In April I travelled to Nashville to attend a conference on networks in economics. Researchers from many fields presented and the quality was extremely high. I made some notes on a few presentations that caught my attention.
Yiqing Xing asked which networks are stable (in a definition that is particular to his paper) if individuals care about the relative rank versus their network neighbours. Although the policy implications are unclear, it was a fun mathematical exercise. The result on homophily was surprising. Keeping average degree constant, the network without homphily is less stable than with homophily. It seems the result is driven by keeping average degree constant—the individuals have a high probability of being in a large clique of $k$, where $k$ is the number of groups in the community.
Mira Frick shows that “assortativity neglect” can persist in all societies. We think our friends are more representative of society than they really are. I got lost in the maths but the result is super interesting.
Suraj Malladi pointed out that random seeding is not that much worse than optimal seeding when seeds are not too costly. After recent interest in targeting using centrality measures in development he takes a step back and questions the value of network targeting.
Francesca Parise introduced Graphon Games. She shows that for larger populations the graphon equilibrium is a good approximation of a whole class of network games. Graphons generalize Erdos Renyi and Stochastic Block Models.
Yan Leng from MIT Media Lab used mobile phone data to study information diffusion. She uses detailed location data to match individuals and randomize into treatment and control. The treatment is “early adoption”, attending an event or visiting a store in the first time period. Individuals of four to six degrees of freedom away are more likely to adopt than those the same distance away from control individuals.