Friday, February 26, 2010

When social search gets personal: ChatRoulette, Peerpong, Aardvark


A couple of items in the news this week got me thinking about the social search space. But not from the usual angle. We have all heard about ChatRoulette by now, and of the random acts of human exhibitionism that take place there. Well, apparently some of those random encounters were too good to let go of. And so some visitors have taken to the a new Missed Connections to find people they met on ChatRoulette.

Cue "I still haven't found what I'm looking for." Yeah, by U2. And maybe that should be "I still haven't found who I'm looking for."

This is a great example of unintended social outcomes, and how in openly-designed social systems, users will find ways of addressing what's not handled by the application. Since ChatRoulette is anonymous by design, we can already anticipate that one of its social facets will be identity. Anonymity and privacy get users in, but on some occasions they will want to find each other again. Anonymity is coupled to identity (who). Just as random is coupled to specific (what).

Missed connections may be where users have to go now to try to re-locate people they met on ChatRoulette. Or ChatRoulette could accommodate this need in the future. It would then in effect be providing more than just random encounters — and would be providing a kind of social search.

Another item in the news this week related to social search was PeerPong, which received funding. (Disclosure: I consulted to PeerPong early on.) Described now as Aardvark for twitter, Peerpong matches user questions to twitter users who may be able to answer them. As aardvark uses one's social network to distribute questions and solicit answers, Peerpong uses twitter. (As you probably know, aardvark was just acquired by Google.)

The social search issue here is obviously different from that happening around ChatRoulette's missed connections. But they have one thing in common worth mentioning. It is: what happens when social search gets personal?

Social search tends to suggest traditional search supplemented with search results qualifed by social relevance. Using, say, social algorithms and user input (ratings, votes, etc) to deliver complementary results. Social search as regular search plus long-tail social data mining.

But there's another kind of social search. This kind, of which aardvark, Peerpong, and missed connections are all examples, uses people to solve search problems. We usually call these question/answer services. And in this area, success can be more elusive. Where in algorithmic social search there is one user experience issue, in question/answer services there are two.

Both questioner and answerer must have a satisfactory experience for the service to work. In fact the service really hangs on the experience of the answerer. The questioner has an immediate and present need or interest — not so the answerer. His or her motives for participation have to be incentivized or contextualized by other means.

The possibility that social search gets personal can be a systemically reinforcing and, as a user experience, much more compelling (and human) means of solving "search" issues. (Question/Answer services are much more than "search".) But this potential for the social to get personal is also a barrier to use — put plainly, people can get freaked out.

ChatRoulette's social search problem will be reciprocity and mutuality — solved only if both parties agree to re-find each other. Presumably the experience these users had on webcam was enough to take care of trust issues (which is not to say it's free of risk). For aardvark and peerpong, the challenge is relational.

What commitments or obligations to ongoing social search will a user have to another user in the future? Users don't know each other, even if they may be connected through twitter, through shared topical interests, or by social/peer networks.

Context of use can address some of this. By contextualizing search experiences and answer contributions, services like these can reduce the freak factor, using social context then to de-personalize perceived obligations, expectations, and commitments. Context can help reduce user fears of expected future participation commitments. And context can be used to supply alternative incentives to use — game contexts, expertise ranking, and the like. In short, using social to absorb some of the personal.

One wouldn't have thought ChatRoulette would have anything to do with social search. But the random selection of users is guaranteed to produce its inverse as an effect and byproduct. When people connect, algorithms become unnecessary.

Cue U2.

PeerPong Raises $2.8M for an Aardvark for Twitter
Calling All Romantics: Chatroulette Now Has Its Own Missed Connections
ChatRoulette, hall of mirrors
ChatRoulette, I'm watching you (watching me)
Google's Aardvark acquisition: Questions for Buzz?

Labels: , ,

Comments: :