I too was wondering how many angels will fit on the head of a pin.
What will happen if chess got solved?
Sort:
Imagine the "atomic computer": after 1. e4 it gives you the results and final position after every possible black answer considering the best play of both players after each black answer in move 1.
A 2024 computer can't go that far. Its horizon is shorten, even so it can chose the best or close to the best move in every position and final result is: 100% draw in the match Atomic Computer vs. 2024 computer.
This scene is what I fell being almost certain. If this is true: top computers and top software of today will keep being unbeatable in the future.
Same way a checkers computer/software of 1994 is unbeatable by a checkers computer/software of 2024.
This don't even look hard to understand.
The strongest Stockfish until now, by testing, is dated 13 June 2023, more than one year old. The most recent version is 16,1 and that is weaker. All those version of Stockfish are drawing everything among them.
https://computerchess.org.uk/ccrl/4040/rating_list_all.html
There is progress if a new version get a positive score against the previous version. If everything is a draw, it may be progress but it has no meaning.
DesperateKingWalk wrote:
Optimissed wrote:
We can be nearly sure chess is a draw: so that solving it won't change a thing.
In the unlikely event that chess is a forced win, that would have an immense effect on the game. That's because if it were a win (which we can be pretty sure it isn't) then since millions of games have been played, that win isn't going to be found by following normal channels. It would instead be found by following extremely abnormal channels and that's the reason why the existence of a forced win would change chess completely. But it won't happen.
These are the ruminations of an extremely high IQ B$ artist.
Here are the facts Mr. 200 IQ boy.
We do not know if chess is a win or a draw. So claiming draw rate as proof of perfect play is just insane. As I have easily demonstrated. Draw rate has nothing to do with showing perfect play.
And say we knew for a fact that chess was a forced draw. Draw rate would still not prove perfect play.
As you can have a 100 percent draw rate with perfect play. And a 100 percent draw rate without perfect play.
As I have clearly demonstrated.
And I know for a fact my computer using Stockfish does not play perfect chess even at 1 second a move....
We do know that chess is a draw. White to play and win on move 1 is just not a feasible expectation, unless of course you want to ignore everything we observe about chess. It is demonstrably certain that the higher the level of chess competition, the more the games are drawn. And the more perfect the moves for both sides, the higher the likelihood that the game is a draw. These are facts.
White doesn't begin with a winning advantage. Black simply has too many resources. The correct evaluation is draw. Anyone with even an elementary understanding of chess knows that this is the case.
International-Grandpatzer wrote:
DesperateKingWalk wrote:
Optimissed wrote:
We can be nearly sure chess is a draw: so that solving it won't change a thing.
In the unlikely event that chess is a forced win, that would have an immense effect on the game. That's because if it were a win (which we can be pretty sure it isn't) then since millions of games have been played, that win isn't going to be found by following normal channels. It would instead be found by following extremely abnormal channels and that's the reason why the existence of a forced win would change chess completely. But it won't happen.
These are the ruminations of an extremely high IQ B$ artist.
Here are the facts Mr. 200 IQ boy.
We do not know if chess is a win or a draw. So claiming draw rate as proof of perfect play is just insane. As I have easily demonstrated. Draw rate has nothing to do with showing perfect play.
And say we knew for a fact that chess was a forced draw. Draw rate would still not prove perfect play.
As you can have a 100 percent draw rate with perfect play. And a 100 percent draw rate without perfect play.
As I have clearly demonstrated.
And I know for a fact my computer using Stockfish does not play perfect chess even at 1 second a move....
We do know that chess is a draw. White to play and win on move 1 is just not a feasible expectation, unless of course you want to ignore everything we observe about chess. It is demonstrably certain that the higher the level of chess competition, the more the games are drawn. And the more perfect the moves for both sides, the higher the likelihood that the game is a draw. These are facts.
White doesn't begin with a winning advantage. Black simply has too many resources. The correct evaluation is draw. Anyone with even an elementary understanding of chess knows that this is the case.
The problem is that knowing chess is a draw doesn't mean it's solved yet tgycx is claiming hat it's "solved " because computers are drawing despite the fact that computers tend to draw pretty quick despite having winning /losing advantages
Jul 11, 2024
0
#189
llama_l wrote:
tygxc wrote:
"what would happen to competitive chess?"
++ It would gradually die out, like competitive checkers.
No... for several reasons no... (I don't feel like writing multiple paragraphs explaining something that should be obvious).
But ok, it's possible... chess could also die out without that, so sure, it's possible.
Chess would not die out.
That would be like claiming that Olympic track events and all lower such events and in schools would die out because cars can go faster.
Doesn't work like that.
@201
"Do you think the very best chess playing ability from 1980 would draw the very best chess playing ability from 2020?" ++ No. The WC11/Final, World Championship 11 Final of 1980 had 62 decisive games. Now we see for the first time 112 draws in 112 games.
You cannot get below 0 errors per game.
You can only achieve 0 errors per game in less time than 5 days per move.
tygxc wrote:
"Do you think the very best chess playing ability from 1980 would draw the very best chess playing ability from 2020?" ++ No. The WC11/Final, World Championship 11 Final of 1980 had 62 decisive games. Now we see for the first time 112 draws in 112 games.
...and somehow you don't think this has anything to do with the #1 engine, Stockfish, being open source and transparent to all its competitors?
@215
"computers tend to draw pretty quick despite having winning /losing advantages"
++ No. If there is a win, then they find it. If there is no win, then they cannot find it and draw.
It is a property of the game of chess, not a property of chess engines.
@204
"Maybe in the future computers can play even a bit slightly more perfect" ++ They cannot get below 0 error/game, they can only achieve 0 error/game in less time than 5 days/move.
"the force of today's computers and software is already enough to draw the game" ++ Yes, only since this year and at 5 days/move.
"if the "quantic computer" or "atomic computer" arrives in 2060 it will still be a 100% draw score againt the 2024 era computers" ++ Yes, at 5 days/move. The 2060 computer will beat the 2024 computer at say 5 hours/move, 5 minutes/move, 5 seconds/move.
@209
"progress still continues and has not stopped...." ++ Yes, progress continues and has not stopped. However at 5 days/move 0 error/game has been reached this year.
You cannot get below 0 error/game.
Future engines can reach 0 error/game in 5 hours/move, 5 minutes/move, 5 seconds/move.
@214
"We do know that chess is a draw." ++ Yes
"White to play and win on move 1 is just not a feasible expectation, unless of course you want to ignore everything we observe about chess." ++ Yes
"It is demonstrably certain that the higher the level of chess competition, the more the games are drawn." ++ Now in ICCF WC Finals 112 draws out of 112 games.
"And the more perfect the moves for both sides, the higher the likelihood that the game is a draw." ++ Yes
"These are facts." ++ Yes
"White doesn't begin with a winning advantage."
++ Yes, an advantage of +1 tempo = + 1/3 pawn is not enough to win.
"Black simply has too many resources." ++ Yes.
"The correct evaluation is draw." ++ Yes.
"Anyone with even an elementary understanding of chess knows that this is the case." ++ Yes.
@219
"why DOES competitive chess always end in a draw?"
++ Because the game is a draw, just like Checkers, Nine Men's Morris...
If neither white nor black make any error, then a draw is inevitable.
The drawing margin is relatively broad.
Black can lose a tempo and still draw, white can lose 2 tempi and still draw.
@220
"you don't think this has anything to do with the #1 engine, Stockfish,
being open source and transparent to all its competitors?"
++ No.
They do not all use Stockfish, also LC0, Komodo...
They tune the engines differently, e.g. the parameter 'contempt'
They use different time per move: average is 5 days/move,
but one may use 2 days on a move and the other 10 days to reply.
They run different hardware. The Russians use inferior hardware because of sanctions,
but 4 of the 17 ICCF WC Finalists are Russian despite their inferior hardware.
In previous years Stockfish was available to all players too,
but there were decisive games, each year fewer.
In present Preliminaries, Semifinals, Candidates the same Stockfish is available to all players,
but there are decisive games, that is how the 17 ICCF WC Finalists qualified.
tygxc wrote:
"you don't think this has anything to do with the #1 engine, Stockfish,
being open source and transparent to all its competitors?"
++ No.
They do not all use Stockfish, also LC0, Komodo...
They tune the engines differently, e.g. the parameter 'contempt'
They use different time per move: average is 5 days/move,
but one may use 2 days on a move and the other 10 days to reply.
They run different hardware. The Russians use inferior hardware because of sanctions,
but 4 of the 17 ICCF WC Finalists are Russian despite their inferior hardware.
In previous years Stockfish was available to all players too,
but there were decisive games, each year fewer.
In present Preliminaries, Semifinals, Candidates the same Stockfish is available to all players,
but there are decisive games, that is how the 17 ICCF WC Finalists qualified.
It's easy to tell you are not a developer.
The point is that all the top engines are using Stockfish's techniques and in some cases, the actual code copied or stolen. Stockfish NNUE borrowed from Leela. So, you're dealing with an incestuous environment where all the engines are fairly uniform copies of each other, with variations evolved only to beat other engines like siblings fighting, but over time this equals out. The amount of draws that has evolved reflects that lack of diversity, and that's why engines are stuck right now in general. Humans have nothing left to offer, which makes the problem worse.
This is evidence of stagnation, not perfect play.
@181
"Those games are incredibly far from being perfect." ++ Based on what?
"Do you seriously believe that those guys would have even the slightest chance against the 2100 ICCF world champion assisted by an engine from the year 2100 running on a computer from the year 2100?" ++ Yes, they would draw just the same.
So you are saying the best chess playing ability today would draw the best chess playing ability from the year 2100. Which means in the next 75 years there will be no improvements to chess computers. We've reached the end of the road when it comes to chess.
To me that seems unlikely. But who knows, maybe you are right. Maybe it's not possible for technology to advance any farther.
Maybe in the future computers can play even a bit slightly more perfect. Questions is: what if the force of today's computers and software is already enough to draw the game? If so, even if the "quantic computer" or "atomic computer" arrives in 2060 it will still be a 100% draw score againt the 2024 era computers.