I guess orkut gets a lot of attention in geek communities because of its affiliation with google. I imagine it is because google has a reputation for being smart and getting it that people are paying attention. I know that 100% of the google employees I have met are brillant.
Nosing around it right now (thanks Chris!) it is like a who's who of open source, blogging, and CS academia. Not sure how I got into that. It's interesting to trace connections, and I have already found a few bloggers that I have added subs to, mostly from the academia grouping) -- but that is entertaining for a morning, and then runs out. I can see where sub-communities may develop that are centered around orkut's groupware tools (messages, lists. etc) but I don't think that these will replace traditional mailing lists. To take a cue from danah (one of the academic oriented bloggers I have since subscribed to), if anyone is going to make somehtign interesting of YASNS google has a good shot of doing it.
What I think will make it work is if it really is available as a service. If I could hook into the authentication, friends, and friends-of-friends data for my gift list thing it would make it more convenient for the users. Don Park sees PKI type possibilities. The value of the relationship information grows vastly as it is able to be used by additional applications. A large, trusted, foaf and identity system is one of the potential killer apps of web services. Sun and MS both see this, but no one trusts them. Maybe google can make it work.
I think there is amazing potential for small (or even large, but service apps favor small, specialized, highly focused, and highly interoperable tools) social network services to benefit from an available social network repository. To just step through an example using the gift list thing. Right now it is somewhat inconvenient to use as you must establish your trust relationships with groups or individuals in it. Once you dothatm the app provides a useful service to you, but it only works if your friends and family also use it. If you could sign up, create a list, and then know that all you have to do is point your friends and family to it and they can get in (authenticating against orkut) and find each other (pulling relationship information from orkut) it works exponentially better.
The problem here is that if google (I think google is doing this, it is a bit unclear if it is just Orkut doing it with google's blessing, or google doing it with Orkut's blessing) provides these services there are costs to them. They should have an expectation of profit from their capital and effort -- and providing free web services doesn;t provide that. On the other hand, if these small, free, tools were required to pay they would simply not be able to use it (I like providing the gift registry for free to friends and family (and anyone else who has wandered in)). The question is then if a large number of satellite services dependent on orkut could add sufficient value to orkut itself that orkut/google/Orkut could reep sufficient reward to keep providing the services.
That finally comes to the question of how they continue to afford this. Selling data mining analysis is one option, but this smells of .com vapor-profit. Targeted advertising is a possibility, but then that is lost when services are provided as machine-client instead of person-client services (which is what would give it legs, in my opinion). What else?