Only considering products made by actual engineers at big companies makes a ton of sense. All the dead web frameworks of the week where made by silly hipsters weren't great choices in retrospect.
> And the issue of trying to prove N=NP is essentially trying to prove that there isn't a magical way to train a neural network with just one image of training data
No it's not. You can't do reasoning on bullshit inaccurate analogies.
> I objected to the game and the rules, and did not want to play. After the game, in class, he commented that "Exchange students always object that the rules are unfair, while local students always accept that the rich get to set them."
Isn't objecting to unfair games super silly as life is inherently unfair.
Plenty of way parents can heavily influence you. I would have turned out much different if my father didn't teach me programming. The current POTUS wouldn't be who he is if he didn't get a small loan of a million dollars.
Looking at the paper the set of games it can build seems rather limited.
It could work for cookie cutter platformers (the sort of games that engines that don't require programming let you build).