Opposite Color Bishop Endgames

[chrisw]

Well, it isn't rocket science. It is the difference between measuring an average on a small, unrepresentative and manipulatable sub-set of problems

by “small”, I guess you mean “all the games it has ever played”.

by “unrepresentative sub-set”, I guess you mean “all positions it has encountered in all fair and according to the rules of chess games that its opponents share 50% of the responsibility for creating”.

by “manipulable”, I guess you mean the stronger of the two entities is able to steer the position towards a win.

or measuring the poor-case behavior on the total of all problems.

This you have not done, nor has anyone else, nor can it be done. Reinforcement learning by self-play is designed to find the sub-set of all positions (the full set being unavailable) which are statistically the most common that occur, and to put maximal effort into learning about those positions. Which is only sensible, if you think about it, as time and technology allows the subset expands in size and the program improves. Stockfish, world’s leading AB program, progresses according to exactly the same paradigm.

Tactical test suites are not a total of all problems. Nor are they a representative sub set of all problems. Your problem is that you are stuck on old paradigm that chess is tactics. It isn’t, as Deepmind and now LC0 prove. We had a lot of chess programmers who were not very good at chess, and people who are not very good at chess become quite convinced that chess is all about tactics, and they play chess constantly on the lookout for tactics. People who are good at chess understand the adage “stop looking for tactics, play positional and the tactics will come by themselves”.

Tactical testsuites are a (non-deliberately) manipulated sub set of positions. A position gets into a test suite for one reason only in reality. It is at or just beyond the boundary of what the historical set of chess programs can solve. Time progresses, those positions get solved, and new positions, again at the boundary, get added.
They are not deliberately designed for next years AB programs to solve, but that is generally the effect.

If you want to advertise that you cannot understand this, you are welcome.

Amusing. And quite the opposite. What was it you said? “LC0 is a good example: it sucks in tactical positions, as is easily demonstrated from its performance in tactical test suits. But in games it just avoids tactically complex positions, so it still has a pretty high Elo for a poor Chess program.”

“avoidance of tactically complex positions” means actually “play positional and the tactics will come by themselves”, but I think you don’t understand at that level.

Zero averaging programs don’t in any case “avoid tactical complexity”, they are quite asymmetric in that regard, they choose, deliberately, to steer to positions where they have the tactics, and the opponent doesn’t. But I don’t think you understand that either.

LC0 has a high Elo because in real world testing by being pitted against another entity it, more often than not, outsmarts it. Isn’t that the idea in chess?
If given a set of positions, designed unwittingly to show progress in AB programs, biased against positional and for tactical, a non-representative sub set of positions likely to be encountered in a normal chess game, stressing the untruth “chess is tactics” and refusing the truth “play positional and the tactics will come by themselves”, you can probably prove anything. That’s what the old paradigm unable to change people do.

I won't waste any more time on it.

Good. Invalidating something you don’t understand by using a bunch of assertions you don’t understand either was a waste of time in the first place.

Bye!!

BTW, it seems you read 'in' as 'is'...

[hgm]

Good! And now that the troll that wants to defend it is a great achievement rather than an embarrassing performance not to recognize such a simple draw has left, we can perhaps face the problem.

[chrisw]

Good! And now that the troll that wants to defend it is a great achievement rather than an embarrassing performance not to recognize such a simple draw has left, we can perhaps face the problem.

We were discussing your claim that “LC0 was a poor chess program”. Okay, I can see you are hopelessly out of your depth here.

[hgm]

I was discussing whether "good at chess" means the same thing as "has high Elo". If you want to hijack the discussion for other purposes, that is your problem.

And indeed, LC0 is a pretty poor chess program; it is usually outperformed by 2400-Elo engines in tactical positions. Centaurs that want to win games don't use it. Why do you think that is?

Games played by one engine are not a representative sample of all chess positions, or even of the negligible fraction of those formed by all positions from all chess games that have ever been played. Even against a variety of opponents, and even if all opponents would not be so similar that they are virtually indistinguishable. As you say, the opponent is only responsible for 50% of the moves, so the other 50% comes from the player itself. Even with only 2 playable moves per turn (that are regularly played by others), after 10 full moves one player could have resticted the sample space to 1/1024 of what is played in general.

[chrisw]

I was discussing whether "good at chess" means the same thing as "has high Elo".

Sure, that's a valid discussion. I remember being in a minority of one arguing just that back in 1990 something or other. Consensus in computer chess fell on the side of Elo, Elo and more Elo.

However, the basic fact does remain, Elo in free and open competition is the only objective measurement there is in a game where the objective is to win the game against another entity trying to do the same.

If you want to hijack the discussion for other purposes, that is your problem.

that's loser language. if you post contentiously on public forums, you'll get argued with.

And indeed, LC0 is a pretty poor chess program;

and that was your really stupid comment that got argued with.

it is usually outperformed by 2400-Elo engines in tactical positions. Centaurs that want to win games don't use it. Why do you think that is?

Games played by one engine are not a representative sample of all chess positions, or even of the negligible fraction of those formed by all positions from all chess games that have ever been played. Even against a variety of opponents, and even if all opponents would not be so similar that they are virtually indistinguishable. As you say, the opponent is only responsible for 50% of the moves, so the other 50% comes from the player itself. Even with only 2 playable moves per turn (that are regularly played by others), after 10 full moves one player could have resticted the sample space to 1/1024 of what is played in general.

Yes, that's opening play and really strong chess players are very good at it.

Clearly this LC0-hate has more deep seated reasons in your case, than the really lame bunch of assertions that you claim to be basing it on. Don't bother, your motivations are unlikely to be of any interest.

[hgm]

Seems more that you are an LC0 fan-boy, who gets all rigged up when someone doesn't praise it into heaven. To reconcile with reality you apparently imagine conspiracies for which people must have motives.

That LC0 is comparatively poor at tactics, and fails to find it in quite ordinary chess positions where mediocre engines do, is just a fact, however.

[chrisw]

Seems more that you are an LC0 fan-boy,

What an incredibly stupid comment.

who gets all rigged up when someone doesn't praise it into heaven.

but you didn't "not praise it to heaven", you said it was a "poor program". If you need to sarcasm into falsity, readers will take note.

To reconcile with reality you apparently imagine conspiracies for which people must have motives.

When I read someone, who really ought to know better, make obviously stupid and objectively false statements like "LC0 is a poor program" then giving lame technical reasons, then I intuit some irrational LC0-hate process going on. Is an obvious very silly negative comment about a piece of work that either is, or is very close to being the strongest chess playing entity in the world.

That LC0 is comparatively poor at tactics, and fails to find it in quite ordinary chess positions where mediocre engines do, is just a fact, however.

Well, there you go again. I would guess what happened is that you tried to emulate or create something yourself and failed. The "LC0 is a poor program" being the reveal.

[hgm]

Yes, you are full of delusions. Fact is that LC0 is poor at tactics. That this makes you think people must hate it, have hidden motives, have embarked on silly projects just doesn't sound like a healthy mental state to me.

[chrisw]

Yes, you are full of delusions. Fact is that LC0 is poor at tactics. That this makes you think people must hate it, have hidden motives, have embarked on silly projects just doesn't sound like a healthy mental state to me.

you said "LC0 is a poor chess program", actually. Which is not a fact. It's in direct contradiction to the facts. It's a hate-statement. Which makes me think you have reasons other than the lame rationalisations that you provided in justification.

[hgm]

It is poor at tactics, which makes it a poor chess program, as tactics is an important part of chess. Is such elementary logic beyond you?

That you suffer from some kind of paranoia by now should be clear to everyone. There is no need to further stress that by telling us what other crazy ideas have invaded your mind...