The distribution idea is getting around

[jducoeur]

[Happy birthday to gyzki!]

Those who are involved in the effort to produce a distributed version of LJ *really* should take a dig into what's going on over on the Twitter side of the fence. Due to Twitter's fundamental instability, there's a movement afoot to create a fully-distributed alternative to it. Given how much momentum Twitter has, if this *does* happen, it's likely to happen big, and those of us trying to produce reasonably open conversational systems will probably need to accomodate it.

(metageek may be pleased to know that some people appear to be talking about just the sort of Jabber-based architecture he advocated in conversation last week...)

Comments

[elizabear.livejournal.com]

Do you read Unshelved? Check out his recent comments: http://syndicated.livejournal.com/unshelved_comic/383657.html

Interesting...

[metageek.livejournal.com]

Thanks for the heads-up. In fact, it seems this protocol is already defined in XEP-163, and implemented in ejabberd. Maybe I should set it up on thibault.org...

Re: Interesting...

[be-well-lowell.livejournal.com]

Is there some kind of summary around? I haven't noticed any indications of the key distribution method, which is the key piece...

Re: Interesting...

[metageek.livejournal.com]

Here's a post about it. XEP-0189 defines a mechanism for key distribution. It sends XMPP keys over XMPP, which doesn't address the bootstrap problem; but I'm not aware of any key distribution system which does.

Re: Interesting...

[be-well-lowell.livejournal.com]

Connecting the key request to an original trigger is still a little unclear to me, but I suppose there are clearly multiple approaches that should work with a bit of effort.

The usual chicken-and-egg problem for trusting keys doesn't bother me particularly, because I think the requirements can be relaxed considerably from the general case.