Mate in 3: 7 minutes for SF 17

[Brunetti]

Normally a mate in 3 would require 7 ply (seen when the King gets captured). And it does not use excessive reductions. So it should see it quickly, or not at all. In this case I had expected the 'not at all', because the main line contains a zugzwang, and Fairy-Max does not do null-move verification.

Hi,
It’s quite clear that null moves and various combinations of extensions and reductions are fantastic winning features for normal gameplay, but they are absolutely detrimental in some checkmate problems and certain "artificial" positions, created specifically for entertainment and to challenge human minds, but totally irrelevant from a practical standpoint. Most of today’s strongest engines are probably (though I would remove this adverb) not the best tool to solve them. As an example, I took my modification of the old Fruit 2.1, made purely brute force, and it finds this checkmate faster than any other program tested here, in just 15 ms:

Code: Select all

id name Fruit 2.1 bruteforce AB
id author Fabien Letouzey
...
uciok
setoption name Hash value 1024
position fen 8/8/5k2/p6p/P1r2n1K/8/8/7R b - - 0 1
go movetime 1000
info depth 1 seldepth 1 score cp 541 time 0 nodes 2 pv c4a4
info depth 2 seldepth 3 score cp 511 time 0 nodes 54 pv c4a4 h1e1
info depth 3 seldepth 5 score cp 552 time 0 nodes 290 pv c4a4 h1d1 a4c4
info depth 3 seldepth 5 score cp 555 time 15 nodes 823 pv f4e2 h4h3 c4a4
info depth 4 seldepth 6 score cp 522 time 15 nodes 1326 pv f4e2 h4h3 c4a4 h1d1
info depth 4 seldepth 6 score cp 539 time 15 nodes 1615 pv c4a4 h1d1 a4e4 h4g3
info depth 5 seldepth 8 score cp 565 time 15 nodes 4043 pv c4a4 h1d1 a4c4 d1d6 f6e5
info depth 5 seldepth 8 score mate 3 time 15 nodes 7138 pv c4c3 h1h2 f6g6 h2h1 f4g2

I therefore think that countless other positions will gradually be found where the newest engine takes longer to find the solution, or doesn’t find it at all, but this should no longer come as a surprise to anyone.

Alex

[hgm]

I think the word 'strong' is being abused a bit in this context. I would not consider these engines strong at chess; they merely have high Elo. And Elo is purely based on an average. It is like asking an engineer what is the best place to make a road cross a river, when I had no money for a bridge. The river is 100m wide everywhere. But in one place it is 30 cm deep over its full width. At the other it is 10 cm deep, except in the central 10m; there it is 2m deep. The 'Elo engineer' would pick the latter place, because the average depth of the crossing there is only 29cm there. Which is 'better' than 30, right?

What these problems show is how you drawn into the hole.

[Ras]

I would not consider these engines strong at chess; they merely have high Elo.

You get high Elo by gaining higher scores, which in turn requires consistent performance over the whole game. That is being better at chess, by definition. The world chess champion is not determined by looking at puzzle positions and have the one earning the title who comes up fastest with the solution. Not sure what other definition of "strong at chess" you're using.

And Elo is purely based on an average.

That is misleading, and your following analogy as well. If an engine plays 30 perfect moves in a game and then throws away the queen, it will lose the game, leading to less Elo.

The issue here is the pruning, and there is no pruning that will always be perfect in every position. It just has gain more than it loses measured over whole games. This specific position has no impact on the winning odds because even without seeing the mate, Black's advantage will secure the win anway. Note that also in human play, only the game outcome counts, not how many moves the winner needs to win.

[hgm]

The world chess champion is not determined by looking at puzzle positions and have the one earning the title who comes up fastest with the solution. Not sure what other definition of "strong at chess" you're using.

But the wold champion is also not determined in a match of 10k games. You only have a very limited number of games in which you should prove yourself, especially in the candidate tournament. So reliability is more relevant than average strength.

In my definition of 'strong' reliability is certainly a factor.

[Ras]

But the wold champion is also not determined in a match of 10k games.

Because humans cannot do that, so they live with high error bars. With so few games, "reliability" is not even measurable. Especially in the candidates where you need to take risk, or else you won't get to the top. Someone like Petrosian would not make it in today's format.

In my definition of 'strong' reliability is certainly a factor.

Reliability at what? At winning games, obviously. Which, in this position, Black can easily do without seeing the mate.

[hgm]

Optimizing Elo would not make you reliably win games. You would happily sacrifice some games if it means it makes you win a marginally larger number of other games.

[Ras]

You would happily sacrifice some games if it means it makes you win a marginally larger number of other games.

Winning more overall score means playing better chess. That's the definition of better chess because winning is the game objective. If you have a tournament where one player wins nine games and loses one, and another player who doesn't lose a single game, but only wins six and draws the other four, the former will be placed higher than the latter. That's what calculated risk is all about.

The whole "higher Elo doesn't mean better chess" idea is nonsense IMO.

[towforce]

I would not consider these engines strong at chess; they merely have high Elo.

Now THERE'S a quote you weren't expecting from a high IQ guy!

The point made in the full post is interesting, though: the engines are often not picking moves that lead to a win in the shortest possible number of moves. This makes the case for my approach: find some deep underlying patterns in chess that unlock the game (I haven't proven that they exist, but I have demonstrated that they are very likely to).

[Ras]

the engines are often not picking moves that lead to a win in the shortest possible number of moves.

Neither do humans. If given the choice between a highly tactical, but shorter win, or just trading down into a trivially won endgame, many humans will do the latter because it means less room for error. Since game length isn't a metric in chess, only win/draw/loss is, both ways are equally strong.

[hgm]

Winning more overall score means playing better chess. That's the definition of better chess because winning is the game objective. If you have a tournament where one player wins nine games and loses one, and another player who doesn't lose a single game, but only wins six and draws the other four, the former will be placed higher than the latter. That's what calculated risk is all about.

The whole "higher Elo doesn't mean better chess" idea is nonsense IMO.

Well, it might be your definition of better...

The point is that it is not always beneficial to take a 'calculated risk'. Risks that are worth it when you have an infinte number of games to recover from the potentially ensuing mishap, might not be worth it in the decisive game against your main competitor in a round robin.