Clearly a GTO strategy can't be exploited. However for one thing people should be very careful about using the phrase GTO in describing poker. Poker is not (anywhere near) solved, so a GTO strategy for playing (say) 3 streets of hold'em is far far far from available. We can only say what is game theory optimal in very very limited circumstances. Most of the time when people say GTO they mean "I did a couple of EV calcs and dc/2+2/most of my skype buddies agree".
I agree. Although people misuse the term GTO is not an excuse for Alex to make the same mistake. Or he at least misapplies it. And although a GTO strategy for 3 streets of poker is no where near being solved doesn't mean that doing your best to use a GTO method to solve a spot in game is unpractical.
Now, to Alex Jacobs point. Say you get to the river with a class of bluff catching hands and all of his value hands beat you and you beat all of his bluffs. Your knowledge of game theory says you should call x% of the time and fold the rest. If you decide "ok, I get here with AK x% of the time so every time I have AK I call". You are definitely not playing a GTO strategy at that point. If the class of hands are all truely equivalent then you need to call at random each time with probability x, and if they are equivalent-ish but varying in strength then you need to call with a weighted probability that adds up to x overall (to handle the case where he might say bluff with AT). If you always call with AK your strategy is exploitable if Prahlad can handread you. If he knows you don't have AK then he shoves more bluffs and if he knows you have AK he shoves more nuts. Alex did not misunderstand what a GTO strategy is, he's saying if the odds say we call x% of the time, we can't just use hand strength alone to decide what makes up that x% (from a pure game theoretic point of view).
I don't really want to get into the details of GTO strategy unless you think it's somehow important to the conversation. But can we both agree that a GTO strategy would say that if your only choices were to call or fold that you need to call some x% of the time and that you should use the best x% of hands in order to make your opponent indifferent to choosing any other action. So you might need to call with AK all the time, fold AK all the time, or calls with AK some % of the time. I think we both agree with this, but you can tell me if you don't.
But I'm not really sure where you think that AJ says any of the points you make in his post.
Where do you think he tries to compare using absolute hand strength vs relative hand strength in deciding which hands are best to call with. Sure he points out that AK and KQ is very similar in terms of relative hand strength, but he doesn't point out that KQ would be a better call because of blockers. Actually in his post he seems to ignore the obvious blockers potential and tries to equate the two hands. "and yet most of you would agree that the vast majority of the time, P has either less than KQ or greater than AK." And so Alex is suggesting that it shouldn't matter if we call with AK or KQ because against Prahlads range they're both bluff catchers. And so he is completely backward in suggesting that a GTO approach would somehow suggest that if you call with AK that you should also call with KQ. In fact, the other user's approach that you should call some x% of time (even if it is based on absolute hand strength vs relative hand strength) is almost surely closer to a GTO approach.
He goes on to make arguments such as Mahatma can exploit a GTO strategy, "it seems to me that this is sort of a game-theoretic approach... THIS is the kind of thinking that makes P so successful, not the guys who are saying to fold. because when you think like this you are giving up already.... you are just saying, my hand falls into the x% when i should call, i hope i win. he knows that you do this."
Instead he says that you need to try and use an exploitative strategy to beat Mahatma and try to "read him", and it is this way that we're going to stop Mahatma from being "so successful". Actually if players were using a more GTO approach, Mahatma would almost surely be LESS successful. It's the people who keep trying to read him who are getting out played by his exploitative strategy, since their ranges will be much much more unbalanced, then if you're using something closer to GTO approach.