By Renee DiResta (Associate, O’Reilly AlphaTech Ventures)
There have been lots of interesting posts lately about how Highlight, Glancee and Sonar performed at SXSW. As a SXSW n00b, I installed all three.

Unfortunately, the battery drain made continued use impossible, but I did notice that there was almost no overlap in the connections they suggested. This could be because smarter people only installed one of the three, but the lack of consistency got me thinking that the way the app parses your social and interest graphs is critical – and that no one really owns the interest graph.

Highlight and Glancee base the connections they show you on overlapping Facebook friends and interests. Sonar uses a combination of Facebook, Twitter, LinkedIn and Foursquare.

Seeding based on Facebook mitigates the identity problem, which is important for an app that’s encouraging folks to connect with strangers. But I don’t think Facebook currently captures interests very well. An “interest” is fluid and broad, bigger than a Like. Back in the day, when people used to type interests into their profile, they were less monetizable but probably more accurate.

I think – I could be totally wrong, and would love to see some data – that most Likes are about brand engagement. When I log into Facebook, I frequently see that three or four friends have Liked Walmart. I’d bet that’s because Walmart had a “Like us and save!” promotion at some point.

A big reason for the introduction of Open Graph (and, more recently, Interest Lists) is that Facebook itself doesn’t think that Likes alone say a whole lot about a person. Gathering actual activity data gives it the potential to piece together the broader interest, and target more effectively.

For ambient social apps to be sticky, they have to get at what’s important. Sharing a random Like or two doesn’t mean much. Glancee pointed out a few people who had “San Francisco” in common… but it’s still doing that even now that I’m back in SF. Sonar, which incorporated Twitter, was the best at surfacing people I cared about*.

Twitter has far fewer users than Facebook. However, the fact that those users actively search for people to follow means that parsing a user’s “Following” list (and tweets) can give pretty good insight into their broader interests. It isn’t a standalone interest graph either, but for active users it’s frequently more accurate than Facebook.

There’s already been a lot of digital ink devoted to the difference between the social and the interest graph. In general, the interest graph is more important for surfacing the people you really want to meet. When a friend introduces you to someone in the real world, they’re almost always doing so because you two have a common interest.

For the moment, a useful ambient social app (besides Grindr) will have to aggregate data across multiple networks to surface good connections… until a winner emerges in the interest graph space.

*this may also be because it only surfaces people who are checked in nearby on Foursquare, and there’s probably a demographic overlap there. (Agora does Foursquare + Twitter as well, and without the battery drain!)

This post was originally posted at No Upside.

About the guest blogger: Renee DiResta is currently an Associate at O’Reilly AlphaTech Ventures, where she researches emerging technology trends and supports portfolio companies. Prior to OATV, she spent six and a half years as a trader at Jane Street Capital, a quantitative proprietary trading firm in New York. Renee holds a B.S. in Computer Science and Political Science from the State University of New York at Stony Brook. Follow her on Twitter at @noupside and her VC firm at @oatv.