I wonder if anyone might be excited to go through a few days of
access_logs and determine the top 1000 most popular images on
wikipedia? I think this would be useful in our discussions on
wikipedia-l as to "fair use" and so on.
My own position, which is not (yet) shared by many, is that we should
take a hardline stance against using anything that isn't pretty
clearly GNU-free. That means (a) licensed under some free license
such as FDL or one of the creative commons free licenses (b) public
domain and (c) possibly some fair use, under the most restrictive law
of the language in question.
It would be informative to know what images our users are actually
seeing, so that we can guess at how difficult it will be to eliminate
them or replace them with free equivalents.
--Jimbo