what is the best engine for big material handicap?

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

Moderators: bob, hgm, Harvey Williamson

Forum rules
This textbox is used to restore diagrams posted with the [d] tag before the upgrade.
User avatar
Ovyron
Posts: 3943
Joined: Tue Jul 03, 2007 2:30 am

Re: what is the best engine for big material handicap?

Post by Ovyron » Tue Oct 22, 2019 11:38 pm

todd wrote:
Tue Oct 22, 2019 11:12 pm
1) In faster time controls, more games are played, and players are more likely to play a series of games against each other, so we can watch live as score gaps open up quickly. In classical it may take years for a noticeable score gap between two players to emerge because they don't play consecutive games very often.
Okay, but a player with 1600 rating against a 1500 one is expected to have a 64% performance against them no matter what, their score gap shouldn't get bigger with more games. I think the point at which ratings break is when draws become very rare (then the superior time management player wins regardless of what moves were played in the game, and chess skills become less relevant.)
todd wrote:
Tue Oct 22, 2019 11:12 pm
2) Score gaps open up more quickly in games with fewer draws. Crazyhouse doesn't have many draws.
Crazyhouse at 3 0 doesn't have many draws, but, say Crazyhouse 10 0 hasn't been tested much, maybe it does go back to more normal drawrates.
todd wrote:
Tue Oct 22, 2019 11:12 pm
3) Crazyhouse is a niche game. When not as many people are searching for a game at the same time, the average game will have a larger skill gap. On a big server like lichess or chess.com, if you seek a 1 0 or 3 0 chess game, as long as you are not a beginner or elite player, you will more or less always quickly get a game with someone near your level, but this isn't the case in crazyhouse. More frequent lopsided matchups -> more big score gaps.
I see. So in this case the gaps would be caused by the players in the intermediate ratings that aren't playing the game. I guess this could be checked in small chess communities and see if elo's robustness depends on a large number of participants.
Your beliefs create your reality, so be careful what you wish for.

todd
Posts: 56
Joined: Thu Apr 19, 2018 7:09 pm

Re: what is the best engine for big material handicap?

Post by todd » Thu Oct 24, 2019 1:39 am

By score gap, I meant (wins-losses) rather than %. That's what would grow over time if the players are unevenly matched.

We are more likely to notice and pass judgment upon a 15-5 score than a 1.5/2 score even though both are 75%.

And we are even more likely to attach significance to the score if we see all 20 games play out in front of our eyes in one sitting, whereas if games are gradually played over the course of years, or if we see two players play two games and then stop, we probably won't remember it.

User avatar
Ovyron
Posts: 3943
Joined: Tue Jul 03, 2007 2:30 am

Re: what is the best engine for big material handicap?

Post by Ovyron » Fri Oct 25, 2019 1:42 am

One interesting fact about crazyhouse 3 0 is that the top 2 players are really close to each other, and it doesn't seem anybody else would have a chance against them in a match, so I wonder if they're already close to the human limit of strength, or if someone else could come with a new approach and easily defeat them, instead of just reaching their level and being really close to them.

Is crazyhouse already at the level of Magnus Carlsen in chess or someone could easily defeat the world champion, but that person hasn't picked up the game yet?

Uri Blass
Posts: 8750
Joined: Wed Mar 08, 2006 11:37 pm
Location: Tel-Aviv Israel

Re: what is the best engine for big material handicap?

Post by Uri Blass » Fri Oct 25, 2019 2:11 am

Ovyron wrote:
Fri Oct 25, 2019 1:42 am
One interesting fact about crazyhouse 3 0 is that the top 2 players are really close to each other, and it doesn't seem anybody else would have a chance against them in a match, so I wonder if they're already close to the human limit of strength, or if someone else could come with a new approach and easily defeat them, instead of just reaching their level and being really close to them.

Is crazyhouse already at the level of Magnus Carlsen in chess or someone could easily defeat the world champion, but that person hasn't picked up the game yet?
I do not understand what is the relevance of magnus carlsen.
I see no reason to assume that the level of magnus carlsen is the human limit of strength in chess.

Maybe with a better software to teach chess some top chess players(more than 2700) can become 200 or 300 elo stronger

Maybe some chess player with some chess talent to calculate faster than carlsen but inferior evaluation of chess positions relative to carlsen may learn from some future teaching software to improve his evaluation function in order to be significantly stronger than carlsen.

lkaufman
Posts: 4048
Joined: Sun Jan 10, 2010 5:15 am
Location: Maryland USA
Contact:

Re: what is the best engine for big material handicap?

Post by lkaufman » Fri Oct 25, 2019 5:22 am

Ovyron wrote:
Fri Oct 25, 2019 1:42 am
One interesting fact about crazyhouse 3 0 is that the top 2 players are really close to each other, and it doesn't seem anybody else would have a chance against them in a match, so I wonder if they're already close to the human limit of strength, or if someone else could come with a new approach and easily defeat them, instead of just reaching their level and being really close to them.

Is crazyhouse already at the level of Magnus Carlsen in chess or someone could easily defeat the world champion, but that person hasn't picked up the game yet?
I am sure that the two best crazyhouse players at 3 0 are nowhere near what we would call grandmaster level in normal chess, just as the best 3 0 player in normal chess who has never played serious tournament chess would not be a GM at standard chess right away. The ability to find tricky moves in a couple seconds and to move quickly and avoid simple traps is not very relevant to who is really the better player. Of course, if no one plays serious games then the best blitz players are probably the best slow players, but I suppose you are asking about whether the best players are near the limit of human potential in serious games, and the answer is certainly no. In 3 0, there is a limit to how fast humans can process and move, so they may be near that limit, but that has little to do with the actual game.
Komodo rules!

User avatar
Ovyron
Posts: 3943
Joined: Tue Jul 03, 2007 2:30 am

Re: what is the best engine for big material handicap?

Post by Ovyron » Fri Oct 25, 2019 8:51 am

Uri Blass wrote:
Fri Oct 25, 2019 2:11 am
I do not understand what is the relevance of magnus carlsen.
I see no reason to assume that the level of magnus carlsen is the human limit of strength in chess.
His relevance is being the current known limit of humanity. Some people have checked the quality of his moves and concluded that he could beat any other human in history if they played a match.

The rest of your post is speculation. Maybe *I* could surpass the human limit if I abandoned everything I'm doing and took chess seriously, everything is possible. By limit I don't mean that it's something that can't be surpassed, I mean it's something that hasn't been surpassed yet, and that we don't know if it can be surpassed.
lkaufman wrote:
Fri Oct 25, 2019 5:22 am
I suppose you are asking about whether the best players are near the limit of human potential in serious games
No, what I'm saying is that 3 0 crazyhouse IS serious crazyhouse. People engaging in it are taking it as seriously as people in chess when they have 100 minutes for the first 40 moves, and additional 50 minutes added after the 40th move.

If slower time controls are never adopted by crazyhouse players then 3 2 is the slowest time control we'll see and it makes sense to only talk about it in the same way we talk about game in 150 minutes of normal chess, because talking about a theoretical level of 150 minutes of crazyhouse would be like talking about a 1500 minute game of chess.

You don't need more than 2 hours on the clock for a game to be serious, on Playchess the engine room tournaments are happening at 5 0 time controls and those entities would destroy any human player (engine 5 0 vs human with 150 mins), so 3 2 time controls of crazyhouse could be enough for it.
Your beliefs create your reality, so be careful what you wish for.

lkaufman
Posts: 4048
Joined: Sun Jan 10, 2010 5:15 am
Location: Maryland USA
Contact:

Re: what is the best engine for big material handicap?

Post by lkaufman » Sat Oct 26, 2019 5:18 am

Ovyron wrote:
Fri Oct 25, 2019 8:51 am
lkaufman wrote:
Fri Oct 25, 2019 5:22 am
I suppose you are asking about whether the best players are near the limit of human potential in serious games
No, what I'm saying is that 3 0 crazyhouse IS serious crazyhouse. People engaging in it are taking it as seriously as people in chess when they have 100 minutes for the first 40 moves, and additional 50 minutes added after the 40th move.

If slower time controls are never adopted by crazyhouse players then 3 2 is the slowest time control we'll see and it makes sense to only talk about it in the same way we talk about game in 150 minutes of normal chess, because talking about a theoretical level of 150 minutes of crazyhouse would be like talking about a 1500 minute game of chess.

You don't need more than 2 hours on the clock for a game to be serious, on Playchess the engine room tournaments are happening at 5 0 time controls and those entities would destroy any human player (engine 5 0 vs human with 150 mins), so 3 2 time controls of crazyhouse could be enough for it.
Of course a blitz game can be "serious", in the sense that the players care about it and do their best, but just as the quality of human blitz chess is pathetic when compared to 4 hour tournament games, this must be true of crazyhouse as well, probably even more so as there are far more tactics which take time to resolve. The fact that a computer might beat the top human giving him 5 minutes to 3 hours odds (I'll trust you on that) has no more relevance to the quality of human blitz play than it does for normal chess. The quality of play by top human shogi players must be vastly above the quality of play of crazyhouse players (at any given time limit), because the game has centuries of history and theory and millions of dollars of sponsorship. In shogi blitz is reasonably popular (not quite as fast as chess blitz), but no one would say that the games compare at all to those played with several hours per side time limit. No one will be able to develop a valid theory about the game of Crazyhouse until the game is played (by humans) at time limits that allow for serious analysis. I wonder if anyone even knows the relative values of the pieces, both on the board and in hand, other than just guessing based on blitz play? The game is in its infancy.
Komodo rules!

User avatar
Nordlandia
Posts: 2588
Joined: Fri Sep 25, 2015 7:38 pm
Location: Sortland, Norway

Re: what is the best engine for big material handicap?

Post by Nordlandia » Sat Oct 26, 2019 8:05 am

Is it possible to play Shogi without the drop rule?

Play Shogi the same way as in normal chess, without crazyhouse drop.

User avatar
Ovyron
Posts: 3943
Joined: Tue Jul 03, 2007 2:30 am

Re: what is the best engine for big material handicap?

Post by Ovyron » Sat Oct 26, 2019 12:34 pm

lkaufman wrote:
Sat Oct 26, 2019 5:18 am
The game is in its infancy.
Being in its infancy would imply that it's going to mature, that "serious" theory is going to be built, and that humans are going to play it at several hours per side time control.

But if that's never going to happen, then the game has already matured and 3 2 is the slowest time control people are willing to try "seriously", because there's no interest from people to try anything slower (Lichess allows the experiment, you can create your own crazyhouse tournament with hours per side and see how nobody joins.)

Don't get me wrong, I'd wish for crazyhouse to flourish and get a similar treatment to chess, with new faces on the leaderboards and new amazing situations discovered on the board that wouldn't be possible in chess. Getting it out of its niche. But if it's all a dream, if crazyhouse has already gone as far as it could go, then I hold it'd be best to just embrace it in its current form, and try to popularize it as it is without trying to enforce longer time controls people don't really want for it.

Because, after all, the compensations comes with games played. The Crazyhouse World Championship was played at 3 2, yes, but they played 60 games. If you ask Elo about it he'd say the amount of games is statistically more significant than the few games that can be played with the Chess World Championship's time control. The argument could be turned on its head, and it could be claimed that it's chess who isn't serious because players should be playing many games until it's statistically significant to know who is better, and that the Chess World Champions could have been others if they continued playing as many games as the blitz time control of crazyhouse allows.

Perhaps crazyhouse's future would be in esports, like video games. A system could be put in place where you win a match after winning 3 games against someone, so the 3 2 time control would facilitate playing many games until a player reaches 3 wins first. Something different from chess, instead of trying to make it fit into chess's shoes, because if chess already is dying some draw death and people are trying to revitalize it with stuff like Armageddon, that's not a path you'd want some other game to emulate.

User avatar
Nordlandia
Posts: 2588
Joined: Fri Sep 25, 2015 7:38 pm
Location: Sortland, Norway

Re: what is the best engine for big material handicap?

Post by Nordlandia » Sat Oct 26, 2019 12:56 pm

There is no official OTB tournaments arranged for Crazyhouse. It's a pity but that's the way it is.

Post Reply