Now that this poll is over -- assuming it is over now and I didn't somehow mess up when scheduling this post -- I suppose I should offer some sort of explanation for the options it presents.
- Yes, I agree that classifying set theory, topology, abstract algebra, ... and so on as merely "applications of category theory" is, to put it mildly, pretty silly. One can, I think, make various more or less as hoc appeals here -- to Lawvere's Elementary Theory of the Category of Sets, say, or to the fact that category theory was originally developed by Mac Lane and Eilenberg as a tool for helping to abstract homology theory, or simply to the apparent ubiquity of commutative diagrams in most modern algebra textbooks -- but I don't think even the most conceited category theorist would ever seriously claim that any of these subjects could be reduced to "applications of category theory". Not in polite company, anyway.
- There actually is an emerging field of "applied category theory", of course, and it amused me to pretend to not know what that is. (The applications people usually have in mind are to fields like computer science and chemistry and epidemiology rather than to what is traditionally thought of as 'pure' mathematics.)
- Equally, it is at best what you might call a non-standard approach to try to justify number theory (or combinatorics, for that matter) as merely a particular subfield of cryptography. G. H. Hardy would not approve, I'm sure. (There's a quote circulating online attributed to Hardy in 1940 in which he took solace in the fact that at least nobody had ever found any military applications for number theory and relativity "and it seems very unlikely that anybody will do so for many years". I would guess it's apocryphal -- it seems a little too on the nose, doesn't it? -- but it does amuse me to think it might be true ...)
- And of course it's also somewhat strange to have an entire section on "probability" and yet relegate statistics to a subfield of optimization alongside "machine learning". (I do know people who work in mathematical optimization who would make the argument for this position; I do not know anybody who works in statistics who would accept it.)
Then why did I structure the poll this way? Why the focus on applications and areas traditionally thought of as "applied mathematics"? Why force whole fields of pure mathematics into a ludicrous classification scheme that their practitioners wouldn't recognize or accept, or create artificial splits between what are widely considered one single unified subject, or focus solely on applications for an area famous for its abstraction and generality?
Well, I assumed most of the people who saw the poll would be pure mathematicians, or at least math majors with a particular interest in pure math. Pure mathematicians certainly created most of the "favorite area of math" polls I'd seen before on this site, and I think most mathblr people would self-identify as pure mathematics (if, that is, they even acknowledge applied mathematics as a coequal branch of mathematics at all and don't just assert that all of math can be reduced to analysis and algebra and number theory and topology and geometry and logic).
And, if you are a pure mathematician, and you were puzzled by the way this poll attempted to force your favorite area of mathematics into a classification system that clearly didn't work, if you were a little bit appalled and offended by its attempts to collapse entire areas of mathematics with their own different histories and objectives and traditions into a single 'oh, everything else I guess' mess at the bottom of the list, if you wondered why I didn't just ask about "favorite area of applied mathematics" and leave the pure mathematics out of it ... well, now you have a little bit more insight into how applied mathematicians feel when you ask them if their research area is part of "continuous math" or "discrete math".