THIS. So much this. Time was when you could actually discover the provenance of an image. Almost every time, when I’m doing a reverse image search, that is my intent. It used to work. It seldom does these days.

Try using tineye.com. It has noise too, but seems to be easier to find the original source than Google these days, at least for me anyway.

And the interesting thing about that is, you'd think it would be (relatively speaking) straightforward for Google to keep track of the first place a given image was indexed (or possibly the first few places, or everywhere it was seen over the first X period of time since you couldn't guarantee the very first would always be the original). Assuming that original was still online, it would seem to be the place to direct searchers to, regardless of pagerank or whatever.