A metadata race?

Wednesday, February 04, 2009 by Dave Winer.

Steve Gillmor perceives a race between Twitter and FriendFeed, which I find interesting, even if I don't think it's as much a race as he does. At this point there is so much distance between the services they provide, it's hard to see them as competitors. 

However, there is one area where all the current providers of social networks are in competition, and I'm not sure they're all aware of it, and so far no one has entered the arena, but I get a sense that at least a couple are poised to -- Twitter and FriendFeed. 

There's an interesting story that came up that's very much in line with the Where's Your Data? and What Happened To NakedJen? threads here on scripting.com. Last week, a bookmarking service, ma.gnolia, lost all its data. All of it. It was bound to happen sooner or later. Everyone who operates a public service back-end knows how fragile these things are, I sure do. At UserLand we came close to melting down a few times, and every one of those times we urged our users, as best we could, to keep local copies of their data. 

Anyway, ma.gnolia crashes, and someone there realizes that FriendFeed has been archiving their data for users that established the connection between the two services. And FriendFeed has an API. So voila -- write a web app to pull the data out of FF and re-populate their database, at least for some of their users.  

This suggest a buddy system for web service providers. Not every service can rely on the same back-end for safety, what happens when that service goes down? And who knows all the dependencies we're creating -- what if FF is using Amazon's back-end and Amazon gets attacked by an asteroid or a crazy terrorist, or whatever.  

All of this leads me to a wish that one of these companies would allow, through their API, for us to store arbitrary data on a per-user basis. I'm working with a small group of users on a new build of the Instant Outliner, and am using FriendFeed's "room" structure to great advantage. It's almost at the point where I don't have to write a back-end at all, I could almost completely depend on theirs. If only. If only they allowed me to store a relatively small amount of XML-based data with each user. Less than a megabyte per user. Probably way less.  

Something like Amazon's SimpleDB only even simpler, would do the job. The equivalent of a Perl hash or a Python dictionary. The same data we pass around in XML-RPC would be very good.  

I know all this sounds super-technical, but the architects at Twitter and FriendFeed know exactly what I'm asking for, and it wouldn't surprise me at all if one or both had this facility almost ready to go.