Efeitos coletivos em eleições e jogos

Abstract

In this work we investigate some collective phenomena emerging from the interaction between individuals in a group or society. The approach is the same one used in statistical mechanics. Basically, we use the techniques and models that are well known in the subject of critical phenomena and statistical mechanics out of equilibrium. First, we studied the collective behavior of a population in an electoral process. In particular, we have analyzed the Brazilian elections from 1998 to 2006. In this study, we analyzed the performance of candidates in a proportional election. We have shown that the vote distributions of the candidates follows a power law for two orders of magnitude, and it is the same no matter the year of the election or different regions of the country. In order to verify the influence of the parties in the election we also analyzed the results using a electoral coefficient and compared our results to legislative elections in some European countries. We also studied majority elections, for that we have used a fragmentation model as an attempt to reproduce the result for mayor elections in the Brazilian cities. The model can gives a good indication about the strategic voting in these kind of election. This study of how people choose their candidates drive us to analyze the emergence of patterns in groups of people playing games with adaptable agents. Therefore, we investigate here the forecasting game, a game where people make prediction about some events. Furthermore, we applied the forecasting game to study ways of how to improve the search tools on the internet. In both cases, we observed a rich scenario where phase transitions happen depending on the fraction of agents using some strategy.