The Next Facebook

A very old e-friend¹ of mine posted a link to Joseph Perla's blog where he was claiming that Facebook is a Ponzi Scheme. He may have a point, although articles I've read elsewhere suggest that selling ad space is only part of Facebook's income stream: they also make money by taking a cut from your spend the online games you can play on Facebook.

Be that as it may it has got me thinking again about the whole Facebook model, what I like about it, and what I don't and how we might be able to engineer something better.

It seems to me that the biggest weakness to Facebook is decent chat: the primitive nature of it, the choice of a web client or nothing, the ephemeral nature of it. Now old hands like me still like mailing lists and Usenet, and for all its weaknesses there are little corners of it that still work. For newer users web forums in the phpBB mould seem to be popular.

So could we combine all of these elements together? I'm thinking of:

a forum model similar to Usenet or phpBB but with content accessible, both read and write, simultaneously via email (so it looks like a mailing list), web (were it looks like a forum) and authenticated NNTP (where it looks like a news group). When read via email or NNTP attachments would appear as links back to the server where the article was originally posted.
a looser model for forums, so for a start each user would essentially have their own forum (the equivalent of their Facebook "wall") and they could be easily added.
a distributed model, so you could either run your own server, or share one with friends, or use one at work, or rely on your ISP providing one, or purchase access to a commercial one. Postings would flow between the servers in the similar way NNTP traffic does now.
the whole concept would be defined only in terms of the standard for that exchange of data: so developers could come up with novel ways of implementing the user interface and, conversely, could run a very restricted server only offering some services or user interfaces.

There's two, related, key elements to this if it's going to work and they're to do with spam. Firstly users have to be authenticated and identified uniquely. Secondly the protocol needs to have a robust way of coping with rogue servers which don't play by the rules.

I rather suspect someone is thinking along these lines already, in the very broadest terms at least. Certainly a non-centralised, and hence potentially ad free and open source model, seems like the way forward to me.

Or am I just wibbling?