btickler Actually at correspondence chess--which is the highest form of chess--the very best correspondence players are FAR stronger than the best human. If you knew more about correspondence chess you would know this and it would be obvious to you?
It really does not matter what you predicted--what matters is your knowledge of chess and your knowledge of of the evidence and your knowledge of top correspondence chess and in these three areas you are quite deficient.
TheDeepMind team is
apparently studying chess Variants which have almost nothing to do with chess that we are discussing in this forum.
From your description of centaur chess you have little idea how it works--just like you have little idea how the very top level of correspondence chess works?
You are speaking out of ignorance.
btickler NO what you say in your last sentence is not inherently obvious.
This is because at the very highest levels of chess there have been perfect games played. There are correspondence players who have not lost a game in a decade and the very top correspondence players are WAY STRONGER than you think.
You couldn't claim to know a "perfect" game. That's the whole point.
And no, the very top correspondence players are not way stronger than I think. They are no stronger than any other human GM...and at least 500 ratings points lower than a decent engine.
Rather, you vastly overestimate your abilities, especially at this point.
As for Lola...well, I have played plenty of correspondence going back to postal cards, probably well before your first mani/pedi. If you meant "CC" as centaur chess, then obviously not as much, but I have experience. More than enough to comment here. Hell, I predicted the emergence of machine learning engines that would knock traditional engines off their perches years before A0/Leela came along, by playing games in their own sandbox, and determining valuations *without* human interference/bias. Human beings accumulated knowledge of chess through thousands of years of history was eclipsed in hours. That's the sum total/worth of human play in determining "best play", or solving chess.
Largely immaterial, though. The engines and their accumulated "knowledge"/wealth of data is far more important to the answer to this question than any human players' musings about it.
Hell, I predicted the emergence of machine learning engines that would knock traditional engines off their perches years before A0/Leela came along, by playing games in their own sandbox, and determining valuations *without* human interference/bias. Human beings accumulated knowledge of chess through thousands of years of history was eclipsed in hours. That's the sum total/worth of human play in determining "best play", or solving chess. Not much.
The DeepMind team is helping Kramnik figure out chess variants currently, for example. It's a form of respect to Kramnik, and I am sure they love to hobknob around with him in the lab...but they don't need him. He needs them. Developers > players in this context. Much the same way that a horsebreeder is a more important person in determining whether a horse has good odds of winning a race than a human steeplechase runner who can't even truly relate to a horserace, but also races in their own fashion. Now in centaur chess, it's more like jockey + plus horse, but the horse is still the deciding factor, and the jockey's main asset is being short and light enough to affect the horse's speed as little as possible. It's a mostly symbolic pairing that isn't really necessary, but it helps the humans feel better . Nobody wants to see the check being handed to the owner and trainers, they want to see the rider as the winner. The rider is incidental. Human beings are already next to worthless in any determination of best play, and that gap is only going to become more pronounced over time, not less. In 50 years from now, there won't be any point in human GMs trying to commentate on TCEC finals, etc. There will still be centaur chess, and human beings will still pretend they matter significantly in the outcome...but their role by then will be more like the spectator that hands a marathon runner a cup of water as they run by.
As for B-level chess, I will clearly have to defer to your knowledge in this arena.
Wow, what a spot-on post. Evidently not many minds will be changed by this discussion, but the argument about perfect chess being put forward by the btickler side of the argument is objectively and mathematically correct, so for a neutral observer this post hits the nail on the head.