This document presents a method for improving social search by using an influence model to identify stronger paths between nodes in a social graph, rather than solely relying on shortest paths. The authors define influence as being proportional to how much a node invests in another. They show that in Twitter and DBLP graphs, the strongest path between random node pairs is often longer but better than the shortest path. Their model provides a simple way to leverage implicit tie strengths for applications like social search.