Brion Vibber wrote in part:
So by analogy "Linus Torvalds style pragmatism" might provide for a third-party filter program that inserts non-FDL images into Wikipedia articles on a reader's computer as they're loaded. ;)
Like a web browser? That's a third-party program that inserts Wikipedia's images into Wikipedia's articles on a reader's computer as they're loaded. The HTML document (the article) that we serve them has no image; we also serve them the image, which they request in reaction to our article (much as a module may be requested in reaction to the kernel's actions), but that's under fair use, not GFDL.
The problem, it seems to me, isn't that we use the images, but that we pretend that we're using them under the GFDL. They're already separated out a bit (not in the download, after all), and we should separate them (or the non-free ones) out further, rather than claiming in some places that we're entirely free and GFDL, while claiming in other places that we use some images through fair use. (I'd even support placing fair use images in a separate namespace and database from the free ones, just to make things ultraclear.) Our fair use images should be treated as *auxiliary*.
(even though we say it's absolutely vital that we include these images to have a legitimate encyclopedia).
It should ''never'' be vital to an article that we include an image, if we can help it, not only for distribution but also for accessibility. We should make our articles as good as possible without the images, whether or not we then also decide to send an image along with it. This is true even for free images, because we may have blind readers.
Unless one of us is a lawyer familiar with the licenses and laws involved, I don't think this discussion is going to go anywhere useful at this point. :)
Of course, IANAL either.
-- Toby