(adapted from my answer on Quora)
With the mass adoption of social networks, I’ve been thinking about the use cases that benefit from analyses of multiple degrees of separation
For sake of clarification, I consider the “social graph” to be one where the nodes are people as opposed to documents (eg, pages). The connection between the nodes do not have to be friends or explicit. They can be links between people due to a common interest (eg, people who follow the same topic on Quora) or implicit connections like the ones Color Labs is trying to create.
The main business cases will be one where you can reasonably infer an answer from related data points.
I think the primary use of the social graph is recommendation. I can’t count how many start-ups promise to suggest to me what films, restaurants, hotels, books, clothes, music and other stuff my friends like.
I think these startups are currently, overly dependent on explicit connections (“friends”). An emerging business case–an important one in my view–is reputation. Klout and PeerIndex are the two players I know in this space.
Brainstorming, other business cases might include: management of contagious diseases; polling; political repression.