Something Hikaru Said

[PK]

It all depends how deep is a game of chess, or in another words: whether the ceiling is, say, 700 or 1700 Elo above current top engines.

[Jhoravi]

A knight Odds is already lost right in the opening so no matter how it plays perfectly it's still lost.

Knight Odds even if equipped with 32man tablebase is easier to beat compared to a less perfect move but with many traps.

In order to beat a human GM in Knight Odds, the engine need special evaluation that involves swindling by choosing a move (sometimes weaker) but contains many landmines that humans will likely fall into. Less effective is playing perfect moves while a piece deficit because it will only result to simplification.

[bob]

A knight Odds is already lost right in the opening so no matter how it plays perfectly it's still lost.

Knight Odds even if equipped with 32man tablebase is easier to beat compared to a less perfect move but with many traps.

In order to beat a human GM in Knight Odds, the engine need special evaluation that involves swindling by choosing a move (sometimes weaker) but contains many landmines that humans will likely fall into. Less effective is playing perfect moves while a piece deficit because it will only result to simplification.

"already lost" is a stretch if the side with the extra knight is 1000+ elo weaker than the other. About 15 years ago I visited one of the scholastic events with Cray Blitz and took on all comers. I eventually ended up playing rook handicap matches, and since there were no players there > 2000 (at least that took the challenge) CB won all of those games handily.

Bottom line depends on (a) time control; (b) motivation to win (it is NOT going to be easy, if it can be done at all, for the weaker side); (c) what the upper bound is on a computer's Elo relative to top-level human GMs. I don't think we can even think about claiming what this might be.

Clearly as the odds value increases, it should be easier and easier to force draws by trading pieces and pawns, since a mistake will generally not return all of the advantage instantly. But after watching the improvement over the last 50 years, nothing suggests to me that there is any absolute upper bound on computer skill.

[syzygy]

A knight Odds is already lost right in the opening so no matter how it plays perfectly it's still lost.

Knight Odds even if equipped with 32man tablebase is easier to beat compared to a less perfect move but with many traps.

In order to beat a human GM in Knight Odds, the engine need special evaluation that involves swindling by choosing a move (sometimes weaker) but contains many landmines that humans will likely fall into. Less effective is playing perfect moves while a piece deficit because it will only result to simplification.

"already lost" is a stretch if the side with the extra knight is 1000+ elo weaker than the other.

Not really. In a lost position, 1000+ Elo is not going to help you find a move that does not lose, simply because there are no such moves.

The only thing the stronger player can do is make it difficult for the weaker player to find correct moves. This is hardly a question of playing "perfect" or "good" moves. Regular chess engines are not designed to look for tricky moves and as hardware gets faster at some point the engines will in fact start to make things easier for their "weak" opponents (as they will see the refutations to all "tricky" moves and assume the opponent sees them too).

Even if the engine is somehow specifically designed to look for tricky moves, it seems unlikely to me that Carlsen, when given an extra knight, will not be able to avoid complications on his way to winning the game.

[lkaufman]

Komodo (latest version, on my 24 core machine, with my handicap book) has already drawn or narrowly won blitz matches (3' +2") giving knight odds to grandmasters (8 games with Lenderman, 13 with Mikhalevski). But at tournament level or even at our standard 45' + 15" I don't believe that any engine will ever win a match (maybe an occasional game) from a reasonably strong grandmaster (maybe an old one like me!) at knight odds. It just comes down to what the human error rate is, and how much that adds up to in an average game. I agree with Kai that something like two good pawns should be about the limit, at least against a WC Candidate. Maybe knight for pawn against an average GM is possible some day. Maybe it's possible to create an engine that is so good at inducing errors that knight odds against an ordinary GM might be possible, but I don't think so. It won't play like any engine today.

[Uri Blass]

Sounds like a challenge to me.

How that challenge has changed over the years:

1970: no computer will be able to beat an IM (David Levy) within 10 years (was actually stretched to maybe 15 years).

1980 or so: No computer will be able to beat a GM. Until Deep Thought.

1980 or so: No computer will ever be able to beat the current world champion (a GM obviously) and backed up by the Fredkin Foundation prize. Again, along came deep thought in 1997.

2016: No computer will ever be able to beat a GM if he is given knight odds.

2025: No computer will ever be able to beat a GM if he is given queen odds.



etc...

The difference is that it is obvious that it is only a question of time until computer beat the best humans in every game and not only in chess(assuming programmers work seriously about developing engines)
so the claims from 1980 were obviously wrong(the claim of david levy was correct because of the time limit of 10 years and he did not say never because he knew that never is not correct).

When we talk about odd matches then it is different and I am not sure even if computer will be able to beat top GM's in the future with a pawn odd at 2 hours/40 moves(maybe nakamura could win against komodo at that time control and we do not know).

[bob]

A knight Odds is already lost right in the opening so no matter how it plays perfectly it's still lost.

Knight Odds even if equipped with 32man tablebase is easier to beat compared to a less perfect move but with many traps.

In order to beat a human GM in Knight Odds, the engine need special evaluation that involves swindling by choosing a move (sometimes weaker) but contains many landmines that humans will likely fall into. Less effective is playing perfect moves while a piece deficit because it will only result to simplification.

"already lost" is a stretch if the side with the extra knight is 1000+ elo weaker than the other.

Not really. In a lost position, 1000+ Elo is not going to help you find a move that does not lose, simply because there are no such moves.

The only thing the stronger player can do is make it difficult for the weaker player to find correct moves. This is hardly a question of playing "perfect" or "good" moves. Regular chess engines are not designed to look for tricky moves and as hardware gets faster at some point the engines will in fact start to make things easier for their "weak" opponents (as they will see the refutations to all "tricky" moves and assume the opponent sees them too).

Even if the engine is somehow specifically designed to look for tricky moves, it seems unlikely to me that Carlsen, when given an extra knight, will not be able to avoid complications on his way to winning the game.

I've heard that for 50 years now. Can't beat a master. OK, can beat a master, can't beat an IM. OK, can beat an IM, but can't beat a GM. OK, can beat a GM, but can't beat the world champion. OK, can beat the world champion but can't beat a top GM with pawn odds. OK, can beat a top GM with pawn odds but can't beat a top GM with knight odds. It is just a matter of years until we go from knight to rook, and so forth. The issue will always be the difference in strength between the two players. If one is 1000 Elo stronger than the other, he is going to see so much more, he will be able to avoid trades, make things as complex as possible, and depend on a fallible opponent to falter at a key point. I would feel much less confident when talking about two computers that are 1000 Elo separated. But humans make mistakes, they get over-confident, they think "I can't lose this, I can afford to play a bit more aggressively than normal" etc. That's fatal against an opponent as strong as a computer with 32 piece EGTBs, not that this will ever happen obviously.

[Uri Blass]

A knight Odds is already lost right in the opening so no matter how it plays perfectly it's still lost.

Knight Odds even if equipped with 32man tablebase is easier to beat compared to a less perfect move but with many traps.

In order to beat a human GM in Knight Odds, the engine need special evaluation that involves swindling by choosing a move (sometimes weaker) but contains many landmines that humans will likely fall into. Less effective is playing perfect moves while a piece deficit because it will only result to simplification.

"already lost" is a stretch if the side with the extra knight is 1000+ elo weaker than the other.

Not really. In a lost position, 1000+ Elo is not going to help you find a move that does not lose, simply because there are no such moves.

The only thing the stronger player can do is make it difficult for the weaker player to find correct moves. This is hardly a question of playing "perfect" or "good" moves. Regular chess engines are not designed to look for tricky moves and as hardware gets faster at some point the engines will in fact start to make things easier for their "weak" opponents (as they will see the refutations to all "tricky" moves and assume the opponent sees them too).

Even if the engine is somehow specifically designed to look for tricky moves, it seems unlikely to me that Carlsen, when given an extra knight, will not be able to avoid complications on his way to winning the game.

I've heard that for 50 years now. Can't beat a master. OK, can beat a master, can't beat an IM. OK, can beat an IM, but can't beat a GM. OK, can beat a GM, but can't beat the world champion. OK, can beat the world champion but can't beat a top GM with pawn odds. OK, can beat a top GM with pawn odds but can't beat a top GM with knight odds. It is just a matter of years until we go from knight to rook, and so forth. The issue will always be the difference in strength between the two players. If one is 1000 Elo stronger than the other, he is going to see so much more, he will be able to avoid trades, make things as complex as possible, and depend on a fallible opponent to falter at a key point. I would feel much less confident when talking about two computers that are 1000 Elo separated. But humans make mistakes, they get over-confident, they think "I can't lose this, I can afford to play a bit more aggressively than normal" etc. That's fatal against an opponent as strong as a computer with 32 piece EGTBs, not that this will ever happen obviously.

I think that it is the opposite and computers who are 1000 elo weaker can lose inspite of getting knight advantage because the weak engine does not know the right strategy to play in these conditions and may go for complications instead of simplifications.

The facts are that even players with fide rating 2200 can beat komodo with knight odds even at 45+15 time control based on experience when engines with the same playing strength cannot do it.

Edit:Note that going for simplifications is not only trading pieces and for example humans are not going to allow unbalanced positions when they win a pawn and the opponent get a king attack for it even if they cannot calculate that the king attack is decisive.

[bob]

A knight Odds is already lost right in the opening so no matter how it plays perfectly it's still lost.

Knight Odds even if equipped with 32man tablebase is easier to beat compared to a less perfect move but with many traps.

In order to beat a human GM in Knight Odds, the engine need special evaluation that involves swindling by choosing a move (sometimes weaker) but contains many landmines that humans will likely fall into. Less effective is playing perfect moves while a piece deficit because it will only result to simplification.

"already lost" is a stretch if the side with the extra knight is 1000+ elo weaker than the other.

Not really. In a lost position, 1000+ Elo is not going to help you find a move that does not lose, simply because there are no such moves.

The only thing the stronger player can do is make it difficult for the weaker player to find correct moves. This is hardly a question of playing "perfect" or "good" moves. Regular chess engines are not designed to look for tricky moves and as hardware gets faster at some point the engines will in fact start to make things easier for their "weak" opponents (as they will see the refutations to all "tricky" moves and assume the opponent sees them too).

Even if the engine is somehow specifically designed to look for tricky moves, it seems unlikely to me that Carlsen, when given an extra knight, will not be able to avoid complications on his way to winning the game.

I've heard that for 50 years now. Can't beat a master. OK, can beat a master, can't beat an IM. OK, can beat an IM, but can't beat a GM. OK, can beat a GM, but can't beat the world champion. OK, can beat the world champion but can't beat a top GM with pawn odds. OK, can beat a top GM with pawn odds but can't beat a top GM with knight odds. It is just a matter of years until we go from knight to rook, and so forth. The issue will always be the difference in strength between the two players. If one is 1000 Elo stronger than the other, he is going to see so much more, he will be able to avoid trades, make things as complex as possible, and depend on a fallible opponent to falter at a key point. I would feel much less confident when talking about two computers that are 1000 Elo separated. But humans make mistakes, they get over-confident, they think "I can't lose this, I can afford to play a bit more aggressively than normal" etc. That's fatal against an opponent as strong as a computer with 32 piece EGTBs, not that this will ever happen obviously.

I think that it is the opposite and computers who are 1000 elo weaker can lose inspite of getting knight advantage because the weak engine does not know the right strategy to play in these conditions and may go for complications instead of simplifications.

The facts are that even players with fide rating 2200 can beat komodo with knight odds even at 45+15 time control based on experience when engines with the same playing strength cannot do it.

Edit:Note that going for simplifications is not only trading pieces and for example humans are not going to allow unbalanced positions when they win a pawn and the opponent get a king attack for it even if they cannot calculate that the king attack is decisive.

I'll say it again. In 1970, A human could beat a computer giving the COMPUTER knight odds, trivially. That has changed. Just because they can't do it today doesn't mean they can't do it next year, or next decade, or next century, or next millennium. Just means they can't today. We won't have 32 piece EGTBs ever, as far as that goes.

[syzygy]

I'll say it again. In 1970, A human could beat a computer giving the COMPUTER knight odds, trivially. That has changed. Just because they can't do it today doesn't mean they can't do it next year, or next decade, or next century, or next millennium. Just means they can't today. We won't have 32 piece EGTBs ever, as far as that goes.

I'll say it again as well. A 1000+ Elo advantage does not allow you to find a winning or drawing move in a position that is lost. Not today, not tomorrow, not ever.

The only real question is whether, say, Carlsen is good enough at chess that he is able to convert, against any play, an easily won position into an actual checkmate position. Since we don't actually know the answer in case of knight odds, this is a matter of opinion. But your analogy with predictions from 1970 is flawed.

Btw, 32 piece EGTBs are of no help here. They'll just show that a lost position is a lost position and that all moves from a lost position lead to equally lost positions.