I could definitely see using this among my close friends if Posterous ever gets too cumbersome.
The real question, though, is how BuddyPress installations might eventually communicate with each other. A federated mesh of BuddyPress instances, with a good UI for managing relationships between them, is the kind of thing that would be needed to take on Facebook.
As a codebase WP tends to have quite a high rate of exploits (or it is a popular exploit target). Right now that's not a huge problem - but for a social network where we want to encourage non-techy types to be involved it could be explosively bad.
Exactly.
And the beauty is that because the central search API is documented and, hopefully, the drop in code app is FOSS companies could host their own networks which support the API for those without the skill or inclination to set it up themselves.
But because it is open there is competition and an incentive to stay on the "right side" etc.
a) people will laden this with too much junk making this a bloated piece of software that is too confusing and takes too long to get out.
b) people will forget the real benefits to people ala drop in easy.
You could have a central database with all the friend relationships, but you don't need that. You can search the way some P2P networks do it. Send the request to all your friends, who forward it to their friends, for a certain number of iterations. When the target is found it can reply to you directly.
Searches could end up taking a looong time - or just timing out.
There could be ways (especially if people optionally added more info to the search profile) to narrow it down. Direct email searches for example would be easiest.
The issue is that it would be a web app not a client - and you have a limit of how long you can run requests for (and, also, if your hosting the app it means pretty much we are limited to PHP - and no fancy stuff like asynchronous requests etc)
Example: if I have 150 friends and I send a "search request" to them, who then send a "search request" to their 150 friends - for, say, 3 levels of iteration that is a lot of work every page is doing - all within one basic request.
To the extent your friends are on the same server, you speed things up more. If you got online by accepting an invitation from a friend, that will tend to happen.
It'd still be better to have asynch requests. I'm not so sure PHP is necessary these days. A nicely packaged virtual machine can work pretty well for a small group. I don't think many people would be setting up web apps just for themselves.
Well what I have specced out is to compete with Facebook and offer a privacy concious service to the masses. So something like that is infeasible for a number of reasons.
In fact, looking again at your proposal, you mention pubsubhubbub. As long as we're using that, we could send search requests to friends, just like we send status updates. Each of them forwards to their own hub. The search data includes the URL for posting results, and a time-to-live that gets decremented each time the search is forwarded.
Now we can get back to the simple PHP site.
- A "drop it in and it works" wordpress like app that acts as someones homepage/social network interface so people can host their own pages
- - include oauth/openid in the app so the homepage can act as a central login too
- - drop pubsubhubub or something similar into the app so that people can share information with their friends and followers directly
- implement a central website to act as a "search" clearing house (search for names and email addresses which resolve to url's). All it will store is name and a hash of your email (with options to store more if the user allows).
- produce a "web bar" like the blogger one to go at the top of the web app - so people logged into their "network" get easy tools when on other peoples pages
All of this exists in parts already (oauth, pubsubhubub etc) so it's just a case of drawing it together.