Information Cascades

Social visualization systems have emerged to support collective intelligence-driven analysis of a growing influx of open data. As with many other online systems, social signals (e.g., forums, polls) are commonly integrated to drive use. Unfortunately, the same social features that can provide rapid, high-accuracy analysis are coupled with the pitfalls of any social system. Through an experiment involving over 300 subjects, we address how social information signals (social proof) affect quantitative judgments in the context of graphical perception. We identify how unbiased social signals lead to fewer errors over non-social settings and conversely, how biased signals lead to more errors. We further reflect on how systematic bias nullifies certain collective intelligence benefits, and we provide evidence of the formation of information cascades. We describe how these findings can be applied to collaborative visualization systems to produce more accurate individual interpretations in social contexts.

We study a sequential trading financial market where there are gains from trade, that is, where informed traders have heterogeneous private values. We show that an informational cascade (i.e., a complete blockage of information) arises and prices fail to aggregate information dispersed among traders. During an informational cascade, all traders with the same preferences choose the same action, following the market (herding) or going against it (contrarianism). We also study financial contagion by extending our model to a two-asset economy. We show that informational cascades in one market can be generated by informational spillovers from the other. Such spillovers have pathological consequences, generating long-lasting misalignments between prices and fundamentals. . We are indebted to our advisor, Douglas Gale, for his invaluable guidance. We also owe a special debt to Marco LiCalzi for a very careful reading of the paper. We thank show that, when there there is multidimensional uncertainty (i.e., uncertainty not only on the direction of a shock to the asset fundamental, but also on the existence of the shock itself), herd behavior can arise even in their framework. Their definition of herding, however, is not the standard one in the literature (see footnote 24). Even with multidimensional uncertainty, informational cascades cannot arise in their study (see their Proposition 2 and their comments at page 733). See also the considerations of

The purpose of this paper is to reflect the behavioral biases that led to this global financial crisis. The paper presents briefly the real causes of the crisis (structural and cyclical factors) and puts a greater accent on the behavioral factors. The authors considered to structure the paper in three main pillars: behavioral factors, the collapse of ethical behavior and the role of behavioral finance in studying, regulating and assessment financial risks. The first pillar consists in a brief presentation of the behavioral factors such as: optimism and wishful thinking, overconfidence, greed, regret, pessimism, passing the responsibility, herding -groupthink, anchoring, representativeness biases, informational cascades and "this time is different" syndrome. The second pillar of the paper presents the collapse of ethical behavior that led to the global financial crisis: predatory lending practices, inappropriate compensation schemes, rating agencies behavior, corporate governance reforms and financial institutions opacity in their reporting. The third pillar presents the mismanagement of risk and regulations that led us into this global mess. The paper concludes with the need of integrating biases of human behavior into regulations in order to make them more effective and people become less financially vulnerable.

Experts tend to herd if they can communicate among themselves. I show that every expert can be induced to do independent research and report truthful results if and only if payments to the expert depend on other experts' reports. If the experts are risk-averse then the prevention of herding is costly.

Online discussion threads are conversational cascades in the form of posted messages that can be generally found in social systems that comprise many-to-many interaction such as blogs, news aggregators or bulletin board systems. We propose a framework based on generative models of growing trees to analyse the structure and evolution of discussion threads. We consider the growth of a discussion to be determined by an interplay between popularity, novelty and a trend (or bias) to reply to the thread originator. The relevance of these features is estimated using a full likelihood approach and allows to characterize the habits and communication patterns of a given platform and/or community.

We examine a model of social learning in networks following the lines of Bala and Goyal (1998, 2001). As a model of agents’ behaviour we have chosen the model of informational cascades of Bikhchandani et al (1992). Similarly to Bala and Goyal we find that the higher the ’degree of integration’ within the society is, the more likely it is that conformity of actions will arise. However, unlike their results our model suggests that in the presence of informational externalities globalisation of informational flows, expressed in the increasing density of communication channels in a network, may drive down the expected social welfare.

This study develops and tests a theoretical framework, which draws on herd behavior literature and explains how and under what conditions tabloids' attention to misinformation drives broadsheets' attention to misinformation. More specifically, the study analyzes all cases of political and business misinformation in Switzerland and the U.K. between 2002 and 2018, which are selected based on corresponding Swiss and U.K. press councils' rulings (N = 114). The findings show that during amplifying events (i.e., election campaigns and economic downturns) tabloids allocate more attention to political and business misinformation, which, in turn, drives broadsheets to allocate more attention to the misinformation as well-and especially if the misinformation serves broadsheets' ideological goals. Moreover, the findings show differences between Swiss and U.K. media markets only in the case of business misinformation and suggest that the attention allocation process depends in particular on the strength of the amplifying event in a media market. Thereby, this study contributes to the understanding of how and under what conditions misinformation spreads in media markets.

Time plays an essential role in the diffusion of information, influence, and disease over networks. In many cases we can only observe when a node is activated by a contagion— when a node learns about a piece of information, makes a decision, adopts a new behavior, or becomes infected with a disease. However, the underlying network connectivity and transmission rates between nodes are unknown. Inferring the underlying diffusion dynamics is important because it leads to new insights and enables forecasting, as well as influencing or containing information propagation. In this paper we model diffusion as a continuous temporal process occurring at different rates over a latent, unobserved network that may change over time. Given information diffusion data, we infer the edges and dynamics of the underlying network. Our model naturally imposes sparse solutions and requires no parameter tuning. We develop an efficient inference algorithm that uses stochastic convex optimization to compute online estimates of the edges and transmission rates. We evaluate our method by tracking information diffusion among 3.3 million mainstream media sites and blogs, and experiment with more than 179 million different instances of information spreading over the network in a one-year period. We apply our network inference algorithm to the top 5,000 media sites and blogs and report several interesting observations. First, information pathways for general recurrent topics are more stable across time than for on- going news events. Second, clusters of news media sites and blogs often emerge and vanish in a matter of days for on-going news events. Finally, major events, for example, large scale civil unrest as in the Libyan civil war or Syrian uprising, increase the number of information pathways among blogs, and also increase the network centrality of blogs and social media sites.

We study a model of learning on social networks in dynamic environments, describing a group of agents who are each trying to estimate an underlying state that varies over time, given access to weak signals and the estimates of their social network neighbors. We study three models of agent behavior. In the fixed response model, agents use a fixed linear combination to incorporate information from their peers into their own estimate. This can be thought of as an extension of the DeGroot model to a dynamic setting. In the best response model, players calculate minimum variance linear estimators of the underlying state. We show that regardless of the initial configuration, fixed response dynamics converge to a steady state, and that the same holds for best response on the complete graph. We show that best response dynamics can, in the long term, lead to estimators with higher variance than is achievable using well chosen fixed responses. The penultimate prediction model is an elaboration of the best response model. While this model only slightly complicates the computations required of the agents, we show that in some cases it greatly increases the efficiency of learning, and on complete graphs is in fact optimal, in a strong sense.

Given a directed social graph and a set of past information cascades observed over the graph, we study the novel problem of detecting modules of the graph (communities of nodes), that also explain the cascades. Our key observation is that both information propagation and social ties formation in a social network can be explained according to the same latent factor, which ultimately guide a user behavior within the network. Based on this observation, we propose the Community-Cascade Network (CCN) model, a stochastic mixture membership generative model that can fit, at the same time, the social graph and the observed set of cascades. Our model produces overlapping communities and for each node, its level of authority and passive interest in each community it belongs.

We investigate the herd behavior of returns for the yen-dollar exchange rate in the Japanese financial market. It is obtained that the probability distribution $P(R)$ of returns $R$ satisfies the power-law behavior $P(R) \simeq R^{-\beta}$ with the exponents $ \beta=3.11$(the time interval $\tau=$ one minute) and 3.36($\tau=$ one day). The informational cascade regime appears in the herding parameter $H\ge 2.33$

An interactive innovation can be defined as an end-user application which is subject to network effects at both the demand and supply sides. As a result of network effects, the diffusion of such an innovation is predicted to follow a take-off curve coinciding with the advent of a critical mass of adopters. The current literature on innovation diffusion, mainly focusing on the demand-side dynamics, such as information cascades and herding behaviors among potential adopters, is not sufficient to explain the take-off (or the failure) of interactive innovations. In this paper, we present and examine a case study of the take-off of an interactive innovation, namely the caller-ring-back-tone (CRBT) and mobile musicmobile value-added services (MVAS)in China. We find that supply-side dynamics, such as choices of platform strategies, helped drive the take-off of this innovation within China's institutional boundaries. The paper makes a contribution in two ways: first, it presents an 'inside-out' view of a unique case of takeoff phenomenon; and, second it provides an integrated view combining factors from both the demand and supply sides to explain the take-off phenomenon, which is rare in empirical studies.

We investigate by numerical simulation a threshold model of social contagion on degree-correlated random networks. We show that the class of networks for which global information cascades occur generally expands as degree-degree correlations become increasingly positive. However, under certain conditions, large-scale information cascades can paradoxically occur when degree-degree correlations are sufficiently positive or negative, but not when correlations are relatively small. We also show that the relationship between the degree of the initially infected vertex and its ability to trigger large cascades is strongly affected by degree-degree correlations.

Modelling accurately financial price variations is an essential step underlying portfolio allocation optimization, derivative pricing and hedging, fund management and trading. The observed complex price fluctuations guide and constraint our theoretical understanding of agent interactions and of the organization of the market. The gaussian paradigm of independent normally distributed price increments has long been known to be incorrect with many

There is a large and rapidly increasing literature which analyses social networks for which substantial amounts of quantitative data are available. Further, there is a growing and related literature on what is referred to by economists as ‘information cascades’ on such networks. However, there is little discussion of the specific events which lead to either the creation of a network in the first instance, or of how a network which is latent in terms of its awareness evolves into a network on which individual behaviour/beliefs are actually altered. In this paper, we describe the process of emergence and evolution of a historical cultural and social network across which the opinions and behaviour of individuals were influenced. We also illustrate how empirical networks can be reconstructed relying principally on information contained in qualitative, historical sources. The specific example we use is religious belief in England in the 1550s. We describe how the burnings of Protestant l...

Abstract: How do blogs cite and influence each other? How do such links evolve? Does the popularity of old blog posts drop exponentially with time? These are some of the questions that we address in this work. Our goal is to build a model that generates realistic cascades, so that it can help us with link prediction and outlier detection.

In settings characterized by imperfect information about an underlying state of nature, but where inferences are made sequentially and are publicly observable, decisions may yield a "cascade" in which everyone herds on a single choice. While cascades potentially play a role in a variety of settings, from technology adoption to social processes such as mate selection, understanding cascade phenomena is imperative for financial markets. Previous empirical efforts studying cascade formation have used both naturally occurring data and laboratory experiments. In this paper, we combine one of the attractive elements of each line of research-observation of market professionals in a controlled environment-to push the investigation of cascade behavior into several new directions. Numerous empirical insights are obtained; perhaps most importantly, we find that market professionals behave quite differently than a control group of student subjects. In particular, market professionals, more so than students, base their decisions on the "quality" of the public signal, leading them to be more likely to disregard "bad" signals. And, unlike in the case with students, for market professionals, the propensity to be Bayesian does not differ significantly across the gain and loss domains. These results have important implications in both a positive and normative sense.

This paper investigates the validity of the Dual Process theory by using eye-tracking methods to trace the process of attention during a non-preference-based problem solving task, i.e. informational cascades. In this setting, gaze direction may convey evidence on how automatic detection is modified or sustained by controlled search. We provide laboratory evidence that gaze direction is driven by cognitive biases, such as overconfidence. In particular, we find a significant statistical correlation between first fixations and subjects' actual choices. Our results suggest that attentional strategies are not necessarily consistent with efficient patterns of information collecting.

In this paper we try to understand the phenomenon whereby a large proportion of the population evades tax payments. We present a model which incorporates elements from the theory of information cascades with the standard model of tax evasion and analyze the connection between the decision of a potential tax evader, the number of tax evaders, and the number caught in previous periods. General conditions exist under which any expected utility maximizing potential tax evaders will decide to emulate other tax evaders.

Initial adoption of technology is examined. Initial adoption refers to the point at which the decision to adopt a technology is made, before the user has formed perceptions of -for instance -how easy it is to use or how useful it is. Adoption behaviour is explained in terms of information cascades. An information cascade exists when a potential technology adopter ignores their private information about the technology and is influenced by the adoption decisions of others. Data were drawn from a microblogging service. Microblogging is a form of social networking where short messages are sent from one blogger to many readers. The system is voluntary and unlike many information systems, gives potential adopters the ability to clearly see the adoption decisions of others. Two empirical analyses support the notion that an individual's decision to adopt is influenced by the decisions of others. A third study examines a possible alternative explanation -that some individuals are lurking, thereby adopting the system although not in the way intended by the developers -and discounts it. We found strong support for the importance of the adoption decision of others in an individual's decision to adopt. A general model of these ideas is one which recognises the importance of what might be termed buzz around a technology and how this can influence adoption decisions.

We study the effect of transaction costs (eg, a trading fee or a transaction tax, like the Tobin tax) on the aggregation of private information in financial markets. We implement a financial market with sequential trading and transaction costs in the laboratory. According to theory, ...

In a variety of important domains, there is considerable correlational evidence suggestive of what are variously referred to as social norm effects, contagion effects, information cascades, or peer effects. It is difficult to statistically identify whether such effects are causal, and there are various non-causal mechanisms that can produce such apparent norm effects. Lab experiments demonstrate that real peer effects occur, but also that apparent cascade or peer effects can be spurious. A curious feature of American local school configuration policy provides an opportunity to identify true peer influences among adolescents. Some school districts send 6th graders to middle school (e.g., 6th-8th grade "junior high"); others retain 6th graders for one additional year in K-6 elementary schools. Using administrative data on public school students in North Carolina, we have found that sixth grade students attending middle schools are much more likely to be cited for discipline problems than those attending elementary school, and the effects appear to persist at least through ninth grade. A plausible explanation is that these effects occur because sixth graders in middle schools are suddenly exposed to two cohorts of older, more delinquent peers.

The availability of high-frequency data for nancial markets has made it possible to study market dynamics on timescales of less than a day 1 ]. F or foreign exchange (FX) rates M uller et al. 2] have s h o wn that there is a net information ow from long to short time scales: the behaviour of long-term traders (who watch the market only from time to time) in uences the behaviour of short-term traders (who watch the market continuously). Motivated by this hierarchical feature, we h a ve studied FX market dynamics in more detail, and report here an analogy between FX market dynamics and hydrodynamic turbulence 3{8]. Speci cally, the relationship between the probability density o f FX price changes (x) and the time delay (t) (Fig. 1a) is much the same as the relationship between the probability density o f t h e v elocity di erences (v) o f t wo points in a turbulent o w and their spatial distance r (Fig. 1b). Guided by this similarity w e claim that there is an information cascade in FX market dynamics that corresponds to the energy cascade in hydrodynamic turbulence. On the basis of this analogy we c a n n o w rationalize the statistics of FX price di erences at di erent time delays, which is important for, for example, option pricing. The analogy also provides a conceptual framework for understanding the short-term dynamics of speculative markets.

This paper identifies a disadvantage to decision making in a team. We show that in some cases available information is lost due to sequential communication that results in informational cascades. Although incentive contracts exist that prevent cascades, in some cases these contracts do not maximize shareholders' expected residual value and cascades are tolerated in equilibrium. Cascades never occur in hierarchies that exogenously prevent commu-(nication. However, when the firm is organized as a hierarchy and the) agents are given the optimal hierarchical contract , in some cases agents will collude and sequentially communicate, admitting the possibility of cascades. In these cases, the principals must monitor and enforce the hierarchical process. When monitoring costs exceed the cost of cascades, the team is the optimal organizational form.

Abstract—The spread of information cascades over social networks forms the diffusion networks. The latent structure of diffusion networks makes the problem of extracting diffusion links difficult. As observing the sources of information is not usually possible, the only available prior knowledge is the infection times of individuals. We confront these challenges by proposing a new method called DNE to extract the diffusion networks by using the time-series data. We model the diffusion process on information networks as a Markov random walk process and develop an algorithm to discover the most probable diffusion links. We validate our model on both synthetic and real data and show the low dependency of our method to the number of transmitting cascades over the underlying networks. Moreover, The proposed model can speed up the extraction process up to 300 times with respect to the existing state of the art method. I.

We examine a model of social learning in networks following the lines of Bala and Goyal (1998, 2001). As a model of agents’ behaviour we have chosen the model of informational cascades of Bikhchandani et al (1992). Similarly to Bala and Goyal we find that the higher the ’degree of integration’ within the society is, the more likely it is that conformity of actions will arise. However, unlike their results our model suggests that in the presence of informational externalities globalisation of informational flows, expressed in the increasing density of communication channels in a network, may drive down the expected social welfare. .

Motivated by the significant amount of successful collaborative problem solving activity on the Web, we ask: Can the accumulated information propagation behavior on the Web be conceived as a giant machine, and reasoned about accordingly? In this paper we elaborate a thesis about the computational capability embodied in information sharing activities that happen on the Web, which we term socio-technical computation, reflecting not only explicitly conditional activities but also the organic potential residing in information on the Web.

We introduce the idea that informational cascades can explain the observed regularity that emigrants from the same locations also tend to choose the same foreign locations. Thus informational cascades generate herd behavior. Herd behavior is compared with the network externalities explanation of the same phenomenon. The relation between social tensions and herd behavior is observed when local populations are xenophobic.

ㅇ Purpose
When purchasing digital content (DC), consumers are typically influenced by various information sources on the website. Prior research has mostly focused on the individual effect of the information sources on the DC choice. To fill the gap in the previous studies, this research includes three main effects: information cascades, recommendations and word of mouth. In particular, the purpose of this paper is to focus on the interaction effect of information cascades and recommendations on the number of software downloads.
ㅇ Design/methodology/approach
The authors use the panel generalized least squares estimation to test the hypotheses by using a panel data set of 2,000 pieces of software at download.cnet.com over a month-long period. Product ranking and recommendation status are used as key independent variables to capture the effects of information cascades and recommendations, respectively.
ㅇ Findings
One of this study's findings is that information cascades positively interact with recommendations to influence the number of software downloads. The authors also show that the impact of information cascades on the number of software downloads is greater than one of the recommendations from a distributor does.
ㅇ Originality/value
Information cascades and recommendations have been considered as the primary effects for online product choices. However, these two effects typically are not considered together in one research. As previous studies have mainly focused on each effect, respectively, the authors believe that this study may fill the gap by examining how these effects are interacted to one other to influence customers' choices. The authors also show that the impact of information cascades on the number of software downloads is greater than one of the recommendations from a system does.

In settings where there is imperfect information about an underlying state of nature, but where inferences are made sequentially and are publicly observable, information cascades can lead to rational herding. Cascade phenomena may be seen in a variety of areas including technology adoption, financial market behavior, as well as in social processes such as mate selection or fads and fashions. Theories of rational herding have found a natural testing ground in experimental environments since the character of private and public information can be readily controlled. In previous experimental studies, behavior consistent with Bayesian benchmarks has been observed in simple contexts, but there are substantial reductions in experimental environments that introduce relevant complications such as costly information. In this paper we make use of a unique subject pool, that of financial market professionals from the floor of the Chicago Board of Trade, to investigate the role of market experience on herding behavior. We find that market professionals behave differently than a control group of college student subjects. In particular the Bayesian behavior of those with market experience does not differ significantly across the gain and loss domains. Cascade formation also differs across the subject pools with market professionals entering fewer reverse cascades.