Let's not use authoritarian language when making requests of authors. And yes, as any experienced historian will confirm, there are often multiple possible attributions. This is famous in the case of art history, where disputes rage whether a painting was by X, or by "school of X" or perhaps by a modern forger. Art works are MUCH more carefully studied than old photographs (because genuine ones are worth millions). It often happens that numerous books reproduce the same historic photograph, for example, allowing for multiple attributions.
I fail to see what this has to do with the point. I think Stephen is suggesting that we follow the practice for all images in order to squelch any problems that already exist in systems that don't provide attribution. I for one think that should be de facto policy anyway.
