In his post What is the Social Web and Why Should I Care? Dan highlighted that today’s social networks are highly isolated introducing many restrictions for their users. The movie below illustrates this.

The various ‘open stack’ protocols that were developed to help open up the social web space, are slowly being embraced by several key players like MySpace, Facebook and Plaxo. The latest and most complete sponsor so far is Google with Buzz. It’s great that big companies like Google embrace the open social web, since it will establish awareness and make it far more likely that other players will follow suit.

However we identified several important elements that the open stack is not yet addressing:

• Federation: the open stack does not include a push based, asynchronous and realtime server-to-server API with a rich semantic. There is for example no way today to become friend and share with someone on another social network.
• Privacy: the current open stack is very much tailored to public information. But many messages sent on social networks, people’s personal profiles and relationships are considered private. At present you will not be able to expose or send private information to friends across networks. It is not straightforward to do this in a simple and secure way with the current approach.
• Synchronization: it is one thing to expose data via one party like Buzz or Facebook. This does not mean that they are fully synchronized. Changes or deletions made in one will not automatically reflect in the other. This will still mean manual labour or outdated items.

Last weekend we presented our onesocialweb project at the Fosdem 2010 in Brussels that addresses these exact issues (get the slides). In our view we need the following building blocks for a truly open and distributed social web:

• A distributed architecture so that both public and private messages can be send across different networks and domains. This will allow you to friend and share with anyone you want, as well as move to another provider without losing your data and friends.
• A means to establish and manage your complete social graph in a single place. This will solve having to re-friend everyone when you go to another social network. Furthermore you decide who in your social graph can see what. And you can expose (parts of) your social graph to different social applications.
• A central repository of all your (links to) social data so it is relatively easy to backup and take along to another provider. You can authorize social applications to add activities on your behalf e.g. Amazon that you bought a book or LinkedIn that you posted a vacancy.
• A means to collaborate with others across different networks, for example on a shared holiday photo book.

We believe that XMPP, the engine behind Instant Messaging, can serve as a solid foundation for such a new social web. (New to XMPP? Have a look at this introductory slidedeck by Remko Tronçon and Peter Saint Andre). There are several similarities that can be leveraged: a unique identifier (e.g. alice@wonderland.lit), a profile with basic information about you, as well as a list of people you like to talk to. This would need to be extended and complemented by activities, the core elements shared on social networks. In addition, we will add support for fine grained access control, real-time notification and collaboration.

Practice makes perfect. That’s why we strongly believe we have to build a reference implementation to understand how this will all work in practice. This means several distributed servers with different clients (i.e. web, desktop and mobile) to replicate experiences people are used to seeing on existing social network sites and clients. We intend to open source all components with an apache 2 license. A first release will be made available end of March 2010. In the meanwhile here is a sneak preview of the web application. For more information on the project please check http://onesocialweb.org and you’re welcome to join the discussion on our mailing list.