[I've changed the subject line.]

2009/3/11 Lars Aronsson lars@aronsson.se:

If the content is free, people don't need to drink from our watertap. It's the water that's important, not the tap. We could have a minimal webserver to receive new edits. Serving replication feeds to a handful of media corporations (who might pay for it!) should be far cheaper than to receive all this web traffic.  Some universities might serve up ad-free mirrors. We could be the Associated Press instead of the New York Times, the producer instead of the retailer. Or is the fact that we spend so much to maintain the 7th most visited website an admission to the fact that the space between the copies actually has a great value to us? A value that will be strengthened by cementing its URL and/or the name Wikipedia (attributing the project) into the new license? I'm not against that. I will go with whatever. I'm very flexible and I still think this is a very fun technical experiment. But I think the change is worth some consideration.

This is somewhat true. MediaWiki still needs a bloody huge central database server (or three) and so it has them.

I suppose the place to ask your question is on wikitech-l.

Being able to duplicate the infrastructure is necessary for forking to be meaningful:

http://davidgerard.co.uk/notes/2007/04/10/disaster-recovery-planning/

I'm not sure anything listed there has meaningfully changed in the last two years.

- d.