True or False Chess is a Draw with Best Play from Both Sides

PATRIOT  The very obvious answer to your question is draws. In fact the best players will see the new information and realize that future play at the highest levels is futile as they will never be able to win even one game. There will not be any White wins or Black wins.

Here are the probable reasons White wins more and more and Black wins less and the draw rate keeps going up:

. 1. At the highest levels the players know that chess is a draw. 

2. They do not even try for a win with Black. They try for a draw with Black. This is because White has the very small practical advantage of the fiirst move --so if Anyone might get lucky and win--it will very probably be White. 

3. There have been several openings which have been analyzed to a draw--example Ruy Lopez. White will avoid such openings.

Your rating is much higher than mine so I respect your evaluation that chess is 99.9999% a draw, but you are talking about high level human players. Perhaps that is true for humans, but it seems that Stockfish most likely will win against any GM and Stockfish itself has been crushed in games by AlphaZero. So if these computers with their superior capabilities can find ways to win against both humans and other computers that proves that the overwhelming proportion of drawn games between humans that you present as evidence that chess is a draw are really mostly a result of faulty play on both sides. Perhaps humans are not capable of the massive analysis required to result in a lot of winning games.

Neither Stockfish nor AlphaZero play perfect chess, although their errors are nearly impossible for humans to exploit. In any case, most of their games end in draws.

ArthurEZ  human high rated players [2500 and up] lose to the best machine for one main reason--speed of calculation  A strong computer can look at tens of millions of positions in one second. Of course humans cannot do this. 

It is like comparing the very best hujman marathoner to a automobile--the automobile il will win.  

However be assured that those high rated players jhave a ton of knowledge about chess.

Ziryab - My point is that if one of these computers were to play a human from some early position in any drawn game by human grand masters I think they most likely would find a win. I also suspect even more advanced programs developed in the future would be able to defeat any current ones in most games, but perhaps would most often draw when playing each other. The question really is can a big enough move tree ever be produced that would show a forced win or draw? That I cannot answer. 

Ziryab  You are correct that the best form of stockfish and also Alphja Zero can make mistakes.  buit now they are fairly rare. 

But there is new information I saw in IICCF correspondence postings  which seems to show a chess application that makes virtually no mistakes, And this will probably be thje end of correspnddence chess at the hiighest levels.  [Becauise nobody will be able to win even one game]

As a former corresponmdence chess player I am a little sad that this may happen.

https://www.chess.com/club/chess-for-the-live/join

Arthur   Maybe in a very early position that was agreed drawn by GMs--some strong chess engine might find a win?  This is unclear as iit has not been tested?

I very much disagree with your idea that super advanced chess engines of the future will be able to win most games from our now super chess engines.

I would suggest that you look at iCCF correspondence chess at the highest levels and also look at the new conversations ?

Thank you. It might have been the first tournament I have won, and all four wins I have achieved have something to do with the opening phase.

The Najdorf win was more or less on autopilot (I have a terrific score as white in this line, which the computers evaluate as equal).

The McCutcheon stuff was decided on the same line that has decided the winner of the Second LSS World Championship- in both games the Black players adopted the computer recommendation to keep Black's king to the kingside, which has proven to be of fatal consequences. The funny thing is that HUMAN high level games show the way for Black to defend properly: Keeping the king in the center with a timely ...Ke8.

The Rossolimo stuff is a line which is for sure quite problematic- the 3 pawns are no real match for the knight, but the position should be holdable- Black fell into a very deep trap which strongly resembles a chess study.

And finally the Italian as Black is a meek line where white allows an early ...d5 for free, with very comfortable play for Black. Still, I was rather surprised to find out that the computer lines were handling Black's 2 bishop advantage remarkably well, something that certainly was not the case 4-5 years ago.

ICCF has very similar rules to LSS, and from my experience the ratings there were inflated compared to LSS (this is something that does not apply anymore). I have not studied the games of the last championship at some detail yet, but it's easy to see that the moves of some previous World Champions deviate from the suggested computer lines very often (best exaple being the legendary 3-times World Champion Aleksandr Surenovich Dronov).

Sure, computers are improving, and the recent AI engine developments may well improve the quality of their play a lot, but I stand by my opinion that they still have a long way to go to reach "perfection" and render correspondence chess obsolete.

you should never play just for a draw. if i look at top gm matches they also do small mistakes and sometimes blunder. so in human chess there is no 'perfect play' and there are in 99.999% percent of the games opportunities for a win.

analytically it has to be a draw, the only difference is white goes first. otherwise, they both start with matched sides and both play perfectly. the only difference is white goes first..

Moasted  I agree in over-the-board play or Chess.com play you should never play for a draw. 

Thus in Chess.com play I put out challenges where I would play with Black against the best humans with computers that I could find and had a very good record--even beating a GM,

But ICCF Correspondence play at the highest levels is a whole different ball of wax. It is higher play than just over-the-board human GM play. In that venue if you have Black you do not have 99,999% opportunities for a win.  You normally have a close to zero percent chance for a win.

pfren   I would love to see "The Rossolimo stuff" where you [apparently] won with that opening?

and/or any of your recent wins in  that very strong tournament!?

You can get the whole tournament pgn from here.

In the Rossolimo, the opponent deviated at move 21 over a line that was scoring poorly, although there is nothing wrong with his novelty itself. The interesting stuff comes 60 moves later, after the very tricky 81.Kd2!? and the bad reply 81...Ba1?

That shows why you still have a lot to learn about this topic. The short history of chess computers shows that later ones always beat earlier ones. At some point, that could stop. But there is no reason to believe that time is now. Given that computers are in their infancy, compared to a couple hundred years from now, there is every reason to believe chess computers of the future will do MUCH better against the very best computers today. 

It should be pretty obvious stockfish of 2020 could beat deep thought of 1988. And it's probably likely the best computers of 1988 drew against each other most of the time. Just like it's likely the best computers of today mostly draw against each other. But the best computer of 2220 isn't going to draw all the time against a dusted off archaic 2020 stockfish. I would bet stockfish would lose every single game. 

I agree chess could be a draw. But it's far too early to know for sure. The answer could lie in that pesky 1% of white wins that aren't in the 99% of draws. 

PATRIOT Sorry I do know about this subject and you are just wrong.

Yes, there is reason to believe the trend you mentioned will stop and the best chess engines of the future Will Not Always Beat The Best Chess Engines of 2020.  

Also regarding the pesky 1% [or whatever the percentage is] Always include a mistake by Black. 

Eventually. Eventually it's proven a draw or proven white wins. But there is no reason to believe NOW (2020) is the time chess computers stop improving. 

"I very much disagree with your idea that super advanced chess engines of the future will be able to win most games from our now super chess engines." 

I dont think anyone can accurately predict 200 years from now. But what makes you think future computers wont beat todays best computers? That's been the history of computers since they were invented. I dont see any reason to believe today we are at the peak of computer development. I would even say the opposite. I think we are only at the very beginning. 

True. End of discussion (again).

ponz111  -  Until you can show that a draw can be forced in each and every variation you cannot claim that chess is a draw. It is possible that there are lines of play that would always lead to a win for one side or the other. Even if every won game ever played was due to a mistake by the other side that may be an indication but not a proof that chess is a draw. I did look up a recent computer chess tournament (Leela vs Stockfish) and in the final 200 game match it seems 174 games were drawn with Leela winning 25 games. I have to imagine a computer fifty years from now playing a current machine would have an even higher proportion of wins. I acknowledge that computers make mistakes and perhaps the percentage of draws will increase dramatically as their play improves.  Anyway, here is a link addressing the subject that you may like to read before further discussion. thttps://en.wikipedia.org/wiki/Solving_chess