Earlier quoted context omitted.
I think it's a good thing in that it's led to the most commercially successful search and ads company in history (which - in large part - does deliver very relevant content, and relatively lower sadness than many competitors). But it's clear there is much room to improve. That point you're addressing was meant to illustrate one way in which it is difficult to find an ad match that doesn't induce sadness. (As opposed…
How about not showing ads in those cases?
1a) That is sub-optimal for ad publishers. I think in building such a system you have to assume they want to make the most money. And while ANY ad may not make much money, it's certainly not capped at the $0 they can expect from NO ad.
1b) Also: imagine if you implement AdCompanyX's html, and you see nothing. Is it broken? Out of ad inventory? Not yet parsed your page? It's not an impossible problem, but it adds UX complexity.
2) As I said in the original post, It's possible it's actually too far of a leap to go from dinosaurs (to kids?) to games. If this is the case, imagine some sort of fitness function scoring how close the best inventory match is to the content of the page. There's a cutoff under which you show no ad. Dinosaur could likely still be close enough to pass this test. Especially given the relatively little text in this example.