal Kwall, Copyrightandthe MoralRight:
Is an American Marriage Possible?, 38 VAND. L. REV. 1 (1985) (urging an amendment of the
Copyright Act to implement the moral rights of authors). But cf. 3 GOLDSTEIN, supra note 120,
§ 15.23, at 15:181 n.10 ("Congress' position on moral rights reflects neutrality, not antipathy.").
While the United States did adopt limited moral rights protection in the Visual Artists Rights Act of
1990, Pub. L. No. 101-650, tit. VI, 104 Stat. 5089, 5128-33 (codified in scattered sections of 17
U.S.C., with the specific rights of attribution and integrity in § 106A), that section is by no means a
general moral rights statute. See 3 GOLDSTEIN, supra note 120, § 15.23, at 15:179 (finding no sys-
"
tematic protection of moral rights in either federal or state law, [alpart from the closely circumscribed
rights" in § 106A). Section 106A is limited to works of visual art, does not give protection to the artist
if the work is made for hire, and is subject to numerous exceptions. 17 U.S.C. § 106A (1994).
Furthermore, the rights granted in § 106A are limited to the life of the author, and may be waived by
the author in writing. Id.
204. Some commentators have taken this position. See, e.g., 2 GoLDSTEIN, supra note 120,
§ 5.3, at 5:80 & n.12 ("The broad scope of derivative rights enables them to secure reputational
interests as well as strictly economic interests by protecting authors against disfigurement of their works
.... The derivative right is in this respect akin to the moral right against distortion of an author's
work."); cf. I.H. Reichman, Electronic Information Tools-The Outer Edge of World Intellectual
PropertyLaw, 17 U. DAYTON L. REv. 797, 813 (1992) ("[T]he recent expansion of derivative work
There are a number of problems with such a justification, however.

First, even if the rule against unauthorized improvements does create a de
facto moral rights rule in the United States, it is not at all clear that such
a rule is a good thing.' 5 The economic and political critiques of moral
rights are powerful ones, and do not stem solely from the Chicago-based
view that all rights should presumptively be alienable.' Establishing a
moral right imposes a real economic and political cost on society. In the
best of circumstances, those who would use a copyrighted work must now
deal with two sets of owners-the economic copyright owner and the
holder of the moral right. Further, since moral rights cannot be disclaimed
in gross, dealing with the author is not a simple matter of signing a
contract, but involves a continuing relationship in which each new use is
at the mercy of the author.' This adds to the difficulty and cost of
rights in United States law has helped to align the 'copyright' countries and the 'authors' rights'
countries without necessarily sacrificing the utilitarian ethos to which the former subscribe."). For an
argument that trademark law might have a similar effect, providing indirect moral rights protection,
see Mark A. Lemley, Rights ofAttribution and Integrity in Online Communications, 1995 3. ONUNE
L. art. 2, 14-16 (June 1995) <http://www.wm.edu/law/publicationsljol/lemley.html>.
205. This debate is long-standing and complex; I do not propose to resolve it here, or indeed to
do more than touch briefly on some of the important issues at stake.
206. Certainly adherents to the Chicago view would seem unlikely to support a strong version of
moral rights, since they are by their nature rights personal to the author that cannot be transferred. See
generally Neil Netanel, Alienability Restrictions and the Enhancementof Author Autonomy in United
States and ContinentalCopyrightLaw, 12 CARDOZO ARTS& ENT. LJ.1 (1994) (explaining that moral
rights in most continental countries are not alienable, though they can be waived in particular
circumstances). But others have criticized moral rights claims on political and artistic grounds. Beyer,
for example, argues that the moral rights position "at its root is actually one of cultural conservatism,
inhibited expression, and unnecessary deference to creators' intentions," as it gives original artists the
power to restrict subsequent interpretations and critiques of their work. Lawrence Adam Beyer,
Intentionalism, Art, and the Suppressionof Innovation:Film Colorizationand the Philosophy of Moral
Rights, 82 Nw. U. L. REV. 1011, 1018 (1988); see also Thomas J.Davis, Jr., Fine Art and Moral
Rights: The Immoral Triumph of Emotionalism, 17 HoFsTRi L. Rfv. 317, 358-59, 363 (1989)
(contending that the public interest is short-changed by allowing authors to assert claims of moral
rights); Geri J.Yonover, The PrecariousBalance:MoralRights, Parody andFair Use, 14 CARDOZO
ARTS & ENT. L.J. 79 (1996); Note, An Author's Artistic Reputation Under the Copyright Act of 1976,
92 HARV. L. REV. 1490, 1494 (1979) (both arguing that the law must balance the rights of original
artists against adaptors and parodists).
207. Consider the few cases brought to date under state and federal moral rights statutes. Most
have involved efforts to move or take down site-specific sculpture years after it was installed in a
building, and frequently when ownership of the building had changed since the art was installed. See,
e.g., Carter v. Helmsley-Spear, Inc., 71 F.3d 77 (2d Cir. 1995); Serra v. United States Gen. Servs.
Admin., 667 F. Supp. 1042 ($.D.N.Y. 1987), affid, 847 F.2d 1045 (2d Cir. 1988); see also Eric M.
Brooks, Comment, "Tilted" Justice: Site-Specific Art and Moral Rights After U.S. Adherence to the
Berne Convention, 77 CAL. L. REV. 1431 (1989) (examining the development and protection of the
moral rights of site-specific visual artists in the United States).
While compelling building owners to display works of art might seem to raise constitutional
problems of compelled speech under cases like Wooley v. Maynard,430 U.S. 705 (1977), and West
Virginia Board ofEducation v. Barnette, 319 U.S. 624 (1943), courts have generally dismissed First
Amendment claims in the copyright context rather summarily. See, e.g., Harper & Row Publishers,
Inc. v. Nation Enters., 471 U.S. 539, 555-60 (1985); Los Angeles News Serv. v. Tullo, 973 F.2d 791,
making any productive use of a copyrighted work? Finally, many of

the best productive uses of a work may be foreclosed entirely under a
moral rights regime, since most authors will consider parodies of their
works (and even many satires based on their works)' to be objec-
tionable "mutilation." 20 Even if one values the moral rights of artists,
it is disturbing that there is no mechanism in the law for measuring that
value or balancing it against the value of the improvements that will be
lost.2 '
Second, even if one is persuaded by the case for moral rights, the rule
on derivative works does not seem a particularly good way of establishing
such rights. For one thing, it seems odd that a legal provision that osten-
sibly exists for the benefit of creators (artists or authors, for example)
should confer rights instead on the owners of intellectual property rights.
As anyone who has ever published a book, a screenplay, or a song can
attest, authorship and ownership are not necessarily the same thing." 2
795-96 (9th Cir. 1992); Sid & Marty Krofft Television Prods., Inc. v. McDonald's Corp., 562 F.2d
1157, 1171 (9th Cir. 1977). See generally Melville B. Nimmer, Does CopyrightAbridge the First
Amendment Guaranteesof Free Speech and Press?, 17 UCLA L. REV. 1180 (1970).
208. But see Netanel, supra note 206, at 6 (suggesting that strong moral rights may be tempered
in practice by an informal social norm against asserting them too strongly).
209. On the distinction between parody and satire in copyright law, see Campbell v. Acuff-Rose
Music, Inc., 510 U.S. 569, 580-81 (1994).
210. See Beyer, supra note 206, at 1016 & n.9. Davis points to proponents of moral rights who
explicitly assert the "right to be free from excessive criticism, to publish a reply to such criticism, and
to be protected from 'a whole category of unpredictable injuries to .. .honor and reputation'" as
elements of the moral right. Davis, supra note 206, at 319 (quoting Martin A. Roeder, The Doctrie
of Moral Right: A Study in the Law of Artists, Authors and Creators, 53 HARV. L. REV. 554, 573
(1940)). Certainly the recognition of such rights would interfere directly with political, literary, and
artistic freedom by giving artists and authors control over what is said about them, whether or not it
was held to violate the First Amendment.
211. It is possible to take some steps toward such a calibration by creating a fair use exception to
the moral rights doctrine. The United States has done so. See 17 U.S.C. § 107 (1994). Nonetheless,
the discussion in Part m suggests that as currently constituted, the fair use doctrine does not effectively
distinguish radical from minor improvements. See supra subpart 11(C). Further, some cases suggest
that the failure to comply with the moral rights of the original author, for example by refusing to
attribute an author's work to her, will weigh heavily against a claim of fair use. See, e.g., Robinson
v. Random House, Inc., 877 F. Supp. 830 (S.D.N.Y. 1995). To the extent that this is true, the fair
use defense to claims of infringement on moral rights may be largely illusory.
212. According to a 1958 study, approximately 40% of all copyrights were works made for hire.
See STAFF OF SENATE SUBCOMM. ON PATENTS, TRADEMARKS & COPYRIGHTS, 86TH CONG., WORKS
MADE FOR HIRE AND ON COMMISSION, STUDY No. 13, at 139 n.49 (Comm. Print 1960) (Borge
Varmer, author). It is reasonable to assume that that percentage has increased as computer programs
havejoined the list of copyrighted works. While the copyright act defines the "author" as the employer
in such a circumstance, see 17 U.S.C. § 201(b) (1994), that is merely a convenient fiction-the actual
authors are almost invariably individual employees. Further, even works for which authorship vests
initially in an individual are frequently assigned to other individuals or to corporations. See, e.g.,
Elkin-Koren, supra note 159, at 280, 279-80 ("The vast majority of copyrighted materials that are on
the market are not owned by the actual author."); Mark A. Lemley, Romantic Authorship and the
Rhetoric ofProperty,75 TEXAs L. REv. 873, 883 (1997) (book review) (recognizing that the doctrines
By allowing the right to produce derivative works to be transferred with

the ownership of the copyright-and indeed separately licensed-the copy-
right law has created what is at base an economic rather than a moral
right. 13 Certainly, a provision designed for the benefit of authors would
have been much more effective if it gave control over alterations of a
copyrighted work directly to the authors.21 Similarly, it is indicative of
the law's concern with economic rather than moral rights that the copyright
rule offers no recognition of the moral rights of improvers, whose works
not only are considered infringing but are captured by the original copy-
right owner. 15 One might expect that a derivative works rule designed
to protect the moral rights of authors would at least afford improvers some
control over the independent expression they have created. But copyright
law does not do so-indeed, it gives the original copyright owner de facto
exclusive rights over the works created by someone else.
In short, whatever one thinks of moral rights, the derivative works
rule in the final analysis does not appear to be such a right. Rather, its
primary purpose seems to be economic-to "enable prospective copyright
owners to proportion their investment to the returns they hope to receive
not only from the market in which their work will first be published, but
from other, derivative, markets as well. " 21 It should stand or fall on its
economic merits. 1 7
B. Different Works Justifications

A second possible reason for treating improvement differently in copy-
right and patent law is that the two laws protect different kinds of works.
Traditionally, copyright protection has been thought to be reserved for
of assignment and transfer ensure that a majority of intellectual property rights end up "not in the hands
of authors or investors, but in the hands of corporate economic interests"); Robert G. Sugarman &
Joseph P. Salvo, Whose Rights? Sampling Gives Law a New Mix, NAT'L L.J., Nov. 11, 1991, at 21,
22 ("Most recording [and publishing] contracts provide that rights in a sound recording inure to the
record label....").
213. Even proponents of moral rights concede that the 1976 Copyright Act is primarily concerned
with the economic incentives of copyright owners, rather than the moral rights of authors. See Kwall,
supra note 203, at 37-3 8.
214. Cf. MARGARET JANE RADIN, REiNTERPRETING PROPERTY 12 (1993) (questioning the effec-
tiveness of the standard ideology of property-individual ownership of land-when applied broadly to
a society in which "the largest proportion of holdings are owned by institutions or entities other than
persons").
216. 2 GOLDSTEIN, supra note 120, § 5.3, at 5:79.
217. Some critics of the economic approach to copyright make a different claim-that economic
analysis is nondeterministic. See Alfred C. Yen, When Authors Won't Sell: Parody, Fair Use and
Efficiency in CopyrightLaw, 62 U. COLO. L. REv. 79, 82-83 (1991). This criticism must be taken
with a grain of salt, given that the alternatives to economic analysis generally advanced-Yen's moral
rights approach, or Lloyd Weinreb's ad hoc "fairness" analysis, see Weinreb, supra note 190, at
1137-are hardly more likely to result in a deterministic outcome.
works exhibiting artistic or literary creativity, while patent protection

applied to works of technical or industrial application." ' While both
laws serve the larger purpose of providing incentives for creation and dis-
semination of new works, perhaps the different nature of the works in-
volved means that the actual incentives needed are different. In particular,
one could argue that patent law ought to be more concerned with improve-
ment on existing works, since improvement is somehow more central to
scientific endeavor than to art. If so, one would expect the law to favor
improvement to a greater extent in patent than in copyright law.219
This explanation too is problematic. First, it grossly oversimplifies
intellectual property law to divide it into artistic works protected by copy-
right and utilitarian works protected by patent.' There have always
been areas of significant overlap between the two. The scope of copyright
encompasses works that have a predominantly utilitarian function, and
works that are entirely factual, though copyright law places some limi-
tations on the protection of such works." Copyright in graphic,
218. See J.H. Reichman, Legal HybridsBetween the Patentand CopyrightParadigms,94 COLUM.
L. REV. 2432, 2448-50 (1994) (documenting this division). As Reichman observes, this division has
never been strictly observed in practice. There have always been certain categories of protection that
crossed over or fell between the bipolar patent-copyright paradigms. See id. at 2453-500.
219. Wiley makes a somewhat different argument along the same general lines. He contends that
the procedural differences between copyright and patent stem from the fact that copyrightable works
are so much more common than patentable works. He points to the fact that while less than six million
patents havebeen issued in the history of the United States, people create millions of "original works
of authorship" every day. Wiley, supra note 197, at 182-83. On the implications of this difference
in the improvement context, see infra notes 231-33 and accompanying text.
220. John Wiley wrote in 1991 that the legal cultures of patent and copyright
divide at the wall of mathematics. High-tech inventors and their lawyers generally
perceive themselves as having little in common with the stylishly literate world of novels,
publishing, and studios, and vice versa. Artists think engineers are nerds. Inventors
think fiction and movies are fun but ultimately useless. You never see anyone wearing
both Gucci loafers and a plastic pocket protector.
Wiley, supra note 197, at 181. Ironically, 1991 is perhaps the last year in which those statements
could have been made with a straight face. If it accomplishes nothing else, the well-hyped convergence
of computers and entertainment media during the 1990s should explode the myth that invention and
creation have nothing in common. See Rochelle Cooper Dreyfuss, A Wiseguy's Approach to
Information Products:Muscling Copyrightand Patentinto a Unitary Theory of Intellectual Property,
1992 Sup. CT. REv. 195, 221-22 ("[A]dvances in production and reproduction make it increasingly
clear that the distinctions that seemed to be inherent in the terms 'patent' and 'copyright' are in fact
illusory becausethey fail to captureanything that is significant about the products of human intellect.").
Dreyfuss suggests that the differences between copyright and patent rules most likely "developed
fortuitously." Id. at 222.
221. See, e.g., Mazer v. Stein, 347 U.S. 201, 219 (1954) (mentioning that copyright law makes
the reward to the owner a secondary consideration); Hoehling v. Universal City Studios, Inc., 618 F.2d
972, 978 (2d Cir. 1980) (holding that copyright law should not be stretched to chill authors from
writing about historical issues or events). To be sure, copyright does not extend to pure collections
of facts which lack originality in the compilation, see Feist Publications, Inc. v. Rural Tel. Serv. Co.,
499 U.S. 340 (1991), or to the separable utilitarian aspects of works that are both useful and artistic,
see Carol Barnhart Inc. v. Economy Cover Corp., 773 F.2d 411, 419 (2d Cir. 1985). See also 17
1036 Texas Law Review [[Vol. 75:989
sculptural, and architectural works in particular may overlap with utility

patent protection for those same works.' The overlap between copy-
right and design patent protection is even more striking.' And the
development of computer software threatens to erase the line between
patentable and copyrightable subject matter entirely, as both copyright and
patent are extended to cover precisely the same sort of material.' A
general statement that improvement is "more important" in patent than in
copyright is meaningless in the context of computer software, where both
doctrines endeavor to create essentially the same incentives. Indeed,
the very fact of overlap between patent and copyright protection may make
having different rules inefficient, since it provides opportunities for intel-
lectual property owners to exploit the system, obtaining protection in the
copyright sphere that nullifies the rights of improvers in the patent sphere.
Further, even if a subject matter distinction can be drawn, it is not at
all clear that improvement should be considered more important in techni-
cal fields than in creative ones. The argument for such a distinction pre-
sumably rests on the idea that scientific endeavor is more nearly linear than
art, and the importance of building on prior work is therefore corres-
pondingly greater. For example, if an original invention is highly con-
strained by the laws of physics, subsequent developers cannot simply
ignore those laws-they must work within the parameters of the physical
laws, and hence may be forced to build on the original inventor's work.
Artists and writers presumably face no similar constraint (or at least a less
restrictive one), since it is possible simply to create a new work "from
scratch" in a completely different genre.'
U.S.C. § 101 (1994). But the very difficulty that courts have had in drawing these lines illustrates one
problem with basing doctrinal differences on them.
222. See Reichman, supra note 218, at 2455-59,2488-90.
223. Id.
224. For example, the Patent and Trademark Office's recently adopted Guidelines for the
Examination of Computer-Related Inventions, 61 Fed. Reg. 7478, 7481 (1996), work hard to dis-
tinguish patentable from copyrightable subject matter, but ultimately fail. Neither an artistic-utilitarian
distinction nor a writing-invention distinction is supportable in the context of computer code, in which
writing is functional in the most basic sense. A full discussion of this convergencebetween the doc-
trines is beyond the scope of this Article. See generally Mark A. Lemley, Convergencein the Law of
Software Copyright?, 10 HIGH TECH. L.J. 1 (1995); Oddi, supra note 118; Samuelson et al., supra
note 118.
225. As the authors of the Manifesto point out, the incentives intellectual property law seeks to
foster in the software industry are for the creation of new and useful software design features.
Samuelson et al., supranote 118, at 2318-19. These design features lie somewhere in between the tra-
ditional subject matter of patent and copyright, and possess some elements of each. Id. at 2351-53.
Thus, the problem is not simply one of disentangling the patent incentives for the production of
software inventions from the copyright incentives for the production of software creations-the two laws
serve to promote the same incentives. Cf. Samuelson, supra note 192, at 210 (suggesting that copy-
right law should be modified to adopt patent law's rules regarding improvements in the software context
because software is more like a patentable invention than it is like other copyrighted works).
226. Some scholars have challenged this assumption. Applying the now-familiar techniques of
literary deconstruction, they argue that all creations are largely a product of communal forces.
But a moment's reflection will demonstrate that this assumption is

unwarranted. There are certainly cases in patent law where several alter-
native approaches will yield roughly equal results, and therefore where
improvement on an original idea is neither the only possible path nor
indeed necessarily the best one. There may be good reason to discourage
scientists from building on an original idea in some cases, on the grounds
that a fresh start is precisely what is needed.' Similarly, in many cases
improvement will be absolutely necessary to creative endeavor. Parodies
are a good example of creative works that "improve" (in the sense of
adding to or transforming) an original work, and that could not survive
without the ability to build on the prior work. Even some nonparodic
works of fiction require a prior work on which to build; consider Tom
Stoppard's play Rosencrantz and GuildensternAre Deadp which builds
on Shakespeare's Hamlet in a particularly creative way. 9 Other exam-
ples include written works in the social sciences, where knowledge is
advanced in large part by analyzing and critiquing the work of others; and
political discourse, which takes place largely in the context of what has
come before.' In short, it is not possible to speak monolithically about
the relative importance of improvement in the patent and copyright
contexts. Absent some persuasive evidence that there is an identifiable
Dividing the stream of intellectual discourse into discrete units-each owned by, and closely associated
with, a particular author-is therefore a logically incoherent exercise subject more to the political force
of asserted authors' groups than to recognition of inherent claims of "personhood." See, e.g., JAMES
BOYLE, SHAMANS, SOFTWARE, AND SPLEENS: LAW AND THE CONSTRUCTION OF THE INFORMATION
SoCIETY (1996); THE CONSTRUCTION OF AUTHORSHIP: TEXTUAL APPROPRIATION IN LAW AND
LITERATURE (Martha Woodmansee & Peter Jaszi eds., 1994); James Boyle, A Theory of Law and
Informadon: Copyright, Spleens, Blackmail, and InsiderTrading, 80 CAL. L. REV. 1413 (1992); Peter
Jaszi, Toward a Theory of Copyright: The Metamorphoses of "Authorship," 1991 DUKE L.J. 455.
While this view is inconsistent with the idea that an author can simply create from scratch, one need
not subscribe to this radical critique of authorship in order to doubt the premise that improvement is
unimportant in the copyright context.
227. Murray Gell-Mann analogizes the situation to traveling over uncertain terrain. SeeMURRAY
GELL-MANN, THE QUARK AND THE JAGUAR: ADVENTURES IN THE SIMPLE AND THE COMPLEX 266-67
(1994). If your goal is to climb as high as possible, you will not be well served by a rule of decision
which says "always travel upwards." That rule may bring you to the top of a low hill and strand you
there, in sight of a much larger mountain. He suggests that some combination of improvement on exis-
ting ideas and the creative use of "noise" will produce the most innovative ideas. Ironically, in such
a case a broad construction of the patent system would seem most conducive to innovation, but only
because it forced subsequent work to deviate from improving on the original work into more productive
avenues. C. Kitch, supra note 13, at 267-71 (arguing for a broad "prospect" theory of patents).
228. TOM STOPPARD, ROSENCRANTZ& GUILDENSTERN ARE DEAD (Henry Popkin ed., 1967).
229. See Elkin-Koren, supra note 159, at 282-83 (stating that copyright's goal of encouraging a
broad range of expression argues for a limited right to control improvements).
230. Cf. Netanel, supra note 12, at 363-64 (asserting that copyright must underwrite and protect
deliberation and debate in the political sphere in order to promote a democratic civil society). Netanel
argues that in the copyright context, "our strong interest in expressive diversity overrides any possible
efficiency advantage in giving one person broad control over all transformative uses of an existing work
of authorship." Id. at 318.
distinction between the two, the different nature of the works involved
cannot justify different categorical rules regarding the treatment of
improvements.
Finally, let us consider John Wiley's point that copyright and patent
differ in large part because so many more things are copyrightable than
patentable."1 Certainly this is true. Wiley uses this difference to explain
why patents face strict and time consuming examination procedures, while
copyrights do not; it would simply not be feasible to examine every new
"work of authorship" for novelty, nonobviousness, and the like." Can
the same fact be used to justify the difference in rules regarding
improvement? It is hard to see how. If anything, one would expect that
the more common a right was, the less each right would constrain the
behavior of subsequent developers. The rarer a right is, and the harder it
is to acquire, the more powerful we should expect it to be. Vesting greater
control over improvement in copyright than in patent runs counter to this
expectation.23
It is certainly true that copyright and patent protect different kinds of
works, though the dichotomy is not as clear as one might initially believe.
But there is no reason to think that those differences can explain the dif-
ferential treatment of improvement between the two statutes.
C. Balance of Power Arguments

In some important respects, copyright is a weaker right than
patent.' Copyright precludes only actual reliance on the copyrighted
work, while patent prohibits even independent development of the same
ideas. 5 Copyright has a number of statutory limitations that allow users
to make copies, notably the fair use doctrine; patent law has no analogous
rules. 6 One possible justification for giving copyright owners broader
231. See Wiley, supra note 197, at 121.

232. Id. There is an element of circularity to this argument. If it were easy to protect patents
without registration, and copyrights required exhaustive examination procedures, it would certainly be
the case that the number of patents would increase and the number of copyrights would decline.
233. Of course, this argument oversimplifies the case. It is possible that copyright is weaker than
patent in other areas, so that the argument in the text is counterbalancedto some extent by other forces.
I discuss this argument in the next subpart. It is also possible to argue that copyrights should be more
powerful than patents precisely because they are more common, on the theory that expanding the con-
trol of the copyright owner will encourage licensing. Evaluation of this neoclassical argument in detail
must wait until Part V.
234. See Netanel, supra note 12, at 318 ("Copyright has traditionally been of much narrower
scope than patent. .. ").
235. See 35 U.S.C. § 271(a) (1994).
236. The closest exception, for "experimental use of a patented technology, is extremely narrow,
and does not protect uses unless the experimentation cannot further the legitimate business interests of
the potential infringer. See Roche Prods., Inc. v. Bolar Pharm. Co., 733 F.2d 858, 863 (Fed. Cir.
1984); Rebecca S. Eisenberg, Patentsand the ProgressofScience: Exclusive Rights and Experimental
control over improvements is to counterbalance these restrictions on

copyright. 7 In particular, one might think that improvers who have
actually copied from the original creator, and whose uses are not "fair"
under section 107, are somehow more culpable than at least that subset of
patent improvers who independently developed the infringing technology.
This is a difficult claim to evaluate, largely because it assumes that some
normative baseline (the "right" balance) has already been defined, when in
fact that is the very enterprise in which we are engaged.
Nonetheless, several things about this argument suggest that it is prob-
lematic as a justification for the different rules regarding improvement.
First, the balance of power justification seems to accept equivalence
between patent and copyright as an overarching goal. Only because copy-
right is perceived as weaker in certain areas are its stronger controls on
improvement justified. Further, its success depends critically on its ability
to demonstrate that patent is in fact a stronger right than copyright. That
is far from evident. There are certainly respects in which patent is
stronger than copyright. But there are also ways (besides the improvement
rules) in which copyright is a more powerful right than patent. Copyrights
last far longer than patents-an average of seventy-five years or so (which
soon may be raised to ninety-five years), versus a maximum of twenty
years for patents.2?8 The standard for protection of a copyright is much
lower than for patents-the copyright owner need only prove a minimum
level of originality, as opposed to the more rigorous patent tests of novelty,
nonobviousness, and enablement. 9 Copyrights do not need to be
examined, which means that they are available to protect a new work
almost immediately,' rather than two to three years down the road."
The scope of rights afforded copyright owners, traditionally thought to
be narrower than patent's all-encompassing "make, use or sell," may
actually be broader for some types of works. 2 And the penalties for
Use, 56 U. CHi. L. REv. 1017, 1019 (1989); Jordan P. Karp, Note, Experimental Use as Patent
Infringement: The Impropriety ofa BroadException, 100 YALE L.J. 2169, 2172-74 (1991).
237. C. Ralph S. Brown, Eligibilityfor CopyrightProtection:A Searchfor PrincipledStandards,
70 MINN. L. REv. 579, 588-89 (1985) (suggesting that the weaker rights conferred by copyright help
justify "the relatively casual approach to granting copyright as opposed to the more searching tests for
patentability").
238. Compare 17 U.S.C. § 302 (1994), with 35 U.S.C. § 154(a)(2) (1994).
239. See supra notes 49-52, 107-10 and accompanying text.
240. Registration with the Copyright Office is not required for copyright protection, but is still
required before bringing an infringement action. See 17 U.S.C. § 411(a) (1994). However, the copy-
right owner need only fill out a simple form to apply for registration-even if the Copyright Office
rejects the application, he can still file suit. See id.
241. See Mark A. Lemley, An EmpiricalStudy of the Twenty-Year PatentTerm, 22 AIPLA Q.J.
369, 386 (1994) (noting that the average patent spends over two years in prosecution).
242. Section 106 enumerates the exclusive rights of a copyright owner. 17 U.S.C. § 106 (1994).
The most significant rights absent from that list are the rights to control use and resale of the work once
copyright infringement are sometimes more severe than those for patent
infringement.4 3 Taking all of these factors into account, at the very least
it is not clear that the "balance of power" is shifted strongly in favor of
patents without the compensating effect of the differing improvement rules.
Finally, I am dubious of the claim that improving on someone else's
work is somehow more culpable than independent development of a work
that happens to be similar. Copyright infringement is a strict liability
offense. It is no defense that the copying was unconscious, or even that
the defendant was merely quoting or selling material that he had no way
of knowing was itself copied from another source.' 4 Copyright
infringers are not necessarily any more at fault than patent infringers,
despite the requirement of "copying" in the former, at least if by fault we
mean mens rea. Both patent and copyright infringement can be willful, or
they can be completely innocent. Instead, the argument seems to be that
taking material from another (as copyright does require) is inherently worse
than developing it on your own.
In the context of improvements, this argument seems misguided. The
value of improvements is precisely that they allow developers to build on
what others have done before them, rather than having to start from
scratch. It would be perverse indeed to require that "improvers" not make
any use of the material they are supposedly improving. It would also be
inefficient, putting improvers to a significant duplication of effort for no
appreciable societal gain. Whatever one thinks is the appropriate treatment
it has first been sold by the copyright owner or a licensee. However, copyright owners have recently
acquired those rights in a backhanded way in the context of digital information, as recent court
decisions have held that any use or transfer of a work in electronic form involves the making of
multiple copies. See, e.g., Triad Sys. Corp. v. Southeastern Express Co., 64 F.3d 1330, 1336-37 (9th
Cir. 1995), cert. denied, 116 S. Ct. 1015 (1996); MAI Sys. Corp. v. Peak Computer, Inc., 991 F.2d
511, 517-19 (9th Cir. 1993); cf. COMMERCE DEP'T INFORMATION INFRASTRUCTURE TASK FORCE,
INTELLECrUAL PROPERTY AND THE NATIONAL INFORMATION INFRASTRUCTURE: THE REPORT OF THE
WORKING GROUP ON INTELLECTUAL PROPERTY RIGHTS 106-07 (1995) (endorsing this result). The
result has been to give copyright owners control over the use of their works even after sale, and to
circumvent the first-sale doctrine by precluding resale of electronic works. See Litman, Revising
CopyrightLawfor the Information Age, supra note 122, at 42-44.
By contrast, patent law's prohibitions on "using" and "selling" patented technology are tempered
by implied license and the doctrine of patent exhaustion, which provide that a license to a supplier to
"make" and "sell" a patented product permits the purchaser of that product to use and resell it. See
Intel Corp. v. United States Int'l Trade Comm'n, 946 F.2d 821, 826-27 (Fed. Cir. 1991); Unidisco,
Inc. v. Schattner, 824 F.2d 965, 967-68 (Fed. Cir. 1987). But see Mallinckrodt, Inc. v. Medipart,
Inc., 976 F.2d 700, 708-09 (Fed. Cir. 1992) (holding that the exhaustion of a patent right can be
limited by licensing agreements). The result is that the patentee has less control over post-sale use of
his patented goods than does the owner of the copyright in works in digital form.
243. For example, willful copyright infringement for commercial gain is a crime punishable by
imprisonment. See 17 U.S.C. § 506(a) (1994). There is no similar criminal penalty for patent
infringement.
244. See, e.g., Lipton v. Nature Co., 71 F.3d 464, 471 (2d Cir. 1995).
of improvements under the law, surely its goal cannot be to encourage the
recreation of identical works by competitors from scratch.
D. Market PowerArguments
A final justification for treating improvements differently under copy-
right law than under patent law might stem from the belief that copyrights
confer less market power than patents. Under this argument, made to me
by Paul Heald, among others, the cost of allowing a copyright owner to
control improvements to his own work is less than the cost of a similar rule
in the patent context because more perfect substitutes usually exist in the
copyright context. If newcomers can compete effectively with an original
creator by designing their own works, they arguably have little or no need
to improve on the original creator's work. By contrast, it may be that
newcomers in patent markets must improve on existing works in order to
compete at all. Thus, giving newcomers rights to their improvements is
more important in patent than in copyright law.
Let me begin by pointing out the limits of the market power rationale.
The argument cannot be that patents always confer market power and copy-
rights never do; that is simply wrong as an empirical matter. Most patents
do not confer market power,' 5 and so the need to allow improvements
should not arise in most patent cases. Further, some copyrights do confer
market power.' More generally, the market power argument is vul-
nerable to many of the same criticisms I have leveled at the "Different
Works" argument above. It is impossible to distinguish "patentable" from
"copyrightable" works cleanly by reference to subject matter, and it is even
harder to distinguish innovations which confer market power from those
that do not by reference to the "patent" or "copyright" label.
Nonetheless, Paul Heald is surely correct that more patents than copy-
rights confer power in an economically relevant market, at least as a statis-
tical matter. I do not think that fact justifies the current legal rules,
however. In the first place, the legal rules that pay attention to market
power-the antitrust laws and the misuse doctrines-already take account
of this difference. The patent misuse doctrine, for example, is much
more vibrant than its copyright counterpart. t 7 And the federal antitrust
245. See supra note 26.

246. See infra notes 403-16 and accompanying text (discussing Lotus v. Borland).
247. See Lemley, supra note 26, at 1608-14 (describing the origins and scope of the patent misuse
doctrinein some detail); Ramsey Hanna, Note, MisusingAntitrust: The SearchforFunctionalCopyright
Misuse Standards,46 STAN. L. REV. 401, 404-06 (1994) (describing the much more limited contours
of the copyright misuse doctrine). Indeed, the first appellate court decision recognizing the existence
of the copyright misuse doctrine was not handed down until 1990. See Lasercomb Am., Inc. v.
Reynolds, 911 F.2d 970 (4th Cir. 1990).
laws' operate with much greater frequency against attempts to extend

patents than against similar attempts to extend copyrights. 9 Thus, the
market power argument is largely self-limiting because efforts to use intel-
lectual property rights to capture market power are punished under existing
legal rules, and those existing rules already punish patent licensing miscon-
duct far more frequently than they do copyright licensing misconduct.
Second, the nature of infringement is fundamentally different in patent
law than it is in copyright law. To infringe a patent, an accused product
must contain every element and limitation identified in the patent claim, or
an equivalent thereof.' Taking only part of a patented invention is per-
fectly legal. By contrast, copyright infringement is far more expansive.
Taking only a tiny portion of a copyrighted work can still be infringement,
so long as more than a de minimis amount of copyrighted expression is
used." 1 This means that even without the differential treatment of deri-
vative works, improvement on a copyrighted work is already significantly
more constrained than improvement on a patented work.
In short, the market power differential between patents and copyrights
is already accounted for by existing law other than the derivative works
rule. The derivative works rule seems to double-count differences in mar-
ket power. Further, since as adumbrated earlier the correlation between
market power and patentability is imperfect at best, a rule that treated mar-
ket power as a relevant factor in determining the need for improvements
would seem preferable to a blanket rule defeating any copyright protection
for such improvements.
V. Optimal Improvement Rules

If there are no persuasive reasons to think that the rule in patent cases
should differ from that in copyright cases, it is reasonable next to ask
whether one of the rules makes more sense than the other. Which rule bet-
ter serves the economic goals of intellectual property law? There are two
basic models of intellectual property that correspond loosely with the copy-
right and patent rules, respectively. Both models operate within the
general rubric of economic theory, attempting to provide optimal incentives
for innovation and creation. The first model may be termed property
rights theory. The fundamental precept of this theory is that efficient
248. See 15 U.S.C. §§ 1-6212 (1994).

249. There are, however, antitrust claims based on the defendant's alleged assertion of a copyright
in a way that captures market power. One example is Professional Real Estate Investors, Inc. v.
Columbia Pictures Industries, 508 U.S. 49 (1993).
250. See, e.g., Wamer-JenkinsonCo. v. HiltonDavis Chem. Co., 65 U.S.L.W. 4162,4163 (U.S.
Mar. 3, 1997); Pennwalt Corp. v. Durand-Wayland, Inc., 833 F.2d 931 (Fed. Cir. 1987) (en bane).
251. See, e.g., Harper & Row Publishers, Inc. v. Nation Enters., 471 U.S. 539, 548-49 (1985)
(holding that taking less than 13 % of a copyrighted work was infringement).
innovation will be encouraged (and wasteful races to invent be prevented)

by assigning broad property rights to original inventors. In opposition to
property rights theory is a theory of "tailored incentives" advanced by
Merges and Nelson. Their fundamental precept is that competition, not
monolithic ownership, most efficiently promotes invention. Merges and
Nelson dispute the presumption of property rights theorists that rivalry in
innovation is wasteful. They suggest that "when it comes to invention and
innovation, faster is better," and that "we are much better off with
considerable rivalry in invention than with too little."2 They offer
empirical evidence to support their position. 3 Merges and Nelson's
approach is consistent with the traditional economic approach, which views
intellectual property as a creation of limited rights by the government for
a specific purpose along the lines described in Part I. Even Landes
and Posner, noted advocates of property rights theory in other contexts,
treat intellectual property as primarily concerned with the balancing
of incentives rather than the initial allocation of property interests.' 5
252. Merges & Nelson, supra note 32, at 877, 878.

253. See id. at 881-908.
254. See WARD S. BOWMAN, JR., PATENT AND ANTITRUST LAW 32 (1973); SCHEP.ER, supra note
21, at 380-84; Martin J. Adelman, The Supreme Court,MarketStructure, andInnovation:Chakrabarty,
Rohm andHaas, 27 ANTITRUST BULL. 457 (1982); Oddi, supra note 13, at 273-85; Rhodes, supra note
13, at 1052-53.
It is important to distinguish the issue discussed in the text from the "property rule-liability rule"
framework for remedies introduced by Calabresi and Melamed in their famous article. See Guido
Calabresi & A. Douglas Melamed, PropertyRules, Liability Rules, andInalienability: One View of the
Cathedral,85 HARV. L. REV. 1089 (1972). As should be evident from even a cursory review of intel-
lectual property cases, successful plaintiffs benefit from a strong "property rule"-they are entitled to
injunctive relief in all but the most extraordinary cases. See Merges, supra note 154, at 2667-73.
Establishing that intellectual property remedies are governed by a "property rule' does not, however,
tell us the extent to which original creators are entitled to real-property-like control over improvements
within the scope of their original work. Kaplow and Shavell suggest that liability rules are appropriate
to protect individuals against negative externalities, while property rules are appropriate to protect indi-
viduals froma (physical) deprivation ofproperty. See Louis Kaplow& Steven Shavell, PropertyRules
Versus Liability Rules: An Economic Analysis, 109 HARV. L. REV. 713 (1996). Where intellectual
property rights fall on this spectrum is arguable.
In any event, there is a stronger argument for the use of property rules in intellectual property
cases: it is extremely difficult for courts to put a value on intellectual property rights. Employing
property rather than liability rules allows courts to avoid such valuation decisions. See generally A.
MITCHELL POuNSKY, AN INTRODUCTION TO LAW AND ECONOMICS 123-26 (1983). I discuss this issue
in more detail infra notes 349-51 and accompanying text.
255. For example, Landes and Posner argue:
Copyright protection-the right of the copyright's owner to prevent others from making
copies-trades off the costs of limiting access to a work against the benefits of providing
incentives to create the work in the first place. Striking the correct balance between
access and incentives is the central problem in copyright law. For copyright law to
promote economic efficiency, its principal legal doctrines must, at least approximately,
maximize the benefits from creating additional works minus both the losses from limiting
access and the costs of administering copyright protection.
William M. Landes & Richard A. Posner, An Economic Analysis of Copyright Law, 18 J. LEGAL
Where property rights theory assigns broad initial rights and then
leaves the parties to bargain to an efficient outcome, the tailored incentives
approach pays closer attention to the particular allocation of rights that is
made. Merges and Nelson's approach, if valid, undermines the fundamen-
tal tenets of a property rights approach to intellectual property, since
(as noted above) invention and creation are unquestionably cumulative
activities."56 My argument diverges from theirs in that my argument
does not depend upon the relative value of competition and property in
motivating invention. I argue in this Part that even if the property rights
approach is efficient in theory, in practice it faces insurmountable
obstacles. Those obstacles can best be overcome by a scheme of divided
entitlements similar to that which currently exists in patent law. The value
of this argument is twofold: it is not highly dependent on one's theory of
the innovation process, and (in part for that reason) it is generalizable to
copyright as well as patent cases.
I will begin to address these issues in subpart A by taking a closer
look at the property rights argument supporting the copyright rule, and
some of the assumptions on which it relies. In subpart B, I will examine
what happens when we relax some of the less realistic assumptions in pro-
perty rights theory.
A. PropertyRights Theory
In 1977, Edmund Kitch offered a new theory of the patent system, one
that he said would "reintegrate[] the patent institution with the general
theory of property rights." 7 The property rights theory of intellectual
property is rooted in many of the same economic traditions as incentive
theory, but its focus is on the ability of intellectual property ownership to
force the efficient use of inventions and creations through licensing."5
The fundamental economic bases of this approach are the "tragedy of the
commons" and the hypothetical Coasean world without transaction costs.
The tragedy of the commons is a classic economic story, in which common
property will be over-used by people who have access to it, since each
individual reaps all of the benefits of his personal use, but shares only a
STUD. 325, 326 (1989); see also Peter Jaszi, Caught in the Net of Copyright, 75 OR. L. REv. 299, 300
(1996) (stating that the property rationale does not accurately characterize intellectual property).
256. See supra note 32 (collecting sources); KENNETH W. DAM, INTELLECTUALPROPERTY IN AN
AGE OF SoFTWARsAND BIOTECHNOLOGY 4 (Chicago Law & Economics Working Paper No. 35, 1995)
("[Iln the overwhelming majority of instances each innovation builds on past innovations.").
257. Kitch, supra note 13, at 265.
258. See Wendy J. Gordon, Of Harms andBenefits: Torts, Restitution, and IntellectualProperty,
21 J. LEGAL STUD. 449, 473 (1992); Kitch, supra note 13, at 276-79; Merges, supra note 154, at
2661.
small portion of the costs.5 9 Thus, lakes open to the public are likely
to be over-fished, with negative consequences for the public (to say nothing
of the fish!) in future years. Common fields will be over-grazed, with
similarly unfortunate consequences. The economic solution to the tragedy
of the commons is private property.m° If everyone owns a small piece
of land (or lake), and can keep others out of it (with real or legal
"fences"), then the private and public incentives are aligned. People will
not over-graze their own land, because if they do they will suffer the full
consequences of their actions. 1 Further, if dealmaking between neigh-
bors is costless, as Coase postulated but did not believe,' transactions
will allow neighbors with large cattle herds to purchase grazing rights from
others with smaller herds. Such transactions should occur until each piece
of land is put to its best possible use.'
In the context of intellectual property, Kitch's article remains one of
the most significant efforts to integrate intellectual property with property
rights theory.' Kitch argues that the patent system operates not (as tra-
ditionally thought) as an incentive-by-reward system, giving exclusive
rights to successful inventors in order to encourage future inventors, but
as a "prospect" system analogous to mineral claims. Of course, one of the
characteristics of ideas is that because they are public goods, there is no
problem of "over-use" analogous to the commons in real property. Rather,
259. See Garrett Hardin, The Tragedy of the Commons, 162 SCIENCE 1243, 1244 (1968).
260. See id. at 1245.
261. While in theory it is possible for cattle owners to agree to limit their grazing in the public
interest, any such effort at agreement is likely to run into insurmountable problems. Not only will
organizing and policing such an agreement take effort that will not be rewarded, but individual grazers
have an incentive to free ride, reaping the benefits of reduced grazing by others while refusing to
reduce their own grazing. For more on these problems, see MANCUR OLSON, JR., THE LoGIc OF
COLLECTIVE ACTION (rev. ed. 1971). One commentator views this internalization of (positive) exter-
nalities as a key function of property. See Harold Demsetz, Toward a Theory of PropertyRights, 57
AM. ECON. REV. 347, 348 (1967).
On the other hand, for a rejection of the tragedy of the commons approach in certain contexts,
see Carol Rose, The Comedy of the Commons: Custom, Commerce, and Inherently Public Property,
53 U. CHI. L. REV. 711, 718-19 (1986). Rose is surely correct that private division of land is not
always efficient. Consider how problematic the task of walking through your neighborhood would be
if every piece of sidewalk were privately owned by a different person, and you were required to obtain
permission to take each step. See also ELINOR OSTROM, GOVERNING THE COMMONS: THE EVOLUTION
OF INSTITUTIONS FOR COLLECTIVE ACTION (1990) (offering examples of successful communal gover-
nance of common spaces).
262. See R.H. Coase, The Problem of Social Cost, 3 J.L. & ECON. 1, 2-7, 15 (1960).
263. See Calabresi & Melamed, supra note 254, at 1095 (discussing this implication of Coase's
theorem).
264. For other property-based views of intellectual property, see, for example, DAM, supra note
256; Trotter Hardy, Property (and Copyright) in Cyberspace, 1996 U. CHI. LEGAL F. 217, 264;
Edmund W. Kitch, Patents:Monopolies or PropertyRights?, 8 RES. L. & ECON. 31 (1986); Michael
Wurzer, Note, Infringement of the Exclusive Right to PrepareDerivative Works: Reducing Uncertainty,
73 MINN. L. REv. 1521, 1544-45 (1989) (delivering an extreme example of the property approach,
under which any part of a work with significant economic value is classed as property).
on Kitch's view, the primary point of the patent system is to encourage

further commercialization and efficient use of as yet unrealized ideas by
patenting them, just as privatizing land will encourage the owner to make
efficient use of it.' Society as a whole should benefit from this equali-
zation of private with social interests.
Fundamental to this conclusion are three assumptions. First, Kitch
argues that
a patent "prospect" increases the efficiency with which investment
in innovation can be managed .... [T]echnological information is
a resource which will not be efficiently used absent exclusive
ownership....
... [T]he patent owner has an incentive to make investments
to maximize the value of the patent without fear that the fruits of the
investment will produce unpatentable information appropriable by
competitors.'
This is analogous to the tragedy of the commons argument that only with
private ownership do private incentives match social incentives. In the
tragedy of the commons, the private incentive to "invest" in a field-for
example by letting it lie fallow, or limiting grazing, in order to permit it
to grow-is less than the social value of such an investment. In the patent
context, Kitch makes an analogous argument: that the private incentive to
improve and market an invention will be less than the social value of such
efforts unless the patent owner is given exclusive control over all such
improvements and marketing efforts.
Second, Kitch argues that
[n]o one is likely to make significant investments searching for ways
to increase the commercial value of a patent unless he has made
previous arrangements with the owner of the patent. This puts the
patent owner in a position to coordinate the search for technological
and market enhancement of the patent's value so that duplicative
investments are not made and so that information is exchanged
among the searchers.'
This is the Coase theorem at work. Under that theory, giving one party
the power to control and orchestrate all subsequent use and research
relating to the patented technology should result in efficient licensing, both
to end users and to potential improvers-assuming, that is, that information
268
is perfect, all parties are rational, and licensing is costless.
265. See Kitch, supra note 13, at 271-75 (making the analogy to mineral claims explicit).
266. Id. at 276.
267. Id.
268. See Anastasia P. Winslow, Rapping on a Revolving Door:An EconomicAnalysis of Parody
and Campbell v. Acuff-Rose Music, Inc., 69 S. CAL. L. REV. 767, 780 (1996) (noting that the Coase
Finally, for social benefit to be maximized, the property owner must

make the invention (and subsequent improvements) available to the public
at a reasonable price-ideally, one that approaches marginal cost as much
as is feasible.' But a property owner will have no incentive to reduce
his prices toward marginal cost unless he faces competition from others.
If the property owner is alone in the market, he may be expected to set a
higher monopoly price for his goods, to the detriment of consumers (and
social welfare). Kitch notes this problem, but does not resolve it. He
merely points out that not all patents confer monopoly rights, and that in
some cases the creators of intellectual property rights will face competition
from the makers of other fungible goods, and therefore that their individual
firm demand curves will be horizontal rather than downward-sloping!'
If one assumes such competition, intellectual property owners may be
expected to price competitively, just as producers of wheat do.
While Kitch makes his argument in the patent context, it is copyright
rather than patent law that seems to have taken his theory to heart. The
prospect theory of intellectual property strongly implies that the way to
obtain efficient investment in improvements is to give all rights to those
improvements to the original intellectual property owner. This should
accomplish two things, according to the theory: give the original owner of
the copyright or patent sufficient incentive to disseminate information about
the original invention, and give her (or her licensees) sufficient incentive
to invest resources in searching for improvements. As we have seen, it is
copyright, not patent law that gives the original creator nearly exclusive
control over alterations or improvements to her work. Indeed, some com-
mentators have argued that copyrights are a species of property rights. 271
theorem suggests that the initial assignment of property rights between original creators and improvers
is irrelevant). For a discussion of what happens when we relax these unrealistic assumptions, see infra
subpartV(B). On the importance of efficient licensing to the case for intellectual property protection,
see Wendy J. Gordon, Asymmetric Market Failure and Prisoner'sDilemma in Intellectual Property,
17 U. DAYTON L. REv. 853, 857 (1992) (arguing that unless markets for protected works develop,
generally through licensing, the economicjustification for protection of intellectual property collapses).
269. It is not possible to price intellectual property at its marginal cost and still stay in the business
of producing new works, since developing those new works requires a fixed investment of resources,
such as time and research money, that frequently dwarfs the marginal cost of making and distributing
copies of the idea once it has been developed.
270. See Kitch, supra note 13, at 274. This is not the same as conceding that demand curves are
always horizontal in patent cases. Compare Oddi, supra note 13, at 281 ("Kitch argues that the normal
assumption of a negatively sloped demand curve should not be made; rather the demand curve should
be substantially horizontal, indicating competitionwith thepatented invention."), with Kitch, supranote
13, at 274-75 ("[ T~he patent case is always visualized as one of a demand curve with a negative slope
and its attendant monopoly effects [when in fact] the demand conditions faced by particular ... patents
vary widely.").
271. SeeKenneth W. Dam, Some Economic Considerationsin the IntellectualPropertyProtection
of Sofiware, 24 J. LEGAL STUD. 321,332 & n.44 (1995); Frank H. Easterbrook, Intellectual Property
Is Still Property, 13 HARV. J.L. & PUB. POL'Y 108, 113 (1990); Hardy, supra note 264, at 218
B. Complicating the BargainingStructure

Unfortunately, as Coase himself recognized, the real world is substan-
tially more complex than the idealized picture painted in some of the
property rights literature.' Each of the fundamental assumptions that
underlies Kitch's model is problematic in the context of actual intellectual
property rights. In this section, I explore each of those assumptions in
more detail. The picture that emerges gives us not only a somewhat deeper
view of the role of improvement and licensing in intellectual property, but
also suggests reasons to prefer patent law's more nuanced treatment of
improvements to copyright law's absolutist view.
1. Identifying and OrganizingImprovers.-At the heart of the pros-

pect theory of intellectual property is the assumption that assigning
property rights to original inventors will promote efficient investment in
improvements by allowing a single party with the proper incentives to
coordinate the search for such improvements, either by investing more
resources itself in research and development or by licensing the right to
make improvements to other companies in a better position to conduct such
research.'
Unfortunately, identifying the right party to improve a technology is
not always easy, and certainly not always costless. Consider for a moment
the curious nature of Kitch's defense of the patent system. He argues that
(arguing for property rights in eyberspace and noting that "something like property rights" already
exists in copyright law). Wendy Gordon also points to detailed similarities between copyright law and
common-law property. See WendyJ. Gordon, An Inquiry into the Merits ofCopyright: The Challenges
of Consistency, Consent, and Encouragement Theory, 41 STAN. L. REV. 1343, 1354-77 (1989)
[hereinafter Gordon, Inquiry]; cf. Gordon, supra note 258, at 460 (arguing that the traditional
reluctance to recognize restitutionary claims in the common law of real property should not be
applicable to intellectual property). To be sure, Gordon's position on this issue is complex. She
acknowledges that the common law places limits on the power of property owners to control their
property. See Gordon, Inquiry, supra, at 1361-64. In addition, her ultimate reasons for favoring
strong copyright protection do not track Kitch's economic arguments. See id. at 1465-68. Further,
in other articles Gordon has suggested that strong copyright protection may not be appropriate in
circumstances where it may inhibit rather than promote improvements by others. See Wendy J.
Gordon, On Owning Information:IntellectualPropertyandthe RestitutionaryImpulse, 78 VA. L. REV.
149, 167, 247 (1992) [hereinafter Gordon, On Owning Information] ("A culture could not exist if all
free riding were prohibited within it."); Gordon, supra note 173, at 1633, 1631-35 (observing that the
owner of a copyright "may not be willing to exploit all of the possible derivative works over which his
copyright would ordinarily give him control," even though such derivative works may have social
value).
Other scholars who adopt the property rights view of copyright, at least in part, include Landes
4 Posner, supra note255, at 354-55 (describing economic distortions, such as the timing of the release
of the original work and transaction costs, as reasons for giving the author exclusive control of
derivative rights).
272. See Coase, supra note 262, at 15.
273. See Kitch, supra note 13, at 276.
awarding patents to those who have identified an invention but might not
yet have commercialized it makes sense, because making those inventors
property owners will give them the right incentive to improve and commer-
cialize their invention.274 But if it is the property incentive that is criti-
cal, why wait for a first inventor to file a detailed patent application, which
after all must reduce the invention to practice and enable others to make
it?275 After all, mineral claims on private land belong to the private land-
owner, whether or not he is the first to discover them.276 In many other
contexts, we assign property rights virtually arbitrarily, and expect trans-
actions to take care of the rest: we did not wait to see who would build a
high-rise building during the Oklahoma land rush, for example. Assigning
property rights in advance of any invention-perhaps to the first person
to request them in a particular field, even if they have not yet invented
anything-seems perfectly consistent with the prospect theory.'m Indeed,
it should be preferable to the current patent system on this theory, since it
will allow efficient investment in initialresearch as well as improvements,
and prevent wasteful races to be the first to invent the initial
technologyY8
The problem with handing out property rights in advance of invention
is the same problem with Kitch's prospect theory-it is unrealistic to expect
that property owners will be uniquely good at identifying potential future
inventors or improvers. Even if assigning property rights ex ante gives the
property owner the proper financial incentives to invest in research (or to
search for the "right" inventor),2 79 there is more to invention than
research dollars. Modern economic literature is replete with evidence of
how hard it is to predict invention in advance, or even to replicate an
invention once it has been made. 2' Further, even companies that
274. See id. at 247, 246-47 ("Absent a patent, firms have less than the optimal incentive to invest
in providing information about and techniques for using the new technology.").
275. See 35 U.S.C. § 112 (1994); Fiers v. Revel, 984 F.2d 1164, 1169 (Fed. Cir. 1993).
276. Kitch discusses mining claims on public rather than private land. See Kitch, supra note 13,
at 271-75.
277. For a proposal along these lines, see Kingston, supra note 55, at 59-86 (proposing an
"innovation warrant" to be awarded in advance of promised innovation); see also Ben T. Yu, A
Contractual Remedy to Premature Innovation: The Vertical Integration of Brand-Name Specific
Research, 22 ECON. INQUIRY 660 (1984) (suggesting that manufacturers might contract with potential
inventors before they invent); Ben T. Yu, PotentialCompetition and Contractingin Innovation, 24 I.L.
& ECON. 215 (1981) (arguing that prior contracting should be adopted to promote competition in
innovation).
278. On wasteful races to invent, see Merges & Nelson, supra note 32, at 868-79.
279. It is not clear that even this assumption is reasonable. Merges points to economic literature
suggesting that a company with a successful invention will choose to focus on one or two basic applica-
tions of that invention, rather than investing money and effort in researching implications of its inven-
tion that lie outside its core competencies. See Merges, supra note 83, at 371-73.
280. For more detail on this issue, see, for example, Steven A. Lippman et al., Heterogeneity
Under Competition, 29 ECON. INQUIRY 774 (1991) (noting factors that may prevent firms that have
innovated successfully from replicating their invention on a broader scale); Michael E. Porter, The
successfully invent may be unable to efficiently commercialize the

invention, and in many cases may not even recognize the significance of
their own invention. Economic history provides some striking examples
of inventors who grossly understated the market value of their own
inventions.281 Expecting one party-original inventor or not-to perfectly
identify the potential uses of a new invention, how it might be improved,
and who can best improve it is simply not realistic.
A much more likely solution is to find a way for improvers-those
with new ideas for how to make the invention better, or for how to market
it-to identify themselves. Of course, as Kitch points out, granting patents
on original inventions is one way to do that. If the original inventor
has the power to control all subsequent uses and improvements on the
invention, he argues, those who are thinking of investing time and money
in improving or marketing the invention will first come to the patent owner
seeking a license.' In this way, granting patents may facilitate the self-
identification of improvers.2
The rather absolute power the original inventor is given to control
improvements means that rational potential improvers will come to the
patent owner before they invest any time or money in improving the patent,
if indeed they do so at all.' If the patent owner has the right to control
Structure Within Industries and Companies' Performance, 51 REv. ECON. & STAT. 214 (1979)
(discussing mobility barriers in specific markets that restrict the ability to innovate or to replicate
successful strategies); David I. Teece, Profiting from Technological Innovation: Implications for
Integration, Collaboration,Licensing and Public Policy, 15 REs. POL. 285 (1986) (theorizing that
successful innovation may depend on a firm's ability to control complementary assets).
281. According to one report, the full commercial potential of many major inventions is not
immediately recognized by the inventor or the marketplace. Nathan Rosenberg cites a number of
examples of such unrecognized potential: Marconi expected the radio to be used for point-to-point
communication, not broadcasting; the transistor was expected to be used primarily in hearing aids for
the deaf; Bell called his telephone a mere "improvement[] in telegraphy."
Kathleen O'Toole, The Future Was 'Obviously Not Obvious," STAN. OBSERVER, May-June 1994, at
13 (recounting a lecture delivered by Nathan Rosenberg).
282. See Kitch, supra note 13, at 276; see also Gordon, supra note 258, at 472-74. Gordon
argues that giving intellectual property rights to creators will encouragethose who wish to use an inven-
tion to come to the intellectual property owner for a license. While this is no doubt true of most users,
the situation is somewhat different in the case of improvers, as explained below.
283. Alternatively, the patent grant may serve a marketing function, identifying original inventors
to companies that might wish to acquire the invention outright, perhaps because they are in a better
position to develop or commercialize it. See Willard F. Mueller, The Origins of the Basic Inventions
UnderlyingDu Pont's Major Product and Process Innovations, 1920 to 1950, in THE RATE AND
DIRECTION OF INVENTIVE AcTivITY (Nat'l Bureau of Econ. Research ed., 1962) (noting that Du Pont
purchased many inventions from outside creators); Roberto Mazzoleni & Richard R. Nelson, Economic
Theories About the Benefits and Costs of Patents 15-16 (1996) (unpublished manuscript, on file with
the Texas Law Review). This is unquestionably a "prospect"-related benefit of the patent system.
However, it does not deal directly with the improvements issue, and, in particular, does not offer any
justification for a rule that gives original inventors the automatic right to control the work of improvers.
284. Rochelle Dreyfuss suggests another option-that potential improvers may simply work on a
different product in the same field, wastefully building from scratch rather than working from the
original invention. See Dreyfuss, supra note 220, at 228.
any improvement made to her invention, there is no point in improving her

invention unless you first get permission from her to do so. To do other-
wise would simply be giving the patent owner the fruits of your labor.
This is a logical result of Arrow's information paradox.' The actual
improver coming to the negotiating table with a patent owner possesses
valuable information (the improvement), which he would like to disclose
to the patent owner in exchange for money. Unfortunately, there is no
way for the patent owner to assess the value of the improvement without
finding out what it is. And if the patent owner finds out what the improve-
ment is, under the prospect theory she is free to use it without compen-
sating the improver." 6
From the perspective of someone trying to identify the optimal
improvers, though, the fact that only potential rather than actual improvers
will seek a license is a real problem. The patent owner has no good way
of evaluating potential improvers and choosing the most efficient one (or
ones).' It would be far more efficient to find a way for improvers to
invent first, and then bargain with the original patent owner over the
allocation of rights to the improved invention. The original inventor would
only have to deal with a small subset of all potential improvers-those that
were successful. Further, both sides would know what they were bar-
gaining over, since the improvement would already have been made. Of
course, for any such bargaining to occur, we must find a way around
Arrow's paradox.
285. See Arrow, supra note 19, at 614-16; see also Gordon, supra note 258, at 475-76
(identifying this problem, though in the analogous context of intellectual property creators trying to
license their work in a world without legal protection). It is worth noting that the problem is signifi-
cantly worse in the intellectual property context than it is in the context of real property. In certain real
property contexts, such as oil and gas leases, oil exploration companies may find it rational to drill first
and negotiate rights later. The transaction costs of signing agreements with multiple property owners
of an unexplored field outweigh the risk of losing the drilling rights should the field be valuable. By
contrast, in the intellectual property context, potential improvers are less likely to win the rights after
disclosure to original inventors, since the original inventor (unlike the real property owners above) pre-
sumably has its own facilities and could develop the improvement in-house. (By way of comparison,
oil exploration companies do not do their exploratory drilling in a competitor's field without an agree-
ment in place.) Further, the value of the improvement is information which would have to be
disclosed, while the value of an oil strike may be partially appropriable by the drilling firm.
286. See Merges, supra note 87, at 81-82. While there are partial solutions to Arrow's paradox
in this context, they all involve giving the improver some sort of legal control over the improvement.
These solutions are described in more detail infra notes 382-87.
287. See Merges & Nelson, supra note 32, at 875 ("In our own research, we have not found a
single case where the holder of a broad patent used it effectively through tailored licensing to coordinate
the R&D of others."). Instead, patent owners sometimes use as a proxy for possible improvers the sub-
set of companies that take a license to work the original patent. By employing a "grantback clause,"
the original inventor can obtain ownership rights to all improvements by licensees, whether or not the
improvement would infringe the original patent. See Timothy J. Engling, Improvements in Patent
Licensing, 78 J. PAT. & TRADEMARK OFF. Soc'Y 739, 740-41, 744-45 (1996) (describing grantback
clauses).
The current rule of blocking patents provides just such a bargaining

mechanism. Improvers have an incentive to invest in research even in the
shadow of an original invention, since they can obtain a patent on their
improvement. And the fact that an improvement patent gives them some
real bargaining power also provides them with an incentive to come to the
bargaining table, and indirectly, an incentive to invest in improvement in
the first place. 8 Furthermore, one implication of Arrow's theorem is
that improvers will be more likely to disclose information that is uniquely
within their knowledge if they have a property right in that information.
In short, where original patent owners have less than perfect information
about improvements, providing improvers rather than possible origi-
nal patent owners with a legal entitlement to the improvement should
encourage the efficient identification of actual rather than merely poten-
tial improvers. 9 The same result should hold for the same reasons in
copyright law.2',
2. Licensing Improvers.-Even assuming that the patent owner in

Kitch's model can perfectly identify potential improvers, he must still
come to terms with them by licensing the right to improve on his original
invention. In theory, the original patent owner should grant such per-
288. The mechanics of such bargaining are considered further in the next section.
289. A proponent of property rights might point to two alternative means of achieving the same
result even when the original inventor controls all rights to improvements. First, one might try to
avoid Arrow's paradox by vesting actual improvers with a trade secret right in their improvement.
Thus improvers might be able to negotiate a confidentiality agreement with the original patent owner
prior to disclosing their invention. Unfortunately, such agreements have proven difficult to enter into
and enforce. Many companies refuse to sign such agreements under any circumstances, because they
fear that the "improver" will disclose an idea that the company has already been working on, and will
later sue for misappropriation. See Whitfield v. Lear, 751 F.2d 90, 93 (2d Cir. 1984) (noting a
general practice of rejecting unsolicited movie scripts); Davis v. General Foods Corp., 21 F. Supp. 445
(S.D.N.Y. 1937) (describing a form letter by which a company reserved all rights to determine if an
unsolicited recipe was deserving of compensation). On the other hand, if the confidentiality agreement
permits such independent development, it arguably does not provide particularly robust protection to
the improver disclosing the information, since it is often hard to prove that an idea was stolen rather
than developed independently.
Alternatively, a patent owner might try to license any and all potential improvers by granting
them a limited property right in whatever they later invent. This would in effect mimic the blocking
patents rle by contract. While in a strict Coasean world such broad agreements would be equivalent
to giving patents to improvers by legal rule, in fact they impose significant and unnecessary transaction
costs (as discussed in the next section) to achieve the same result patent law already does.
290. See Goldstein, supra note 37, at 252 (suggesting that the creators of unauthorized derivative
works should have rights in their original creations). While the interests at stake are somewhat dif-
ferent in copyright cases, where the "improvement" is more likely to consist of a number of small
expressions rather than a single idea, the same problem with predisclosure bargaining still occurs.
Indeed, the existence of Hollywood script registries testifies to one copyright-driven industry's attempts
to deal with the problem. See generallyRobert M. Winteringham, Note, Stolen from StardustandAir:
Idea Theft in the EntertainmentIndustry anda Proposalfora ConceptInitiator Credit,46 FED. COMM.
L.J. 373 (1994).
mission in exchange for a royalty based upon the value of the new
improvement. In practice, however, efficient licensing negotiations will
not always take place, for a number of reasons.
a. Transaction Costs.-First,it is simply not true that licensing

transactions are costless. At a minimum, all transactions cost time and
money-the purchaser must figure out what they need, determine who has
it, contact them, negotiate a payment, and exchange the payment for the
good or service they require. In some contexts, most of these costs are
trivial-you know you will find apples at your corner grocery, and the
grocery has already fixed a price and set up a mechanism (the cash
register) to take your money in exchange for the apples. Intellectual
property licenses are more complex and certainly more costly than simple
purchases of goods. For one thing, licenses may involve complex assign-
ments of partial legal rights, and they therefore typically involve lawyers
drafting the licenses. 9 ' For another, these licenses frequently create
an ongoing relationship between the parties, one which may require
monitoring. 2' Further, the scope of most intellectual property rights is
not evident from the face of the patent (and impossible to determine from
the existence of a copyright). This means that even identifying who the
proper parties to a license are can be a significant problem, 2' and some-
times an insurmountable one (as where the best improver does not yet exist
or is working in a garage somewhere, which was quite common in the
early days of the software industry). Finally, while the parties ideally
would base the cost of a license on the value of the right licensed, that
value will likely be difficult to determine accurately in the case of unique
goods like intellectual property rights.2' This problem is exacerbated in
the context of licensing potential improvements, since if it is hard to value
an invention that has already been made, it is well-nigh impossible to value
one that might be made in the future.
The result of all these factors is that the transaction costs of intel-
lectual property licenses are significant.295 Various commentators have
291. See JAY DRATLER, JR., LICENSING OF INTELLECTUAL PROPERTY § 1.01(2), 1-6 to 1-7
(1996).
292. See id. § 1.04(1).
294. SeeJ. PeterKilling, TechnologyAcquisition:License Agreement orJoint Venture, 15 COLUM.
J. WORLD Bus., Fall 1980, at 38, 44-45 ("Neither buyers nor sellers of technology seem to have a
clear idea of the value of the commodity in which they are trading."); David J. Teece, The Marketfor
Know-How and the Efficient InternationalTransfer of Technology, 458 ANNALS AM. ACAD. POL. &
SOC. Sl. 81, 87-88 (1981).
295. But see Dam, supranote 271, at 353 n.108 (assuming that transaction costs are relatively low
in the software arena and that likely new developers can easily be identified, but offering no evidence
to support this assumption).
estimated that cost at as much as twenty percent of the total value of the
underlying technology license, 2' or in excess of $100,000 per trans-
action.2' These data are of limited value, since they are taken from
international technology licenses, where the costs are no doubt higher than
for the average license." Nonetheless, even the average transaction
costs associated with an intellectual property license are unlikely to be
trivial.
To be sure, there are institutions such as the American Society of
Composers, Authors and Publishers (ASCAP) that have developed to
reduce the transaction costs of intellectual property licensing, normally by
licensing in bulk. 2' There have also been attempts to reduce the per-unit
cost through standardization of contracts, notably in the distribution of
computer software through "shrinkwrap licenses."" But these mecha-
nisms for reducing transaction costs are far more effective at licensing
standard end uses than at licensing the creation of improvements. Improve-
ments are hard to categorize and assign set fees; different uses will incor-
porate different amounts of the original work and will occur in different
markets. Further, copyright owners are likely to want the power to review
each individual use by an improver, while they may be willing to waive
such rights in the case of standardized, consumptive end uses. Thus, an
"improvers' ASCAP" may well face insurmountable problems. Further,
to the extent such a standard "improver's license" is created, it will likely
be accompanied by a standardized "per-image" or "per-use" fee, one
which is suboptimal because it does not reflect the value of the use pro-
posed by that particular improver.0 1
The presence of these costs in intellectual property licensing transac-
tions leads to two types of first-order deviations from the efficient behavior
296. See DAVID J. TEECE, THE MULTINATIONAL CORPORATION AND THE RESOURCE COST OF
INTERNATIONAL TECHNOLOGY TRANSFER 43-44 (1976).
297. See FAROK J.CONTRACTOR, INTERNATIONAL TECHNOLOGY LICENSING: COMPENSATION,
COSTS, AND NEGOTIATION 105 (1981).
298. See Richard E. Caves et al., The Imperfect Marketfor Technology Licenses, 45 OXFORD
BULL. ECON. & STATS. 249, 260-62 (1983); cf. Hardy, supra note 264, at 236-39 (discussing the
transaction costs in the computerization of information and arguing for a property entitlement in
cyberspace); Mark A. Lemley, Shrinkwraps in Cyberspace, 35 JURIMErRICs J. 311, 323 (1995)
(suggesting that the growth of the Internet may reduce transaction costs for many types of licenses).
299. See Stanley M. Besen et al., An Economic Analysis of Copyright Collectives, 78 VA. L. REV.
383, 385 (1992); Robert P. Merges, Contractinginto LiabilityRules: Intellectual PropertyRights and
Collective Rights Organizations, 84 CAL. L. REV. 1293, 1296-97 (1996); Merges, supra note 154, at
2669-70.
300. For more detail on such licenses, see Mark A. Lemley, Intellectual PropertyandShrinkwrap
Licenses, 68 S. CAL. L. REV. 1239 (1995).
301. For a detailed discussion of this problem in the multimedia context, see Merges, supra note
299, at 1273-75; Fara Daun, Comment, The Content Shop: Toward an Economic Legal Structurefor
ClearingandLicensing Multimedia Content, 30 LOYOLA L.A. L. REV. 215 (1996).
predicted by economic models that do not account for transaction costs.

First, some original inventors will inefficiently choose not to license poten-
tial improvers for their technology. This may happen either because the
perceived value of the improvements is sufficiently small that it is over-
whelmed by the transaction costs of licensing, or because the marginal
value of having a third party (rather than the original inventor) develop the
improvements does not outweigh the transaction costs of licensing.'
Second, some potential improvers who would seek a license for their
improvements will no longer do so because of transaction costs. A subset
of these potential improvers who forego licenses will nonetheless proceed
to improve without a license,' opening themselves up to a suit for
infringement. A second subset will forego the improvements process
entirely because they could not secure a license, even though it would have
been efficient for them to engage in the effort of seeking to improve the
original work.
b. Uncertainty.-Some of the problems that increase the cost of

transacting in intellectual property licenses, particularly the difficulty of
valuing both original inventions and improvements, may also prevent bar-
gaining parties from coming to terms.' In the Coasean world, two par-
ties may be expected to agree if there is a surplus to be gained from the
agreement. The value of the surplus (the "gains from trade") is what will
be divided during the negotiation. For example, if I can commercialize my
invention to produce a profit of $500, and you could produce a profit of
$700 commercializing the same invention, the logical thing for me to do
is to sell you the right to commercialize my invention for some price
greater than $500 but less than $700. The gains from trade in this
transaction are $200.
But in order for the parties to divide the gains from trade, they must
know what those gains are. Uncertainty over the value of an invention (or
in our case, over the value of potential improvements to the invention) can
prevent the parties from agreeing to an efficient license transaction. So can
uncertainty over the scope of a patent-that is, over whether the licensing
transaction is necessary at all. 5 This may happen in one of two ways.
302. Indeed, Merges declares that "it seems whimsical to assume that all improvers and potential
improvers will be able to bargain with the holders of pioneering patents." Merges, supra note 83, at
374.
303. For example, consider an author who wishes to excerpt from the work of another, but is
unable to track down the successors in interest of a dead author or a defunct publisher.
304. See Merges, supra note 154, at 2657-59 (explaining why uncertainty over the relative value
of original inventions and improvements is likely).
305. Some have suggested that the scope of a patent is "clearly demarcated in the patent itself."
See, e.g. Kenneth W. Dam, The Economic Underpinnings of PatentLaw, 23 J. LEGAL STUD. 247,267
(1994); see also Markman v. Westview Instruments, Inc., 52 F.3d 967, 986 (Fed. Cir. 1995) ("If the
First, uncertainty may prevent the parties from recognizing that there are
any gains to be had from trade. In the example above, if neither of us
recognize that you would be better at commercializing the invention than
me, we will not enter into a license agreement. Second, even where the
parties recognize that licensing would be beneficial, uncertainty may cause
them to disagree over the appropriate price in a way that prevents the bar-
gain from taking place. Again using the example above, suppose that
while you would actually be able to commercialize the invention for a $700
profit, because of uncertainty I think your efforts will result in profits of
$900, while you think they will produce profits of only $600. You will be
unwilling to accept my demand that you pay me $700 for the exclusive
right to commercialize the product, even though to me that demand repre-
sents a fair division of what I perceive to be the gains from trade. Rational
Coasean bargaining assumes perfect information, a characteristic strikingly
absent from negotiations involving intellectual property.'
c. Externalities.-Improvements that produce positive

externalities-that is, social benefits produced by a work that cannot be
captured by the work's owner-may not be efficiently produced under a
property rights licensing scheme. The willingness of an original copyright
owner to license his work to an improver under a property rights scheme
depends, not surprisingly, on the improver's willingness to pay. The
improver's willingness to pay, in turn, depends (among other things) on the
revenue the improver expects to receive from the improvement. Where the
improver's expected revenue matches society's expected benefit, exter-
nalities do not interfere with efficient licensing between the parties.
However, where the improver's work would create a significant social
benefit that the improver cannot capture, the improver (and thus the origi-
nal copyright owner) will undervalue that work. The result is that the
patent's claims are sufficiently unambiguous for the Pro, there should exist no factual ambiguity
... .'), aff'd, 116 S. Ct. 1384 (1996). But that statement is clearly wrong as a factual matter. There
is considerable uncertainty over the scope of most patents, for many reasons: because of drafting
ambiguities, because of the doctrine of equivalents, and because of uncertainty about the validity of the
patent. See supra subpart H(A); cf. Paul N. Higbee Jr., The Jury's Role in Patent Cases: Markman
v. Westview Instruments, Inc., 3 J.INTELL. PROP. L. 407, 422 (1996) (criticizing the Markman court
for reasoning that statutory requirements and administrative supervision removed all ambiguity sur-
rounding the scope of patent claims). All of these uncertainties complicate license negotiations.
306. See Kenneth J. Arrow, The Property Rights Doctrine and Demand Revelation Under
Incomplete Information, in ECONOMICS AND HUMAN WELFARE: ESSAYS IN HONOR OF TIBOR
ScrrovsKy 23, 29-31 (Michael J.Boskin ed., 1979), reprintedin 4 KENNETH J.ARROW, COLLECTED
PAPERS OF KENNETH J.ARROW: THE ECONOMICS OF INFORMATION 216, 222-24 (1984); cf. Steven
N. Wiggins & Gary D. Libecap, Oil Field Unitization:ContractualFailurein the PresenceofImperfect
Information, 75 Am.ECON. REv. 368, 368 (1985) (providing empirical evidencethat divergentvalua-
tions of oil field revenue resulting from "imperfect and asymmetric information" led to the breakdown
of oil field development negotiations in a significant number of cases).
improver may be unwilling or unable to pay for a license to use an original

work, even though the aggregate benefit to society would more than make
up for the license fee demanded.
As an example, consider the efficient production of law review
articles.' Many law review articles, this one included, rely heavily on
both the ideas and the words of scholars who have written before. In an
absolutist property rights world, those prior scholars might well hold
the right to prevent any discussion or quotation of their work without
permission, on the theory that they would be willing to license their works
for use by subsequent scholars. That theory might work if law professors
were actually paid to publish law review articles that contained valuable
contributions to legal theory. 8 But because law review authors are not
compensated on the basis of how valuable their ideas are (at least not
directly), they may not be willing to pay for the right to quote prior
scholars, notwithstanding the social value that such a quotation would add
to the article. Less trivial examples of positive externalities in improve-
ments include news reporting, speech on matters of public concern, and
explanatory material.
At a more general level, one might characterize competition itself
as an uncompensated positive externality in a property rights model.
Consider the example of reverse engineering of software. Potential
competitors might well wish to reverse engineer a computer program in
order to produce a competing program that is independently created, but
which is interoperable with the first.' Because even intermediate copies
that do not show up in final products can violate the Copyright Act,"'0
such a potential competitor would need a license from the original creator
in order to enter the market in competition with that original creator. A
Chicago school economist might suggest that the original creator will grant
such a license to the competitor if and only if the value of the new use
exceeds the cost, and therefore that a failure to license is conclusive evi-
dence that the competitor's proposed use would be inefficient. This
307. I will ignore the cynic's view that the socially efficient number of law review articles might
be zero.
308. See Gordon, supra note 173, at 1630 ("[R]evenues from scholarly articles are arguably
smaller than such [social] benefit would warrant."). For a treatment of the issue in some detail, see
Howard P. Tuckman & Jack Leahey, What Is an Article Worth?, 83 J. POL. ECON. 951 (1975).
309. For examples of actual cases presenting such issues, see DSC Communicationsv. DGI Techs.,
Inc., 81 F.3d 597 (5th Cir. 1996); Sega Enters.Ltd. v. Accolade, Inc., 977 F.2d 1510 (9th Cir. 1992);
and Atari Games Corp. v. Nintendo of America Inc., 975 F.2d 832 (Fed. Cir. 1992). On the general
process of software reverse engineering, see Andrew Johnson-Laird, Reverse Engineering of Software:
Separating Legal Mythology from Actual Technology, 5 SoFrWARE L.J. 331, 342-45 (1992). For a
general discussion of the economic parameters of the problem, see Julie E. Cohen, Reverse Engineering
and the Rise of Electronic Vigilantism:IntellectualProperty Implicationsof "Lock-Out"Programs,68
S. CAL. L. REV. 1091 (1995).
310. See Sega, 977 F.2d at 1518-19.
explanation rings hollow, and not merely because competitors might have
noneconomic reasons not to agree.31 In this case, the proposed use
might be socially valuable precisely because it encourages more compe-
tition in a market than would exist without the use, and therefore reduces
the profits of all competitors while benefiting consumers. 3 ' The
enhanced consumer surplus associated with a move from monopoly to
competition by definition cannot be "captured" by the competitor, and so
the competitor cannot possibly offer to pay the original creator the full
social value of the license. Nor can consumers effectively band together
to pay for the reverse engineering, because of collective action
problems. 1 3 This form of licensing market failure is endemic in
transactions where the proposed licensee will compete directly with the
original copyright owner. Such transactions rarely occur. 14
d. StrategicBehavior.-Even if the parties have sufficient infor-

mation so that they can rationally sit down to bargain, strategic behavior
on the part of one or both may prevent them from reaching an efficient
agreement. The Coase theorem postulates that parties with a surplus to
divide will divide it, but it does not specify how the surplus will be
divided. One might think that dividing the surplus down the middle
is "fair," but in point of fact a myriad of factors may enter into this
calculation. To a party trying to divide a $200 surplus, there is a big
difference between getting $1 of the surplus and getting $199, or even
between getting $100 and getting $25.
How does one maximize one's share of the gains from trade? One
way is to lie, or at the very least to conceal information. If you can con-
vince your opponent that your contribution is worth more than it is, or per-
haps just that you believe it is worth as much as you say, you may be able
to extort more money out of him, skewing the gains from trade in your
favor. Alternatively, you might play a game of "chicken," convincing
your opponent that you are irrational enough to kill the whole deal unless
311. On such noneconomic factors, see infra notes 323-26 and accompanying text.
312. See Jerry R. Green& Suzanne Scotchmer, On the Division of Profit in SequentialInnovation,
26 RAND J.ECON. 20, 21-22 (1995) (arguing that the first innovator cannot capture all the profit from
later innovation by a different improver through licensing transactions). Green and Scotchmer view
this as a problem, but in fact it seems central to the limited nature of the intellectual property
enterprise.
313. Cf. RAFAEL ROB & ARTHUR FISHMAN, THE FREQUENCY AND PRICING OF PRODUCT
INNOVATIONS (University of Pa. Institute for Law and Econ. Discussion Working Paper No. 226,
1995).
314. For example, virtually all software licensing contracts contain provisions purporting to forbid
reverse engineering. See Lemley, supra note 300, at 1239; David A. Rice, Public Goods, Private
Contract, and Public Policy: FederalPreemption of Software License ProhibitionsAgainst Reverse
Engineering, 53 U. PrIr. L. REv. 543 (1992).
you get more than your "fair" share. 15 Such strategies are rational in
the sense that they may produce asymmetric gains skewed in favor of the
party that employs them. However, they may also result in the parties
failing to come to terms at all, particularly if both parties aggressively seek
the lion's share of the gains from trade.316
Robert Cooter has modeled the consequences of strategic behavior for
such negotiations,3 17 and Robert Merges has applied that model in the
patent context. 1 8 Merges treats a hypothetical situation involving two
negotiating parties, one an original inventor whose invention produces
profits of $100, and the second an improver whose improvement will
produce a total of $1000 in profits if used in conjunction with the original
invention.3 19 He suggests that giving absolute control over improvements
to the original inventor encourages that original inventor to seek a larger
share of the gains from trade than the improver, even though the improver
has contributed the majority of the value of the improved invention?' °
This strategic behavior threatens to prevent the parties from concluding a
deal, and indeed Merges notes several examples of what he calls
"bargaining breakdown" in patent history.321 As with the problems of
transaction costs and uncertainty, the effect of strategic behavior is that
parties in some cases will not enter into efficient licensing agreements. 3'
e. Noneconomic Incentives.-Finally, some efficient licensing

arrangements may not occur simply because the parties take account of
other, noneconomic considerations. Some of these considerations may be
315. See WILLIAM POUNDsTONE, PRISONER'S DILEMMA212-13 (1992) (discussing the advantages
of perceived irrationality in game-theoretic models of "chicken").
316. While one might object that one or both parties should back down if they see that the entire
negotiation is at risk, that is not always the case. Because of information uncertainties it will not
always be evident that a party is playing chicken. Further, where one party is a repeat player in license
negotiations, it may profit that player to develop a reputation for refusing to back down, since that
reputation will in turn influence the behavior of its opponents in subsequent negotiations. See JAMES
W. FRIEDMAN, GAME THEORY WITH APPLICATIONS TO ECONOMICS 136-38 (1986) (arguing that
"tough" responses early in the bargaining process are rewarded later by discouraging other competitors
from entering the market).
317. See Robert Cooter, The Cost of Coase, 11 J. LEGAL STUD. 1, 20-21 (1982) [hereinafter
Cooter, Cost of Coase]; Robert Cooter, DecentralizedLawfor a Complex Economy: The Structural
Approach to Adjudicating the New Law Merchant, 144 U. PA. L. REv. 1643, 1675-77 (1996)
[hereinafter Cooter, Law Merchant]; Robert Cooter & Stephen Marks, Bargainingin the Shadow of
the Law: A Testable Model of Strategic Behavior, 11 J. LEGAL STUD. 225 (1982).
318. See Merges, supra note 87, at 82-83; Merges, supra note 154, at 2659-60; Merges &
Nelson, supra note 32, at 865-66.
319. See Merges, supra note 87, at 79.
320. See id. at 81.
321. See id. at 84, 84-89 (listing examples involving Marconi Wireless Telegraph Company,
AT&T, RCA, and Bessemer).
322. Indeed, Cooter goes so far as to characterize hard bargaining as a negative externality. See
Cooter, Law Merchant,supra note 317, at 1676.
idiosyncratic. For example, a patent owner might be unwilling to license

her work to a Californian, for no other reason than that she does not like
Californians. Perhaps more likely, she may be unwilling to license her
work to her ex-husband, for reasons that have nothing to do with his
willingness to pay. Depending on how broadly one is willing to define
rationality, such refusals might be termed "irrational."' But certainly
they do happen all the time. In the normal goods market, we do not worry
too much about them-if your corner grocery will not sell to Californians,
the one down the street likely will. But in the context of intellectual
property, where the "goods" are unique and the markets therefore "thin,"
such refusals may be more problematic.
There are some cases where such noneconomic refusals to license are
endemic. For example, very few novelists or playwrights will be willing
to grant the right to excerpt from their material to critics who intend to
write a devastating review of their work. Similarly, few people would be
willing to license their books or songs (or for that matter their names or
images) a' to those who would ridicule them, for example by parodying
their earlier work. Several scholars have identified parody and criticism
as areas where markets fail because the intellectual property owner cannot
be expected to act "rationally" in licensing his work. 3' While this is not
necessarily market failure in a strict sense-from the intellectual property
owner's perspective, in some cases it may make perfect economic sense to
refuse to license parody or criticism if one fears the effects on one's repu-
tation or sales; at the least, an author might be expected to charge more
for parodies than for other types of licenses-what frequently happens in
parody and criticism cases is that intellectual property owners refuse to
license such works on terms that are substantially equal to the terms they
would give nonparodic uses of similar type and quantity.3"
323. If one is willing to take enough noneconomic factors into account in defining rationality, vir-
tually all refusals to license will be rational. For example, licensing your work to someoneyou do not
like may impose a "psychic cost" that would have to be taken into account in determining the market-
clearing license price.
324. The expansion of intellectual property has led to numerous efforts to control references to,
or criticism of, a person, using trademark law and the right of publicity. See, e.g., Waits v. Frito-Lay,
Inc., 978 F.2d 1093, 1096 (9th Cir. 1992); White v. Samsung Elecs. Am., Inc., 971 F.2d 1395 (9th
Cir. 1992); Midler v. Ford Motor Co., 849 F.2d 460 (9th Cir. 1988). For criticism of this trend, see
White v. Samsung Electronics America, Inc., 989 F.2d 1512 (9th Cir. 1993) (Kozinski, J., dissenting
from denial of rehearing en bane).
325. See, e.g., Gordon, supra note 173, at 1632; Robert P. Merges, Are You Making Fun of Me?:
Notes on Market Failureand the Parody Defense in Copyright, 21 AIPLA Q.J. 305 (1993); Richard
A. Posner, When Is Parody Fair Use?, 21 J. LEGAL STUD. 67 (1992); Winslow, supra note 268;
Landy, supra note 36, at 250-51.
326. An example of a refusal to license a parody on any terms is the Campbell v. Acuff-Rose
Music, Inc. case, in which the copyright holder to Roy Orbison's song Pretty Woman refused to grant
a license for a parodic rap version of the song by 2 Live Crew. Campbell v. Acuff-Rose Music, Inc.,
510 U.S. 569, 572-73 (1994); see also Merges, supra note 325, at 308-09 (noting the irony of subse-
Noneconomic factors may also prevent licensing even in obviously

commercial contexts. Economists may presume that corporations always
act rationally, but any lawyer can recite cases where a corporation refused
to do a deal or settle a lawsuit with a rival for no better reason than the
bad feelings past experience had left. Corporate intellectual property
owners may refuse to license patent rights to a competitor not because
there is anything wrong with the licensing deal, but simply because the
proposed licensee is a competitor. From a social perspective, the effect is
the same regardless of the reason: licensing that "should" take place from
a strict economic perspective does not.
f. Implicationsof Imperfect Licensing.-The material in this sub-

section suggests that, for a variety of reasons, licensing intellectual
property is costly and uncertain. Despite the property rights model of effi-
cient transactions, in the real world one cannot always expect that efficient
transactions will occur. This problem has two sorts of implications for
intellectual property law. First, because the efficiency of intellectual
property is so closely tied to the assumption of efficient licensing, intellec-
tual property law should be structured in such a way as to encourage effi-
cient licensing transactions whenever possible. Second, because efficient
licensing will not always occur, intellectual property law must do more
than determine that someone owns a particular intellectual property right:
it must give some thought to who ought to own that right. Unfortunately,
there is some tension between these goals, as we shall see.
Cooter's basic model of bargaining suggests one way to encourage
more licensing. Cooter suggests that negotiation in voluntary transactions
is about dividing up the surplus profits to be gained from doing the
deal. 3" Many of the difficulties identified in this subsection, notably the
problems of transaction costs and uncertainty, interfere with licensing
because they reduce (in the case of transaction costs) or obscure (in the
quent arguments by the plaintiff that the song parody "undermines economic incentives"). Gordon
refers to this as an "antidissemination motive" and includes it in her catalogue of potential market
failures. Gordon, supra note 173, at 1632-35; see also Landy, supra note 36, at 250-51 (contending
that refusing to grant permission for the use of copyrighted material in parody cases to avoid criticism
causes market failure).
327. See Cooter, Cost of Coase, supra note 317, at 21; see also Merges, supra note 87, at 81
(discussing the division of the "cooperative surplus"). But see Merges, supra note 299, at 1391-93
(suggesting that unified property rights will encourage the development of licensing organizations, such
as ASCAP, designed to overcome the transaction costs of bargaining). The ASCAP model of market
facilitation is based on statistical estimation of the uses that are made of copyrighted works. This
model works well for market transactions which are frequent, low in individual value, and susceptible
to standardization. Copyright licenses to end users may possess those characteristics, but licenses to
likely improvers do not. It is difficult to imagine how ASCAP or its equivalent could come up with
a statistical model and royalty structure that would capture the complexities of potential improvements
to the satisfaction of all parties involved.
case of uncertainty) the size of the surplus profit to be divided. Thus, if

the surplus from a transaction is $500, transaction costs of $400 will make
the deal much less attractive, and transaction costs of $600 will make it
entirely infeasible. Similarly, if each party is uncertain about the value of
the surplus by ± $250, such uncertainty could reduce or even eliminate the
perceived gains from trade. If it is possible to increase the gains from
trade without substantially increasing the transaction costs or uncertainty
associated with trading, doing so should increase the likelihood that effi-
cient bargaining will occur.'s Transaction costs of $400 or even $600
will be much less likely to block a deal with a surplus of $2000 than one
with a surplus of $500.
How can intellectual property law affect the gains from trade? One
way is to increase the costs of failing to come to an agreement. To some
extent this seems to be what the blocking patents rule is designed to do.
As Merges explains, a rule of absolute control by the original inventor over
improvements will leave very little for original inventors and improvers to
bargain over-the original inventor is entitled by law to capture the full
value of the improvement, and the improver's only bargaining leverage is
the threat to withhold the discovery of the improvement, a threat that is
hard to translate into money because of Arrow's paradox.329 The result
is fewer licenses with improvers, but also fewer unsolicited improvements.
Granting patents to both the original inventor and the improver, by
contrast, gives each a much larger stake in the success of the licensing
negotiation. Unless the parties agree, neither party can use the improved
invention. Thus, the blocking patents rule increases the surplus being bar-
gained over by imposing a cost on original inventors who fail to come to
terms with improvers. 3" The analysis of Cooter's model above suggests
that the result of blocking patents doctrine should be an increase in the
number of deals struck between original inventors and improvers.331
328. Such an increase in the surplus will not eliminate the incentive to engage in strategic
behavior, and may even increase it (because more money will be at stake in each transaction). Nor
is it likely to eliminate all noneconomic barriers to trade, though at least some of those barriers should
disappear when a sufficient amount of money is at stake.
329. See Merges, supra note 87, at 80-81.
330. How much the gains from trade increase depends on how important the improvement was
in relation to the original invention. I consider the implications of this fact infra subpart VI(A).
331. In a recent article, Ian Ayres and Eric Talley argue that dividing legal entitlements may
encourage negotiation between the parties owning partial entitlements. See Ian Ayres & Eric Talley,
SolomonicBargaining:Dividinga LegalEntidtlementto FacilitateCoasean Trade, 104 YALE L.J. 1027,
1092-94 (1995) [hereinafter Ayres & Talley, Solomonic Bargaining]. Their argument is not the one
presented in the text-that increasing the gains from trade will increase the incentive to come to terms.
Rather, they suggest that using liability rather than property rules (and thus giving compulsory licenses
to improvers on the payment of a reasonable royalty set by a court) would encourage the parties to
forego strategic behavior in the bargaining process, primarily because it increases uncertainty as to the
outcome of the negotiation. See id. at 1093. For discussion of their model, and of possible counter-
Furthermore, the blocking patents rule will encourage improvers to come

forward with more information than they otherwise would, decreasing the
uncertainty that would otherwise exist about the relative value of the
invention and the improvement.
Unfortunately, dividing the entitlement in order to increase the gains
from bargaining is a risky strategy. While doing so may make licensing
more likely, it cannot guarantee that it will occur.33 And where
licensing does not occur, the result under a blocking patents rule is more
problematic for society than under either of the two possible absolute
property rules-no one gets to use the improvement. This danger suggests
that it is not always appropriate to divide the property entitlement in cases
of improvement. In those cases where the property entitlement is not
divided, who gets the property right (the original inventor or the improver)
may be of considerable importance.
In Parts II and III, I distinguished three classes of improvements,
termed "minor," "significant," and "radical," respectively.333 For
illustration, let us define the value of the original invention or creation as
$200, and imagine a minor improvement worth an additional $10, a signifi-
cant improvement worth an additional $200, and a radical improvement
worth an additional $1200. The copyright system presumptively gives con-
trol over all three classes of improvements to the original creator. By
contrast, the patent law has different rules in each of the three cases, which
roughly correspond to the value of the improvement in relation to the origi-
nal invention. Minor improvements are captured by the original inventor,
vailing factors, see Louis Kaplow & Steven Shavell, Do LiabilityRules FacilitateBargaining?A Reply
to Ayres and Talley, 105 YALE L.J. 221,227 (1995); Ian Ayres & Eric Talley, DistinguishingBetween
Consensualand NonconsensualAdvantagesof Liability Rules, 105 YALE LJ.235 (1995) [hereinafter
Ayres & Talley, Distinguishing].
The recent addition of § 104A to the Copyright Act should provide a test of the Ayres-Talley
thesis. As part of its effort to restore copyright ownership to certain works in the public domain,
Congress provided an exception for good faith creators of derivative works. Those who have created
derivative works based on a once-public-domain work in which copyright has been restored can con-
tinue to produce the derivative work, as long as they pay "reasonable compensation" to the owner of
the restored copyright. 17 U.S.C. § 104A(d)(3)(A) (1994). If the parties cannot agree on compen-
sation, either party may go to court to obtain a court order setting compensation. 17 U.S.C.
§ 104A(d)(3)(B) (1994). If Ayres and Talley are correct, this use of liability rules should reduce stra-
tegic behavior among parties negotiating "reasonable compensation."
I discuss objections to the use of liability rules in intellectual property cases in more detail infra
notes 360-62 and accompanying text.
332. Indeed, Michael Heller has suggested that splintering a property entitlement among too many
factions could lead to a "tragedy of the anticommons," in which the parties all hold out and so refuse
to come to agreement. See Michael Heller, Property in Transition 3 (Nov. 12, 1996) (unpublished
manuscript, on file with the Texas Law Review). Heller does note, though, that increasing the cost of
holding out is more likely to bring the parties to agreement. See id. at 49; see also DOUGLAS
LAYCOCK, MODERN AMEImcAN REMEDIES 371 (2d ed. 1994) (suggesting that "bilateral monopoly"
increases transaction costs over multi-party bargaining cases).
333. See supra notes 78-92, 148-68 and accompanying text.
just as in copyright law, while radical improvements are the property of the
improver.' (The rule for significant improvements in patent law is
somewhere in between-the parties divide the property entitlement.335)
Not only is the radical improver excused from patent liability to the origi-
nal inventor under the reverse doctrine of equivalents, but she can pre-
sumably patent her improvement, thus excluding the original inventor (and
everyone else) from practicing the improvement without a license.
The copyright approach of assigning a property right in radical
improvements to the original creator will produce an efficient result in
circumstances where the parties can be expected to come to terms-the
copyright owner will presumably license the right to make the valuable
improvement to the radical improver. If licensing is imperfect, however,
patent law's treatment of radical improvers makes much more sense than
copyright's.33 6 If the parties cannot be expected to come to terms in all
cases, a default rule giving all rights in a radical improvement to the origi-
nal intellectual property owner is likely to encourage overinvestment in
initial works and underinvestment in radical improvements. To borrow an
example from Grady and Alexander,
Allowing someone who has invented a crude telephone to control the
entire communications industry creates such a large reward that
inventors would dissipate rents by trying to be the pioneer. In the
perverse equilibrium that would result from a system awarding full
control to the inventor who is first, the costs of developing dreams
that ultimately fail would equal or exceed the benefit to society of
those that succeed.337
334. While Merges refers to this as an exception to property-rule treatment of intellectual property
rights, see Merges, supra note 154, at 2668, that characterization seems inaccurate to me. Radical
improvers can sell their improvement and prevent others from doing so; original inventors have no
power over them. This is as much a property rule as giving original inventors the same power would
be, though of course the party holding the property right is different.
336. Mazzoleni and Nelson reach a similar conclusion:
If one assumes that, in general, potential licensees and patent holders have little difficulty
in reaching a license agreement (that is that the transaction costs of patent licensing are
small) then one may take a relatively relaxed view of the costs of granting a large prospect
controlling patent, even when one believes that potential explorers of the prospect are
diverse in terms of what they would do. On the other hand, if one believes that the
transaction costs often are high, and patent holders are prone to litigation, one is less
sanguine about this.
Mazzoleni & Nelson, supra note 283, at 20.
337. Grady & Alexander, supranote 13, at 307-08. While Grady and Alexander offer Bell's tele-
phone patent as an example, perhaps a better example is the crude incandescent light bulb patent issued
to Sawyer and Man. See The Incandescent Lamp Patent, 159 U.S. 465 (1895). If a copyright rule
were to prevail, Sawyer and Man would have had the right to prevent Edison from marketing his
greatly improved light bulb.
At the same time, potential radical improvers would have little or no

incentive to improve on original works, since the original intellectual
property owner would ultimately be able to control their improvement in
the case of bargaining breakdown.
The reverse doctrine of equivalents in patent law provides a way to
avoid this problem. In the small subset of cases where the improver's
invention is such a major advance over the original invention that its value
is much greater than the original, the incentive theory of intellectual
property suggests that we should be more concerned about encouraging
improvements than we are about encouraging original inventions. While
it is true that the reverse doctrine of equivalents reduces the incentives for
original inventors somewhat from what they would be under a copyright-
style regime, the loss is small. 38 and it is outweighed by the encourage-
ment of improvements which are themselves more valuable than the
original.
Why then have a blocking patents rule at all? Why not simply divide
the world into "minor" and "radical" improvements, and assign property
rights based on whether the pioneer's or the improver's invention is more
valuable? I suggested one justification for the blocking patents doctrine
above-it may encourage efficient licensing transactions in situations where
they otherwise would not occur due to market imperfections. A second
reason is that in the real world it is not simple to determine even the rela-
tive value of original inventions and improvements. Some improvements
are clearly minor in relation to the underlying invention; others are clearly
radical. In between, there may be a significant degree of uncertainty as to
which invention is actually more valuable. In such a case, where the
proper default rule is unclear, the intermediate approach of the blocking
patents doctrine seems appropriate.
3. Market Power and Holdup Problems.-A third set of problems

with efficient licensing of intellectual property has to do with the relation-
ship between intellectual property rights and market power. Fundamental
economic theory instructs us that consumers of a product (and improvers,
who are after all consumers too) will make efficient use of that product
only if it is priced at marginal cost.339 It also tells us that producers will
price at marginal cost only if they are forced to by the existence of compe-
tition. A producer who controls a market will cut output and raise prices,
increasing its profits but reducing both consumer and aggregate social
338. See Merges, supra note 87, at 101-03 (arguing that ex ante incentives to invent will hardly
be affected by the reverse doctrine of equivalents, since its application in any particularcase is highly
contingent and extremely unlikely).
339. See PAUL A. SAMUELSON, ECONOMICS 470 (11th ed. 1980).
welfare.' Because intellectual property rights generally give their

owner the power to prevent competitors from using the technology that is
the subject of the right, one might conclude that intellectual property rights
confer a "monopoly" on their owner."l If so, intellectual property rights
should be expected to result in prices that are above marginal cost, and
therefore in an inefficiently low use of the patented technology.' 4
It is well established, however (at least among economists), that
intellectual property rights do not automatically confer market power" 3
Patents confer exclusive rights to use a particular technology, but that
technology need not be coextensive with an economic market. Patented
technologies can and frequently do compete with each other, and that com-
petition will constrain monopoly pricing. Competition is even more likely
in the case of other intellectual property rights, since those other rights do
not restrict all uses of the intellectual property by competitors.'
Unfortunately, it is possible to take this important insight too far.
The prospect model of the patent system works best under the opposite
assumption, sometimes made by commentators-that intellectual property
rights can never confer market power on their owner.' This assumption
is wrong, however, as even Kitch acknowledges.' Intellectual property
rights can confer market power in certain circumstances; think of Morse's
telegraph patent, 7 for example, or Bell's telephone patent,'" both of
which controlled the most effective means of distance communication that
existed at the time. Further, intellectual property has the potential in cer-
tain specialized circumstances to give its owner market power that extends
beyond the scope of the intellectual property right itself. This is particu-
340. See id. at 470-71.

341. Several early court decisions came to precisely this conclusion. See supra note 26.
342. See Fisher, supra note 190, at 1700-02 (explaining how an intellectual property owner with
market power can charge supracompetitive prices).
343. See supra note 26.
344. For example, copyright permits independent derivation in most circumstances, see Gaste v.
Kaiserman, 863 F.2d 1061, 1066 (2d Cir. 1988), and permits some forms of reverse engineering in
computer software cases. See infra note 415. Trade secrets law permits both independent discovery
and reverse engineering. See UNIF. TRADE SECRETS ACr § 1, cmt. (amended 1985), 14 U.L.A. 438
(1990); RESTATEMENT (THIRD) OF UNFAIR COMPETMON § 43 (1995).
345. Judge Easterbrook, for example, suggests that patent's "right to exclude is no different in
principle from General Motors' right to exclude Ford from using its assembly line." Easterbrook,
supranote 271, at 109; see also Dam, supra note 305, at 249-50 ("[W]ithout the benefit of empirical
research, it is entirely plausible to conclude that in the great bulk of instances no significant market
power is granted [by patents]."). Dam's assertion may be correct, though he offers no reason to
believe it is. Judge Easterbrook's statement is clearly wrong-patent law extends far beyond the use
of a certain set of tangible assets, and indeed extends beyond the realm of copying of ideas, as we have
seen.
346. See Kitch, supra note 13, at 274-75.
347. See O'Reilly v. Morse, 56 U.S. (15 How.) 62, 64-65, 124 (1853).
348. See The Telephone Cases, 126 U.S. 1, 534 (1887).
larly likely where there are strong standardization effects in a market, so

that a standard once adopted is likely to be durable. Intellectual property
rights reinforce market power in such circumstances. If the intellectual
property right includes copyright-style control on the making of
improvements, it may make it difficult or impossible for competitors to
break free of the standard even after the standard has become obsolete.3 9
Original intellectual property owners in such a market are in a position
to hold up the innovation process. Because improvement in such a market
virtually requires the improver to build on the technology that has become
a market standard, the owner of the standard can preclude any such
improvements, thus maintaining her market share. 3' At the very least,
she may be expected to charge supracompetitive prices to improvers who
wish to license her intellectual property. Basic economic theory suggests
that the effect of charging "monopoly license fees" will be that consumers
of those licenses (improvers) will not buy enough of them. Thus,
monopoly pricing for licenses to improve original works should result in
underproduction of improvements.
Patent law's reverse doctrine of equivalents offers a way out for
improvers in such a market. Under this doctrine, original inventors cannot
prevent improvers from using their basic invention where the improvement
is so radical that it fundamentally changes the nature of the invention. By
permitting radical improvers to appropriate the original invention so long
as their contribution is sufficiently radical, the reverse doctrine of equiva-
lents provides at least some relief from the holdup problem.351
349. For more on such markets and the effect of intellectual property, see Mark A. Lemley,
Antitrust and the Internet StandardizationProblem, 28 CONN. L. REV. 1041 (1996). For an example
of how the copyright rule can produce such a result, see infra notes 403-16 and accompanying text.
350. Whether it is economically rational for original inventors to suppress such improvements,
rather than licensing them for a fee, is a matter of considerable debate among economists. For histori-
cal evidence of such suppression, see Mark Clark, Suppressing Innovation: Bell Laboratoriesand
Magnetic Recording, 34 TEcH. & CULTURE 516, 532 (1993), and Richard Dunford, Suppression of
Technology as a Strategy for Controlling Resource Dependence, 32 ADMIN. SCI. Q. 512 (1987).
While it might seem irrational to think that an original inventor would suppress an improvement within
her control if it truly was valuable, several circumstances might induce her to do so. If the improve-
ment requires a new manufacturing technology or a different market approach, there may be substantial
fixed costs associated with switching over production from the old to the new way. The further
removed the improvement is from the original invention, the worse this problem is likely to be. Cf.
Teece, supra note 280, at 288, 288-90 (suggesting that much of the profit from innovation may come
from a firm's ability to capture "complementary assets" that relate uniquely to the particular technology
used). The alternative to switching over production facilities-licensing the improver to compete with
the original inventor, for a fee-is also unlikely to be attractive to the original inventor. Even if the
licensor could extract the full value of the improvement in a licensing transaction, which seems
unlikely, its market control will disappear along with the intellectual property right. Innovationis asso-
ciated with strong first-mover advantages, so the first company to manufacture and sell a product is
likely to maintain its dominant position even after the patent expires. Thus, the improver, rather than
the original inventor, would likely come out the victor in the long run.
351. Other doctrines common to both patent and copyright cases, notably the misuse defenses, may
also be called upon to provide relief in this situation. See Dam, supra note 305, at 260-61; see also
C. Conclusions
The copyright rules governing improvements closely track the property
rights model of innovation, under which a strong unified grant of property
rights leads to efficient licensing of the invention and efficient incentives
to produce improvements. Unfortunately, the property rights story is
incomplete. Problems of imperfect information, transaction costs, strategic
behavior, and market power all impose barriers to the hypothetical efficient
license. 52
There is no reason to believe that these problems are uniquely con-
fined to patent rather than copyright licensing. If anything, some of the
problems that prevent efficient licensing are worse in the copyright context
than in the patent context. There are more potential authors and artists
who could improve a particular work than there are engineers who could
improve a particular invention, largely because the degree of specialization
required is less in the artistic world than in engineering. Thus, the prob-
lem of identifying improvers is at least as large for copyrighted works as
it is for patented works. Similarly, the uncertainty associated with the
success of a work of art or literature is probably greater than with
inventions, which at least function (or fail to function) according to defined
principles. While the bargaining costs per transaction are probably less in
absolute terms in copyright licenses than in patent licenses, the underlying
value of the transaction is also less. There is no reason to believe that
transaction costs as a percent of value are any less in the copyright context.
Indeed, for at least one important class of copyrighted works (multimedia
works), the transaction costs of the required licensing threaten to stifle an
entire industry.3 53 Finally, copyright owners are more likely than patent
Cohen, supra note 309, at 1181-98 (suggesting that such holdups can be cured by applications of the
misuse doctrines). However, applying antitrust and related doctrines to achieve structural relief in stan-
dardized markets may be futile, as I have suggested elsewhere. See Lemley, supra note 349, at 1073.
352. I am sensitive to Wendy Gordon's concern that "occasional cases of market failure should
not immediately trigger judicial exceptions," given the high administrative cost of case-by-case
decisionmaking. Gordon, supra note 258, at 474. Nonetheless, two factors lead me to believe that
varying property rights in accordance with the patent model is the more appropriate solution. First,
that variance operates by dividing cases into classes, rather than treating each one individually, a fact
which should reduce the administrative costs of decisionmaking. Second, I do not believe the circum-
stances detailed in this Part can be dismissed as "occasional" instances of market failure.
353. See generally Robert D. Sprague, Multimedia: The Convergence of New Technologies and
TraditionalCopyrightIssues, 71 DENV. U. L. REV. 635, 659-69 (1994) (examining the effects of cur-
rent copyright law when applied to multimedia); Daun, supra note 301, at 217, 217-18 ("When the
product's content consists of hundreds or even thousands of separate items of intellectual property, the
tasks and costs associated with negotiating and administering individual licenses can be overwhelming
and even paralyzing."); Heather J. Meeker, Note, Multimedia and Copyright, 20 RUTGEes COMP. &
TECH. L.J. 375, 397-400 (1994) (arguing that the difficulty in persuading copyright holders to license
their intellectual property is a contributing factor to the slow growth of the multimedia industry).
owners to object to uses of their work for noneconomic reasons, perhaps

because they are more personally involved with their works.35" Thus,
there is no reason to think that copyright owners are better able to
overcome these licensing problems than patent owners.
Patent law seems better equipped than copyright law to deal with these
real-world barriers to efficient licensing. The blocking patents and reverse
doctrine of equivalents rules ameliorate licensing problems in comple-
mentary ways. The doctrine of blocking patents helps to overcome search
problems by giving actual improvers a limited property right in their
improvement, thus encouraging them to self-identify to original inventors.
By contrast, because of Arrow's information paradox, copyright law must
rely on the ability of original creators to effectively identify potential
improvers. As a result, the blocking patents doctrine should reduce the
effects of transaction costs and uncertainty relative to the copyright rule.
An existing improvement created by a known improver can be valued much
more easily than a hypothetical improvement. Further, to the extent that
transaction costs and uncertainty threaten to undermine licensing
transactions that are only marginally value-producing, the blocking patents
rule may actually encourage agreement-not by reducing transaction costs
or uncertainty, but by increasing the value to both parties of coming to an
agreement.355 The blocking patents rule may also reduce incentives to
engage in certain sorts of strategic behavior.
Some recent scholarship has endorsed the use of liability rules as a
means of forcing revelation of preferences during bargaining.356 Even
if these scholars are correct that uncertainty as to legal entitlements will
force parties to reveal their preferences, that does not necessarily suggest
that a liability rule is preferable to a property rule. Rather, in such
354. See generally Margaret Jane Radin, PropertyandPersonhood,34 STAN. L. REv. 957(1982)
(suggesting that the personal economievalue of some forms of property predominates over its economic
value); see also Neil Netanel, Copyright Alienability Restrictions and the Enhancement of Author
Autonomy: A Normative Evaluation, 24 RUTGERS L.J. 347, 422-23 (1993) (applying Radin's theory in
the copyright context).
355. See Hardy, supra note 264, at 239 ("The 'absolute' amount of transaction costs, however,
does not provide a complete picture. We really care about 'high' or 'low' costs relative to the value
of the contract undertaken. That is, the important figure is the ratio of the transaction costs to the value
of the rights being sought."). In the context I am discussing, I would amend Professor Hardy's analy-
sis only slightly, to suggest that the relevant question for our purposes is the ratio of the transaction
costs to the potential value added by the licensing transaction.
356. See, e.g., Ayres & Talley, Solomonic Bargaining, supra note 331, at 1032-33, 1029-30
(asserting that "liability rules possess an 'information-forcing' quality" that "may induce more
contracting and more efficient contracting"); Netanel, supra note 12, at 334-35 & nn.247-48
(discussing Ayres and Talley's Solomonic Bargaining). It is interesting that the GATT Agreement on
Trade-Related Aspects of Intellectual Property Rights (TRIPs) contemplates a compulsory license in
the situation of significant or radical improvement. Agreement on Trade-Related Aspects of Intellectual
Property Rights, Apr. 15, 1994, art. 3(L)(i).
circumstances a divided property 357 entitlement such as the one patent law
offers should have similar effects.
The reverse doctrine of equivalents serves as an efficient default rule.
It reflects a recognition that real-world economic problems will prevent
efficient licensing from taking place in some instances. Where efficient
licensing does not occur, the reverse doctrine of equivalents ensures that
it is the radical improver, rather than the (relatively more modest) original
inventor, who retains a property right in the improvement. The reverse
doctrine of equivalents also serves to counteract some of the holdup prob-
lems introduced-by intellectual property rights which happen to confer
market power. It does this by allowing a radical improver to use an origi-
nal inventor's idea in certain circumstances in which the original inventor
could otherwise have held the radical improvement hostage.
The basic structure of the patent model is therefore quite simple. The
treatment of improvements is a function of the value and significance of the
improvement in relation to the original invention. Improvements which are
minor in relation to the original invention are likely to be found to infringe,
either literally or under the "insubstantial differences" test for doctrine of
equivalents infringement."' More significant improvements within the
range of the original claims still infringe the original patent, but in such
cases the improver is entitled to a patent of his own. This not only encour-
ages the development of significant improvements, but gives improvers bar-
gaining leverage in licensing negotiations in direct proportion to the relative
value of their improvements. Finally, truly radical improvements are ex-
empted from liability to the original patent owner, regardless of whether
they fall within the literal scope of the original claims, in order to encour-
age such improvements and to ensure that they reach the market.359
357. See Kaplow & Shavell, supra note 331, at 222 n.5. Ayres and Talley acknowledge that
divided ownership of property entitlements may sometimes serve the same preference-revealingfunction
as liability rules. Ayres& Talley, SolomonicBargaining,supra note 331, at 1072-82,1116-17. They
identify several types of property division, including probabilistic assignment of property rights and
division into scope of possible uses. See id. Arguably, blocking patents involve both division of
property rights into different possible uses (since the original inventor retains the old use, and the
improver owns the right to the new use) and probabilistic assignment of property rights (since the exis-
tence of blocking patents will depend on an administrative and/or judicial assessment that both inven-
tions warrant patent protection, and that the patents block). While Ayres and Talley argue that a com-
pulsory license should replace the divided property entitlement of blocking patents, see id. at 1092-94,
they appear to concede that a divided property entitlement is preferable to an undivided one for their
purposes.
358. See Hilton Davis Chem. Co. v. Wamer-JenkinsonCo., 62 F.3d 1512, 1517 (Fed. Cir. 1995)
(en bane), rev'd on other grounds, 117 S. Ct. 1040 (1997). Judge Newman suggests that making
minor improvers liable for infringement under the doctrine of equivalents will not only serve the
function of encouraging original inventors, but may also encourage better improvements, since
improvers will be forced into "leapfrogging" rather than incremental advances over the prior art. See
id. at 1532-33 (Newman, J., concurring).
359. This model is in line with the conclusions Merges and Nelson draw regarding patent law.
See Merges & Nelson, supra note 32, at 909 ("Once a court completes its assessment of the
The model also has advantages over the compulsory-licensing schemes

that are sometimes proposed to limit the property rights of intellectual
property owners." Unlike the compulsory licensing approach, the
model still treats licensing as something to be worked out in the market,
rather than by the courts. It merely suggests a set of default parameters
that will maximize the chance that bargaining will occur, and minimize the
damage if it does not occur. It is therefore consistent with the warnings of
Wendy Gordon and others that judicial or administrative intervention is
doubtful as a way to reach efficient results,361 and with the suggestion of
Kaplow and Shavell that property rules may be appropriate in this
situation. 62
This simple model is complicated to some extent by the uncertainty of
the litigation process. Unlike the licensing uncertainty discussed earlier,
the uncertainty here is over how the Patent Office and/or the courts will
evaluate a party's claim to be a significant or radical improver. Because
courts may imperfectly identify such improvers, the value calculations
engaged in by parties to a licensing transaction should, strictly speaking,
be modified to take account of the probable success of a legal claim to
significant or radical improver status. In theory, the patent rule ought to
be significantly more uncertain than the copyright rule, which requires no
such judicial determination. Arguably, this is a disadvantage to the patent
rule relative to the copyright rule.3
In practice, however, litigation uncertainty should present problems
for licensing transactions only to the extent it is skewed in favor of one
side or the other. There is no reason to expect such asymmetry in this
case. Absent asymmetry, litigation uncertainty is at least as likely to
encourage a licensing transaction (for example, where both parties are risk
averse) as it is to discourage one. Further, if one accepts Ayres and
Talley's view that uncertainty as to litigation outcome reduces the likeli-
significance of the patented device, it should consider in addition the importance of the advance
represented in the accused device." (emphasis in original)).
360. See Fisher, supra note 190, at 1725-26; Merges, supra note 325, at 307-12.
361. See Gordon, supra note 268, at 868-69. While high transaction costs have been offered as
a justification for the imposition of liability rules (such as compulsory licensing in the intellectual
property context), Merges has persuasively argued that liability rules may not be justified in intellectual
property cases to the extent that private organizations will efficiently contract into liability rules
themselves. See Merges, supra note 299, at 1391-92; Merges, supra note 254, at 2662.
362. See Kaplow & Shavell, supra note 254, at 762 (arguing that property rules are superior to
liability rules when the initial possessors of a thing place greater idiosyncratic value on it than those
who take it from them and when courts have difficulty in valuing the thing). For the reasons discussed
in the last section, both of these circumstances are likely in the case of intellectual property. Of course,
deciding that a property entitlement is appropriate does not answer the question of who should receive
that property entitlement. See supra note 334.
363. C. Merges, supra note 299, at 1306-07 (preferring undivided property entitlements to
liability rules in part because of their lower administrative cost).
hood of strategic behavior in bargaining, 3" the litigation uncertainty

encouraged by the patent model is actually a feature of the system, not a
bug.
VI. Implications of the Improvements Model

Application of the improvements model just described is generally
consistent with existing patent law doctrine. However, its application in
the copyright law context would necessitate some changes to existing law,
at least as currently interpreted by the courts. In this section, I consider
how the law could be made consistent with the model, and offer some
examples of how the new rules would work.
A. Patent Law
While the model presented in Part V is derived from patent law
doctrine, and therefore is consistent with that doctrine, it does have
implications for one controversial aspect of patent law: the treatment of
pioneer inventions. A pioneer invention is "a wholly novel device, or one
of such novelty and importance as to mark a distinct step in the progress
of the art, as distinguished from a mere improvement or perfection of what
had gone before."365 Other courts have used various terms in an attempt
to distinguish pioneering from normal inventions. 3" However defined,
the basic idea is that pioneer inventions are somehow more fundamental
than the average patentable invention. They open up a new industry, or
at least a new avenue of approach.367 Under long-standing judicial
doctrine, pioneer patents receive a broader range of protection under the
doctrine of equivalents than other patented inventions." 8 Further, the
range of equivalents is calibrated even more finely than that, with
the effective scope of a patent depending on the significance of the
patented invention."6 While some commentators have defended the
364. See supra notes 331-32 and accompanying text (discussing the Ayres-Talley model).
365. Westinghousev. Boyden Power Brake Co., 170 U.S. 537, 561-62 (1898).
366. See, e.g., Sun Studs, Inc. v. ATA Equip. Leasing, Inc., 872 F.2d 978, 987 (Fed. Cir. 1989)
(using the term "broad breakthrough'); Universal Gym Equip., Inc. v. ERWA Exercise Equip. Ltd.,
827 F.2d 1542, 1548 (Fed. Cir. 1987) (using the term "major advance'); see also John R. Thomas,
The Question Concerning PatentLaw and PioneerInventions, 10 HIGH TECH. L.J. 35, 48 (1995)
(collecting cases describing pioneer patents in various ways).
367. It is important to distinguish pioneer inventions, as that term is used in patent law, from the
"original inventions" at issue in this Article. My model assumes a "first" invention and subsequent
improvements on that invention purely for the sake of simplicity. In practice, the "original' invention
in the model is likely itself to be an improvement on what came before it, and it may or may not be
a pioneering invention.
368. See Graver Tank & Mfg. Co. v. Linde Air Prods. Co., 339 U.S. 605, 608 (1950).
369. Thus, inventions that are important but not pioneering are entitled to fairly broad scope,
though less than pioneering patents. See Hughes Aircraft Co. v. United States, 717 F.2d 1351, 1354
doctrine3 " and even suggested ways to extend it,37 others have
attacked some of its implications. 3" The Federal Circuit cast doubt on
the status of the doctrine in its 1995 en bane decision in Hilton Davis,
which reformulated the doctrine of equivalents in a way that did not
include the pioneer status of the invention."
The pioneer invention doctrine is entirely consistent with the model of
improvements derived in Part V. That model identifies the relative impor-
tance of the original invention and the improvement as the critical factor
in determining how the improvement will be treated once it is determined
to be infringing?7" This approach implicitly accounts for pioneering
inventions in one respect: the significance of the original invention will
help determine how important an improvement must be to qualify as
"significant" or "radical." An improvement worth $500 may be radical if
it is made to an original invention worth only $100, but not if made to
an original invention worth $2000. The fundamental insight of the
pioneer invention doctrine is the same as that of the improvements
model-efficiency is best served by some sort of calibration, however
rough, between the importance of the invention and the scope of the patent.
B. Copyright Law
Copyright law would need to change in two ways to accommodate the
model. First, the current copyright rules regarding the ownership of
derivative works are inconsistent with the model's treatment of significant
improvements. Second, as currently interpreted, copyright's fair use
doctrine does not provide effective protection for radical improvers in all
cases. In this subpart, I suggest how copyright law could be modified to
(Fed. Cir. 1983). Inventions that are narrow advances within a highly developed art receive a very
limited range of equivalents. See Thomas & Betts Corp. v. Litton Sys., Inc., 720 F.2d 1572, 1580
(Fed. Cir. 1983).
370. See, e.g., Esther Steinhauser, Note, Using the Doctrine of Equivalents to Provide Broad
ProtectionforPioneerPatents:Limited ProtectionforImprovementPatents, 12 PACE L. REV. 491,494
(1992) (advocating a "return to the original raison d'etre for the doctrine of equivalents by providing
broad protection for pioneer patents").
371. See, e.g., A. Samuel Oddi, Beyond Obviousness: Invention Protection in the Twenty-First
Century, 38 AM. U. L. REv. 1097, 1128-37 (1989) (proposing a new "revolutionary patent" for pio-
neer inventions that would have broader scope and extended duration); Harold C. Wegner, Equitable
Equivalents: Weighing the Equities to Determine Patent Inftingement in Biotechnology and Other
Emerging Technologies, 18 RUTGERS COMP. & TECH. L.J. 1, 46 (1992) (recommending that pioneer
status be relevant in deciding whether to apply the doctrine of equivalents at all).
372. See, e.g., Thomas, supra note 366, at 95 (suggesting that pioneer status should depend on
the social implications of a new invention, rather than merely its technological significance).
373. Hilton Davis Chem. Co. v. Wamer-JenkinsonCo., 62 F.3d 1512, 1519-20 (Fed. Cir. 1995)
(en banc) (listing other relevant factors, including "known interchangeability of the accused and claimed
elements," copying, "designing around," and independent development), rev'd on other grounds, 117
S. Ct. 1040 (1997).
374. By contrast, the pioneer invention doctrine works towards the same goal as the model in a
different way, by basing the infringement determination itself in part on the significance of the patent.
fit the improvements model, and offer some examples of how such modifi-
cations would affect actual copyright cases.
1. Derivative Work.-Current copyright doctrine effectively assigns

the rights in unlicensed improvements made by third parties to original
creators, by denying the improver copyright protection in his original
expression and declaring any creation of such derivative works to be
infringement.3 7 The rule that derivative works infringe the copyright in
the original stems from section 106(2) of the Copyright Act,376 and is
consistent with the rule in patent law that even significant improvements
can infringe the patent if they fall within the scope of the claims. The rule
precluding improvers from owning their original expression stems from
section 103(a)," and is inconsistent with the blocking patents rule. For
the reasons explained in Parts IV and V, the copyright rule cannot be
justified, and should be replaced by a rule permitting improvers to own
their original expression. 7 8 Such a rule would protect substantial
improvers-those who have contributed significant copyrightable expression
of their own. It.will not protect minor improvers, however.379
375. See supra notes 152-62 and accompanying text (discussing copyright's treatment of derivative
works).
376. 17 U.S.C. § 106(2) (1994).
377. Section 103(a) provides that "Ithe subject matter of copyright as specified by section 102
includes compilations and derivative works, but protection for a work employing preexisting material
in which copyright subsists does not extend to any part of the work in which such material has been
used unlawfully." 17 U.S.C. § 103(a) (1994). Based on the languageof this section, it is possible for
infringing improvers to copyright their derivative works so long as the infringing material is separable
from the original expression. Thus, an infringing song on a CD of otherwise original songs will pre-
sumably not bar copyright in the original songs. But when the original and derivative contributions
are intertwined, as is the case in many derivative works (such as movies based on books, translations,
and computer programs), the improver loses any claim to his original contributions to the derivative
work.
378. Goldstein has taken a similar position. He writes:
The [derivative works] rule is, however, hard to justify when applied to derivative works
such as the motion picture in Sheldon v. MGM in which the underlying work represents
only a small part of the value of the derivative work but, because it underlies the whole,
will defeat copyright protection for the entire derivative work.
Goldstein, supra note 37, at 244; see also Gordon, supra note 268, at 865-66 (criticizing the broad
control copyright owners have over derivative works); Gordon, supranote 162, at 10 19-20 (advocating
a patent-like model of improvements for copyright); Nadan, supra note 146, at 1660-72 (offering an
alternative proposal to recognize component works as noninfringing and separately copyrightable).
379. Courts have required that substantial original material be contributed by an improver in order
to make the derivative work separately copyrightablerand therefore eligible for the proposed blocking
copyrights rule. See, e.g., Gracen v. Bradford Exch., 698 F.2d 300, 305 (7th Cir. 1983); Durham
Indus., Inc. v. Tomy Corp., 630 F.2d 905, 910 (2d Cir. 1980); L. Batlin & Son, Inc. v. Snyder, 536
F.2d 486, 489-92 (2d Cir. 1976); cf. Landes & Posner, supra note 255, at 356 (justifying this require-
ment on the grounds that it would otherwise be difficult to distinguish original from derivative works).
But see Goldstein, supra note 37, at 243 (criticizing the heightened originality standard for derivative
work copyrights).
Netanel objects that creating a "blocking copyrights" doctrine will

have little practical effect, since the original copyright owner will still be
able to bring suit for infringement against the creator of derivative
works." s It is certainly true that a blocking copyrights rule will not give
substantial improvers freedom to act. Nonetheless, it will significantly
improve their bargaining position vis-A-vis original creators in at least two
38
ways. '
First, it will give improvers something to bargain with, particularly
where the value of their improvement is significant. As an example, con-
sider the owner of a copyright in a book and the "improver" who makes
a movie based on the book. Under the current copyright regime, the book
owners have no incentive to bargain with those who have created a movie
based on the book, since those derivatives effectively belong to the book
owners already. 3' A blocking copyrights rule will force original creators
to deal with improvers if they wish to use the improvement. Even if the
book owners are willing to go to the trouble of creating their own movie,
a blocking copyrights rule will place certain constraints on their ability to
do so, since substantial similarity between the original expression added by
the two movies may lead to a finding that the book owners have infringed
the movie-maker's copyright.3s Both effects will encourage the book
owners to bargain with those who have already added to their original
work.
An example of how such a system might work can be seen in the case
of Harper Bros. v. Klaw.3 In that case, the plaintiff (Wallace) owned
380. See Netanel, supra note 12, at 318 n. 155 (suggesting that "by making unauthorized derivative
works an infringement and subjecting infringing derivative authors to property rule damages and
injunction, the Copyright Act so substantially impairs such derivative authors' incentive to create such
new works that depriving them of their own copyright, while probably overkill, has little additional
incentive-impairing effect").
381. See Reichman, supra note 204, at 815 (suggesting that broad derivative rights protection
allows original developers of computer programs to capture functional attributes of those programs).
383. See Harper Bros. v. Klaw, 232 F. 609, 610 (S.D.N.Y. 1916) (finding that a dramatic work
produced by a licensee based on a novel "follows the novel very closely, and it would (I think) be quite
impossible to make another play that really told the story of Ben Hur without presenting substantially
the same sequence of ideas as is presented by the copyrighted version"); cf. Cohen, supra note 44
(distinguishing similarity probative of copying and similarity constituting unlawful appropriation). Of
course, if it truly is impossible to make another movie version of the book without using material from
the first adaptation, the doctrine of merger might come into play to justify the use by the book owner
of expression from the movie. See Morrissey v. Procter & Gamble Co., 379 F.2d 675, 678-79 (Ist
Cir. 1967) (maintaining that the merger doctrine prevents copyright protection when granting a copy-
right would exhaust all future use of the substance of the forms). Because of the merger doctrine, it
should be difficult for opportunistic "improvers" to interfere with a company's existing plans to create
derivative works (for example, by being first to market a line of stuffed animals based on a new Disney
film).
384. 232 F. 609 (S.D.N.Y. 1916). Because there is no blocking copyrights rule in current law,
this case is not a perfect example. The dual injunctions in the case were the function of a prior license
agreement between the parties. See id. at 610.
the copyright in the 1899 novel Ben Hur. Wallace had licensed to Harper
Brothers the rights to produce a dramatic version of the novel for the
purposes of stage production.3" After the development of motion
pictures, both sides to the license claimed that they owned the exclusive
rights to produce a screenplay version of Ben Hur. The court found that
Harper had not been granted the right to produce a screenplay, but that
Wallace could not do so without infringing on the rights that had been
licensed to Harper.386 Noting that "it does not always follow that,
because one owns a certain thing, he may use it to the detriment of
another," the court enjoined both parties from producing a screenplay
without the agreement of the other.3" This dual injunction-and the
obvious incentive it provides for the parties to come to terms-are the
essential features of the blocking patents regime.
A blocking copyrights rule may also relieve pressure on improvers in
a second way: by reducing the incentives for opportunistic copyright suits
by plaintiffs claiming, for example, that a successful movie is based on
their book or screenplay. Under current law, plaintiffs may be enticed to
sue by the knowledge that if they are successful in asserting that the movie
is a derivative work, they will effectively own the movie. Examples of
suits in this category include Litchfield v. Spielberg,85 where the plain-
tiffs claimed that the movie E.T. was based on her screenplay Lokey from
Maldomar; Sheldon v. Metro-Goldwyn Pictures Corp.,389 where the
plaintiffs successfully alleged that the movie Letty Lynton was based upon
their play; and Abend v. MCA,390 where the plaintiffs sought to terminate
defendant's right to distribute the movie Rear Window, which was based
on a license from their book.
Some suits of this sort undoubtedly have merit-indeed, some have
been successful-but others are brought primarily in the hope that the
plaintiff will capture the value contributed by the defendants, even though
the value of the improvement vastly exceeds the value contributed by the
original work. 39 1 A blocking copyrights rule will not eliminate such
suits, whether baseless or not-plaintiffs may still hope to get money out
of a defendant through the threat of an injunction, or by an apportionment
of the profits that the court decides is attributable to the original work.3'
385. Id.
386. Id.
387. Id. at 613.
388. 736 F.2d 1352 (9th Cir. 1984).
389. 309 U.S. 390 (1940).
390. 863 F.2d 1465 (9th Cir. 1988), aff'd sub nom. Stewart v. Abend, 495 U.S. 207 (1990).
391. By citing these cases as examples, I do not intend to suggest :hat any of these plaintiffs acted
in bad faith, nor to pass on the merits of their claims. I merely identify them as representative of the
sort of lawsuit that is likely under the current rule.
392. For example, in Sheldon, the plaintiffs were entitled to enjoin the defendants from marketing
their film, even though the court found that 80% of the value of the film was contributed by the defen-
But it should reduce the incentive to bring such suits; plaintiffs cannot hope
to capture the value of the defendant's work directly, because the defendant
would own the copyright in her contributions.
2. Fair Use.-Copyright's fair use doctrine has the potential to

protect radical improvers by exempting them from infringement liability.
That potential is evident in the "transformative use" doctrine which has
recently gained currency among courts and commentators. Unfortunately,
for the reasons explained in Part 1I, the fair use doctrine has generally
not been interpreted in a way that would effectively protect radical
improvers. 3" The problem has been the tendency of the courts to focus
primarily on market harm to the copyright owner, to the exclusion of all
else.3"
Reformulating the fair use doctrine to protect radical improvers does
not require legislative change, or even the repudiation of prior court
precedent. Nor does it mean that fair use will be available only to radical
improvers, thus depriving other forms of copying or adaptation of protec-
tions they have long expected.395 What it requires is a more careful
balancing of the relative contributions made by the original copyright
owner and the improver. The transformative use doctrine already recog-
nizes contributions made by improvers as an element favoring a finding of
dants themselves. Sheldon, 309 U.S. at 408-09 (upholding apportionment of profits); Sheldon v.
Metro-Goldwyn Pictures Corp., 81 F.2d 49, 56 (2d Cir. 1936). But see Abend, 863 F.2d at 1479
(refusing to enjoin distribution of Rear Window despite finding a valid copyright termination);
Goldstein, supra note 37, at 238 (criticizing the injunction granted to the plaintiffs in Sheldon).
394. Indeed, fair use itself is sometimes justified merely as a mechanism for implying agreement
where the parties would have agreed but for transaction costs. See Harper & Row Publishers, Inc. v.
Nation Enters., 471 U.S. 539, 563 (1985); Gordon, supra note 173, at 1629-30 & n.162. The implied
agreement theory cannot explain fair use doctrine, however. It is hard to imagine a plaintiff bringing
a lawsuit (much less going all the way to the Supreme Court) over a use small enough that she would
have been willing to license it but for transaction costs. The absurd implication of this theory is that
in any case important enough to be litigated, fair use should never apply! See 3 NIMMER & NIMMER,
supra note 9, § 13.05, at 13:153 (asserting that the theory of implied consent is "fictional"); Thau,
supra note 192, at 195-96 ("[W]here the costs of copyright enforcement exceed the plaintiff's potential
gain from bringing suit, there is no need for a special doctrine in copyright law to say that the
defendant's use is allowed.").
395. Some commentators fear that the transformative use doctrine will end up restricting fair use.
See, e.g., Nicole B. Casarez, Deconstructing the Fair Use Doctrine: The Cost of Personal and
Workplace Copying After American Geophysical Union v. Texaco, Inc., 6 FoRDHAM INTELL. PROP.
MEDIA & ENT. L.J. 641, 680-81 (1996); Gordon, supra note 173, at 1601; Lape, supra note 190, at
712-13. This fear is unwarranted. See Campbell v. Acuff-RoseMusic, Inc., 510 U.S. 569, 580 (1994)
(indicating that transformative use operates in addition to other factors favoring fair use, rather than
as a replacement for them); cf. Casarez, supra at 682-83 (suggesting that transformative use and public
benefit analyses can coexist). Further, in some cases other statutory provisions (such as 17 U.S.C.
§ 117 (1994)) may protect certain adaptations or copies from infringement liability. Samuelson, for
example, argues that even minor changes should be permitted in computer software where necessary
to correct errors. See Samuelson, supra note 192, at 205 & n.130.
fair use. But courts must be willing to permit a use in circumstances

where it adds a great deal of value relative to what has been copied, even
if the result is to harm the market for the original copyrighted work."
The transformative use doctrine will serve as an effective vehicle for
representing the interests of radical improvers only if there is some chance
that the purpose or character of the accused use can outweigh a finding of
market harm to the copyright owner. The structure of section 107 cer-
tainly suggests that the four listed factors are to be weighed together, with
no one factor dominating the inquiry.3" However, some courts have
given primacy to the market effect test,398 with the result that a produc-
tive use that competes with the original work or actual or potential licensed
derivatives of that work can never be a fair use. In my view, this
approach defeats the purpose of the fair use inquiry and inefficiently
prevents radical improvers from marketing their valuable works in circum-
stances where the original copyright owner prefers not to see the improve-
ment sold. A better approach would be to acknowledge that a strong
showing of transformation by a radical improver can outweigh even a
showing of direct harm to the original copyright owner, particularly where
the amount of the original work taken is relatively small.3 Of course,
396. An alternative proposal that would accomplish some of the same goals as protecting radical
improvers directly is to grant fair use where the court has determined that one of the market failures
identified in Part V is likely to prevent effective licensing. See Dratler, supra note 190, at 296-97
(proposing to treat market effect under the first statutory factor, by determining whether "market
failure" will prevent the efficient licensing of productive works); Elkin-Koren, supra note 159, at 293
(arguing that fair use should apply where licensing fails, but successful licensing should not always
preclude a finding of fair use). Dratler's approach would solve the problem of licensing failure, at least
if courts can be counted on to identify and respond to such failures accurately. It would not solve the
market power holdup problem, however.
Still others suggest that the fair use inquiry should be used to promote licensing by requiring fair
use claimants to offer a reasonable royalty to the copyright owner before using the work. See, e.g.,
Melanie A. Clemons, Author v. Parodist:Striking a Compromise, 46 OHio ST. L.J. 3, 17-19 (1985);
Landy, supra note 36, at 251; Winslow, supra note 268, at 805-06; cf. Merges, supra note 325, at 309-
10 (suggesting that a copyright owner's failure to accept an offered royalty should weigh in favor of
finding fair use). Introducing such evidence is intended to promote licensing and alleviate the perceived
unfairness of the "all-or-nothing" fair use inquiry. However, such a scheme would introduce a host
of problematic issues into the legal process: What happens when the author cannot be found? When
secrecy is an issue? How much money is "reasonable"? Further, given that many refusals to license
in the parody context are noneconomic, it is not clear that such an offer would actually lead to much
additional licensing.
397. See 17 U.S.C. § 107 (1994).
399. This is in accord with the rule in Germany, where under the doctrine of "free use" a use may
be assumed to be free if "the individuality of the used work is thrust into the background and pales
beside the individuality of the newly created work." See Jan-Hendrik Brinink, Computer Software
Users' Rights in Germany and in the United States: A Comparison 8-9 & n.21 (1996) (unpublished
manuscript, on file with the Texas Law Review) (citing Helmut Haberstumpf, Der urheberrechtliche
Schutz von Computerprogrammen,in RECHTSSCHUTZUNDVERWERTUNGVON COMPUTEtPROGRAMMEN
139-46 (2. Auflage 1993)).
not all transformative uses will be radical improvements.' For

example, while it may be accurate to characterize a photocopy of an
excerpt from a book as "transformative" because it is more portable and
can be written on,"° such a use is surely not a "radical improvement" of
the original work sufficient to set aside a finding of market harm.'
An example of how the radical improver rule might be applied in fair
use is the First Circuit case of Lotus Development Corp. v. Borland
International,Inc.' In that case, plaintiff Lotus sued Borland for copy-
right infringement of the menu command hierarchy of its Lotus 1-2-3
spreadsheet. The Lotus spreadsheet program was adapted from the earlier
Visicalc program, and Lotus's program quickly became the industry
standard. Borland did not copy the literal code of the Lotus program, but
did copy the menu structure of the Lotus program in creating its own
spreadsheet, called Quattro Pro. Quattro Pro was not a mere clone of
Lotus 1-2-3; it was independently developed by Borland and by all
accounts was a much better program than 1-2-3. Borland copied the Lotus
menu structure for two reasons: in order to allow customers who had
previously used Lotus to translate their files and macros into the new
Borland program, and to allow customers who were more comfortable with
the Lotus user interface to continue using that interface.'
The district court held that the menu command hierarchies were
copyrightable, and that Borland infringed Lotus's copyright by incorpo-
rating them into its program. 5 It rejected Borland's justifications for the
copying.' Judge Keeton's opinion relied heavily on his prior decision
in Lotus v. Paperback Software International, which had found
Paperback liable for selling a clone of the Lotus 1-2-3 program. In both
400. See Winslow, supra note 268, at 811-12 (arguing that trivial changes should not trigger a fair
use defense).
401. See American Geophysical Union v. Texaco Inc., 60 F.3d 913, 932-36 (2d Cir. 1994)
(Jacobs, J., dissenting) (arguing that photocopying an article for use in scientific research is
transformative).
402. Whether there ws market harm in the American Geophysical Union case is another matter
entirely. See id. at 937 (Jacobs, J., dissenting) (pointing out that the majority's theory of market harm
is circular because any harm is predicated on a finding of no fair use).
403. 49 F.3d 807 (1st Cir. 1995) ("Borland V'), aff'd by an equally divided court, 116 S. Ct. 804
(1996). The litigation between Lotus and Borland arrived before the First Circuit as the culmination
of extended litigation. No fewer than four reported opinions dealt with various aspects of the case in
the district court. See Lotus Dev. Corp. v. Borland Int'l, Inc., 788 F. Supp. 78 (D. Mass. 1992)
("Borland 1"); Lotus Dev. Corp. v. Borland Int'l, Inc., 799 F. Supp. 203 (D. Mass. 1992) ("Borland
1"); Lotus Dev. Corp. v. Borland Int'l, Inc., 831 F. Supp. 202 (D. Mass. 1993) ("Borland 1IP');
Lotus Dev. Corp. v. Borland Int'l, Inc., 831 F. Supp. 223 (D. Mass. 1993) ("Borland IV').
404. For varying accounts of the facts of the case, see Borland V, 49 F.3d at 809-12; id. at 821
(Boudin, J., concurring); and Borland IV, 831 F. Supp. at 226-29.
405. BorlandIlI, 831 F. Supp. at 217-21.
406. BorlandIV, 831 F. Supp. at 231-33.
407. 740 F. Supp. 37 (D. Mass. 1990).
cases, Judge Keeton reasoned that the menu command hierarchy was
copyrightable since it was detailed enough to include original expression,
and was not completely dictated by functional considerations." 5 On
appeal, the First Circuit reversed. It held that Lotus's menu command
hierarchy was entirely uncopyrightable as a "method of operation" of the
program itself.'
Neither approach is particularly satisfactory.41 Judge Keeton's
approach gives Lotus the power not only to control inherently functional
aspects of its program under copyright, but also to lock up the industry
with an inefficient program. Because the Lotus program represented an
industry standard at the time Borland wrote its competing program (though
not by the time of the First Circuit's decision), Borland would have been
at a significant competitive disadvantage if it could not copy enough of
Lotus's program to allow users to continue to use their stored files. At the
same time, the First Circuit's reasoning is problematic. Nof only is it in
tension with decisions from other courts construing the "method of
operation" exception narrowly (or ignoring it altogether),411 but it treats
Borland and Paperback Software identically, even though one has invested
a great deal in advancing the art of spreadsheet programming and the other
has not.412 Judge Boudin's concurrence expressed his dissatisfaction with
the majority's rationale, but failed to articulate a coherent alternative
ground for decision.4 3
This case seems a good candidate for applying a radical improver
doctrine in copyright. While it is possible to argue that Lotus's menu
command hierarchy should not be copyrightable, there are problems
408. Id. at 68; Borland I, 799 F. Supp. at 217-19; BorlandIf, 831 F. Supp. at 214-15.
409. Borland V, 49 F.3d at 815.
410. See David R. Owen, Note, Interfaces and Interoperabilityin Lotus v. Borland: A Market-
OrientedApproach to the FairUse Doctrine, 64 FoRDHAM L. REv. 2381,2403-10 (1996) (criticizing
both opinions).
411. See, e.g., Engineering Dynamics, Inc. v. Structural Software, Inc., 26 F.3d 1335, 1348 (5th
Cir. 1994); Autoskill, Inc. v. National Educ. Support Sys., Inc., 994 F.2d 1476, 1495 n.23 (10th Cir.
1993); Brown Bag Software v. Symantec Corp., 960 F.2d 1465, 1477 (9th Cir. 1992).
412. See Cohen, supra note 309, at 1148-52; Lemley, supra note 224, at 31-32 (both identifying
this problem).
413. Borland V, 49 F.3d at 821-22 (Boudin, J., concurring); see also Lunney, supra note 37, at
2418 (characterizing the Lotus case as a choice between underprotecting and overprotecting Lotus's
spreadsheet).
414. See Dennis S. Karjala & Peter S. Menell, Brief Amicus Curiae: Applying Fundamental
CopyrightPrinciplesto Lotus Development Corp. v. Borland International, Inc., 10 HIGH TECH. LJ.
177 (1995) (supportingthe First Circuit's rationale); Lemley, supra note 224, at 21-22 (suggesting that
the First Circuit could have reached the conclusion that it did under the rationale of Computer
Associates Internationalv. Altai, Inc., 982 F.2d 693 (2d Cir. 1992)); Pamela Samuelson et al., Brief
Amicus Curiae of CopyrightLaw Professorsin Support of Respondent in Lotus Development Corp. v.
Borland International, Inc., 3 J. INTELt. PRoP. L. 103 (1996) (offering several alternative rationales
for a decision in Borland's favor).
with that argument. What does seem clear is that Borland has taken only
a portion of the Lotus program, and at least part of that taking is critical
to Borland's ability to bring its own radically improved spreadsheet
program to market. As with the series of reverse engineering cases in the
computer industry, 415 the market impact on the original copyright owner
is not really the result of the copying, but rather of the improvements that
the copier has made. The fact that the improver has chosen to compete
directly with the original copyright owner should not render it liable for
copying where the improvements are sufficiently radical that the improver's
product is for all intents and purposes a new one. This is particularly true
where, as in Lotus, enforcing the copyright would permit the copyright
owner to engage in holdup tactics.416
The same rationale may also explain the application of fair use in the
context of parody, biography, literary criticism, and even certain fictional
works in the same genre. Like the Borlandcase or the reverse engineering
cases, these types of uses require the improver to start with some part of
the original program,417 but in many cases the improved product is suffi-
ciently different from the original that it stands or falls on its own merits
rather than the merits of the work it copied. As an example, consider Tom
Stoppard's play Rosencrantz and Guildenstern Are Dead.415 The entire
play focuses on two minor characters from Shakespeare's Hamlet, telling
their story (and in the process, telling portions of Hamlet from a different
perspective). Hamlet is of course in the public domain, at least for the
time being.419 But if it were not, or if the play were instead based on
Margaret Mitchell's Gone with the Wind,' one can imagine the author
of the underlying work wishing to capture the value of Stoppard's work as
well, on the theory that it is after all derivative of his masterpiece. Such
a copyright claim should fail, not because there is no copying (there is),
415. See, e.g., Batemanv. Mnemonics, Inc., 79 F.3d 1532, 1539 (11th Cir. 1996); Sega Enters.,
Ltd. v. Accolade, Inc., 977 F.2d 1510, 1522-23 (9th Cir. 1992); Atari Games Corp. v. Nintendo of
Am. Inc., 975 F.2d 832, 843 (Fed. Cir. 1992).
416. See Fisher, supra note 190, at 1728 (suggesting that courts should be more willing to find
fair use where the alternative is anticompetitive control of the market).
417. For example, biographers may need to quote from the papers of their subjects in order to
prove a point in their book. See Leval, supra note 11, at 1113-14 (discussing New Era Publications
International,ApS v. Henry Holt & Co., 873 F.2d 576 (2d Cir. 1989)).
418. STOPPARD, supra note 228.
419. Recent efforts to "resurrect" copyright in public domain works, see 17 U.S.C. § 104A
(1994), combined with the ever-lengthening term of copyright protection, see S. 483, 104th Cong.
(1996), and the growth of non-copyright means of protecting literary works, see ProCD, Inc. v.
Zeidenberg, 86 F.3d 1447, 1454 (7th Cir. 1996) (allowing parties to protect uncopyrightable works
by contract); H.R. 3531, 104th Cong. (1996) (proposing sui generis protection for databases, and
defining databases to include any collection of information), suggest that we cannot guarantee that it
will stay there.
420. MARGARET MITCHELL, GONE WITH THE WIND (1936).
but because the principal value of Stoppard's play lies not in what he has
taken, but in what he has added. A similar argument might be made
regarding Woods v. Universal City Studios, Inc., where the owner of a
copyrighted chair shown in a film gave the copyright owner the power to
obtain an injunction against the showing of the film.42
The radical improver doctrine should also justify fair use protection
of independent works that merely interoperate with a copyrighted
work,4' though it seems likely that the fair use doctrine as currently
interpreted would already protect such works. 4' In many of the software
and video game compatibility cases, for example, the plaintiff seeks to
prevent the defendant from copying the small portion of its game cartridge
object code which serves as a "key," allowing the game cartridge to
operate on the plaintiff's game system.' Where all that is copied in the
final product is the "key" code, the defendant's final product seems likely
to qualify as a "radical improvement. " ' The value of the defendant's
independently created computer game comes from their own software, not
from the functional key code that they copied from the plaintiffs.'
Finally, one might also argue that intermediate copying ought to
be excused to the extent that it is a necessary step towards the creation
of an improvement which will itself be relieved from liability under the
doctrine I have proposed. The creation of intermediate or laboratory
copies is unquestionably infringement under section 106.427 However,
recent courts have been relatively lenient towards such intermediate
copying under the fair use doctrine, reasoning that a defendant should not
be foreclosed from undertaking lawful analysis or competition by the fact
that a copy will necessarily be made in the course of preparations for that
lawful activity.' It should be relatively straightforward to apply the
421. Woods v. Universal City Studios, Inc., 920 F. Supp. 62, 65 (S.D.N.Y. 1996).
422. See Nadan, supra note 146, at 1653.
423. See supra note 147 and accompanying text.
424. See, e.g., Sega Enters. Ltd. v. Accolade, Inc., 977 F.2d 1510 (9th Cir. 1992); Atari Games
Corp. v. Nintendo of Am. Inc., 975 F.2d 832 (Fed. Cir. 1992).
425, Intermediate copying to obtain the key code is a different issue. It was this issue that was
actually treated by the courts in Sega and Atari. See Cohen, supra note 309, at 1091.
426. While copying the key code does add value to the defendant's program, the value comes not
from the plaintiff's copyrighted work, but from compatibility with the plaintiff's game system, which
the defendanthas not copied. See DSC Communications Corp. v. DGI Techs., Inc., 81 F.3d 597, 600
(5th Cir. 1996) (noting that the plaintiff's attempt to block defendant from copying operating systems
to test its processor compatibility could constitute a misuse of copyrighted software to stifle competition
with plaintiff's unpatented processors).
427. See Sega, 977 F.2d at 1519.
428. See, e.g., id. at 1527-28; Vault Corp. v. Quaid Software Ltd., 847 F.2d 255, 261 (5th Cir.
1988); Batemanv. Mnemonics, Inc., 79 F.3d 1532, 1547 (11th Cir. 1995); see also Cohen, supra note
309, at 1134 (approving of the Sega decision and opining that "where copying is necessary to gain
access to and understanding of the ideas and principles embodied in a work, copying solely to gain
understanding should be deemed fair use").
same rationale to intermediate copying necessary to produce a radical

improvement.
The scope of the radical improver doctrine in copyright as I have
envisioned it seems significantly broader than the reverse doctrine of
equivalents in patent law.429 There are two reasons for this apparent
discrepancy. First, it is much harder to get a patent than it is to "get" a
copyright. Indeed, copyright protection is automatic for many works that
meet fairly minimal standards. Since under the model I put forward in
Part V the value of an improvement is theoretically to be measured against
the value of the original work, it is reasonable to assume that works may
be "radical improvements" more easily in copyright than in patent law, just
as original works are more easily protectable in copyright than in patent
law. Second, while a patent is infringed only if all the claim elements are
present in the accused device, a copyrighted work can be infringed by the
copying of even a quantitatively small part of the work.4' In deciding
whether an improvement is radical in copyright law, one does not compare
it to the whole of the original work, but merely to the portion of that work
which has been copied. Thus, we might define Stoppard's play as a
"radical improvement" not over Hamlet taken as a whole, but merely over
those small elements of Hamlet's plot which are included in Rosencrantz
and Guildenstern Are Dead. Further, the term "improvement" should not
be read to imply a value judgment about the relative merit of the two copy-
righted works, but simply to refer to the new material produced by the
"improver." Thus, a court need not conclude that Stoppard's play is more
important than Shakespeare's in order to excuse Stoppard's copying. It
need only find that Stoppard has transformed the copied portion of
Shakespeare's work in such a major way that the value of Stoppard's play
comes primarily from his original contributions, and not from
Shakespeare's.
VII. Conclusion
Intellectual property law represents a "delicate balance" between the
rights of intellectual property owners and the rights of users, among them
the next generation of owners.43 That balance does not reflect merely
a legislative compromise between interest groups. 2 Rather, it can be
429. See supra notes 96-106 and accompanying text (noting the small number of cases in which
the reverse doctrine of equivalents has been applied).
430. See 1 GOLDSTEIN, supra note 120, § 1.6, at 1:18.
431. See Fogerty v. Fantasy, Inc., 510 U.S. 517, 526-27 (1994); Stewart v. Abend, 495 U.S. 207,
228 (1990).
432. See, e.g., Julie E. Cohen, A Right to Read Anonymously: A Closer Look at "Copyright
Management in Cyberspace, 28 CONN. L. REv. 981, 994-1002 (1996) (discussing the inadequacies
of the private ordering model in intellectual property). Indeed, interest group theory suggests a rather
1084 Texas Law Review
justified in economic terms by the need to ensure not only the creation of
entirely new works of intellectual property, but also the creation of
improvements to existing works. Patent law has a number of doctrines
which balance the rights of original inventors with those of improvers,
including the blocking patents rule and the reverse doctrine of equivalents.
Surprisingly, copyright has no such doctrines, relying instead on a rule that
gives original creators property rights over improvements made by others
in most circumstances.
The copyright rule cannot be explained on efficiency grounds, for the
patent approach is more efficient than the copyright approach. This is true
largely because copyright law assumes that strong, unified property rights
will lead to efficient licensing, while patent doctrine .accommodates the
imperfections that exist in the market for licensing improvers. Nor can the
copyright rule be explained by other substantive differences between the
two laws. I recommend some changes to copyright law which are fairly
easily accomplished, and which would bring copyright law in line with a
more realistic model of intellectual property licensing.
more robust imbalance, since intellectual property owners have far more political power and organizing
ability than users. See generally Jessica Litman, CopyrightLegislation and Technological Change, 68
OR. L. REv. 275 (1989).

75TexLRev989 2

Uploaded by

Copyright:

Available Formats

75TexLRev989 2

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

75TexLRev989 2

Uploaded by

Copyright:

Available Formats

DATE DOWNLOADED: Fri Sep 30 00:51:32 2022

SOURCE: Content Downloaded from HeinOnline

Bluebook 21st ed.

ALWD 7th ed.

APA 7th ed.

Chicago 17th ed.

McGill Guide 9th ed.

AGLC 4th ed.

MLA 9th ed.

OSCOLA 4th ed.

I. The Economics of Invention and Improvement .......... 993

II. Treatment of Improvements in Patent Law. 1000

I. Treatment of Improvements in Copyright Law .......... 1013

IV. Is Differential Treatment of Improvement Justified? ....... 1029

V. Optimal Improvement Rules ..................... 1042

B. Complicating the BargainingStructure ... ........ 1048

VI. Implications of the Improvements Model ..... ........ 1072

VII. Conclusion ................................. 1083

I have a colleague in practice who claims that as a trade secrets

improvements. 6 Improvers are therefore in a position to bargain with

improvement under the patent and copyright laws is justifiable. For a

I. The Economics of Invention and Improvement

15. 347 U.S. 201 (1954).

prevent them from using the information.19 To adapt an old parable, if

intellectual property rights in an effort to give authors and inventors control

licensing is to subsequent innovation, and the more sensitive the industry

38. See 35 U.S.C. § 154(a)(2) (1994).

II. Treatment of Improvements in Patent Law

"instrument for writing" if there are already other writing instruments in

57. See 35 U.S.C. § 112 para. 1 (1994).

textile material, when in fact an examination of over six thousand

59. Id. at 472-74.

Finally, patent claims may reach new and unanticipated inventions

protection afforded the invention under the doctrine of equivalents. In

70. See Keith A. Robb, Hilton Davis and the DoctrineofEquivalents-AnInsubstantialDfference

1. Minor Improvers.-As noted above, inventors are entitled to a

effort to avoid infringement. As noted above, those changes may result in

2. Significant Improvers.-Subsequent developers whose improve-

It may seem odd at first that an improvement which is nonobvious in

3. Radical Improvers.-In some sense, any invention, no matter how

material issues of fact precluded summary judgment on this issue, and

of the original patentee, and so encourages radical improvements, just as

III. Treatment of Improvements in Copyright Law

right than patent law. The power of copyrights is tempered in practice,

claimed invention (or an equivalent) must be present in the accused

In addition to the broad definition of a copy subject to the constraints

134. 17 U.S.C. § 106 (1994).

In light of this, it is not evident what section 106(2) adds to the

containing different sounds, but which was designed to be played in the

1. Minor Improvers.-In Part II, I defined a minor patent improver

2. Significant Improvers.-This class of improvers is quite broad,

power of capture may influence license negotiations between copyright

3. Radical Improvers.-The final case involves improvers who have

the value of their contributions." Infringement under the derivative

as "undoubtedly the single most important element of fair use,"" 7 and

far as to create a presumption that commercially motivated uses are

the arts, is generally furthered by the creation of transformative

IV. Is Differential Treatment of Improvement Justified?

Because I find those reasons ultimately unpersuasive, I proceed in Part V