@FischyVishy said in #28:
> www.reddit.com/r/chess/comments/199mr2s/mikhail_tal_and_modern_chess_engines/
From the second highest rated comment from this thread:
"Many (maybe most?) of Tal’s sacrifices were unsound. (...)"
That's the myth.
If we compare the percentage of correct sacrifices amongst all of Tal's sacrifices with the related statistics from other players, we may see that Tal had a lower percentage of correct sacrifices — because he sacrificed more often!
But if we compare the percentage of correct sacrifices in the moves in all his games, with that of other players, then we will probably see that he played as many or more correct sacrifices than the other players.
For example, Lets say:
Karpov has 100 sacs per 10000 moves, all of them are correct.
Tal has 250 sacs per 10000 moves, 50% of them are correct.
Then indeed 100% of Karpovs sacs are correct and just 50% of Tals sacs are correct.
But Karpov has just 1% correct sacs in his games, while Tal has 1.25% correct sacs (and 1.25% incorrect sacs).
@mojo_jojo_1985 But then how do we know if it's a correct sacrifice? Also, many sacrifices you'll see are evaluated as being very bad for Tal, but I give them a full score on the human evaluation, so this kind of flag would miss that point. Furthermore, eval drops could be on moves which aren't sacrifices, so manual curation is still needed!
@Herr_Linnemann That is actually exactly what I was going for! Excellent, you're on the right track :).
Very long post, I think you had to research so much, thank you very much! Very nice!!!
Now, who other than @FischyVishy actually looked through every single one of those games?
@FischyVishy how would one distinguish between a sacrifice and a hung piece anyway?
Correctness is an objective notion. Whether a sacrifice is really really hard do prove wrong or not, if there is a line where said sac is refuted, then it is an incorrect sacrifice. Whether you or Magnus or a board of all the hall of fame chess players reincarnated at their peak gives them a full score is irrelevant in terms of correctness.
@Herr_Linnemann Let's say a correct sac brings a full point, a wrong one causes a loss and no sac leads to a draw.
Then Karpov would have 5050/10000 thanks to his 100 correct sacs,
Tal would have 5000/10000 thanks to his 125 correct sacs and 125 wrong ones.
@mojo_jojo_1985 What you're saying is true, of course, but in that sense I agree with the others who stated that there needs to be a human element to sacrifices. The reason is that risk-takers like Tal would get punished unnecessarily then. If a positional player only sacrifices when they see a clear line that wins, all of their sacrifices would be correct. Let's say, though, that someone like Tal sacrifices when they evaluate the position as "unclear", but an engine, with its immense strength, can correctly evaluate the "truth" of the position and evaluates the resulting sacrifice as "bad"; in that case, that would be very unfair for Tal. If no human could possibly correctly analyze the position because the refutation of the sacrifice is predicated on some really long sequence of moves + a subtle, only move on move 7 of the line for example, should players who sacrifice in that position really be punished? I believe not, and as the articles get published, you'd be surprised how frequently these kinds of moves arise. Tal really was a magician.
I'm just gonna point our that no matter the stones you turned and you did quite an extensive research apparently, you can't analyze Tal's games just with an engine. You need GMs with you to help you out, and even judge those engine lines and analysis. Engines are overrated, no matter their rating. Ratings are overrated too, that's a different story. And btw, so is analysis, Tal would have been the better person to prove the latter to you, he's not around, he left us his games. His games alone are proof enough of this, if you can see it that is, though if you need to perform an extensive research, maybe you are completely blind to what his games show so evidently to the rest of us, and you got more things in common with an engine rather than a human. Not all people are the same, and some, like you, do function this way, very much like an engine. Tal was obviously not. I'm personally not a fan of engines, they have caused harm to chess since they came about. Their lines lack everything a human does and it shows. Their lines and games are incredibly boring or completely out of touch. And I might even question your introductory praise for them as "capable of humiliating human players". Maybe it is so, cause humans got too depended on them, they became complacent, they stopped using their imagination and intuition, and stopped inventing stuff themselves. What could have been a subpar but useful tool, almost replaced them as superior, and as a result beat them in their own game to a point an engine minded person like you writes such a sentence for those who do not think alike, like Tal or others. At least in the beginning of the whole engine thing, IBM would need a rematch, staffed with the best human GMs they could get, to have their "engine" beat Kasparov. We all know how that went without those human GMs. Tal's moves objectively unsound? He'd still beat the crap out of you and your engines and make a mockery out of your analysis without even much of an effort.
@vstrider said in #37:
> I'm just gonna point our that no matter the stones you turned and you did quite an extensive research apparently, you can't analyze Tal's games just with an engine. You need GMs with you to help you out, and even judge those engine lines and analysis.
>
> Engines are overrated, no matter their rating.
>
> Ratings are overrated too, that's a different story.
>
> And btw, so is analysis.
>
> Tal would have been the better person to prove the latter to you, he's not around, he left us his games. His games alone are proof enough of this, if you can see it that is.
>
> Though if you need to perform an extensive research, maybe you are completely blind to what his games show so evidently to the rest of us.
>
> And you got more things in common with an engine rather than a human.
>
> Not all people are the same, and some, like you, do function this way, very much like an engine.
>
> Tal was obviously not.
>
> I'm personally not a fan of engines.
>
> They have caused harm to chess since they came about. Their lines lack everything a human does and it shows. Their lines and games are incredibly boring or completely out of touch.
>
> And I might even question your introductory praise for them as "capable of humiliating human players". Maybe it is so, because humans got too depended on them.
>
> They became complacent, they stopped using their imagination and intuition, and stopped inventing stuff themselves.
>
> What could have been a subpar but useful tool, almost replaced them as superior, and as a result beat them in their own game.
>
> To a point an engine minded person like you writes such a sentence for those who do not think alike, like Tal or others.
>
> At least in the beginning of the whole engine thing, IBM would need a rematch, staffed with the best human GMs they could get, to have their "engine" beat Kasparov. We all know how that went without those human GMs.
>
> Tal's moves objectively unsound? He'd still beat the crap out of you and your engines and make a mockery out of your analysis without even much of an effort.
>
Reformatted your text for readability.
I am not going to argue with you over your personal attacks.
All I say, if you think Stockfish playing style is ugly, I agree. But Lc0 actually plays a pretty, human like chess. You should try it out.
@mojo_jojo_1985 said in #35:
> @Herr_Linnemann Let's say a correct sac brings a full point, a wrong one causes a loss and no sac leads to a draw.
> Then Karpov would have 5050/10000 thanks to his 100 correct sacs,
> Tal would have 5000/10000 thanks to his 125 correct sacs and 125 wrong ones.
That's a good point. It is of interest how well the player scored with his correct and incorrect sacs.
You should publish a book, this is simply incredible