Hi Rob,
The SSO activity has been somewhat dormant for a couple months now, but is probably can be resurrected if someone (you?) volunteers to shepherd the effort.
As it turns out, Dan Libby of videntity.org just 20 minutes ago published a MediaWiki patch to allow for OpenID-based SSO.
http://wiki.www.videntity.org/wiki/MediaWiki_OpenID_Patch
If someone (you?) were to say, implement a LID server and client for MediaWiki, that would give it a big head start over other potential solutions.
You are probably right, but from my perspective, this isn't a matter of a "land grab" before "the other guy" gets too much market share with their protocol, whatever it may be ;-) [I might be exaggerating your point here ...]
In my mind, the question is "what can we do to reduce the number of usernames and passwords that users have to use, how can be increase user convenience, how can we reduce spam and other bad stuff, how can we build cool new social stuff" on top of what hopefully will be a globally interoperable, privacy-protected, user-controlled identity infrastructure that pretty much everybody can buy into. We're trying to proactively do our part here ...
It wouldn't be the simplest solution to intra-Wikimedia SSO, but it would work, assuming that the LID libraries are mature enough to deal with Wikimedia's demands. If such a solution were to get substantial testing outside of the Wikimedia realm of servers, that would be a big argument for the maturity of the solution.
You are making a good point, and this is one of the reasons I posted to the list here -- how would one get any of this deployed with acceptably low risk to the operation of wikipedia? And for my own better understanding: how has that been done in the past by this project?
Cheers,
Johannes Ernst