COREY REECE taltàl mahàitha Thoughts about the web, from the big picture to minutiae.

Problem: Current popular social networks are closed systems.

By closed I mean that the information that you add to the system (biological and relationships) is not usually exportable. Additionally, if a friend a has an account on another service that implements similar functionality, it is not currently possible to add that friend without having them join the network you are on. These sites do this to keep you being exposed to their ads. Not allowing you to add remote friends forces friends all to join the same network causing a snowball effect in a peer network– it is essential to the growth of these networks.

So a few weeks ago as I succumbed to joining Facebook, I grumbled about having rode a friend-join-wave to friendster and orkut previously and then never really using them again. I predicted that Facebook will eventually suffer the same fate as some new site du jour coming down the pipe takes its place.*

Everyone is now suffering signupitis. Some people have have attempted to solve this by having each new service you sign up with pull in your basic info in a microformat. This does not thrill me because these sites still benefit from my info and my relationships. You could possibly become paranoid when you think about how all this information about you could be abused by a government or major corporation.

I propose the following: An open source profile server implementing a protocol for friend notification and profile exchange.

Individuals need to be addressable (read: identifiable) so OpenID is a natural fit. There is even a way to pass attributes along with an OpenID authentication which will be formalized in the next spec release.

To befriend someone, all you need is their OpenID. Then you use the OpenID process to verify the friend passing along attributes that request friendship. Next time the friend logs in to the profile server that they have an account on, they can grant or deny the friendship. If they friend chooses grant then their server reciprocates the request to the friend that originated the confirmation.

Then profile data and relationship data could be transmitted (according to preference settings) to the friend in a microformat or in FOAF. I am leaning towards FOAF because its been around forever, it is a good first stab (not perfect, e.g. naming), and it is RDF so maybe the Semantic Web prophecy has some chance of becoming a reality, all hail the prophet.

Next steps after that would be making an open standard for allowing apps a la Facebook a.k.a “deep integration“. Maybe even starting with their API so that all the existing apps would not have to be rewritten much.

The stakes on this is huge. OpenID is about to explode on the internet and this is just an obvious development after that. This open model puts the power the relationship and profile data back into the users hands. Profile servers can evolve and compete on features, not network size. The entire Internet is one unified social network– just like we thought before these sites tried to convince us otherwise. There is no business model behind this yet that I can see. (If you figure one out, please email me.) I will have a prototype released this week in rails. Watch this space for updates.

Walled gardens no more. Oh yeah, and no more signups.

* Facebook had to do something like an Apps API that allows them to scale functionality past what they could have done themselves. It looks more like a act of desperation than innovation to me now. Every App now made adds value to Facebook and they get the revenue from it.