ETech: FOAF

Related link: http://www.foaf-project.org

As it often happens, I have a basic understanding of a technology before I show up at ETech and then I attend a talk and my whole world changes. Once you see the enthusiasm and wild gesticulations of the speaker(s) you catch their enthusiasm and the deeper meaning and context of the technology becomes clear.

Listening to Dan Brickley and Edd Dumbill's talk about FOAF really drove home the key points and the greater context behind FOAF. FOAF stands for Friend Of A Friend and is a distributed approach to social software. FOAF uses XML/RDF to express information about a person:

I am Dan and I work at W3C and I know Libby who works at ILRT and her FOAF profile is [over here] and here are the titles and descriptions we wrote together.

FOAF is essentially a machine readable version of a person's home page. Interpreting and understanding an HTML based homepage for a person is still too complicated a task for computers. FOAF tackles this by expressing relationships and information about yourself in XML/RDF. Given this machine readable format, FOAF lets you describe anything about yourself or anyone else.

Unlike centralized social software sites like Friendster, Tribe.net and Orkut, FAOF takes a completely decentralized approach. FOAF profiles are normal XML text files that are stored on your website -- just as normal home pages are. This means that the data about yourself belongs to you and not to the social software site. Orkut's terms of service state that the data about you is not owned by you -- it is owned by Orkut (Google). This is a fundamentally bad idea -- my data should be owned by no one but myself.

FOAF's distributed nature gives you the power to put the data back in your control, and on a server/website that is controlled by you. This freedom comes at a cost that is typical when centralized services become distributed. Authentication is just one of the difficult problems to solve -- how do you know that a given FOAF profile really does belong to the person who claims it? Centralized services can easily solve this problem by authenticating users and restricting the user to edit only their own data.

The top-down approach used by central server based social software systems will ultimately limit the scope of what can be done with these systems. The decentralized bottom up approach of the web that enables anyone to do anything has proven to be a great approach. If the web was controlled by any one entitiy (benevolent or evil) it would not have grown at the amazing rate that we've witnessed.

I think FOAF is the right approach for the future, but many difficult problems remain -- how will FOAF handle lying, trust, reputation? I'm hoping that we can solve these difficult problems soon -- if nothing else, FOAF will expose these problems to the greater public which will get more brains thinking about the problem. And that's certainly a step in the right direction.

Do you have a FOAF profile? What are your experiences with it so far?