Free ESG report : Seamless data management with Avere FXT

Comment Benjamin Birt, a CompSci student at the University of St Andrews, has announced a new type of social network, which he calls PeerBook. We note that there are some folks in Hamburg who might want to chat to him about the name and an international academic working group who might like a quiet word about this being a whole new concept, too.

Birt's Peerbook already has a homepage up, but it's not yet available to the general public, although he claims to have completed extensive testing, especially of the system's security. It does reveal that he's already replicated some of the basic functionality of Facebook – but also that it's written in Java, which will rule out access from several types of modern Internet client. One of the big enabling factors of the existing social-network sites is that they're just websites and no client software is required.

The idea isn't new, it's just that nobody has quite made it work yet. Peer-to-peer networks are everywhere these days, from file-sharing setups such as good ol' Napster to Bittorrent today, but it's also pervasive in other forms – Skype and Spotify both use people's PCs for distributing their data and some collaborative office systems such as Microsoft's Sharepoint can interoperate in P2P fashion. There's even a new distributed-portal system that appeared this year called Osiris, which is a step in the direction of a P2P web of user-generated content.

And yet, all the main social networks are still classical server-based systems: Facebook, MySpace, Orkut, Bebo, Friendster - even outliers such as Livejournal and the pure-play blogging sites. Each has its own interface, its own community and few links from one to another. The closest to any kind of federated system is arguably Livejournal, by dint of the fact that its software is open source and has been turned into a plethora of related sites, including Dreamwidth, Deadjournal and the defunct Greatestjournal, most of which allowed some form of linking between sites. LJ's former owner, Six Apart, creator of the Movable Type blogging platforms and TypePad site, also started another blog-oriented social network, Vox, which tried to integrate more Web 2.0 facilities, such as fine-grained access control on posts and links to Flickr, YouTube and so on.

One of the first steps towards a P2P social network is distributed authentication, and one of the biggest steps here was taken by Livejournal and Deadjournal when they implemented LJ creator Brad Fitzpatrick's OpenID to allow cross-site linking. OpenID is now widely supported by all sorts of companies, from AOL and the BBC to Yahoo. OpenID's functionality is intentionally limited, though, allowing you to sign in to sites you've never previously visited using an ID from a different site – for instance, signing comments on Blogger.com using a Livejournal ID.

RSS allows stories to be syndicated from websites into readers, which can of course mean other websites – Google Reader being a notable example of this. It doesn't allow different sites to work together in collaborative discussions, though.

For peer-to-peer social networking to work, multiple different sites need to be able to exchange rich data – not only authentication and XML-marked-up stories, but photos, comment threads, friend relationships and so on. This kind of rich communication between websites is becoming more common across protocols such as REST and JSON, but it's still mainly confined to in-house enterprise applications rather than the public Interweb.

Historically, these kind of open interoperability systems, from TCP/IP itself to HTML, HTTP, DNS, Kerberos, POP3, IMAP and much of the glue that holds the web together have come about through open collaborative processes such as Internet RFCs rather than the adoption of proprietary code. New sites are not likely to do it, as hundreds of millions of users are quite happy with what they've got on the current ones. What we need is lots of existing sites to start coming up with multiple different open mechanisms to build links between them, then the market and natural selection can take its course.

On the other hand, we could be completely wrong and Ben Birt could be poised to become the next Mark Zuckerberg. Which could be a mixed blessing for him.®

5 ways to reduce advertising network latency