Chess will never be solved, here's why

@14920

"we don’t know if the computers ACTUALLY played perfectly"
++ This is not computer vs. computer,
but ICCF (grand)master + engines vs. ICCF (grand)master + engines

bro really cant understand that that makes zero difference?

"we just know they are the SAME STRENGTH"
++ In previous years they were also the same strength, but there were decisive games.

tygxc, im not surprised that you cant comprehend the difference between if --> then and if and only if <-> then statements.

@14923

"Current engines have shown past ICCF WC draws to be fallible"
++ As expected: when there were 20 decisive games,
2 draws with a pair of errors are expected and you identified one of these.
When there were 10 decisive games, 1 draw with a pair of errors are expected.
Now there are 0 decisive games, no draw with a pair of errors is expected.
Even if there were a few (2-3) draws with a pair of errors, then the present ICCF WC Finals draws are still part of a weak solution of chess, as there are 5 lines of defense that draw.

"future engines will show current draws to be fallible" ++ No.

@14905

"What we could examine is errors made by engines in tablebase positions."
++ I proposed a relevant position:

Black to play and draw.
It is a 7-men endgame table base position.

KRPPvKRP already done years ago in the games I posted for you but you've been studiously ignoring.

I repeat the results (courtesy @cobra91) here.

SF15's average blunder rate per ply in KRPPvKRP in that sample was around 1% in the winning position and 2% in the drawn position. About 1/3 of the games were perfect.

But KRPPvKRP positions are relatively easy 7 man position for Stockfish. It doesn't find the other 7 man position easy. Zero perfect games and an average blunder rate per ply of nearly 5% (around 1 blunder per 10 moves).

There is no disputing that Stockfish can play easy positions perfectly . For all practical purposes I believe you can say it plays 2-man, 3-man and 4-man chess perfectly (but not accurately even in KRvK). It begins to fail in difficult 5 man positions such as the one reported. 

As it is a draw, it could result from the initial position with optimal play from both sides.

Unless the initial position is winning for one side or other - another possibility that you studiously ignore.

Humans in endgame books had this wrong and thought it a white win, so it is not trivial. As is is a rook ending, that is the most common type of endgame.

You don't have the slightest idea what would be the most common type of endgame with perfect play. There are probably more weak solutions to chess under either basic or competition rules than there are basic rules positions. Each of of which could generally finish up in any number of the thousands of 7 man endgames.

@14925

"You did not give any reason for the last point. Please demonstrate how it is incorrect"
++ Future engines cannot go below 0 error/game.
If the present 114 draws out of 114 games contained more than a few (2-3) draws with a pair of errors, then there would be at least 1 decisive game with an unpaired error.
Even if a few (2-3) draws with a pair of errors were found, then still the present 114 draws out of 114 games are part of a weak solution of chess, as 5 redundant lines of defense were found to draw.

@14928

"KRPPvKRP already done years ago" ++ We are talking modern engines but OK.
Drawn KRPP vs. KRP is relevant.
I see the error rate go down from 7.7% @ 1 s/move to 0% @ 64-1024 s/move.

why do you conveniently leave out the error at 2048s/move, does that hurt your feelings or something?

"Unless the initial position is winning for one side or other" ++ That is invalid.

ah yes, false claims without justification, the tygxc logic.

"what would be the most common type of endgame with perfect play"
++ Rook endings, as seen in ICCF WC draws which are perfect play, and as there are ways to draw some rook endgames even a pawn or two down, and as rooks get traded less than other pieces because they always control 14 squares regardless of where they stand.

this entire argument is you trying to claim that the ICCF is perfect play, you cant use what happens in it to argue what perfect play looks like LMFAO

and YOUR rigor is infallible ?...plz. awhile back u said the whiteball-blackballs werent addition. wutta (nvm). both can be use2solve it [Removed: Offensive] ~W

@14928

"KRPPvKRP already done years ago" ++ We are talking modern engines but OK.

SF15 isn't exactly ancient history.

Drawn KRPP vs. KRP is relevant.
I see the error rate go down from 7.7% @ 1 s/move to 0% @ 64-1024 s/move.

You have your own way of presenting figures. It's 0% at 64 and 1024 s/move (only), not between. And it's back up to 2.5% at 2048 seconds a move.

It is a fact that the linear regression line in both KRPPvKRP examples shows a general decrease in blunder rate with increasing think time. That may be because it's a relatively easy endgame. 

"Unless the initial position is winning for one side or other" ++ That is invalid.

No you're invalid.

"[You don't have the slightest idea ]what would be the most common type of endgame with perfect play" (See you're still editing out anything you don't like.)

++ Rook endings, as seen in ICCF WC draws which are perfect play,

Wotta lotta b*llocks. You can't rely on SF to do perfect play when there's more than 4 men on the board and all ICCF moves probably come from SF. And the question is not about ICCF tournaments or what happens at your local chess club. I repeat, you don't have the slightest idea what would be the most common type of endgame with perfect play. The question can't even make sense until you specify exactly which version of perfect play and hardly would even then.

and as there are ways to draw some rook endgames even a pawn or two down, and as rooks get traded less than other pieces because they always control 14 squares regardless of where they stand.

Very interesting, but SF doesn't do very well with the endgame even though it's far from the most difficult of tablebased endings. Minimal relevance to the topic.

(And actually whether or not rooks control 14 squares is rather more dependent on where the other pieces stand.)

the fallacies he makes and his complete lack of mathematical rigor

and YOUR rigor is infallible ?...burst ! awhile back u said the whiteball-blackballs werent addition. wutta (nvm). both can be use2solve it [Removed: Offensive] ~W

it's not learning, it's updating its playstyle to yours, better said adapting its moves, which means it is learning. a move that could have been a blunder can now be a brilliant move, so it is learning.