I'll save a bit of time and repost what I said in the Newgrounds profiles thread just a moment ago, since it actually does appear to be something that can be applied to this topic in general:
"I think this should be treated as a case-by-case sort of thing.
Madness Combat, for example, has a somewhat clearly-defined canon and story (albeit one that gets absolutely ridiculous at times and wasn't exactly present in the first game), and is mostly confined to one creator from what I can remember. It's not exactly written as well as some of the other verses we have on here, but that's no reason to dismiss it.
On the other hand, works like Pico's School and its many, many, many follow-ups have no properly linear story, thousands of wildly different incarnations across several random creators (especially so once "Pico Day" became a thing), and almost no canon to speak of outside of the original Pico's School flash game. Most of the characters die multiple times across the different works they appear in with no explanation for why they're alive in later games and animations, and scaling them would be an absolute nightmare. Those kinds of characters and works should be excluded entirely.
Crossovers like Newgrounds Rumble should also be disregarded."
If a work from YouTube, Newgrounds, et cetera has a coherent story/cast of characters and establishes its canon well enough, it should be fine for us to use and index. However, if a work is impossible to manage due to linear story being relatively nonexistent, multiple conflicting incarnations of the characters existing with no indication of which ones are more canon than the next (See: Everyone from Pico's School), and/or having a general lack of regard for maintaining consistency, then we should ignore it entirely. Characters from advertisements almost always fall into the latter category, so take that however you want.
Things like high sexual content and works having their own version of Jesus are another matter entirely, though...