Information Diffusion Between Two Networks

The speed of reaching an agreement between distinct groups is heavily affected by the way people are processing information. Many papers focus on the speed of converging to the consensus, and a basic idea is that segregation (such as ethnic segregation) slows the convergence, but very few explored how to accelerate it. Furthermore, existing network learning theories depend heavily upon principles outside of economics (like graph theory and combinatorial techniques) without explicitly modelling behaviours arising from incentives. My research proposal documents how I aim to fill this gap by providing a thorough understanding of how information processing mechanisms affect the speed of convergence and designing an efficient mechanism to deduce the effect of segregation and speed up consensus in which individuals'incentives and behaviours are taken into account. The speed to consensus is important for social welfare. Consider for example a Pew Research Center (2006) study that found only about 41% of people believe human activities are the main reason for global warming, whereas 21% believe natural patterns cause it. Such disagreement concerning factual questions stems from differences in information. Individuals in different groups have similarly biased incomplete information, and based on the information, they form divergent beliefs leading to different reactions. Solving these problems is in the interest of everyone butcan only be reached through convergence of information and beliefs (consensus). An important part of this research is also what role incentives play in the process of convergence. Many social learning theories are only based on mathematical tools in explaining the micro-level stochastic process of information diffusion. The omission of individual incentives in those theories precludes welfare and policy analysis. However, once incentives are taken into account, efficient information processing methods canincrease social welfare by avoiding unnecessary costs. I intend to study segregated networks and think about information processing that makes use of information effectively to restrict the convergence process to a reasonable time. Incorporating these features into a novel theoretical framework is the first part of this research. The basic idea is that knowing the true state of the world is socially optimal but people can only convey information through social ties. This framework helps me in investigating the time of reaching an agreement using a myopic best-responding strategy and a rationalfarsighted strategy. The second part of this research will explore an efficient information processing strategy that facilitates consensus by taking incentives into account in practical problems. The first step is to filter out valid information to avoidrepeated accounting when people are exposed to overlapping information. The next step is to define a relaxed consensus that people's beliefs can still have distance. While most theories arbitrarily set a mathematical parameter, I will use incentives to set up the distance.