Soccer-like chess

[cdani]

For my experience with Andscacs, it's 0.00 means I don't know how to improve, i.e. lack of knowledge. I'm not sure if this is applicable to Stockfish and Komodo.

I think I found a clearly better way of explaining the 0.00. For me it means that the "mathematical function" (search + eval) of the engine has found a local minimum.

[corres]

I also congratulate Komodo team on their success. But I think the developers of Stockfish and all contributors earn every recognition for their effort making Stockfish more and more stronger. To be the second between the strongest chess engines is a reason to be proud. In particular this case when the team of Stockfish works very hard without any salary. In a World which walks around the many it is a very praiseworthy action. So very much thanks to them!

[corres]

I think that draws are the inherent feature of chess. At least 50 Elo needs
to reduce draws. But big draw rate is not an evidence to perfect play rather to the same number of mistakes. In chess always the last large mistake causes defeat.
In my opinion for a match like TCEC is needed balanced and unbalanced but materially equalized openings too. One would like to know how the chess programs behave between these two circumstances - naturally with reversed color. I think the main issue about the openings of TCEC was that
these set of openings is not reproduces the repertoir of the chess of World.
Mainly i miss more opened openings, classical semi opened openings and more indian defences.

[corres]

Sorry, but of course ...World walks around the MONEY...
Oh, my "fluent" English....

[syzygy]

The 'unbalancing' of the openings was obviously done very poorly, and a complete bust. They were not anywhere near unbalanced enough. Pawn odds is known to produce around 20% advantage (i.e. a 30% score for the handicapped side), and even this falls short of the mark, as a 50-50 mix of draws and wins would cause a 75% score.

So it was predictable the out-of book advantage would have to be a Pawn and then something, and the chosen lines only provided a fraction of that.

I'm not sure where you got your above data, but a 70% score for a healthy extra pawn for Komodo at the start sounds pretty low, assuming you are averaging White and Black. Keep in mind that in the initial position any pawn removed other than the "f" pawn automatically gives some compensation in mobility. A score of about 0.7 to 0.75 in the opening for Komodo seems to be about the score which produces equal White wins and draws, which is roughly what an extra pawn in the opening would show if all else is equal. But yes, the positions were not unbalanced enough to avoid excessive draws.

The problem is that there is no right amount of imbalance once the engines start to play close to "perfection".

If 0.72 out of the opening is still a theoretical draw, then perfect engines will draw it, no matter what. If the engines are only close to perfect you might have a few wins and losses, but they will be rare.

Similarly, if 0.73 out of the opening is a theoretical win for white, then perfect engines will win it as white, no matter what.

A high win rate is very easy to achieve: just start with positions that have a winning advantage for white or for black. But this will not guarantee more interesting games than games starting from an equal position.

[Leto]

The 'unbalancing' of the openings was obviously done very poorly, and a complete bust. They were not anywhere near unbalanced enough. Pawn odds is known to produce around 20% advantage (i.e. a 30% score for the handicapped side), and even this falls short of the mark, as a 50-50 mix of draws and wins would cause a 75% score.

So it was predictable the out-of book advantage would have to be a Pawn and then something, and the chosen lines only provided a fraction of that.

I'm not sure where you got your above data, but a 70% score for a healthy extra pawn for Komodo at the start sounds pretty low, assuming you are averaging White and Black. Keep in mind that in the initial position any pawn removed other than the "f" pawn automatically gives some compensation in mobility. A score of about 0.7 to 0.75 in the opening for Komodo seems to be about the score which produces equal White wins and draws, which is roughly what an extra pawn in the opening would show if all else is equal. But yes, the positions were not unbalanced enough to avoid excessive draws.

The problem is that there is no right amount of imbalance once the engines start to play close to "perfection".

If 0.72 out of the opening is still a theoretical draw, then perfect engines will draw it, no matter what. If the engines are only close to perfect you might have a few wins and losses, but they will be rare.

Similarly, if 0.73 out of the opening is a theoretical win for white, then perfect engines will win it as white, no matter what.

A high win rate is very easy to achieve: just start with positions that have a winning advantage for white or for black. But this will not guarantee more interesting games than games starting from an equal position.

What if .71 advantage out of the opening is the perfect amount to get low draw rate and the better engine can draw with the disadvantaged side while the weaker engine loses? Wouldn't that be ideal for TCEC9?

.71 is just a number I made up but I think it'd be great for TCEC9 if we could find the perfect imbalance amount to get low draw rate and the stronger engine can draw with the disadvantaged side.

edit: I think Komodo and Stockfish are still not close to perfection, probably 400 elo more can be added still.

[lkaufman]

The 'unbalancing' of the openings was obviously done very poorly, and a complete bust. They were not anywhere near unbalanced enough. Pawn odds is known to produce around 20% advantage (i.e. a 30% score for the handicapped side), and even this falls short of the mark, as a 50-50 mix of draws and wins would cause a 75% score.

So it was predictable the out-of book advantage would have to be a Pawn and then something, and the chosen lines only provided a fraction of that.

I'm not sure where you got your above data, but a 70% score for a healthy extra pawn for Komodo at the start sounds pretty low, assuming you are averaging White and Black. Keep in mind that in the initial position any pawn removed other than the "f" pawn automatically gives some compensation in mobility. A score of about 0.7 to 0.75 in the opening for Komodo seems to be about the score which produces equal White wins and draws, which is roughly what an extra pawn in the opening would show if all else is equal. But yes, the positions were not unbalanced enough to avoid excessive draws.

The problem is that there is no right amount of imbalance once the engines start to play close to "perfection".

If 0.72 out of the opening is still a theoretical draw, then perfect engines will draw it, no matter what. If the engines are only close to perfect you might have a few wins and losses, but they will be rare.

Similarly, if 0.73 out of the opening is a theoretical win for white, then perfect engines will win it as white, no matter what.

A high win rate is very easy to achieve: just start with positions that have a winning advantage for white or for black. But this will not guarantee more interesting games than games starting from an equal position.

I don't think any engine today is close enough to perfection for your point to be relevant. If .72 is theoretically drawn and .73 is theoretically won, if we start with .72 positions the better engine will generally win as White and generally draw as Black; the errors both engines make will total WAY more than .01 in a game. If the engines are equally strong we'll get about half wins and half draws if we start anywhere near the line.

[carldaman]

Is it possible to come up with openings that are unbalanced enough to significantly reduce the draw rate and yet allow for the participants to create winning chances rather than just winning a won game? Is there a sweetspot into how unbalanced it can get without making it unfair?

Unbalanced opening sure and there is nothing wrong with draws it is only a 1/2-1/2 result, what mattered is how the engines played the game, those deep maneuvers that is very difficult to understand, you can only understand it after couple of moves in the game. There are draw results that are beautiful and thrilling that are better than a game with one side won. As a player I like to see how an engine solves a problematic position in the eyes of a human player. How to handle a pawn weakness like backward pawns and isolated pawns? How to counter a knight outpost?

I think the challenge to create opening is to produce an interesting game. Not all won games are interesting.

Agreed, not all drawn games are boring, either. Just go with opening positions that are 'rich in play' and enjoy the chess, no matter the result.

[whereagles]

I don't think any engine today is close enough to perfection for your point to be relevant. (...)

During TCEC8 I've seen a lot of close calls on what "best move" is. If you get like 5-10 of these calls during a game, at 50% chance to pick the right one, the odds for perfect play are like 0.5^5 at most, which is... hum... small

But then again, this assumes "best" and "2nd best" to have different chess consequence, i.e. "best" maintains or improves the win/draw/loss status whereas "2nd best" degrades it (if you see what I mean). Normally when engines disagree on what "best" and "2nd best" are, these moves have equal chess consequence.

[Laskos]

The 'unbalancing' of the openings was obviously done very poorly, and a complete bust. They were not anywhere near unbalanced enough. Pawn odds is known to produce around 20% advantage (i.e. a 30% score for the handicapped side), and even this falls short of the mark, as a 50-50 mix of draws and wins would cause a 75% score.

So it was predictable the out-of book advantage would have to be a Pawn and then something, and the chosen lines only provided a fraction of that.

I'm not sure where you got your above data, but a 70% score for a healthy extra pawn for Komodo at the start sounds pretty low, assuming you are averaging White and Black. Keep in mind that in the initial position any pawn removed other than the "f" pawn automatically gives some compensation in mobility. A score of about 0.7 to 0.75 in the opening for Komodo seems to be about the score which produces equal White wins and draws, which is roughly what an extra pawn in the opening would show if all else is equal. But yes, the positions were not unbalanced enough to avoid excessive draws.

The problem is that there is no right amount of imbalance once the engines start to play close to "perfection".

If 0.72 out of the opening is still a theoretical draw, then perfect engines will draw it, no matter what. If the engines are only close to perfect you might have a few wins and losses, but they will be rare.

Similarly, if 0.73 out of the opening is a theoretical win for white, then perfect engines will win it as white, no matter what.

A high win rate is very easy to achieve: just start with positions that have a winning advantage for white or for black. But this will not guarantee more interesting games than games starting from an equal position.

We are very far away from this scenario. To show this I plotted expected performance as a function of eval value of a recent Stockfish for time controls separated by a factor of 60:



TCEC is roughly another factor of 60 longer TC than 2' + 1.2'', so imagine TCEC line somewhere there higher, distanced by a similar amount to the seen separation. Nowhere near yet your scenario, and observe that on the relevant eval range [0.50, 1.00] the variation is quite small. So, choosing good unbalanced openings for TCEC is feasible. And we are very far away from perfect play.