Is the social graph a moving target?

Yes.

Yes it is.

He's right, un-friending is hard and socially awkward.

A dynamic social graph seems really complicated, though:

Facebook and other social graphs represent a rolling set of relationships that are out of date the instant they are created. I have a fight w/ a friend, I lose touch w/ an old college roommate, a work relationship grows stronger – existing graphs have very little visibility into this with a boolean friendship model.

Maybe the next great graph isn't based on people; maybe it's based on interests.

Also, Kevin writes:

Friendster to MySpace to Facebook to Instagram/Path to ?

Why do we move from one service to another every few years? Maybe it’s because these services don’t have the visibility into our real social lives. Maybe it’s the allure of a new product. I’m not sure.

First off, a lot of people I know never left Facebook.

Second, Instagram and Path are great but don't fulfill the parts of my life that are served by LinkedIn, Twitter, Tumblr, Pinterest, and Google+.

Perhaps social graphs aren't as useful as we originally thought?

I keep thinking of this in the context of Twitter not letting Instagram users and Tumblr users find their Twitter friends on those services automatically.

This is the opposite of Facebook, which wants you to find your Facebook friends on every platform and app.

Twitter doesn't want you to take its social graph anywhere; Facebook wants you to take its social graph everywhere.

Both these approaches can't be right, can they?

Commentary on Hacker News: http://news.ycombinator.com/item?id=4429388

Several people pointed out: Isn't this what Google+ Circles are supposed to do?

So the question is: Who do you know RIGHT NOW?

For what purpose: So I can find stuff, or so they can find stuff?

I guess it's a real-time relevancy graph. Or at least a more formal, software manipulable model of such.

Too theoretical.

What would actually be useful?