I applaud this fantastically succinct point: "Some foods taste good but kill you. Some drugs make you feel good but rot your brain. And some ads lead the curious away from knowledge."
I do think, however, that it's incredibly difficult to construct an entirely machine-learning- and auction-based ad network without these sorts of problems popping up. Especially with presumably low dinosaur-related ad inventory to choose from. (Perhaps dinosaurs -> kids -> games isn't a far leap in terms of correlation?)
Based on physorg's deceptive ad placement, text styling, and terrible anti-aging ad selection for the bottom, I would argue that our collective sadness is better directed at them than at Google.
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 to defend the current approach as the "end-state")
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.
FYI your cutoff idea is what Google actually does. Ad quality is a problem Google works on. The main complaint people have is that Google doesn't do a good enough job of overruling users "for their own good" when those users are consistently actively telling Google what they want to see.
If that were true, why would there be a need for Adblock? Google would just stop showing ads to people who don't like them, or would be showing us all such great stuff that we'd be grateful for them.
I do think, however, that it's incredibly difficult to construct an entirely machine-learning- and auction-based ad network without these sorts of problems popping up. Especially with presumably low dinosaur-related ad inventory to choose from. (Perhaps dinosaurs -> kids -> games isn't a far leap in terms of correlation?)
Based on physorg's deceptive ad placement, text styling, and terrible anti-aging ad selection for the bottom, I would argue that our collective sadness is better directed at them than at Google.