This is a great set of examples of this problem.

Of course, the offered solution is "hire a bunch of people to check the facts", which seems to be underestimating the scale of this issue for humans, and perhaps overestimating the difficulty of classifying credibility.

Considering that Google's bread and butter is general site credibility, having some "truthiness vector" added into the mix doesn't seem impossible?

Someone might say "why does Google decide who is credible", but they already do this through the search results anyways. They just don't seem to be able to differentiate between something matching a search, and something being factually accurate.