Individual Altruism Cannot Overcome Congestion Effects in a Global Pandemic Game

A key challenge in responding to public health crises such as COVID-19 is the difficulty of predicting the results of feedback interconnections between the disease and society. As a step towards understanding these interconnections, we pose a simple game-theoretic model of a global pandemic in which individuals can choose where to live, and we investigate the global behavior that may emerge as a result of individuals reacting locally to the competing costs of isolation and infection. We study the game-theoretic equilibria that emerge from this setup when the population is composed of either selfish or altruistic individuals. First, we demonstrate that as is typical in these types of games, selfish equilibria are in general not optimal, but that all stable selfish equilibria are within a constant factor of optimal. Second, there exist infinitely-many stable altruistic equilibria;all but finitely-many of these are worse than the worst selfish equilibrium, and the social cost of altruistic equilibria is unbounded. Our work is in sharp contrast to recent work in network congestion games in which all altruistic equilibria are socially optimal. This suggests that a population without central coordination may react very poorly to a pandemic, and that individual altruism could even exacerbate the problem. © 2022 IEEE.

Robust Stochastic Stability in Dynamic and Reactive Environments

The theory of learning in games has extensively studied situations where agents respond dynamically to each other by optimizing a fixed utility function. However, in many settings of interest, agent utility functions themselves vary as a result of past agent choices. The ongoing COVID-19 pandemic provides an example: a highly prevalent virus may incentivize individuals to wear masks, but extensive adoption of mask-wearing reduces virus prevalence which in turn reduces individual incentives for mask-wearing. This paper develops a general framework using probabilistic coupling methods that can be used to derive the stochastically stable states of log-linear learning in certain games which feature such game-environment feedback. As a case study, we apply this framework to a simple dynamic game-theoretic model of social precautions in an epidemic and give conditions under which maximally-cautious social behavior in this model is stochastically stable.