Your network blocks the Lichess assets!

lichess.org
Donate

Chess Engines, Centaur Chess and an idea for a new kind of tournament

this a great idea as its own variant.

this a great idea as its own variant.

Interesting idea. I’ll just add that correspondence chess organizations allow the use of engines. USCF doesn’t, but the world championship of correspondence chess is played with computers.

Interesting idea. I’ll just add that correspondence chess organizations allow the use of engines. USCF doesn’t, but the world championship of correspondence chess is played with computers.

Could we reverse the problem in being forced to consider engine suggestions at each turn and then being the human brain and the hand. But both sides would have a small godepth only AB pruning only of exactly the same parameters on both side, and the task would be to only thing about the beyond such go depth and accept or do a human move instead. Also the evaluation would be known to be the SF simple evaluation only, no NNue compression of that from deep entrail of the board. (Engine theory poetry of mine, I have a problem with oracle stuff, it has ruin many a days of many people over history, and it seems we are falling into such trap the whole lot of us, again).

Having engine as half gift, half poison of the mind. (I know in real chess learning that would be something I would practice, maybe not watch and faint in awe of prowesses being displayed in perfomance spotlight).

Could we reverse the problem in being forced to consider engine suggestions at each turn and then being the human brain and the hand. But both sides would have a small godepth only AB pruning only of exactly the same parameters on both side, and the task would be to only thing about the beyond such go depth and accept or do a human move instead. Also the evaluation would be known to be the SF simple evaluation only, no NNue compression of that from deep entrail of the board. (Engine theory poetry of mine, I have a problem with oracle stuff, it has ruin many a days of many people over history, and it seems we are falling into such trap the whole lot of us, again). Having engine as half gift, half poison of the mind. (I know in real chess learning that would be something I would practice, maybe not watch and faint in awe of prowesses being displayed in perfomance spotlight).

You could restrict the engine analysis to specific chess pieces to encourage the players to go into certain endgames. For example, one player gets the rook engine and the other gets the knight engine and both try to convert into an endgame that contains their "stronger" piece aka. the piece they can evaluate with their engine.

You could restrict the engine analysis to specific chess pieces to encourage the players to go into certain endgames. For example, one player gets the rook engine and the other gets the knight engine and both try to convert into an endgame that contains their "stronger" piece aka. the piece they can evaluate with their engine.

Good idea, but in my opinion, it might be impractical. Asking engine for the best move may not always be the best idea. Unlike video games, chess is very complicated and requires a very heuristic approach. In other words, their is no definite way to win or lose. Hence, the game heavily relies on ideas and technique. Ideas and techniques are not based on one move. Hence asking for one move from stockfish may only result us messing up later because we may not have the same idea stockfish might have had. Although it's a good idea, implementation is not as easy.

The eval bar is technically useless because it only shows us how the position is. With a bit of training, we might be able to judge it ourselves. Besides, if it is bad, it will only demotivate us. If it's good, we most likely knew it before. If we don't know, it's most likely that the position is not extremely advantageous for either side.

Good idea, but in my opinion, it might be impractical. Asking engine for the best move may not always be the best idea. Unlike video games, chess is very complicated and requires a very heuristic approach. In other words, their is no definite way to win or lose. Hence, the game heavily relies on ideas and technique. Ideas and techniques are not based on one move. Hence asking for one move from stockfish may only result us messing up later because we may not have the same idea stockfish might have had. Although it's a good idea, implementation is not as easy. The eval bar is technically useless because it only shows us how the position is. With a bit of training, we might be able to judge it ourselves. Besides, if it is bad, it will only demotivate us. If it's good, we most likely knew it before. If we don't know, it's most likely that the position is not extremely advantageous for either side.

@NFS_Rivals said in #35:

Good idea, but in my opinion, it might be impractical. Asking engine for the best move may not always be the best idea. Unlike video games, chess is very complicated and requires a very heuristic approach. In other words, their is no definite way to win or lose. Hence, the game heavily relies on ideas and technique. Ideas and techniques are not based on one move. Hence asking for one move from stockfish may only result us messing up later because we may not have the same idea stockfish might have had. Although it's a good idea, implementation is not as easy.

Thanks for reading! I agree with you, but I think that's part of what a player in this variant would have to consider: Can I learn something useful after spending resources to get an engine answer? You're right that the answer may be that it will be hard for the user to understand the engine's recommendation, but that's the risk a player takes when asking for help this way. That's part of what might make this variant interesting : Because there is no guarantee, players must calculate the cost/benefit of their choice. I've definitely had games where I had a sense there was probably an attacking chance I was having a hard time calculating, or a saving defensive move I wasn't seeing, and asking for help in those circumstances might provide the saving resource you need.

The eval bar is technically useless because it only shows us how the position is. With a bit of training, we might be able to judge it ourselves. Besides, if it is bad, it will only demotivate us. If it's good, we most likely knew it before. If we don't know, it's most likely that the position is not extremely advantageous for either side.

This is definitely something that depends on ability, but again, I think that potentially makes the variant interesting. Even at my relatively weak level, there are times where I think a particular sacrifice might lead to a quick mating attack, but if you told me that the eval bar was actually 0.0 I know not to trust my calculation.

@NFS_Rivals said in #35: > Good idea, but in my opinion, it might be impractical. Asking engine for the best move may not always be the best idea. Unlike video games, chess is very complicated and requires a very heuristic approach. In other words, their is no definite way to win or lose. Hence, the game heavily relies on ideas and technique. Ideas and techniques are not based on one move. Hence asking for one move from stockfish may only result us messing up later because we may not have the same idea stockfish might have had. Although it's a good idea, implementation is not as easy. Thanks for reading! I agree with you, but I think that's part of what a player in this variant would have to consider: Can I learn something useful after spending resources to get an engine answer? You're right that the answer may be that it will be hard for the user to understand the engine's recommendation, but that's the risk a player takes when asking for help this way. That's part of what might make this variant interesting : Because there is no guarantee, players must calculate the cost/benefit of their choice. I've definitely had games where I had a sense there was probably an attacking chance I was having a hard time calculating, or a saving defensive move I wasn't seeing, and asking for help in those circumstances might provide the saving resource you need. > > The eval bar is technically useless because it only shows us how the position is. With a bit of training, we might be able to judge it ourselves. Besides, if it is bad, it will only demotivate us. If it's good, we most likely knew it before. If we don't know, it's most likely that the position is not extremely advantageous for either side. This is definitely something that depends on ability, but again, I think that potentially makes the variant interesting. Even at my relatively weak level, there are times where I think a particular sacrifice might lead to a quick mating attack, but if you told me that the eval bar was actually 0.0 I know not to trust my calculation.

An interesting variation would be to allow players to consult the engine to get the top move+eval in the position, but forbid them from actually playing it. You can gain useful insight from it but will be forced to play a worse variation.

An interesting variation would be to allow players to consult the engine to get the top move+eval in the position, but forbid them from actually playing it. You can gain useful insight from it but will be forced to play a worse variation.

I think there are 2 philosophies of engine use attitude. As tool for humans to learn or experience new chess, and as oracle. The blog might have been only about the performance oracle. But then the author naturally is also concerned about the human effect of that. But it is not clear what the scope of the human experience is. Or did I read so far with lots of holes?

The single game performance versus the life time scale experience of many games or mini-games. These bring diffrent perspectives, and it seems we are dangling in the local learning perfomance versus the many games scope most of the time in chess learning theories, for the target question discussion part at least. I might have to explain my words but those are my thoughts. I might have to read parts I missed as well. I would welcome to be pointed back at those, for dedicated (with question in mind) attention.

I think there are 2 philosophies of engine use attitude. As tool for humans to learn or experience new chess, and as oracle. The blog might have been only about the performance oracle. But then the author naturally is also concerned about the human effect of that. But it is not clear what the scope of the human experience is. Or did I read so far with lots of holes? The single game performance versus the life time scale experience of many games or mini-games. These bring diffrent perspectives, and it seems we are dangling in the local learning perfomance versus the many games scope most of the time in chess learning theories, for the target question discussion part at least. I might have to explain my words but those are my thoughts. I might have to read parts I missed as well. I would welcome to be pointed back at those, for dedicated (with question in mind) attention.

a few points:

  1. having negative points sounds like a realy bad Idea to me. that means you win the game but loos points. I say if you reache '1' point, then ypu just can consult the computer again.

  2. the idea itswlf sound really intriguing to me. as a board game designer, I like the approach.

  3. there are other ways to limit usage of engine advice. for example, you can say that asking for a move cost a win point. asking for evaluation does not, but each player is only allowed 3 such requests per game. and when you get the evaluation, the other player also sees it. There can even be a rule that if the evaluation is above 5 points against you, you loose the game - so if you have a probable mate in 6, but you secreficed your quin for it, asking for evaluation may cost you the game

a few points: 1) having negative points sounds like a realy bad Idea to me. that means you win the game but loos points. I say if you reache '1' point, then ypu just can consult the computer again. 2) the idea itswlf sound really intriguing to me. as a board game designer, I like the approach. 3) there are other ways to limit usage of engine advice. for example, you can say that asking for a move cost a win point. asking for evaluation does not, but each player is only allowed 3 such requests per game. and when you get the evaluation, the other player also sees it. There can even be a rule that if the evaluation is above 5 points against you, you loose the game - so if you have a probable mate in 6, but you secreficed your quin for it, asking for evaluation may cost you the game

Local region of chess space of performance chess, is what I mean in previous post, versus the whole of chess experience.

Which includes studying it. Not always high stakes. That which allows one to explore and not fear losing some status, etc.... The non-human chess goal is not really what I seek in chess, seems to me like giving up on it. That might be why I don't get this oracle intrusion layer. And debating handicaps, as new strategic layers not on the board, seems like fleeing from the chess activity itself.

I also meant local optimal cavity of sorts. Being trapped in a technology bubble (there are some population tendencies of that sort of possibility in my very warped view, or perspective (not always, but recurrent), and chess does not seem to be immune).

perhaps it is question of some kind of humility in front of chess. Seems like a lot of chess culture in relation to engine use is about that. Like it should come as a surprise that we can't own or domiate the whole of chess. So give up to engine the understanding of it?

Local region of chess space of performance chess, is what I mean in previous post, versus the whole of chess experience. Which includes studying it. Not always high stakes. That which allows one to explore and not fear losing some status, etc.... The non-human chess goal is not really what I seek in chess, seems to me like giving up on it. That might be why I don't get this oracle intrusion layer. And debating handicaps, as new strategic layers not on the board, seems like fleeing from the chess activity itself. I also meant local optimal cavity of sorts. Being trapped in a technology bubble (there are some population tendencies of that sort of possibility in my very warped view, or perspective (not always, but recurrent), and chess does not seem to be immune). perhaps it is question of some kind of humility in front of chess. Seems like a lot of chess culture in relation to engine use is about that. Like it should come as a surprise that we can't own or domiate the whole of chess. So give up to engine the understanding of it?