I really enjoyed Brad Fitzpatrick’s talk about the open social graph today. Brad’s talk was essentially a distilled version of this lengthy piece he wrote two months ago. Rather than retread his entire argument, I want to highlight the juicy bit that really popped for me. If you’re interested in more then you should really read Brad’s original post.

Brad said that decentralization of the social graph is inevitable. Part of his argument was that throughout the history of internet technology, a initial popular idea is developed quickly in a centralized fashion, and then later, a decentralized version is created that is more difficult to build, but often worth the effort.

For example, it used to be that HOSTS files resided on all users computers that mapped names to IP addresses for all the available sites on a network. The file frequently became out of date and had to be updated from a centralized source. This solution wasn’t scalable, and so eventually a much more complicated, decentralized system was built to replace the function of HOSTS files: the DNS system. The DNS system would have been a mess to build from the start in a decentralized fashion, but as networks grew and became interconnected, a decentralized system was essential to support continued growth.

Another example is email networks. In the beginning of email, there were a bunch of small island networks that could only communicate internally. This was due to a centralized message delivery system in each network, which had no awareness or connections to the other networks around it. Eventually, as inter-network email communication became a desired feature, email routing in a decentralized fashion emerged. So, no email server knows the location and destination of every email box in the world, yet users on any email server can send message to users on any other email server. Technology, yet again, trended toward decentralization.

The inevitable march away from centralized network technology towards decentralized solutions is a theme we talk about around the office a lot, and I was glad to see this opinion represented (and well-articulated) at Web 2.0 Summit.