Hardest chess variant for human

Discussion of anything and everything relating to chess playing software and machines.

Moderators: hgm, Rebel, chrisw

Madeleine Birchfield
Posts: 512
Joined: Tue Sep 29, 2020 4:29 pm
Location: Dublin, Ireland
Full name: Madeleine Birchfield

Re: Hardest chess variant for human

Post by Madeleine Birchfield »

duncan wrote: Fri Apr 16, 2021 12:07 am Which chess variants do humans find hardest, leading to mate against humans in less moves than regular chess?

Are there any variants where humans including grand masters make frequent tactical blunders, as the calculations are far too complex for a human, but simple for the computer.
Any chess variant involving 4 or more spatial dimensions, for the reason that humans are tertible at visualising higher spatial dimensions, while it is trivial to code up a board representation and move generation.of a variant in higher spatial dimensions for a chess engine.
Dann Corbit
Posts: 12537
Joined: Wed Mar 08, 2006 8:57 pm
Location: Redmond, WA USA

Re: Hardest chess variant for human

Post by Dann Corbit »

lkaufman wrote: Thu Apr 22, 2021 4:33 am
Dann Corbit wrote: Thu Apr 22, 2021 12:52 am Fourth dimensional Go, played on a tesseract.
Hard to do in ordinary 3-space, but computers can always expand the inner dimension to enable one more.
Go is not a "chess variant".
Well, a springer is not a chess man. And an Archbishop does not move like any particular chess piece.
It is a strategy game with two kinds of chess men. Dark stones and light stones.
Rules are different, and there is no king.
But, what (exactly) defines a chess variant?
Sometimes, after all, even the objective is to lose.
Taking ideas is not a vice, it is a virtue. We have another word for this. It is called learning.
But sharing ideas is an even greater virtue. We have another word for this. It is called teaching.
lkaufman
Posts: 5960
Joined: Sun Jan 10, 2010 6:15 am
Location: Maryland USA

Re: Hardest chess variant for human

Post by lkaufman »

Dann Corbit wrote: Thu Apr 22, 2021 5:51 am
lkaufman wrote: Thu Apr 22, 2021 4:33 am
Dann Corbit wrote: Thu Apr 22, 2021 12:52 am Fourth dimensional Go, played on a tesseract.
Hard to do in ordinary 3-space, but computers can always expand the inner dimension to enable one more.
Go is not a "chess variant".
Well, a springer is not a chess man. And an Archbishop does not move like any particular chess piece.
It is a strategy game with two kinds of chess men. Dark stones and light stones.
Rules are different, and there is no king.
But, what (exactly) defines a chess variant?
Sometimes, after all, even the objective is to lose.
I've always considered that if you win the game by capturing/trapping a specific enemy piece (which we call "king" in western chess), that makes it a chess variant (even if you can also win in other ways, like the old bare-king rule). So checkers/draughts/dame is not a chess variant, the idea is to capture all of the men. Neither is Othello (reversi), Hex, Renju, or Go. Losing chess does seem not to qualify as a chess variant unless it is played with the rule that checkmate still wins, although it is generally called a chess variant since it is played on the chess board with chess pieces. So I suppose a more inclusive definition would be either a game which you win by capturing/trapping a specific enemy piece or any game played with a standard or near-standard chess set and board. Some common sense is necessary, there will always be borderline calls with such a definition.
Komodo rules!
duncan
Posts: 12038
Joined: Mon Jul 07, 2008 10:50 pm

Re: Hardest chess variant for human

Post by duncan »

lkaufman wrote: Wed Apr 21, 2021 5:44 pm
duncan wrote: Wed Apr 21, 2021 12:34 pm
lkaufman wrote: Wed Apr 21, 2021 4:27 am
duncan wrote: Tue Apr 20, 2021 10:23 pm
lkaufman wrote: Fri Apr 16, 2021 8:14 am
duncan wrote: Fri Apr 16, 2021 12:07 am Which chess variants do humans find hardest, leading to mate against humans in less moves than regular chess?

Are there any variants where humans including grand masters make frequent tactical blunders, as the calculations are far too complex for a human, but simple for the computer.



How to Play 4D Chess

https://codeprojects.org/XL22-IlnhXtfp3 ... jJFcx-uHU/
I would expect that CrazyHouse chess would be such a variant, although of course this depends on how much effort goes into making a strong engine. But even the existing modified SF versions should be good enough to crush the best human in fewer moves than in normal chess. Another criterion could be how much handicap the best human needs. In normal chess, knight odds is enough for a top human (except with very short time limit); in Crazyhouse I don't think knight odds would be nearly enough.
Do you have an estimate of how big a handicap you would need to defeat a strong CrazyHouse chess, in tournament conditions ?
I am not a strong enough CrazyHouse player to answer your question, but in Shogi, the game that presumably inspired CrazyHouse chess with similar drop rules, I need rook plus bishop handicap to beat the strongest engine, and I am still technically the top-ranked American shogi player on the last pre-pandemic Pan-Atlantic rating list (no U.S. events since the Pandemic). Those are the two strongest pieces in shogi!
and do you have an estimate what handicap the world's top shogi player would need to defeat the strongest engine?
I think that such a match at bishop handicap would be very close and interesting, though as with chess the human's chances go up with the time limit. In a game with a fast time limit I think that the human champ would lose with bishop handicap, with something like 3 hours per side he might have good chances. I haven't heard of any such games having been played; I'm sure that top pros have tried this privately, but not on record.
Do you know if there any shogi/chess variants actual or hypothetical which a world champion would need a queen advantage to defeat it in tournament conditions?

Would 4 spatial dimensions be enough?
lkaufman
Posts: 5960
Joined: Sun Jan 10, 2010 6:15 am
Location: Maryland USA

Re: Hardest chess variant for human

Post by lkaufman »

duncan wrote: Thu Apr 22, 2021 2:34 pm
lkaufman wrote: Wed Apr 21, 2021 5:44 pm
duncan wrote: Wed Apr 21, 2021 12:34 pm
lkaufman wrote: Wed Apr 21, 2021 4:27 am
duncan wrote: Tue Apr 20, 2021 10:23 pm
lkaufman wrote: Fri Apr 16, 2021 8:14 am
duncan wrote: Fri Apr 16, 2021 12:07 am Which chess variants do humans find hardest, leading to mate against humans in less moves than regular chess?

Are there any variants where humans including grand masters make frequent tactical blunders, as the calculations are far too complex for a human, but simple for the computer.



How to Play 4D Chess

https://codeprojects.org/XL22-IlnhXtfp3 ... jJFcx-uHU/
I would expect that CrazyHouse chess would be such a variant, although of course this depends on how much effort goes into making a strong engine. But even the existing modified SF versions should be good enough to crush the best human in fewer moves than in normal chess. Another criterion could be how much handicap the best human needs. In normal chess, knight odds is enough for a top human (except with very short time limit); in Crazyhouse I don't think knight odds would be nearly enough.
Do you have an estimate of how big a handicap you would need to defeat a strong CrazyHouse chess, in tournament conditions ?
I am not a strong enough CrazyHouse player to answer your question, but in Shogi, the game that presumably inspired CrazyHouse chess with similar drop rules, I need rook plus bishop handicap to beat the strongest engine, and I am still technically the top-ranked American shogi player on the last pre-pandemic Pan-Atlantic rating list (no U.S. events since the Pandemic). Those are the two strongest pieces in shogi!
and do you have an estimate what handicap the world's top shogi player would need to defeat the strongest engine?
I think that such a match at bishop handicap would be very close and interesting, though as with chess the human's chances go up with the time limit. In a game with a fast time limit I think that the human champ would lose with bishop handicap, with something like 3 hours per side he might have good chances. I haven't heard of any such games having been played; I'm sure that top pros have tried this privately, but not on record.
Do you know if there any shogi/chess variants actual or hypothetical which a world champion would need a queen advantage to defeat it in tournament conditions?

Would 4 spatial dimensions be enough?
There is no queen or equivalent in shogi, the strongest piece is the rook. If the world champion were to lose a match with the top program at bishop handicap (which I rate as a tossup, but either side might win decisively), then the next larger handicap is rook, which would fulfill your criterion as the largest one piece handicap. There's no point in talking about giving queen handicap in games which are so dissimilar to standard ones that no human player would be likely to play anywhere remotely near "world champion" level. World chess champions have devoted their lives to mastering the game (same with shogi), but the best player of some newly made up game like 4 dimension chess would probably not have devoted even 1% of his life to it, nor is there any literature or theory from which to study or other strong players to play against. The game has to be close enough to normal chess or shogi for the skills to be translated. Clearly chess960 (FRC) is such a game, while CrazyHouse chess is a borderline call, there are clearly significant similarities with chess but still big differences. I would imagine that no human could beat a really strong engine in crazyhouse chess with rook odds, while queen odds is probably still a fairly easy win for the top human player of the game.
Komodo rules!
User avatar
towforce
Posts: 11542
Joined: Thu Mar 09, 2006 12:57 am
Location: Birmingham UK

Re: Hardest chess variant for human

Post by towforce »

lkaufman wrote: Thu Apr 22, 2021 4:33 amGo is not a "chess variant".

From an AI perspective, my personal definition of "chess style games" would be:

* two player board game

* turn based

* one move per turn

* both players have full information

* no randomising factors

* amenable to computer play by either game tree building or pattern recognition

* the best humans are not able to consistently get the best result (a draw in chess IMO)
Writing is the antidote to confusion.
It's not "how smart you are", it's "how are you smart".
Your brain doesn't work the way you want, so train it!