Stowe Boyd covered an interesting paper on social networks and concluded "the apparent, superficial social network based on following and followers conceals a deeper, sparser social network". Every current service has an incredibly primitive representation of your relationships. You're either friends with somebody, or you're not. Here's what''s missing:
– Strength. There's no way to specify how close you are to somebody else.
– Time. Is the friendship long-lasting? Have you talked recently?
– Context. What other friends is this friend close to? Which circles do they move in?
It's well known that you can use communication data to answer these questions. This implicit approach is better than trying to get people to enter this information manually because:
– Convenience. Nobody wants to spend time doing data entry and house-keeping on their network. Doing it automatically solves that problem.
– Reliability. You can objectively measure how many emails somebody has sent you, and how many you've returned to them. This removes the subjective element that creeps in if you're asked to rate the strength of a relationship on an arbitrary scale. It also removes the temptation to exaggerate your closeness to someone influential.
So why hasn't anyone done this? There's massive technical barriers to overcome before you can access large stores of email, and big privacy issues. I'm convinced they can be overcome, and that's what I'm doing with Mailana. If you want to see the sort of detailed social graph I'm talking about, Boulder Twits is using the same backend as my email analysis system.
Pete Warden's blog 