Rowan Collins wrote:
The ideal solution would be [...] to store licence information as language-independent metadata, descriptions in multiple languages, and use something like the "content negotiation" in HTTP to determine what to include or display... :)
Sounds good to me. ;-)
Seriously, image licences are meta-data, and should be treated as such (as should lots of other stuff - interwikis, categories, etc.), and we on the Wikimedia Research Team^WNetwork are looking into the plausibility of modifying MediaWiki to have meta-data accessed separately.
Quite apart from anything else, the use of proper meta-data for licensing would (will? ;-)) allow useful queries to be made, such as "grab articles in the category 'British monarchs' where all elements are either Public Domain worldwide, GFDL, or CC-BY", which would allow semi-automatic creation of non-infringing items for publication anywhere (though there still might be problems, such as local blasphemy laws, etc., which would be difficult to cover).
Note that this would need a quite advanced tagging system (c.f. the discussion on wikien-l over the fact that our 'PD' images of Lindisfarne scrolls are only PD in the United States, and are infringing in the UK), which would be, erm, 'fun'.
Yours,