What Makes Social Search Efficient.

The idea of the small world first put forth by Milgram in the 1960u0027s shows empirically how people knowing reliably only connections to their direct contacts can leverage their knowledge to perform an efficient global search, referred to as social search, in surprisingly few steps. Later, it was established that social networks are often interconnected in such a way that knowledge of all the edges enables a search in even a smaller number of step; such networks are often called small-world networks. Yet, despite a diverse body of work on the social search and its efficiency, it has been unclear why nodes with limited knowledge of just direct links are able to route efficiently. To probe this question, here we use a real location-based social network, Gowalla, to emulate a synthetic social search task. The results demonstrate that the spatial distributions of friends, and friends of friends (FoF) as well as the types of information utilized for search play a key role in effective social search. We also establish that neither the ways nodes are embedded into space nor edges distributed among nodes are important for social search efficiency. Moreover, we show that even very limited knowledge of friends of friends significantly improves social search performance with gains growing most rapidly for small fractions of FoF knowledge.