While I completely agree with Soljashy's above comment, I would like to point out that Glasnost formulated the clearest theoretical basis for action that has been brought up in this discussion so far:
glasnost said:
[T]he two factors that should go into determining whether a photoshop belongs on Danbooru are quality and triviality.
...This one being probably number two:
Bapabooiee said:
[A major concern:] the original artist's name is still used in the tags, causing content on the site to become "muddled-up" between original work, and third-party edits (...).
True. An (exceptionally good) edited work is actually a derivative work, similar to a heavily altered remix in the field of music. Credits go to both the original artist and the 'shopper, else is wrong.
EDIT:
The more I think about this, the more Jxh2154 makes sense -- and, on the other hand, the more complicated it gets. Sorry if I sound pseudo-academic...
Much of the problem is made up by a fundamental difference between western and Japanese author's rights cultures. The Japanese seem to especially respect the integrity of individual works of art. That has more to do with moral principles than with making money. In the west, over the decades, franchisable things, such as "logos, characters, names, and distinctive likenesses thereof", have become the crucial point of copyright.
The existence of Danbooru is largely based on the idea that the credits information of (certain types of) Japanese works wildly shared on the Internet should be easily available to everybody. This is a reaction against 4chan-type imageboard culture. What is that culture, then? It is a kind of postmodern folklore, enabled by the Internet and various data processing technologies -- such as "photoshopping".
Folklore was there millennia before copyright. In folklore, knowledge of the author is the exception rather than the rule. Instead, works appealing to different audiences cherish, spread, and keep being modified.
Basically, we are discussing what to collect, organize and share: works that can be unambiguously copyrighted to one or more authors, or high quality products of a new wave of folklore still poorly understood.
If we want exact author and source information, the latter task is vastly more complicated than the former.
Updated