Reasoning in Bayesian Opinion Exchange Networks Is PSPACE-Hard

We study the Bayesian model of opinion exchange of fully rational agents arranged on a network. In this model, the agents receive private signals that are indicative of an unkown state of the world. Then, they repeatedly announce the state of the world they consider most likely to their neighbors, at the same time updating their beliefs based on their neighbors' announcements. This model is extensively studied in economics since the work of Aumann (1976) and Geanakoplos and Polemarchakis (1982). It is known that the agents eventually agree with high probability on any network. It is often argued that the computations needed by agents in this model are difficult, but prior to our results there was no rigorous work showing this hardness. We show that it is PSPACE-hard for the agents to compute their actions in this model. Furthermore, we show that it is equally difficult even to approximate an agent's posterior: It is PSPACE-hard to distinguish between the posterior being almost entirely concentrated on one state of the world or another.

[1]  J. Geanakoplos,et al.  We Can't Disagree Forever , 1982 .

[2]  P. Clifford,et al.  A model for spatial conflict , 1973 .

[3]  N. Pimlott.  Learning from the neighbours. , 2012, Canadian family physician Medecin de famille canadien.

[4]  Michael Athans,et al.  Convergence and asymptotic agreement in distributed decision problems , 1982, 1982 21st IEEE Conference on Decision and Control.

[5]  Joan Feigenbaum,et al.  Random debaters and the hardness of approximating stochastic functions , 1994, Proceedings of IEEE 9th Annual Conference on Structure in Complexity Theory.

[6]  Matthew O. Jackson,et al.  Naïve Learning in Social Networks and the Wisdom of Crowds , 2010 .

[7]  Peter Jonsson.  A Nonapproximability Result for Finite Function Generation , 1997, Inf. Process. Lett..

[8]  Peter Jonsson.  Strong bounds on the approximability of two Pspace-hard problems in propositional planning , 2004, Annals of Mathematics and Artificial Intelligence.

[9]  M. Amin Rahimian,et al.  Bayesian Decision Making in Groups is Hard , 2017, Oper. Res..

[10]  Leslie G. Valiant,et al.  NP is as easy as detecting unique solutions , 1985, STOC '85.

[11]  Elchanan Mossel,et al.  Opinion Exchange Dynamics , 2014, Probability Surveys.

[12]  J. van Leeuwen,et al.  Theoretical Computer Science , 2003, Lecture Notes in Computer Science.

[13]  Chris Breward,et al.  Fashion , 2019, Textile History.

[14]  K. Roeder,et al.  Journal of the American Statistical Association: Comment , 2006 .

[15]  John Geanakoplos,et al.  Common Knowledge , 1992, TARK.

[16]  Oriol Carbonell-Nicolau.  Games and Economic Behavior , 2011 .

[17]  Dock Bumpers,et al.  Volume 2 , 2005, Proceedings of the Ninth International Conference on Computer Supported Cooperative Work in Design, 2005..

[18]  Asuman E. Ozdaglar,et al.  Opinion Dynamics and Learning in Social Networks , 2010, Dyn. Games Appl..

[19]  Redaktionen.  THE REVIEW OF ECONOMIC STUDIES , 1960 .

[20]  Asuman E. Ozdaglar,et al.  Spread of (Mis)Information in Social Networks , 2009, Games Econ. Behav..

[21]  John N. Tsitsiklis,et al.  The Complexity of Markov Decision Processes , 1987, Math. Oper. Res..

[22]  S. Bikhchandani,et al.  You have printed the following article : A Theory of Fads , Fashion , Custom , and Cultural Change as Informational Cascades , 2007 .

[23]  Johan Kwisthout,et al.  Approximate inference in Bayesian networks: Parameterized complexity results , 2018, Int. J. Approx. Reason..

[24]  V. Borkar,et al.  Asymptotic agreement in distributed estimation , 1982 .

[25]  Joan Feigenbaum,et al.  Probabilistically Checkable Debate Systems and Nonapproximability of PSPACE-Hard Functions , 1995, Chic. J. Theor. Comput. Sci..

[26]  Elchanan Mossel,et al.  Asymptotic learning on Bayesian social networks , 2012, Probability Theory and Related Fields.

[27]  Robin Hanson,et al.  Are Disagreements Honest , 2004 .

[28]  Harry B. Hunt,et al.  The Complexity of Approximation PSPACE-Complete Problems for Hierarchical Specifications , 1993, Nord. J. Comput..

[29]  Elchanan Mossel,et al.  Efficient Bayesian Learning in Social Networks with Gaussian Estimators , 2010, 2016 54th Annual Allerton Conference on Communication, Control, and Computing (Allerton).

[30]  M. Degroot.  Reaching a Consensus , 1974 .

[31]  R. Holley,et al.  Ergodic Theorems for Weakly Interacting Infinite Systems and the Voter Model , 1975 .

[32]  Douglas Gale,et al.  Bayesian learning in social networks , 2003, Games Econ. Behav..

[33]  Lones Smith,et al.  Pathological Outcomes of Observational Learning , 2000 .

[34]  Robin Hanson,et al.  For Bayesian Wannabes, Are Disagreements Not About Information? , 2003 .

[35]  Polly S Nichols,et al.  Agreeing to disagree. , 2005, General dentistry.

[36]  Harry B. Hunt,et al.  Generalized CNF satisfiability problems and non-efficient approximability , 1994, Proceedings of IEEE 9th Annual Conference on Structure in Complexity Theory.

[37]  K. Pearson,et al.  Biometrika , 1902, The American Naturalist.

[38]  Nicolas Vieille,et al.  Informational externalities and emergence of consensus , 2009, Games Econ. Behav..

[39]  Fernando Paganini,et al.  IEEE Transactions on Automatic Control , 2006 .

[40]  A. Banerjee,et al.  A Simple Model of Herd Behavior , 1992 .

[41]  Scott Aaronson,et al.  The complexity of agreement , 2004, STOC '05.