Duncan J. Watts

Recent work on the structure of social networks and the internet has focused attention on graphs with distributions of vertex degree that are significantly different from the Poisson degree distributions that have been widely studied in the past. In this paper we develop in detail the theory of random graphs with arbitrary degree distributions. In addition to simple undirected, unipartite graphs, we examine the properties of directed and bipartite graphs. Among other results, we derive exact expressions for the position of the phase transition at which a giant component first forms, the mean component size, the size of the giant component if there is one, the mean number of vertices a certain distance away from a randomly chosen vertex, and the average vertex-vertex distance within a graph. We apply our theory to some real-world graphs, including the world-wide web and collaboration graphs of scientists and Fortune 1000 company directors. We demonstrate that in some cases random graphs with appropriate distributions of vertex degree predict with surprising accuracy the behavior of the real world, while in others there is a measurable discrepancy between theory and reality, perhaps indicating the presence of additional social structure in the network that is not captured by the random graph.

Addressing fake news requires a multidisciplinary effort

The origin of large but rare cascades that are triggered by small initial shocks is a phenomenon that manifests itself as diversely as cultural fads, collective action, the diffusion of norms and innovations, and cascading failures in infrastructure and organizational networks. This paper presents a possible explanation of this phenomenon in terms of a sparse, random network of interacting agents whose decisions are determined by the actions of their neighbors according to a simple threshold rule. Two regimes are identified in which the network is susceptible to very large cascades-herein called global cascades-that occur very rarely. When cascade propagation is limited by the connectivity of the network, a power law distribution of cascade sizes is observed, analogous to the cluster size distribution in standard percolation theory and avalanches in self-organized criticality. But when the network is highly connected, cascade propagation is limited instead by the local stability of the nodes themselves, and the size distribution of cascades is bimodal, implying a more extreme kind of instability that is correspondingly harder to anticipate. In the first regime, where the distribution of network neighbors is highly skewed, it is found that the most connected nodes are far more likely than average nodes to trigger cascades, but not in the second regime. Finally, it is shown that heterogeneity plays an ambiguous role in determining a system's stability: increasingly heterogeneous thresholds make the system more vulnerable to global cascades; but an increasingly heterogeneous degree distribution makes it less vulnerable.

Recent work on the Internet, social networks, and the power grid has addressed the resilience of these networks to either random or targeted deletion of network nodes or links. Such deletions include, for example, the failure of Internet routers or power transmission lines. Percolation models on random graphs provide a simple representation of this process but have typically been limited to graphs with Poisson degree distribution at their vertices. Such graphs are quite unlike real-world networks, which often possess power-law or other highly skewed degree distributions. In this paper we study percolation on graphs with completely general degree distribution, giving exact solutions for a variety of cases, including site percolation, bond percolation, and models in which occupation probabilities depend on vertex degree. We discuss the application of our theory to the understanding of network resilience.

We propose to change the default P-value threshold for statistical significance from 0.05 to 0.005 for claims of new discoveries.

Hit songs, books, and movies are many times more successful than average, suggesting that "the best" alternatives are qualitatively different from "the rest"; yet experts routinely fail to predict which products will succeed. We investigated this paradox experimentally, by creating an artificial "music market" in which 14,341 participants downloaded previously unknown songs either with or without knowledge of previous participants' choices. Increasing the strength of social influence increased both inequality and unpredictability of success. Success was also only partly determined by quality: The best songs rarely did poorly, and the worst rarely did well, but any other result was possible.

A central idea in marketing and diffusion research is that influentials—a minority of individuals who influence an exceptional number of their peers—are important to the formation of public opinion. Here we examine this idea, which we call the “influentials hypothesis,” using a series of computer simulations of interpersonal influence processes. Under most conditions that we consider, we find that large cascades of influence are driven not by influentials but by a critical mass of easily influenced individuals. Although our results do not exclude the possibility that influentials can be important, they suggest that the influentials hypothesis requires more careful specification and testing than it has received.

Social networks evolve over time, driven by the shared activities and affiliations of their members, by similarity of individuals' attributes, and by the closure of short network cycles. We analyzed a dynamic social network comprising 43,553 students, faculty, and staff at a large university, in which interactions between individuals are inferred from time-stamped e-mail headers recorded over one academic year and are matched with affiliations and attributes. We found that network evolution is dominated by a combination of effects arising from network topology itself and the organizational structure in which the network is embedded. In the absence of global perturbations, average network properties appear to approach an equilibrium state, whereas individual properties are unstable.

In this paper we investigate the attributes and relative influence of 1.6M Twitter users by tracking 74 million diffusion events that took place on the Twitter follower graph over a two month interval in 2009. Unsurprisingly, we find that the largest cascades tend to be generated by users who have been influential in the past and who have a large number of followers. We also find that URLs that were rated more interesting and/or elicited more positive feelings by workers on Mechanical Turk were more likely to spread. In spite of these intuitive results, however, we find that predictions of which particular user or URL will generate large cascades are relatively unreliable. We conclude, therefore, that word-of-mouth diffusion can only be harnessed reliably by targeting large numbers of potential influencers, thereby capturing average effects. Finally, we consider a family of hypothetical marketing strategies, defined by the relative cost of identifying versus compensating potential "influencers." We find that although under some circumstances, the most influential users are also the most cost-effective, under a wide range of plausible assumptions the most cost-effective performance can be realized using "ordinary influencers"---individuals who exert average or even less-than-average influence.

The small‐world phenomenon formalized in this article as the coincidence of high local clustering and short global separation, is shown to be a general feature of sparse, decentralized networks that are neither completely ordered nor completely random. Networks of this kind have received little attention, yet they appear to be widespread in the social and natural sciences, as is indicated here by three distinct examples. Furthermore, small admixtures of randomness to an otherwise ordered network can have a dramatic impact on its dynamical, as well as structural, properties‐a feature illustrated by a simple model of disease transmission.

We study the small-world network model, which mimics the transition between regular-lattice and random-lattice behavior in social networks of increasing size. We contend that the model displays a critical point with a divergent characteristic length as the degree of randomness tends to zero. We propose a real-space renormalization group transformation for the model and demonstrate that the transformation is exact in the limit of large system size. We use this result to calculate the exact value of the single critical exponent for the system, and to derive the scaling form for the average number of `degrees of separation' between two nodes on the network as a function of the three independent variables. We confirm our results by extensive numerical simulation.

We describe some new exactly solvable models of the structure of social networks, based on random graphs with arbitrary degree distributions. We give models both for simple unipartite networks, such as acquaintance networks, and bipartite networks, such as affiliation networks. We compare the predictions of our models to data for a number of real-world social networks and find that in some cases, the models are in remarkable agreement with the data, whereas in others the agreement is poorer, perhaps indicating the presence of additional social structure in the network that is not captured by the random graph.

From the Internet to networks of friendship, disease transmission, and even terrorism, the concept--and the reality--of networks has come to pervade modern society. But what exactly is a network? What different types of networks are there? Why are they interesting, and what can they tell us? In recent years, scientists from a range of fields--including mathematics, physics, computer science, sociology, and biology--have been pursuing these questions and building a new "science of networks." This book brings together for the first time a set of seminal articles representing research from across these disciplines. It is an ideal sourcebook for the key research in this fast-growing field. The book is organized into four sections, each preceded by an editors' introduction summarizing its contents and general theme. The first section sets the stage by discussing some of the historical antecedents of contemporary research in the area. From there the book moves to the empirical side of the science of networks before turning to the foundational modeling ideas that have been the focus of much subsequent activity. The book closes by taking the reader to the cutting edge of network science--the relationship between network structure and system dynamics. From network robustness to the spread of disease, this section offers a potpourri of topics on this rapidly expanding frontier of the new science.

In this paper we study the small-world network model of Watts and Strogatz, which mimics some aspects of the structure of networks of social interactions. We argue that there is one nontrivial length-scale in the model, analogous to the correlation length in other systems, which is well-defined in the limit of infinite system size and which diverges continuously as the randomness in the network tends to zero, giving a normal critical point in this limit. This length-scale governs the crossover from large- to small-world behavior in the model, as well as the number of vertices in a neighborhood of given radius on the network. We derive the value of the single critical exponent controlling behavior in the critical region and the finite size scaling form for the average vertex-vertex distance on the network, and, using series expansion and Padé approximants, find an approximate analytic form for the scaling function. We calculate the effective dimension of small-world graphs and show that this dimension varies as a function of the length-scale on which it is measured, in a manner reminiscent of multifractals. We also study the problem of site percolation on small-world networks as a simple model of disease propagation, and derive an approximate expression for the percolation probability at which a giant component of connected vertices first forms (in epidemiological terms, the point at which an epidemic occurs). The typical cluster radius satisfies the expected finite size scaling form with a cluster size exponent close to that for a random graph. All our analytic results are confirmed by extensive numerical simulations of the model.

Social networks have the surprising property of being “searchable”: Ordinary people are capable of directing messages through their network of acquaintances to reach a specific but distant target person in only a few steps. We present a model that offers an explanation of social network searchability in terms of recognizable personal identities: sets of characteristics measured along a number of social dimensions. Our model defines a class of searchable networks and a method for searching them that may be applicable to many network search problems, including the location of data files in peer-to-peer networks, pages on the World Wide Web, and information in distributed databases.

In recent years, the analysis and modeling of networks, and also networked dynamical systems, have been the subject of considerable interdisciplinary interest, yielding several hundred papers in physics, mathematics, computer science, biology, economics, and sociology journals ( Newman 2003c ), as well as a number of books ( Barabasi 2002 , Buchanan 2002 , Watts 2003 ). Here I review the major findings of this emerging field and discuss briefly their relationship with previous work in the social and mathematical sciences.

'Six degrees of separation' is a cliche, as is 'it's a small world', both cliches of the language and cliches of everyone's experience. We all live in tightly bonded social networks, yet linked to vast numbers of other people more closely than we sometimes think. Only in recent years, however, have scientists begun to apply insights from the theoretical study of networks to understand forms of network as superficially different as social networks and electrical networks, computer networks and economic networks, and to show how common principles underlie them all. Duncan Watts explores the science of networks and its implications, ranging from the Dutch tulipmania of the 17th century to the success of Harry Potter, from the impact of September 11 on Manhattan to the brain of the sea-slug, from the processes that lead to stockmarket crashes to the structure of the world wide web. As stimulating and life-changing as James Gleick's Chaos, Six Degrees is a ground-breaking and important book.

Everyone knows the small-world phenomenon: soon after meeting a stranger, we are surprised to discover that we have a mutual friend, or we are connected through a short chain of acquaintances. In his book, Duncan Watts uses this intriguing phenomenon--colloquially called six degrees of separation--as a prelude to a more general exploration: under what conditions can a small world arise in any kind of network?The networks of this story are everywhere: the brain is a network of neurons; organisations are people networks; the global economy is a network of national economies, which are networks of markets, which are in turn networks of interacting producers and consumers. Food webs, ecosystems, and the Internet can all be represented as networks, as can strategies for solving a problem, topics in a conversation, and even words in a language. Many of these networks, the author claims, will turn out to be small worlds.How do such networks matter? Simply put, local actions can have global consequences, and the relationship between local and global dynamics depends critically on the network's structure. Watts illustrates the subtleties of this relationship using a variety of simple models---the spread of infectious disease through a structured population; the evolution of cooperation in game theory; the computational capacity of cellular automata; and the sychronisation of coupled phase-oscillators.Watts's novel approach is relevant to many problems that deal with network connectivity and complex systems' behaviour in general: How do diseases (or rumours) spread through social networks? How does cooperation evolve in large groups? How do cascading failures propagate through large power grids, or financial systems? What is the most efficient architecture for an organisation, or for a communications network? This fascinating exploration will be fruitful in a remarkable variety of fields, including physics and mathematics, as well as sociology, economics, and biology.

We report on a global social-search experiment in which more than 60,000 e-mail users attempted to reach one of 18 target persons in 13 countries by forwarding messages to acquaintances. We find that successful social search is conducted primarily through intermediate to weak strength ties, does not require highly connected "hubs" to succeed, and, in contrast to unsuccessful social search, disproportionately relies on professional relationships. By accounting for the attrition of message chains, we estimate that social searches can reach their targets in a median of five to seven steps, depending on the separation of source and target, although small variations in chain lengths and participation rates generate large differences in target reachability. We conclude that although global social networks are, in principle, searchable, actual success depends sensitively on individual incentives.

We study several longstanding questions in media communications research, in the context of the microblogging service Twitter, regarding the production, flow, and consumption of information. To do so, we exploit a recently introduced feature of Twitter known as "lists" to distinguish between elite users - by which we mean celebrities, bloggers, and representatives of media outlets and other formal organizations - and ordinary users. Based on this classification, we find a striking concentration of attention on Twitter, in that roughly 50% of URLs consumed are generated by just 20K elite users, where the media produces the most information, but celebrities are the most followed. We also find significant homophily within categories: celebrities listen to celebrities, while bloggers listen to bloggers etc; however, bloggers in general rebroadcast more information than the other categories. Next we re-examine the classical "two-step flow" theory of communications, finding considerable support for it on Twitter. Third, we find that URLs broadcast by different categories of users or containing different types of content exhibit systematically different lifespans. And finally, we examine the attention paid by the different user categories to different news topics.

The authors investigate the origins of homophily in a large university community, using network data in which interactions, attributes, and affiliations are all recorded over time. The analysis indicates that highly similar pairs do show greater than average propensity to form new ties; however, it also finds that tie formation is heavily biased by triadic closure and focal closure, which effectively constrain the opportunities among which individuals may select. In the case of triadic closure, moreover, selection to “friend of a friend” status is determined by an analogous combination of individual preference and structural proximity. The authors conclude that the dynamic interplay of choice homophily and induced homophily, compounded over many “generations” of biased selection of similar individuals to structurally proximate positions, can amplify even a modest preference for similar others, via a cumulative advantage–like process, to produce striking patterns of observed homophily.

Recent work has demonstrated that Web search volume can "predict the present," meaning that it can be used to accurately track outcomes such as unemployment levels, auto and home sales, and disease prevalence in near real time. Here we show that what consumers are searching for online can also predict their collective future behavior days or even weeks in advance. Specifically we use search query volume to forecast the opening weekend box-office revenue for feature films, first-month sales of video games, and the rank of songs on the Billboard Hot 100 chart, finding in all cases that search counts are highly predictive of future outcomes. We also find that search counts generally boost the performance of baseline models fit on other publicly available data, where the boost varies from modest to dramatic, depending on the application in question. Finally, we reexamine previous work on tracking flu trends and show that, perhaps surprisingly, the utility of search data relative to a simple autoregressive model is modest. We conclude that in the absence of other data sources, or where small improvements in predictive performance are material, search queries provide a useful guide to the near future.

Models of contagion arise broadly both in the biological and social sciences, with applications ranging from the transmission of infectious diseases to the diffusion of innovations and the spread of cultural fads. In this Letter, we introduce a general model of contagion which, by explicitly incorporating memory of past exposures to, for example, an infectious agent, rumor, or new product, includes the main features of existing contagion models and interpolates between them. We obtain exact solutions for a simple version of the model, finding that under general conditions only three classes of collective dynamics exist, two of which correspond to familiar epidemic threshold and critical mass dynamics, while the third is a distinct intermediate case. We find that for a given length of memory, the class into which a particular system falls is determined by two parameters, each of which ought to be measurable empirically. Our model suggests novel measures for assessing the susceptibility of a population to large contagion events, and also a possible strategy for inhibiting or facilitating them.

Viral products and ideas are intuitively understood to grow through a person-to-person diffusion process analogous to the spread of an infectious disease; however, until recently it has been prohibitively difficult to directly observe purportedly viral events, and thus to rigorously quantify or characterize their structural properties. Here we propose a formal measure of what we label “structural virality” that interpolates between two conceptual extremes: content that gains its popularity through a single, large broadcast and that which grows through multiple generations with any one individual directly responsible for only a fraction of the total adoption. We use this notion of structural virality to analyze a unique data set of a billion diffusion events on Twitter, including the propagation of news stories, videos, images, and petitions. We find that across all domains and all sizes of events, online diffusion is characterized by surprising structural diversity; that is, popular events regularly grow via both broadcast and viral mechanisms, as well as essentially all conceivable combinations of the two. Nevertheless, we find that structural virality is typically low, and remains so independent of size, suggesting that popularity is largely driven by the size of the largest broadcast. Finally, we attempt to replicate these findings with a model of contagion characterized by a low infection rate spreading on a scale-free network. We find that although several of our empirical findings are consistent with such a model, it fails to replicate the observed diversity of structural virality, thereby suggesting new directions for future modeling efforts. This paper was accepted by Lorin Hitt, information systems.

The relationship between financial incentives and performance, long of interest to social scientists, has gained new relevance with the advent of web-based "crowd-sourcing" models of production. Here we investigate the effect of compensation on performance in the context of two experiments, conducted on Amazon's Mechanical Turk (AMT). We find that increased financial incentives increase the quantity, but not the quality, of work performed by participants, where the difference appears to be due to an "anchoring" effect: workers who were paid more also perceived the value of their work to be greater, and thus were no more motivated than workers paid less. In contrast with compensation levels, we find the details of the compensation scheme do matter---specifically, a "quota" system results in better work for less pay than an equivalent "piece rate" system. Although counterintuitive, these findings are consistent with previous laboratory studies, and may have real-world analogs as well.

The small-world network model is a simple model of the structure of social networks, which possesses characteristics of both regular lattices and random graphs. The model consists of a one-dimensional lattice with a low density of shortcuts added between randomly selected pairs of points. These shortcuts greatly reduce the typical path length between any two points on the lattice. We present a mean-field solution for the average path length and for the distribution of path lengths in the model. This solution is exact in the limit of large system size and either a large or small number of shortcuts.

The relationship between financial incentives and performance, long of interest to social scientists, has gained new relevance with the advent of web-based "crowd-sourcing" models of production. Here we investigate the effect of compensation on performance in the context of two experiments, conducted on Amazon's Mechanical Turk (AMT). We find that increased financial incentives increase the quantity, but not the quality, of work performed by participants, where the difference appears to be due to an "anchoring" effect: workers who were paid more also perceived the value of their work to be greater, and thus were no more motivated than workers paid less. In contrast with compensation levels, we find the details of the compensation scheme do matter--specifically, a "quota" system results in better work for less pay than an equivalent "piece rate" system. Although counterintuitive, these findings are consistent with previous laboratory studies, and may have real-world analogs as well.

We present a model of contagion that unifies and generalizes existing models of the spread of social influences and micro-organismal infections. Our model incorporates individual memory of exposure to a contagious entity (e.g., a rumor or disease), variable magnitudes of exposure (dose sizes), and heterogeneity in the susceptibility of individuals. Through analysis and simulation, we examine in detail the case where individuals may recover from an infection and then immediately become susceptible again (analogous to the so-called SIS model). We identify three basic classes of contagion models which we call \textit{epidemic threshold}, \textit{vanishing critical mass}, and \textit{critical mass} classes, where each class of models corresponds to different strategies for prevention or facilitation. We find that the conditions for a particular contagion model to belong to one of the these three classes depend only on memory length and the probabilities of being infected by one and two exposures respectively. These parameters are in principle measurable for real contagious influences or entities, thus yielding empirical implications for our model. We also study the case where individuals attain permanent immunity once recovered, finding that epidemics inevitably die out but may be surprisingly persistent when individuals possess memory.

A longstanding idea in the literature on human cooperation is that cooperation should be reinforced when conditional cooperators are more likely to interact. In the context of social networks, this idea implies that cooperation should fare better in highly clustered networks such as cliques than in networks with low clustering such as random networks. To test this hypothesis, we conducted a series of web-based experiments, in which 24 individuals played a local public goods game arranged on one of five network topologies that varied between disconnected cliques and a random regular graph. In contrast with previous theoretical work, we found that network topology had no significant effect on average contributions. This result implies either that individuals are not conditional cooperators, or else that cooperation does not benefit from positive reinforcement between connected neighbors. We then tested both of these possibilities in two subsequent series of experiments in which artificial seed players were introduced, making either full or zero contributions. First, we found that although players did generally behave like conditional cooperators, they were as likely to decrease their contributions in response to low contributing neighbors as they were to increase their contributions in response to high contributing neighbors. Second, we found that positive effects of cooperation were contagious only to direct neighbors in the network. In total we report on 113 human subjects experiments, highlighting the speed, flexibility, and cost-effectiveness of web-based experiments over those conducted in physical labs.

Models of networked diffusion that are motivated by analogy with the spread of infectious disease have been applied to a wide range of social and economic adoption processes, including those related to new products, ideas, norms and behaviors. However, it is unknown how accurately these models account for the empirical structure of diffusion over networks. Here we describe the diffusion patterns arising from seven online domains, ranging from communications platforms to networked games to microblogging services, each involving distinct types of content and modes of sharing. We find strikingly similar patterns across all domains.

Social networks are of interest to researchers in part because they are thought to mediate the flow of information in communities and organizations. Here we study the temporal dynamics of communication using on-line data, including e-mail communication among the faculty and staff of a large university over a two-year period. We formulate a temporal notion of "distance" in the underlying social network by measuring the minimum time required for information to spread from one node to another - a concept that draws on the notion of vector-clocks from the study of distributed computing systems. We find that such temporal measures provide structural insights that are not apparent from analyses of the pure social network topology. In particular, we define the network backbone to be the subgraph consisting of edges on which information has the potential to flow the quickest. We find that the backbone is a sparse graph with a concentration of both highly embedded edges and long-range bridges - a finding that sheds new light on the relationship between tie strength and connectivity in social networks.

Complex problems in science, business, and engineering typically require some tradeoff between exploitation of known solutions and exploration for novel ones, where, in many cases, information about known solutions can also disseminate among individual problem solvers through formal or informal networks. Prior research on complex problem solving by collectives has found the counterintuitive result that inefficient networks, meaning networks that disseminate information relatively slowly, can perform better than efficient networks for problems that require extended exploration. In this paper, we report on a series of 256 Web-based experiments in which groups of 16 individuals collectively solved a complex problem and shared information through different communication networks. As expected, we found that collective exploration improved average success over independent exploration because good solutions could diffuse through the network. In contrast to prior work, however, we found that efficient networks outperformed inefficient networks, even in a problem space with qualitative properties thought to favor inefficient networks. We explain this result in terms of individual-level explore-exploit decisions, which we find were influenced by the network structure as well as by strategic considerations and the relative payoff between maxima. We conclude by discussing implications for real-world problem solving and possible extensions.

If handled appropriately, data about Internet-based communication and interactivity could revolutionize our understanding of collective human behaviour. Duncan Watts, who with Steven Strogatz published the Nature paper that introduced the 'small world' phenomenon in 1998 (later familiar as 'six degrees of separation'), puts in a word for the social sciences in this week's Essay. Starting from the premise that the major problems facing humanity today are social and economic, Watts constructs an early bid for social science as the science of the twenty-first century. Work on social networks and the data collecting power of the Internet are strong foundations for a new science that could revolutionize our understanding of collective human behaviour.

Historically, social scientists have sought out explanations of human and social phenomena that provide interpretable causal mechanisms, while often ignoring their predictive accuracy. We argue that the increasingly computational nature of social science is beginning to reverse this traditional bias against prediction; however, it has also highlighted three important issues that require resolution. First, current practices for evaluating predictions must be better standardized. Second, theoretical limits to predictive accuracy in complex social systems must be better characterized, thereby setting expectations for what can be predicted or explained. Third, predictive accuracy and interpretability must be recognized as complements, not substitutes, when evaluating explanations. Resolving these three issues will lead to better, more replicable, and more useful social science.

Mainstream news, mainly on television, vastly outweighs fake news, and news itself is a small fraction of U.S. media consumption.

Data sharing, research ethics, and incentives must improve

The natural tendency for humans to make and break relationships is thought to facilitate the emergence of cooperation. In particular, allowing conditional cooperators to choose with whom they interact is believed to reinforce the rewards accruing to mutual cooperation while simultaneously excluding defectors. Here we report on a series of human subjects experiments in which groups of 24 participants played an iterated prisoner's dilemma game where, critically, they were also allowed to propose and delete links to players of their own choosing at some variable rate. Over a wide variety of parameter settings and initial conditions, we found that dynamic partner updating significantly increased the level of cooperation, the average payoffs to players, and the assortativity between cooperators. Even relatively slow update rates were sufficient to produce large effects, while subsequent increases to the update rate had progressively smaller, but still positive, effects. For standard prisoner's dilemma payoffs, we also found that assortativity resulted predominantly from cooperators avoiding defectors, not by severing ties with defecting partners, and that cooperation correspondingly suffered. Finally, by modifying the payoffs to satisfy two novel conditions, we found that cooperators did punish defectors by severing ties, leading to higher levels of cooperation that persisted for longer.

Although it is under-studied relative to other social media platforms, YouTube is arguably the largest and most engaging online media consumption platform in the world. Recently, YouTube's scale has fueled concerns that YouTube users are being radicalized via a combination of biased recommendations and ostensibly apolitical "anti-woke" channels, both of which have been claimed to direct attention to radical political content. Here we test this hypothesis using a representative panel of more than 300,000 Americans and their individual-level browsing behavior, on and off YouTube, from January 2016 through December 2019. Using a labeled set of political news channels, we find that news consumption on YouTube is dominated by mainstream and largely centrist sources. Consumers of far-right content, while more engaged than average, represent a small and stable percentage of news consumers. However, consumption of "anti-woke" content, defined in terms of its opposition to progressive intellectual and political agendas, grew steadily in popularity and is correlated with consumption of far-right content off-platform. We find no evidence that engagement with far-right content is caused by YouTube recommendations systematically, nor do we find clear evidence that anti-woke channels serve as a gateway to the far right. Rather, consumption of political content on YouTube appears to reflect individual preferences that extend across the web as a whole.

Although population structure has long been recognized as relevant to the spread of infectious disease, traditional mathematical models have understated the role of nonhomogenous mixing in populations with geographical and social structure. Recently, a wide variety of spatial and network models have been proposed that incorporate various aspects of interaction structure among individuals. However, these more complex models necessarily suffer from limited tractability, rendering general conclusions difficult to draw. In seeking a compromise between parsimony and realism, we introduce a class of metapopulation models in which we assume homogeneous mixing holds within local contexts, and that these contexts are embedded in a nested hierarchy of successively larger domains. We model the movement of individuals between contexts via simple transport parameters and allow diseases to spread stochastically. Our model exhibits some important stylized features of real epidemics, including extreme size variation and temporal heterogeneity, that are difficult to characterize with traditional measures. In particular, our results suggest that when epidemics do occur the basic reproduction number R 0 may bear little relation to their final size. Informed by our model's behavior, we suggest measures for characterizing epidemic thresholds and discuss implications for the control of epidemics.

We study the problem of cooperative behavior emerging in an environment where individual behaviors and interaction structures coevolve. Players not only learn which strategy to adopt by imitating the strategy of the best-performing player they observe, but also choose with whom they should interact by selectively creating and/or severing ties with other players based on a myopic cost-benefit comparison. We find that scalable cooperation—that is, high levels of cooperation in large populations—can be achieved in sparse networks, assuming that individuals are able to sever ties unilaterally and that new ties can only be created with the mutual consent of both parties. Detailed examination shows that there is an important trade-off between local reinforcement and global expansion in achieving cooperation in dynamic networks. As a result, networks in which ties are costly and local structure is largely absent tend to generate higher levels of cooperation than those in which ties are made easily and friends of friends interact with high probability, where the latter result contrasts strongly with the usual intuition.

Individuals influence each others' decisions about cultural products such as songs, books, and movies; but to what extent can the perception of success become a "self-fulfilling prophecy"? We have explored this question experimentally by artificially inverting the true popularity of songs in an online "music market," in which 12,207 participants listened to and downloaded songs by unknown bands. We found that most songs experienced self-fulfilling prophecies, in which perceived-but initially false-popularity became real over time. We also found, however, that the inversion was not self-fulfilling for the market as a whole, in part because the very best songs recovered their popularity in the long run. Moreover, the distortion of market information reduced the correlation between appeal and popularity, and led to fewer overall downloads. These results, although partial and speculative, suggest a new approach to the study of cultural markets, and indicate the potential of web-based experiments to explore the social psychological origin of other macro-sociological phenomena.

The dynamics of information exchange is an important but understudied aspect of collective communication, coordination, and problem solving in a wide range of distributed systems, both physical (e.g., the Internet) and social (e.g., business firms). In this paper, we introduce a model of organizational networks according to which links are added incrementally to a hierarchical backbone and test the resulting networks under variable conditions of information exchange. Our main result is the identification of a class of multiscale networks that reduce, over a wide range of environments, the likelihood that individual nodes will suffer congestion-related failure and that the network as a whole will disintegrate when failures do occur. We call this dual robustness property of multiscale networks “ultrarobustness.” Furthermore, we find that multiscale networks attain most of their robustness with surprisingly few link additions, suggesting that ultrarobust organizational networks can be generated in an efficient and scalable manner. Our results are directly relevant to the relief of congestion in communication networks and also more broadly to activities, like distributed problem solving, that require individuals to exchange information in an unpredictable manner.

It is often asserted that friends and acquaintances have more similar beliefs and attitudes than do strangers; yet empirical studies disagree over exactly how much diversity of opinion exists within local social networks and, relatedly, how much awareness individuals have of their neighbors' views. This article reports results from a network survey, conducted on the Facebook social networking platform, in which participants were asked about their own political attitudes, as well as their beliefs about their friends' attitudes. Although considerable attitude similarity exists among friends, the results show that friends disagree more than they think they do. In particular, friends are typically unaware of their disagreements, even when they say they discuss the topic, suggesting that discussion is not the primary means by which friends infer each other's views on particular issues. Rather, it appears that respondents infer opinions in part by relying on stereotypes of their friends and in part by projecting their own views. The resulting gap between real and perceived agreement may have implications for the dynamics of political polarization and theories of social influence in general.

Researchers increasingly use electronic communication data to construct and study large social networks, effectively inferring unobserved ties (e.g. i is connected to j) from observed communication events (e.g. i emails j). Often overlooked, however, is the impact of tie definition on the corresponding network, and in turn the relevance of the inferred network to the research question of interest. Here we study the problem of network inference and relevance for two email data sets of different size and origin. In each case, we generate a family of networks parameterized by a threshold condition on the frequency of emails exchanged between pairs of individuals. After demonstrating that different choices of the threshold correspond to dramatically different network structures, we then formulate the relevance of these networks in terms of a series of prediction tasks that depend on various network features. In general, we find: a) that prediction accuracy is maximized over a non-trivial range of thresholds corresponding to 5-10 reciprocated emails per year; b) that for any prediction task, choosing the optimal value of the threshold yields a sizable (~30%) boost in accuracy over naive choices; and c) that the optimal threshold value appears to be (somewhat surprisingly) consistent across data sets and prediction tasks. We emphasize the practical utility in defining ties via their relevance to the prediction task(s) at hand and discuss implications of our empirical results.

A longstanding idea in the literature on human cooperation is that cooperation should be reinforced when conditional cooperators are more likely to interact. In the context of social networks, this idea implies that cooperation should fare better in highly clustered networks such as cliques than in networks with low clustering such as random networks. To test this hypothesis, we conducted a series of web-based experiments, in which 24 individuals played a local public goods game arranged on one of five network topologies that varied between disconnected cliques and a random regular graph. In contrast with previous theoretical work, we found that network topology had no significant effect on average contributions. This result implies either that individuals are not conditional cooperators, or else that cooperation does not benefit from positive reinforcement between connected neighbors. We then tested both of these possibilities in two subsequent series of experiments in which artificial seed players were introduced, making either full or zero contributions. First, we found that although players did generally behave like conditional cooperators, they were as likely to decrease their contributions in response to low contributing neighbors as they were to increase their contributions in response to high contributing neighbors. Second, we found that positive effects of cooperation were contagious only to direct neighbors in the network. In total we report on 113 human subjects experiments, highlighting the speed, flexibility, and cost-effectiveness of web-based experiments over those conducted in physical labs.

Sociologists have long advocated a sociological approach to explanation by contrasting it with common sense. The argument of this article, however, is that sociologists rely on common sense more than they realize. Moreover, this unacknowledged reliance causes serious problems for their explanations of social action, that is, for why people do what they do. Many such explanations, it is argued, conflate understandability with causality in ways that are not valid by the standards of scientific explanation. It follows that if sociologists want their explanations to be scientifically valid, they must evaluate them specifically on those grounds—in particular, by forcing them to make predictions. In becoming more scientific, however, it is predicted that sociologists’ explanations will also become less satisfying from an intuitive, sense-making perspective. Even as novel sources of data and improved methods open exciting new directions for sociological research, therefore, sociologists will increasingly have to choose between unsatisfying scientific explanations and satisfying but unscientific stories.

The relationship between team size and productivity is a question of broad relevance across economics, psychology, and management science. For complex tasks, however, where both the potential benefits and costs of coordinated work increase with the number of workers, neither theoretical arguments nor empirical evidence consistently favor larger vs. smaller teams. Experimental findings, meanwhile, have relied on small groups and highly stylized tasks, hence are hard to generalize to realistic settings. Here we narrow the gap between real-world task complexity and experimental control, reporting results from an online experiment in which 47 teams of size ranging from n = 1 to 32 collaborated on a realistic crisis mapping task. We find that individuals in teams exerted lower overall effort than independent workers, in part by allocating their effort to less demanding (and less productive) sub-tasks; however, we also find that individuals in teams collaborated more with increasing team size. Directly comparing these competing effects, we find that the largest teams outperformed an equivalent number of independent workers, suggesting that gains to collaboration dominated losses to effort. Importantly, these teams also performed comparably to a field deployment of crisis mappers, suggesting that experiments of the type described here can help solve practical problems as well as advancing the science of collective intelligence.

How predictable is success in complex social systems? In spite of a recent profusion of prediction studies that exploit online social and information network data, this question remains unanswered, in part because it has not been adequately specified. In this paper we attempt to clarify the question by presenting a simple stylized model of success that attributes prediction error to one of two generic sources: insufficiency of available data and/or models on the one hand; and inherent unpredictability of complex social systems on the other. We then use this model to motivate an illustrative empirical study of information cascade size prediction on Twitter. Despite an unprecedented volume of information about users, content, and past performance, our best performing models can explain less than half of the variance in cascade sizes. In turn, this result suggests that even with unlimited data predictive performance would be bounded well below deterministic accuracy. Finally, we explore this potential bound theoretically using simulations of a diffusion process on a random scale free network similar to Twitter. We show that although higher predictive power is possible in theory, such performance requires a homogeneous system and perfect ex-ante knowledge of it: even a small degree of uncertainty in estimating product quality or slight variation in quality across products leads to substantially more restrictive bounds on predictability. We conclude that realistic bounds on predictive accuracy are not dissimilar from those we have obtained empirically, and that such bounds for other complex social systems for which data is more difficult to obtain are likely even lower.

Since the 2016 US presidential election, the deliberate spread of misinformation online, and on social media in particular, has generated extraordinary concern, in large part because of its potential effects on public opinion, political polarization, and ultimately democratic decision making. Recently, however, a handful of papers have argued that both the prevalence and consumption of “fake news” per se is extremely low compared with other types of news and news-relevant content. Although neither prevalence nor consumption is a direct measure of influence, this work suggests that proper understanding of misinformation and its effects requires a much broader view of the problem, encompassing biased and misleading—but not necessarily factually incorrect—information that is routinely produced or amplified by mainstream news organizations. In this paper, we propose an ambitious collective research agenda to measure the origins, nature, and prevalence of misinformation, broadly construed, as well as its impact on democracy. We also sketch out some illustrative examples of completed, ongoing, or planned research projects that contribute to this agenda.

In a large-scale, preregistered experiment on informal political communication, we algorithmically matched participants, varying two dimensions: 1) the degree of incidental similarity on nonpolitical features; and 2) their stance agreement on a contentious political topic. Matched participants were first shown a computer-generated social media profile of their match highlighting all the shared nonpolitical features; then, they read a short, personal, but argumentative, essay written by their match about the reduction of inequality via redistribution of wealth by the government. We show that support for redistribution increased and polarization decreased for participants with both mild and strong views, regardless of their political leaning. We further show that feeling close to the match is associated with an 86% increase in the probability of assimilation of political views. Our analysis also uncovers an asymmetry: Interacting with someone with opposite views greatly reduced feelings of closeness; however, interacting with someone with consistent views only moderately increased them. By extending previous work about the effects of incidental similarity and shared identity on affect into the domain of political opinion change, our results bear real-world implications for the (re)-design of social media platforms. Because many people prefer to keep politics outside of their social networks, encouraging cross-cutting political communication based on nonpolitical commonalities is a potential solution for fostering consensus on potentially divisive and partisan topics.

Page 1 of 2 http://harvardbusinessonline.hbsp.harvard.edu/hbsp/hbr/articles/article.jsp?articleID=F0705A&ml_action=get-article&print=true Duncan J. Watts (djw24@columbia.edu) is a professor of sociology at Columbia University, in New York, where he directs the Collective Dynamics Group. He is the author of Six Degrees: The Science of a Connected Age (Norton, 2003). Jonah Peretti (jonah@buzzfeed.com) is a founding partner of the Huffington Post, BuzzFeed, and ContagiousMedia.org.

Social scientists are often interested in understanding how the dynamics of social systems are driven by the behavior of individuals that make up those systems. However, this process is hindered by the difficulty of experimentally studying how individual behavioral tendencies lead to collective social dynamics in large groups of people interacting over time. In this study, we investigate the role of social influence, a process well studied at the individual level, on the puzzling nature of success for cultural products such as books, movies, and music. Using a "multiple-worlds" experimental design, we are able to isolate the causal effect of an individual-level mechanism on collective social outcomes. We employ this design in a Web-based experiment in which 2,930 participants listened to, rated, and downloaded 48 songs by up-and-coming bands. Surprisingly, despite relatively large differences in the demographics, behavior, and preferences of participants, the experimental results at both the individual and collective levels were similar to those found in Salganik, Dodds, and Watts (2006). Further, by comparing results from two distinct pools of participants, we are able to gain new insights into the role of individual behavior on collective outcomes. We conclude with a discussion of the strengths and weaknesses of Web-based experiments to address questions of collective social dynamics.

In this paper we address the general question of how social influence determines collective outcomes for large populations of individuals faced with binary decisions. First, we define conditions under which the behavior of individuals making binary decisions can be described in terms of what we call an influence-response function: a one-dimensional function of the (weighted) number of individuals choosing each of the alternatives. And second, we demonstrate that, under the assumptions of global and anonymous interactions, general knowledge of the influence-response functions is sufficient to compute equilibrium, and even non-equilibrium, properties of the collective dynamics. By enabling us to treat in a consistent manner classes of decisions that have previously been analyzed separately, our framework allows us to find similarities between apparently quite different kinds of decision situations, and conversely to identify important differences between decisions that would otherwise appear very similar.

Recommendation systems are an increasingly prominent part of the web, accounting for up to a third of all traffic on several of the world's most popular sites. Nevertheless, little is known about how much activity such systems actually cause over and above activity that would have occurred via other means (e.g., search) if recommendations were absent. Although the ideal way to estimate the causal impact of recommendations is via randomized experiments, such experiments are costly and may inconvenience users. In this paper, therefore, we present a method for estimating causal effects from purely observational data. Specifically, we show that causal identification through an instrumental variable is possible when a product experiences an instantaneous shock in direct traffic and the products recommended next to it do not. We then apply our method to browsing logs containing anonymized activity for 2.1 million users on Amazon.com over a 9 month period and analyze over 4,000 unique products that experience such shocks. We find that although recommendation click-throughs do account for a large fraction of traffic among these products, at least 75% of this activity would likely occur in the absence of recommendations. We conclude with a discussion about the assumptions under which the method is appropriate and caveats around extrapolating results to other products, sites, or settings.

Randomized experiments have enormous potential to improve human welfare in many domains, including healthcare, education, finance, and public policy. However, such "A/B tests" are often criticized on ethical grounds even as similar, untested interventions are implemented without objection. We find robust evidence across 16 studies of 5,873 participants from three diverse populations spanning nine domains-from healthcare to autonomous vehicle design to poverty reduction-that people frequently rate A/B tests designed to establish the comparative effectiveness of two policies or treatments as inappropriate even when universally implementing either A or B, untested, is seen as appropriate. This "A/B effect" is as strong among those with higher educational attainment and science literacy and among relevant professionals. It persists even when there is no reason to prefer A to B and even when recipients are treated unequally and randomly in all conditions (A, B, and A/B). Several remaining explanations for the effect-a belief that consent is required to impose a policy on half of a population but not on the entire population; an aversion to controlled but not to uncontrolled experiments; and a proxy form of the illusion of knowledge (according to which randomized evaluations are unnecessary because experts already do or should know "what works")-appear to contribute to the effect, but none dominates or fully accounts for it. We conclude that rigorously evaluating policies or treatments via pragmatic randomized trials may provoke greater objection than simply implementing those same policies or treatments untested.

Complexity-defined in terms of the number of components and the nature of the interdependencies between them-is clearly a relevant feature of all tasks that groups perform. Yet the role that task complexity plays in determining group performance remains poorly understood, in part because no clear language exists to express complexity in a way that allows for straightforward comparisons across tasks. Here we avoid this analytical difficulty by identifying a class of tasks for which complexity can be varied systematically while keeping all other elements of the task unchanged. We then test the effects of task complexity in a preregistered two-phase experiment in which 1,200 individuals were evaluated on a series of tasks of varying complexity (phase 1) and then randomly assigned to solve similar tasks either in interacting groups or as independent individuals (phase 2). We find that interacting groups are as fast as the fastest individual and more efficient than the most efficient individual for complex tasks but not for simpler ones. Leveraging our highly granular digital data, we define and precisely measure group process losses and synergistic gains and show that the balance between the two switches signs at intermediate values of task complexity. Finally, we find that interacting groups generate more solutions more rapidly and explore the solution space more broadly than independent problem solvers, finding higher-quality solutions than all but the highest-scoring individuals.

Partisan segregation within the news audience buffers many Americans from countervailing political views, posing a risk to democracy. Empirical studies of the online media ecosystem suggest that only a small minority of Americans, driven by a mix of demand and algorithms, are siloed according to their political ideology. However, such research omits the comparatively larger television audience and often ignores temporal dynamics underlying news consumption. By analyzing billions of browsing and viewing events between 2016 and 2019, with a novel framework for measuring partisan audiences, we first estimate that 17% of Americans are partisan-segregated through television versus roughly 4% online. Second, television news consumers are several times more likely to maintain their partisan news diets month-over-month. Third, TV viewers' news diets are far more concentrated on preferred sources. Last, partisan news channels' audiences are growing even as the TV news audience is shrinking. Our results suggest that television is the top driver of partisan audience segregation among Americans.

Abstract The dominant paradigm of experiments in the social and behavioral sciences views an experiment as a test of a theory, where the theory is assumed to generalize beyond the experiment's specific conditions. According to this view, which Alan Newell once characterized as “playing twenty questions with nature,” theory is advanced one experiment at a time, and the integration of disparate findings is assumed to happen via the scientific publishing process. In this article, we argue that the process of integration is at best inefficient, and at worst it does not, in fact, occur. We further show that the challenge of integration cannot be adequately addressed by recently proposed reforms that focus on the reliability and replicability of individual findings, nor simply by conducting more or larger experiments. Rather, the problem arises from the imprecise nature of social and behavioral theories and, consequently, a lack of commensurability across experiments conducted under different conditions. Therefore, researchers must fundamentally rethink how they design experiments and how the experiments relate to theory. We specifically describe an alternative framework, integrative experiment design, which intrinsically promotes commensurability and continuous integration of knowledge. In this paradigm, researchers explicitly map the design space of possible experiments associated with a given research question, embracing many potentially relevant theories rather than focusing on just one. The researchers then iteratively generate theories and test them with experiments explicitly sampled from the design space, allowing results to be integrated across experiments. Given recent methodological and technological developments, we conclude that this approach is feasible and would generate more-reliable, more-cumulative empirical and theoretical knowledge than the current paradigm—and with far greater efficiency.

The notion of common sense is invoked so frequently in contexts as diverse as everyday conversation, political debates, and evaluations of artificial intelligence that its meaning might be surmised to be unproblematic. Surprisingly, however, neither the intrinsic properties of common sense knowledge (what makes a claim commonsensical) nor the degree to which it is shared by people (its “commonness”) have been characterized empirically. In this paper, we introduce an analytical framework for quantifying both these elements of common sense. First, we define the commonsensicality of individual claims and people in terms of the latter’s propensity to agree on the former and their awareness of one another’s agreement. Second, we formalize the commonness of common sense as a clique detection problem on a bipartite belief graph of people and claims, defining <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" display="inline" overflow="scroll"> <mml:mrow> <mml:mi mathvariant="italic">pq</mml:mi> </mml:mrow> </mml:math> common sense as the fraction <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" display="inline" overflow="scroll"> <mml:mi>q</mml:mi> </mml:math> of claims shared by a fraction <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" display="inline" overflow="scroll"> <mml:mi>p</mml:mi> </mml:math> of people. Evaluating our framework on a dataset of <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" display="inline" overflow="scroll"> <mml:mrow> <mml:mn>2</mml:mn> <mml:mo>,</mml:mo> <mml:mn>046</mml:mn> </mml:mrow> </mml:math> raters evaluating <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" display="inline" overflow="scroll"> <mml:mrow> <mml:mn>4</mml:mn> <mml:mo>,</mml:mo> <mml:mn>407</mml:mn> </mml:mrow> </mml:math> diverse claims, we find that commonsensicality aligns most closely with plainly worded, fact-like statements about everyday physical reality. Psychometric attributes such as social perceptiveness influence individual common sense, but surprisingly demographic factors such as age or gender do not. Finally, we find that collective common sense is rare: At most, a small fraction <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" display="inline" overflow="scroll"> <mml:mi>p</mml:mi> </mml:math> of people agree on more than a small fraction <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" display="inline" overflow="scroll"> <mml:mi>q</mml:mi> </mml:math> of claims. Together, these results undercut universalistic beliefs about common sense and raise questions about its variability that are relevant both to human and artificial intelligence.

The increasing availability of electronic communication data, such as that arising from e-mail exchange, presents social and information scientists with new possibilities for characterizing individual behavior and, by extension, identifying latent structure in human populations. Here, we propose a model of individual e-mail communication that is sufficiently rich to capture meaningful variability across individuals, while remaining simple enough to be interpretable. We show that the model, a cascading non-homogeneous Poisson process, can be formulated as a double-chain hidden Markov model, allowing us to use an efficient inference algorithm to estimate the model parameters from observed data. We then apply this model to two e-mail data sets consisting of 404 and 6,164 users, respectively, that were collected from two universities in different countries and years. We find that the resulting best-estimate parameter distributions for both data sets are surprisingly similar, indicating that at least some features of communication dynamics generalize beyond specific contexts. We also find that variability of individual behavior over time is significantly less than variability across the population, suggesting that individuals can be classified into persistent "types". We conclude that communication patterns may prove useful as an additional class of attribute data, complementing demographic and network data, for user classification and outlier detection-a point that we illustrate with an interpretable clustering of users based on their inferred model parameters.

Do social movements actively shape the opinions and attitudes of participants by bringing together diverse groups that subsequently influence one another?Ethnographic studies of the 2013 Gezi uprising seem to answer "yes, " pointing to solidarity among groups that were traditionally indifferent, or even hostile, to one another.We argue that two mechanisms with differing implications may generate this observed outcome: "influence" (change in attitude caused by interacting with other participants); and "selection" (individuals who participated in the movement were generally more supportive of other groups beforehand).We tease out the relative importance of these mechanisms by constructing a panel of over 30,000 Twitter users and analyzing their support for the main Turkish opposition parties before, during, and after the movement.We find that although individuals changed in significant ways, becoming in general more supportive of the other opposition parties, those who participated in the movement were also significantly more supportive of the other parties all along.These findings suggest that both mechanisms were important, but that selection dominated.In addition to our substantive findings, our paper also makes a methodological contribution that we believe could be useful to studies of social movements and mass opinion change more generally.In contrast with traditional panel studies, which must be designed and implemented prior to the event of interest, our method relies on ex post panel construction, and hence can be used to study unanticipated or otherwise inaccessible events.We conclude that despite the well known limitations of social media, their "always on" nature and their widespread availability offer an important source of public opinion data.

Learning in finitely repeated games of cooperation remains poorly understood in part because their dynamics play out over a timescale exceeding that of traditional lab experiments. Here, we report results of a virtual lab experiment in which 94 subjects play up to 400 ten-round games of Prisoner's Dilemma over the course of twenty consecutive weekdays. Consistent with previous work, the typical round of first defection moves earlier for several days; however, this unravelling process stabilizes after roughly one week. Analysing individual strategies, we find that approximately 40% of players behave as resilient cooperators who avoid unravelling even at significant cost to themselves. Finally, using a standard learning model we predict that a sufficiently large minority of resilient cooperators can permanently stabilize unravelling among a majority of rational players. These results shed hopeful light on the long-term dynamics of cooperation, and demonstrate the importance of long-run experiments.

Virtual labs allow researchers to design high-throughput and macro-level experiments that are not feasible in traditional in-person physical lab settings. Despite the increasing popularity of online research, researchers still face many technical and logistical barriers when designing and deploying virtual lab experiments. While several platforms exist to facilitate the development of virtual lab experiments, they typically present researchers with a stark trade-off between usability and functionality. We introduce Empirica: a modular virtual lab that offers a solution to the usability-functionality trade-off by employing a "flexible defaults" design strategy. This strategy enables us to maintain complete "build anything" flexibility while offering a development platform that is accessible to novice programmers. Empirica's architecture is designed to allow for parameterizable experimental designs, reusable protocols, and rapid development. These features will increase the accessibility of virtual lab experiments, remove barriers to innovation in experiment design, and enable rapid progress in the understanding of distributed human computation.

Citing recent successes in forecasting elections, movies, products, and other outcomes, prediction market advocates call for widespread use of market-based methods for government and corporate decision making. Though theoretical and empirical evidence suggests that markets do often outperform alternative mechanisms, less attention has been paid to the magnitude of improvement. Here we compare the performance of prediction markets to conventional methods of prediction, namely polls and statistical models. Examining thousands of sporting and movie events, we find that the relative advantage of prediction markets is surprisingly small, as measured by squared error, calibration, and discrimination. Moreover, these domains also exhibit remarkably steep diminishing returns to information, with nearly all the predictive power captured by only two or three parameters. As policy makers consider adoption of prediction markets, costs should be weighed against potentially modest benefits.

What explains the relative persistence of same-race romantic relationships?One possible explanation is structural-this phenomenon could reflect the fact that social interactions are already stratified along racial lines-while another attributes these patterns to individual-level preferences.We present novel evidence from an online dating community involving more than 250,000 people in the United States about the frequency with which individuals both express a preference for same-race romantic partners and act to choose same-race partners.Prior work suggests that political ideology is an important correlate of conservative attitudes about race in the United States, and we find that conservatives, including both men and women and blacks and whites, are much more likely than liberals to state a preference for same-race partners.Further, conservatives are not simply more selective in general; they are specifically selective with regard to race.Do these stated preferences predict real behaviors?In general, we find that stated preferences are a strong predictor of a behavioral preference for same-race partners, and that this pattern persists across ideological groups.At the same time, both men and women of all political persuasions act as if they prefer same-race relationships even when they claim not to.As a result, the gap between conservatives and liberals in revealed same-race preferences, while still substantial, is not as pronounced as their stated attitudes would suggest.We conclude by discussing some implications of our findings for the broader issues of racial homogamy and segregation.

Online platforms have banned ("deplatformed") influencers, communities, and even entire websites to reduce content deemed harmful. Deplatformed users often migrate to alternative platforms, which raises concerns about the effectiveness of deplatforming. Here, we study the deplatforming of Parler, a fringe social media platform, between 2020 January 11 and 2021 February 25, in the aftermath of the US Capitol riot. Using two large panels that capture longitudinal user-level activity across mainstream and fringe social media content (N = 112, 705, adjusted to be representative of US desktop and mobile users), we find that other fringe social media, such as Gab and Rumble, prospered after Parler's deplatforming. Further, the overall activity on fringe social media increased while Parler was offline. Using a difference-in-differences analysis (N = 996), we then identify the causal effect of deplatforming on active Parler users, finding that deplatforming increased the probability of daily activity across other fringe social media in early 2021 by 10.9 percentage points (pp) (95% CI [5.9 pp, 15.9 pp]) on desktop devices, and by 15.9 pp (95% CI [10.2 pp, 21.7 pp]) on mobile devices, without decreasing activity on fringe social media in general (including Parler). Our results indicate that the isolated deplatforming of a major fringe platform was ineffective at reducing overall user activity on fringe social media.

Many companies have proprietary resources and/or data that are indispensable for research, and academics provide the creative fuel for much early-stage research that leads to industrial innovation. It is essential to the health of the research enterprise that collaborations between industrial and university researchers flourish. This system of collaboration is under strain. Financial motivations driving product development have led to concerns that industry-sponsored research comes at the expense of transparency (1). Yet many industry researchers distrust quality control in academia (2) and question whether academics value reproducibility as much as rapid publication. Cultural differences between industry and academia can create or increase difficulties in reproducing research findings. We discuss key aspects of this problem that industry-academia collaborations must address and for which other stakeholders, from funding agencies to journals, can provide leadership and support.

Abstract Surveys are a vital tool for understanding public opinion and knowledge, but they can also yield biased estimates of behavior. Here we explore a popular and important behavior that is frequently measured in public opinion surveys: news consumption. Previous studies have shown that television news consumption is consistently overreported in surveys relative to passively collected behavioral data. We validate these earlier findings, showing that they continue to hold despite large shifts in news consumption habits over time, while also adding some new nuance regarding question wording. We extend these findings to survey reports of online and social media news consumption, with respect to both levels and trends. Third, we demonstrate the usefulness of passively collected data for measuring a quantity such as “consuming news” for which different researchers might reasonably choose different definitions. Finally, recognizing that passively collected data suffers from its own limitations, we outline a framework for using a mix of passively collected behavioral and survey-generated attitudinal data to accurately estimate consumption of news and related effects on public opinion and knowledge, conditional on media consumption.

The states that not only are pairs of individuals in a large social network connected by short paths, but that ordinary individuals can find these paths. Although theoretically plausible, empirical evidence for the hypothesis is limited, as most chains in small-world experiments fail to complete, thereby biasing estimates of chain lengths. Using data from two recent experiments, comprising a total of 162,328 message chains, and directed at one of 30 targets spread across 19 countries, we model heterogeneity in chain attrition rates as a function of individual attributes. We then introduce a rigorous way of estimating true chain lengths that is provably unbiased, and can account for empirically-observed variation in attrition rates. Our findings provide mixed support for the algorithmic hypothesis. On the one hand, it appears that roughly half of all chains can be completed in 6-7 steps--thus supporting the six degrees of separation assertion--but on the other hand, estimates of the mean are much longer, suggesting that for at least some of the population, the world is not small in the algorithmic sense. We conclude that search distances in social networks are fundamentally different from topological distances, for which the mean and median of the shortest path lengths between nodes tend to be similar.

"We propose to change the default P-value threshold forstatistical significance for claims of new discoveries from 0.05 to 0.005."

Researchers and public health officials have attributed low uptake of the COVID-19 vaccine in the US to social media misinformation. To evaluate this claim, we introduce a framework combining lab experiments, crowdsourcing, and machine learning to estimate the effect of 13,206 vaccine-related URLs shared on Facebook on US vaccination intentions. Strikingly, we estimate the impact of misinformation was 50X less than that of content not flagged by fact-checkers that nonetheless expressed vaccine skepticism. Although misinformation was significantly more harmful when viewed, its exposure on Facebook was limited. In contrast, mainstream stories highlighting rare vaccine deaths both increased vaccine hesitancy and were among Facebook’s most-viewed stories. Our work suggests that curbing misinformation benefits public health, but highlights the need to scrutinize factually correct but potentially misleading content.

In recent years, critics of online platforms have raised concerns about the ability of recommendation algorithms to amplify problematic content, with potentially radicalizing consequences. However, attempts to evaluate the effect of recommenders have suffered from a lack of appropriate counterfactuals-what a user would have viewed in the absence of algorithmic recommendations-and hence cannot disentangle the effects of the algorithm from a user's intentions. Here we propose a method that we call "counterfactual bots" to causally estimate the role of algorithmic recommendations on the consumption of highly partisan content on YouTube. By comparing bots that replicate real users' consumption patterns with "counterfactual" bots that follow rule-based trajectories, we show that, on average, relying exclusively on the YouTube recommender results in less partisan consumption, where the effect is most pronounced for heavy partisan consumers. Following a similar method, we also show that if partisan consumers switch to moderate content, YouTube's sidebar recommender "forgets" their partisan preference within roughly 30 videos regardless of their prior history, while homepage recommendations shift more gradually toward moderate content. Overall, our findings indicate that, at least since the algorithm changes that YouTube implemented in 2019, individual consumption patterns mostly reflect individual preferences, where algorithmic recommendations play, if anything, a moderating role.

Commentaries on the target article offer diverse perspectives on integrative experiment design. Our responses engage three themes: (1) Disputes of our characterization of the problem, (2) skepticism toward our proposed solution, and (3) endorsement of the solution, with accompanying discussions of its implementation in existing work and its potential for other domains. Collectively, the commentaries enhance our confidence in the promise and viability of integrative experiment design, while highlighting important considerations about how it is used.

Theories of organizations are sympathetic to long-standing ideas from network science that organizational networks should be regarded as multiscale and capable of displaying emergent properties. However, the historical difficulty of collecting individual-level network data for many (N ≫ 1) organizations, each of which comprises many (n ≫ 1) individuals, has hobbled efforts to develop specific, theoretically motivated hypotheses connecting micro- (i.e., individual-level) network structure with macro-organizational properties. In this paper we seek to stimulate such efforts with an exploratory analysis of a unique data set of aggregated, anonymized email data from an enterprise email system that includes 1.8 billion messages sent by 1.4 million users from 65 publicly traded U.S. firms spanning a wide range of sizes and 7 industrial sectors. We uncover wide heterogeneity among firms with respect to all measured network characteristics, and we find robust network and organizational variation as a result of size. Interestingly, we find no clear associations between organizational network structure and firm age, industry, or performance; however, we do find that centralization increases with geographical dispersion—a result that is not explained by network size. Although preliminary, these results raise new questions for organizational theory as well as new issues for collecting, processing, and interpreting digital network data. This paper was accepted by David Simchi-Levi, Special Section of Management Science: 65th Anniversary.

Agent-based models are a popular way to explore the dynamics of human interactions, but rarely are these models based on empirical observations of actual human behavior. Here we exploit data collected in an experimental setting where over 150 human players played in a series of almost a hundred public goods games. First, we fit a series of deterministic models to the data, finding that a reasonably parsimonious model with just three parameters performs extremely well on the standard test of predicting average contributions. This same model, however, performs extremely poorly when predicting the full distribution of contributions, which is strongly bimodal. In response, we introduce and test a corresponding series of stochastic models, thus identifying a model that both predicts average contribution and also the full distribution. Finally, we deploy this model to explore hypotheses about regions of the parameter space outside of what was experimentally accessible. In particular, we investigate (a) whether a previous conclusion that network topology does not impact contribution levels holds for much larger networks than could be studied in a lab; (b) to what extent observed contributions depend on average network degree and variance in the degree distribution, and (c) the dependency of contributions on degree assortativity as well as the correlation between the generosity of players and the degree of the nodes to which they are assigned.

Since the 2016 presidential election, an increasingly familiar narrative has emerged concerning the unexpected victory of Donald Trump. Fake news, much of it produced by Russian sources, was amplified on social networks such as Facebook and Twitter, generating millions of views among a segment of the electorate eager to hear stories about Hillary Clinton’s untrustworthiness, unlikeability, and possibly even criminality. “Alt-right” news sites like Breitbart and The Daily Caller supplemented the outright manufactured information with highly slanted and misleading coverage of their own. The continuing fragmentation of the media and the increasing ability of Americans to self-select into like-minded “filter bubbles”…

The past 15 years have witnessed a remarkable increase in both the scale and scope of social and behavioral data available to researchers, leading some to herald the emergence of a new field: "computational social science." Against these exciting developments stands a stubborn fact: that in spite of many thousands of published papers, there has been surprisingly little progress on the "big" questions that motivated the field in the first place?questions concerning systemic risk in financial systems, problem solving in complex organizations, and the dynamics of epidemics or social movements, among others. In this talk I highlight some examples of research that would not have been possible just a handful of years ago and that illustrate the promise of CSS. At the same time, they illustrate its limitations. I then conclude with some thoughts on how CSS can bridge the gap between its current state and its potential.

Why is the Mona Lisa the most famous painting in the world? Why did Facebook succeed when other social networking sites failed? Did the surge in Iraq really lead to less violence? How much can CEO’s impact the performance of their companies? And does higher pay incentivize people to work harder? If you think the answers to these questions are a matter of common sense, think again. As sociologist and network science pioneer Duncan Watts explains in this provocative book, the explanations that we give for the outcomes that we observe in life-explanations that seem obvious once we know the answer-are less useful than they seem. Drawing on the latest scientific research, along with a wealth of historical and contemporary examples, Watts shows how commonsense reasoning and history conspire to mislead us into thinking that we understand more about the world of human behavior than we do; and in turn, why attempts to predict, manage, or manipulate social and economic systems so often go awry. It seems obvious, for example, that people respond to incentives; yet policy makers and managers alike frequently fail to anticipate how people will respond to the incentives they create. Social trends often seem to be driven by certain influential people; yet marketers have been unable to identify these “influencers” in advance. And although successful products or companies always seem in retrospect to have succeeded because of their unique qualities, predicting the qualities of the next hit product or hot company is notoriously difficult even for experienced professionals. Only by understanding how and when common sense fails, Watts argues, can we improve how we plan for the future, as well as understand the present -an argument that has important implications in politics, business, and marketing, as well as in science and everyday life.

In this essay we make four interrelated points. First, we reiterate previous arguments (Kleinberg et al 2015) that forecasting problems are more common in social science than is often appreciated. From this observation it follows that social scientists should care about predictive accuracy in addition to unbiased or consistent estimation of causal relationships. Second, we argue that social scientists should be interested in prediction even if they have no interest in forecasting per se. Whether they do so explicitly or not, that is, causal claims necessarily make predictions; thus it is both fair and arguably useful to hold them accountable for the accuracy of the predictions they make. Third, we argue that prediction, used in either of the above two senses, is a useful metric for quantifying progress. Important differences between social science explanations and machine learning algorithms notwithstanding, social scientists can still learn from approaches like the Common Task Framework (CTF) which have successfully driven progress in certain fields of AI over the past 30 years (Donoho, 2015). Finally, we anticipate that as the predictive performance of forecasting models and explanations alike receives more attention, it will become clear that it is subject to some upper limit which lies well below deterministic accuracy for many applications of interest (Martin et al 2016). Characterizing the properties of complex social systems that lead to higher or lower predictive limits therefore poses an interesting challenge for computational social science.

We examine potential bias in Facebook’s 10-trillion cell URLs dataset, consisting of URLs shared on its platform and their engagement metrics. Despite the unprecedented size of the dataset, it was altered to protect user privacy in two ways: 1) by adding differentially private noise to engagement counts, and 2) by censoring the data with a 100-public-share threshold for a URL’s inclusion. To understand how these alterations affect conclusions drawn from the data, we estimate the preva-lence of fake news in the massive, censored URLs dataset and compare it to an estimate from a smaller, representative dataset. We show that censoring can substantially alter conclusions that are drawn from the Facebook dataset. Because of this 100-public-share threshold, descriptive statis-tics from the Facebook URLs dataset overestimate the share of fake news and news overall by as much as 4X. We conclude with more general implications for censoring data.

We explore a model of the interaction between banks and outside investors in which the ability of banks to issue inside money (short-term liabilities believed to be convertible into currency at par) can generate a collapse in asset prices and widespread bank insolvency. The banks and investors share a common belief about the future value of certain long-term assets, but they have different objective functions; changes to this common belief result in portfolio adjustments and trade. Positive belief shocks induce banks to buy risky assets from investors, and the banks finance those purchases by issuing new short-term liabilities. Negative belief shocks induce banks to sell assets in order to reduce their chance of insolvency to a tolerably low level, and they supply more assets at lower prices, which can result in multiple market-clearing prices. A sufficiently severe negative shock causes the set of equilibrium prices to contract (in a manner given by a cusp catastrophe), causing prices to plummet discontinuously and banks to become insolvent. Successive positive and negative shocks of equal magnitude do not cancel; rather, a banking catastrophe can occur even if beliefs simply return to their initial state. Capital requirements can prevent crises by curtailing the expansion of balance sheets when beliefs become more optimistic, but they can also force larger price declines. Emergency asset price supports can be understood as attempts by a central bank to coordinate expectations on an equilibrium with solvency.

Cooperation in repeated games has been widely studied in experimental settings; however, the duration over which players participate in such experiments is typically confined to at most hours, and often to a single game. Given that in real world settings people may have years of experience, it is natural to ask how behavior in cooperative games evolves over the long run. Here we analyze behavioral data from three distinct games involving 571 individual experiments conducted over a two-year interval. First, in the case of a standard linear public goods game we show that as players gain experience, they become less generous both on average and in particular towards the end of each game. Second, we analyze a multiplayer prisoner's dilemma where players are also allowed to make and break ties with their neighbors, finding that experienced players show an increase in cooperativeness early on in the game, but exhibit sharper "endgame" effects. Third, and finally, we analyze a collaborative search game in which players can choose to act selfishly or cooperatively, finding again that experienced players exhibit more cooperative behavior as well as sharper endgame effects. Together these results show consistent evidence of long-run learning, but also highlight directions for future theoretical work that may account for the observed direction and magnitude of the effects.

We resolve a controversy over two competing hypotheses about why people object to randomized experiments: 1) People unsurprisingly object to experiments only when they object to a policy or treatment the experiment contains, or 2) people can paradoxically object to experiments even when they approve of implementing either condition for everyone. Using multiple measures of preference and test criteria in five preregistered within-subjects studies with 1,955 participants, we find that people often disapprove of experiments involving randomization despite approving of the policies or treatments to be tested.

Abstract As organizations gravitate to group‐based structures, the problem of improving performance through judicious selection of group members has preoccupied scientists and managers alike. However, which individual attributes best predict group performance remains poorly understood. Here, we describe a preregistered experiment in which we simultaneously manipulated four widely studied attributes of group compositions: skill level, skill diversity, social perceptiveness, and cognitive style diversity. We find that while the average skill level of group members, skill diversity, and social perceptiveness are significant predictors of group performance, skill level dominates all other factors combined. Additionally, we explore the relationship between patterns of collaborative behavior and performance outcomes and find that any potential gains in solution quality from additional communication between the group members are outweighed by the overhead time cost, leading to lower overall efficiency. However, groups exhibiting more “turn‐taking” behavior are considerably faster and thus more efficient. Finally, contrary to our expectation, we find that group compositional factors (i.e., skill level and social perceptiveness) are not associated with the amount of communication between group members nor turn‐taking dynamics.

No abstract available.

In spite of the recent popularity of viral marketing, we argue that most big companies should not rely on it to spread the word about their products and brands. Instead, we propose a new model called “Big Seed Marketing” that combines the power of traditional advertising with the extra punch provided by viral propagation. Between traditional advertising and viral marketing is an important gap that can be filled by big companies looking for an advantage in the market place and a better return on their advertising and marketing dollar.

Why is the Mona Lisa the most famous painting in the world? Why did Facebook succeed when other social networking sites failed? Did the surge in Iraq really lead to less violence? And does higher pay incentivize people to work harder? If you think the answers to these questions are a matter of common sense, think again. As sociologist and network science pioneer Duncan Watts explains in this provocative book, the explanations that we give for the outcomes that we observe in life-explanations that seem obvious once we know the answer-are less useful than they seem. Watts shows how commonsense reasoning and history conspire to mislead us into thinking that we understand more about the world of human behavior than we do; and in turn, why attempts to predict, manage, or manipulate social and economic systems so often go awry. Only by understanding how and when common sense fails can we improve how we plan for the future, as well as understand the present-an argument that has important implications in politics, business, marketing, and even everyday life.

Can events be accurately described as historic at the time they are happening? Claims of this sort are in effect predictions about the evaluations of future historians; that is, that they will regard the events in question as significant. Here we provide empirical evidence in support of earlier philosophical arguments1 that such claims are likely to be spurious and that, conversely, many events that will one day be viewed as historic attract little attention at the time. We introduce a conceptual and methodological framework for applying machine learning prediction models to large corpora of digitized historical archives. We find that although such models can correctly identify some historically important documents, they tend to overpredict historical significance while also failing to identify many documents that will later be deemed important, where both types of error increase monotonically with the number of documents under consideration. On balance, we conclude that historical significance is extremely difficult to predict, consistent with other recent work on intrinsic limits to predictability in complex social systems2,3. However, the results also indicate the feasibility of developing 'artificial archivists' to identify potentially historic documents in very large digital corpora.

Many complex problems in science, business, and engineering require a trade-off between exploitation of known solutions and exploration of new possibilities. When complex problems are solved by collectives rather than individuals, this explore-exploit trade-off is complicated by the presence of communication networks, which can accelerate collective learning, but can also lead to convergence on suboptimal solutions. In this paper, we report on a series of 195 web-based experiments in which groups of 16 individuals collectively solved a complex problem and shared information through different communication networks. We found that network structure affected collective performance indirectly, via its impact on individual search strategies, as well as directly, by impacting the speed of information diffusion. We also found that networks in general suppress individual exploration, but greatly amplify the benefits of the exploration that takes place. Finally, we identified two ways in which individual and collective performance were in tension, consistent with longstanding theoretical claims.

Agent-based models are a popular way to explore the dynamics of human interactions, but rarely are these models based on empirical observations of actual human behavior. Here we exploit data collected in an experimental setting where over 150 human players played in a series of almost a hundred public goods games. First, we fit a series of deterministic models to the data, finding that a reasonably parsimonious model with just three parameters performs extremely well on the standard test of predicting average contributions. This same model, however, performs extremely poorly when predicting the full distribution of contributions, which is strongly bimodal. In response, we introduce and test a corresponding series of stochastic models, thus identifying a model that both predicts average contribution and also the full distribution. Finally, we deploy this model to explore hypotheses about regions of the parameter space outside of what was experimentally accessible. In particular, we investigate (a) whether a previous conclusion that network topology does not impact contribution levels holds for much larger networks than could be studied in a lab; (b) to what extent observed contributions depend on average network degree and variance in the degree distribution, and (c) the dependency of contributions on degree assortativity as well as the correlation between the generosity of players and the degree of the nodes to which they are assigned.

This article explores threshold models of social influence, with a particular focus on the consequences of simple heuristics in the context of social influence on collective decision-making processes. It first provides an overview of social-influence and threshold models before discussing influence cascades on complete and random networks. It then considers cascades in networks that emphasize the importance of social groups in the formation of influence networks, namely random-group networks and generalized-affiliation networks. It also describes cascade-seeding strategies and demonstrates how the structure of influence networks shapes the roles of individual actors (prominent or not) in the generation of collective behavior, including wild cascades of influence.