Worst Move Chess

[vittyvirus]

Can we modify alpha-beta algorithm so that it searches for the worst move in instead of the best? If it's easy (I think it is easy), then what about conducting a tournament where both engines play worst moves in a position, and the one which loses wins?

[Evert]

Can we modify alpha-beta algorithm so that it searches for the worst move in instead of the best?

Yeah. Flip the sign of the evaluation (and the mate score) so that white's goal is to find the position that has the best score for black.

If it's easy (I think it is easy), then what about conducting a tournament where both engines play worst moves in a position, and the one which loses wins?

I don't think this works, since the situation where a side cannot avoid mating the other side has to be extremely rare (I think there are problems that have this as their point, where every legal move is mate in two or something) which means that in practice both sides will aim to trade down to a draw.

If you want to play Losers Chess you need to be a bit more clever about it, I think, but I've never tried to implement it myself. Perhaps having negative piece values (and appropriate victory conditions) is all you really need.

[Henk]

Perhaps that problem is even more difficult than writing for normal chess. For there is not much knowledge about evaluation. But in the land of the blind one eye is king.

Fairest tournament is inventing a total new game. And give the developers a few months to develop a program(engine? I even don't know what it is). After that games are played and results published.

Of course there should be enough money for each developer to compensate the amount of work done. So again money, time and motivation are bottlenecks.

[jorose]

Yes, this works as a modification. Not only that, but when I first implemented alpha beta for the first time (after I decided reinventing the wheel completely was a bad idea and back when I had trouble passing 1k nodes/second) this was a bug I had. I found it when I let the engine think in the position after
[pgn]
1.e4 e5 2.Nf3 Nf6
[/pgn]
I had realised before that I had a bug but was somewhat annoyed at how absolutely terrible he was playing. In fact the main line in that position was 3.Ke2 Ke7 4.Ne1 Ne8 I thought to myself those must be the worst possible moves outside of putting own pieces en prise. So I tried to think about how alpha beta is supposed to work and what could be wrong and I realised that putting a piece en prise is only bad if you think the opponent will take the piece. Then I realised the moves musst be the worst possible moves under the assumption the opponent is trying to play as badly as possible (3.Ba6 is really active if black has no intention of taking the bishop). So that's how I found the bug.

[Lyudmil Tsvetkov]

Can we modify alpha-beta algorithm so that it searches for the worst move in instead of the best? If it's easy (I think it is easy), then what about conducting a tournament where both engines play worst moves in a position, and the one which loses wins?

Syed, Syed,

1st, this thread belongs to the programming section, I hope the mods will move it, as it mentions specific algorithms, this is not a general discussion already. I am not sure if you did not do it on purpose, maybe you did.

2nd, completely pointless thread, as any move that is not the best, is simply the worst, as your performance chances only decrease

and 3rd, no one wants weak/worst moves, as no one wants weak/worst/unoriginal ideas. They are all around us, actually, this very same forum is FULL of weak/worst posts, diluting the quality of the forum and, maybe purposefully, distracting the attention from the real good things.

[Henk]

Perhaps that problem is even more difficult than writing for normal chess. For there is not much knowledge about evaluation. But in the land of the blind one eye is king.

Fairest tournament is inventing a total new game. And give the developers a few months to develop a program(engine? I even don't know what it is). After that games are played and results published.

Of course there should be enough money for each developer to compensate the amount of work done. So again money, time and motivation are bottlenecks.

Again this make no sense for a fairest competition does not exist. There are always participants that have advantage.

[vittyvirus]

1st, this thread belongs to the programming section, I hope the mods will move it, as it mentions specific algorithms, this is not a general discussion already. I am not sure if you did not do it on purpose, maybe you did.

Well, maybe you're right... It anyway wasn't intentional, although I didn't think twice before posting it here.

2nd, completely pointless thread, as any move that is not the best, is simply the worst, as your performance chances only decrease

Anymove that is not the best does make it a worse move, but not the worst! By worst move, I technically mean that the move which receives the worst score. You have to be more clear about what 'performance chances' really means.

and 3rd, no one wants weak/worst moves, as no one wants weak/worst/unoriginal ideas. They are all around us, actually, this very same forum is FULL of weak/worst posts, diluting the quality of the forum and, maybe purposefully, distracting the attention from the real good things.

If you aren't interested in the weak/worst, that doesn't mean no one is interested. There are threads about worst engines etc. Why would anyone want to distract our attention from "real good things"?
No attacking here...

[Uri Blass]

Can we modify alpha-beta algorithm so that it searches for the worst move in instead of the best? If it's easy (I think it is easy), then what about conducting a tournament where both engines play worst moves in a position, and the one which loses wins?

If you want to lose the game against an opponent who also want to lose the the right technique is not to play the worst moves.

It is better to win material except some pawns and later to use your material advantage to force the opponent to push the pawns and make checkmate.

for example see the following endgame
[pgn][Event "Computer chess game"]
[Site "URIBLASS-THINK"]
[Date "2015.03.13"]
[Round "?"]
[White "UriBlass"]
[Black "UriBlass"]
[Result "1-0"]
[BlackElo "2400"]
[Time "11:59:00"]
[WhiteElo "2400"]
[TimeControl "2400"]
[SetUp "1"]
[FEN "k2b4/P7/K7/1P6/3n4/1q6/8/8 b - - 0 1"]
[Termination "normal"]
[PlyCount "4"]
[WhiteType "human"]
[BlackType "human"]

1. ... Qb4 2. b6 Nb5 3. b7# 1-0
[/pgn]

[Lyudmil Tsvetkov]

1st, this thread belongs to the programming section, I hope the mods will move it, as it mentions specific algorithms, this is not a general discussion already. I am not sure if you did not do it on purpose, maybe you did.

Well, maybe you're right... It anyway wasn't intentional, although I didn't think twice before posting it here.

2nd, completely pointless thread, as any move that is not the best, is simply the worst, as your performance chances only decrease

Anymove that is not the best does make it a worse move, but not the worst! By worst move, I technically mean that the move which receives the worst score. You have to be more clear about what 'performance chances' really means.

and 3rd, no one wants weak/worst moves, as no one wants weak/worst/unoriginal ideas. They are all around us, actually, this very same forum is FULL of weak/worst posts, diluting the quality of the forum and, maybe purposefully, distracting the attention from the real good things.

If you aren't interested in the weak/worst, that doesn't mean no one is interested. There are threads about worst engines etc. Why would anyone want to distract our attention from "real good things"?
No attacking here...

The point is that there is simply no end to those weak/worst posts on this forum, simply no end.

From the posts started in the last 24 hours, 2 or 3 are of this kind, so 50%.
This repeats all over again.

I really like normal threads, with people discussing normal things, strictly computer chess related. Even if I do not understand them, I really like them.

I love the threads that contribute, containing some interesting info, usefuls stats, specific examples, etc. Even if I do not understand them, I really love them, as I know they are of value.

But what about threads like: I want a list of original engines, but please do not discuss which engines are original and which not? Can you make sense of what the questions is all about?

Or, Worst move chess. What is the purpose of this thread? What kind of specific information do you provide to the community? Any specific examples on this? How will that help authors to improve their engines?

Is not the worst move simply the move with the lowest score instead of the move with the highest score? Does not that end the discussion before it has even started? I am surprised people write here at all.

Henk posts a thread Worst engine 2015, but he has a point about it, it is at least cute, as I am sure our community likes all engines alike, strong and weaker, and looking at a very weak engine play is kind of fun.

What is the point about the thread Worst move chess?

In a garden you can not have at the same time flowers, and weeds. It is either the flowers, or the weeds. Some people like flowers, and other like weeds. At some point, even if there is a mix of flowers and weeds, one of the species outlasts the other. People should decide what they want, flowers, or weeds.

[jefk]

[quote="Lyudmil Tsvetkov"]
What is the point about the thread Worst move chess?
[/quote]

well there is a chess variant 'losing chess', some engines
have been made already for this, it can be played on some
servers, and with some good books i thought the
first mover can win (=lose) the game (although i'm not so
completely sure about this (but you could check it yourself)

http://en.wikipedia.org/wiki/Losing_chess
http://www.chessvariants.org/diffobject ... eaway.html

jef