We don't have enough information to be certain that best play by both sides results in a draw. There may come a day where a computer shows a forced win, probably for white, no matter what black does.
True or False Chess is a Draw with Best Play from Both Sides
Sort:
Aug 19, 2019
0
#2383
GreenyKnight2001 wrote:
It is also possible that at the beginning of the game white is in a very complicated zugzang and it is black who can force the win...
Agreed, which is why I made a point of saying "probably for white", I recognize that there is a possibility that chess is a forced win for black.
I hope that chess is never 'solved', I'm not sure what that would do to the future of the game and I don't want to find out.
Nwap111 wrote:
As you know Ponz, very often a draw is a result of mistakes by both sides, not of perfect play by both sides. So is it possible to say perfect play(which no one knows what it is) results in a draw?
Exactly. Nobody has any idea what perfect play is. If they did, they would do it every time. Even the worlds best computers dont play perfect. He just means that as far as he knows, based on his extremely limited knowledge (which we all suffer from) he believes chess is a draw.
Actually very strong players do have an idea of what perfect play is in chess just as checker players had an idea of what would be perfect play in checkers'
It is a fallacy to assume that if you have an idea of what perfect play is in chess that you would always be able to play a perfect game.
Yes, a draw is often a result of mistakes played by both players. And of course it is possible to play a perfect game either knowing or not knowing you have played a perfect game.
Some very short games are often perfect games.
It is true that chess may never be solved via math. It is possible that mankind will die out without chess being solved via math.
However, it is very evident to all the strongest players that chess is a draw with perfect play. When Magnus plays a game of chess--he knows before the game starts that if he and his opponent play perfectly--the game should end in a draw.
Magnus and other strong players base their chess theory--especially their opening theory on the assumption that chess is a draw when played perfectly.
Yes, Ponz, they may know what perfect play looks like. But didn't you just say that they cannot play it? Isn't that the problem. Even if you know what perfect play is, no human can play perfectly. In fact, all human contests are just a contest of error. We call skill the ability to minimize error.
ponz111 wrote:
Yes, a draw is often a result of mistakes played by both players. And of course it is possible to play a perfect game either knowing or not knowing you have played a perfect game.
Some very short games are often perfect games.
You mean games like this...1.e4, black resigns? How do you know any very short game is perfect? Have you made sure every possible move is not better? You are simply guessing. You don't know, I dont know. Nobody knows. You have this notion that perfect must mean a draw. But you dont know. Perfect might mean a win for white or black. Nobody knows what perfect is, it's never been done.
When you guess that a perfect game has been played, you are assuming that no moves exist that could make the game better for either side. The only way that's possible is the game 1. draw agreed. Even then, you are still just guessing.
There are perfect moves, or perfect play, like a mate in one. But there are not perfect games because all the possibilities have not been explored.
ponz111 wrote:
It is true that chess may never be solved via math. It is possible that mankind will die out without chess being solved via math.
However, it is very evident to all the strongest players that chess is a draw with perfect play. When Magnus plays a game of chess--he knows before the game starts that if he and his opponent play perfectly--the game should end in a draw.
Magnus and other strong players base their chess theory--especially their opening theory on the assumption that chess is a draw when played perfectly.
But you know that's not true. If the world champion plays "perfectly" which can only be the best of his ability and no better, and the worlds best computer plays "perfectly" which is also just the best of it's ability, it will not end in a draw every time. One perfect is better than the other perfect. Which is why actual perfect has never been done.
It is complete bunk that a perfect game has never been played. Actually thousands of perfect games have been played.
"Perfect play" does not mean play to the best of your ability. Your ability to play varies from day to day and sometimes from hour to hour. "Perfect play" means not to make a mistake which would change the outcome of the game. There is even a chance that 2 experts will play a perfect game together.
There are zillions of ways to play a perfect game. Zillions of different move sequences.
If you play a perfect game--there are still variations that could have been played which also would have resulted in a perfect game.
Some people assume there are only a few ways to play a perfect game--this is not true at all.
It is not true that when you play a perfect game you should assume that there were no other moves or series of moves which would have resulted ion a perfect game.
1. d4 or 1. c4 or 1 Nf3 or 1. e4 or 1. e3 are some of the moves which will lead to a perfect game if both players play well. GMs and other strong players know this.
These opening sequence will also lead to a perfect game if both players play without error.
1. d4 Nf6 1. c4 Nf6 1. e4 e5 1. e4 c5 1. Nf3 Nf6. and many many others....
One "perfect" is not better than another "perfect" I have in the past shown many positions where there are 10-15 different moves which could lead to the rest of the game being played perfectly.
Aug 19, 2019
0
#2395
I've watched games between alpha zero and stockfish where neither side has made a mistake, but rather alpha zero just plays better moves. It's hard to understand, but agadmator has a good analysis on it.
I believe with humans, yes chess is a draw with perfect play. But with engines, no.
Then again, perfect play is a weird term. To one it might be a certain move, to another a different move.
ponz111 wrote:
It is complete bunk that a perfect game has never been played. Actually thousands of perfect games have been played.
"Perfect play" does not mean play to the best of your ability. Your ability to play varies from day to day and sometimes from hour to hour. "Perfect play" means not to make a mistake which would change the outcome of the game. There is even a chance that 2 experts will play a perfect game together.
To me, that sounds ridiculous. If perfect isn't the best, what is it? Better than best? You are defining perfect as a move that doesn't change the outcome of the game. And that's true for mate in one type positions. But we are talking about the game from move one. You dont know what the perfect move or perfect game is because you dont know if a move changes the outcome of the game. You dont know that because you dont know all the possibilities. Nobody does.
Keep in mind that your idea of a perfect game is just your opinion. Your idea of perfect is probably very different from someone else. If a game ends in a draw, you believe it was perfect. For other people, perfection in chess doesn't exist, at least not yet, because computers and people are still improving. Perfection hasn't been reached yet, and probably never will.
I am defining "perfect game" as one where neither side makes a mistake which would change the outcome of the game.
I have given many positions where there are many moves which could lead to a perfect game.
It is very silly and a type of "strawman" for you to say that if a game ends in a draw--I believe it is a perfect game. I never said that. In fact I acknowledged that just because a game ends in a draw THAT DOES NOT MEAN THAT GAME WAS A PERFECT GAME!
It is true that we are only human and have not reached being a perfect player--but that does not mean we have never played a perfect game. Most GMs have played perfectly from their side of the board and if they encounter a player who also plays a game perfectly--then a perfect game will be reached.
This has happened to many GMs. Many so called "grandmaster draws" are, in fact "perfect games"
And while perfectionism ion chess has not been reached yet.
SpiderUnicorn wrote:
We cannot yet prove if a 'perfect' game is actually perfect or not.
According to ponz, that's easy. Just play perfect moves, and when the result is any number of different draws, that's proof it was perfect. I think "obvious" was the term he used.
Perfect play and best play is theoretically different.
Perfect play is 100% perfect mathematic where a
1. winning condition must be 100% winning
2. a draw must be 100% draw
3. a losing game ( doesnt matter ).
We have 7 men tablebase that valid above conditions.
You will be surprised to see that perfect player will have lower elo rating than best play!!!
Why?
This is a tablebase position which is 100% draw. A perfect will play random moves including giving away the rook. cos perfect player knows KRB vs KR is draw and KB vs KR is still draw.
But best play will try to get the best chances to keep trying until opponent does mistake. So, best player will have higher winning chance if another opponent is imperfect.
Theoretically today stockfish is 3500 rating, perfect chess player may be even 3400 rating because of giving away winning chances against imperfect players.
Stockfish had similar problem with perfect play( of course if position is simple, Stokfish can 100% correctly calculate like perfect player). Cos regular Stockfish knows similar positions leads to draw and doest not try anything to win ( in fact theoretically draw , you cant do anything). However, in 2018 May, programmers introduced imperfect evaluation to stockfish ( contempt 24). This way stockfish is virtually fooled with +0.24 evaluation without reason. So, stockfish think many draw positions are not 0.00 anymore but +0.24 and try to win a perfectly draw positions( that works against weaker players).
Surprisingly, that imperfect play( you can call best play as it is trying every possible chances to win) increased stockfish elo approximately +25.
As you know Ponz, very often a draw is a result of mistakes by both sides, not of perfect play by both sides. So is it possible to say perfect play(which no one knows what it is) results in a draw?