Chess will never be solved, here's why

This is for optimissed not you lmao

Rubbish. Perfect means not being in a position in which there is a forced result worse you should have got. 106 draws backs up nothing. If you give 212 people who have just learned the moves a mate in 33 in a KBNvK position to play against each other, it's practically certain you'll get 106 draws. It doesn't somehow mean it wasn't actually a winning position in the first place and they've all played perfectly.

Or try letting any version of SF attempt to mate Syzygy from an average depth (58) white to win KNNvKP position (just 5 pieces) 106 times and you'll also get 106 draws.

Confidence in one's self does not lead to arrogance, that's faulty logic. Looking down on others leads to arrogance. Democracy and ability work just fine together.

You might be a Hans Niemann, but there are more Naroditskys out there...

Is there any example of this accuracy that can be objectively tested?

No advertising

Well said, but the fact remains SF hasn't been programmed to play even KRK accurately.

My "playing around" was just to demonstrate it can blunder points under competition rules in the simplest of 3 man positions. With at least 34 moves spare in KRK from a ply count 0 position it won't actually blunder from such a position without some previous playing around - but it is routinely inaccurate.

By the time you get to a 5 man KNNvKP position it has become so inaccurate that I can reliably beat it under competition rules from "frustrated wins" as well as many tablebase drawn positions. Here is an old example of both against SF12 from a frustrated mate in 90.

It seems inevitable that it can only get worse in closely matched positions as the number of men on the board increases. Problem is humans can only get much worse, so it can beat everybody.

You're very good at telling people how to solve problems without any concrete details. I believe SF development is happy to accept volunteers, so why don't you just go and program the engine accordingly. It might just tie you up long enough for our problem to be over.

O is constantly dishonest.
Like in his recent pretense that he's the only one discussing the forum topic.
Gross crass blatant falsehood by him.

He's actually the one getting it off topic

That's includes you btw

You also cause it to go off topic

@11834

"106 draws tends to back that up." ++ 108 now.

"the perfect games are, I think, around 50 moves long" ++ Average 39 moves.

"games that would last 150 to 200 moves" ++ Perfect games are drawn long before.
Average 39 moves, standard deviation 11, shortest 15, longest 73.

Any sequence of 150 moves must contain a series of forced moves.
Just assuming a branching factor w of 2 non-transposing choices per move and 1 side only would lead to
2^150 = 10^45 positions, i.e. more than the 10^44 legal positions and much more than the 10^38 legal positions without underpromotions to pieces not previously captured.
Games of 150 moves are a myth, theoretically as well as practically.
Even more if games are adjudicated by the 7-men endgame table base once 7 men are reached.

tygxc it could be 10,000 draws and it wouldnt matter.

why do you continue to prattle off about these games when they mean nothing to solving chess.

its a faulty engine until mathematically proven otherwise.

"Games of 150 moves are a myth, theoretically"

by definition of "theoretically" it isnt a myth theoretically. you have no idea how basic definitions work do you?

tygxc why dont you just respond instead of downvoting like a coward.

@11825

"you can solve checkers with a number of years of CPU time that is practical."
++ Schaeffer weakly solved Checkers in 2 years (2005 to 2007) with 50 CPU, i.e. 100 CPU years and considered 10^7 * 10^7 = 10^14 positions of the 5*10^20 legal positions.

The ongoing ICCF World Championship finals considered:
90*10^6 positions/s/server * 2 servers/finalist * 17 finalists * 3600 s/h * 24 d/h * 365.25 d/a * 2 a = 1.9*10^17 positions
of the 10^38 legal positions without underpromotions to pieces not previously captured.

That is commensurate with the required
Sqrt (10^38 / 10,000) = 10^17 positions to weakly solve Chess.

So weakly solving Chess now in 2024 is just as practical as weakly solving Checkers was in 2007.

"That is commensurate with the required
Sqrt (10^38 / 10,000) = 10^17 positions to weakly solve Chess."

LMFAO BRO SAID THAT ITS THE SAME NUMBER OF POSITIONS SO IT'S THE SAME AS GETTING THE SOLUTION!?!?!?

thats the number of positions of the weak solution, not the number of positions needed to FIND the weak solution. and this is still ASSUMING that your faulty heuristics are accurate.

this has been pointed out to you repeatedly, and you continue to ignore it.

tygxc we all know you are the one doing the downvoting. you are probably embarrassed at the ease of which your arguments are torn apart so you downvote to create a false sense of contention, to try to make it seem like others disagree with what im saying. there is no contention.

you are just wrong.

why havent you addressed the fact that I brought your arguments up to dozens of math majors/professors and they all found the same flaws that I pointed out to you?

tygxc, why havent you addressed the fact that I brought your arguments up to dozens of math majors/professors and they all found the same flaws that I pointed out to you?

i can see your downvotes.

Malaria is the one of the only feasible explanations for your fevered delusions...

Engines are flawed and are not "evaluating" for perfect play ala tablebases, so your 1.9*10^17 positions is meaningless. You could evaluate all 10^44 positions with the current Stockfish and the results would not be a solution for chess in any way. Throw in every ICCF master and every other engine for good measure and it changes nothing.

dude i told him about the math majors and professors i brought his claims up to (due to his disbelief that I am also a math major) for over a month. he's just been ignoring it this entire time. I think he's still trying to figure out a rationalization for it to fit into his fantasy.

the longer you stay on this thread, the more you realize how intellectually dishonest tygxc is.

he's been literally just ignoring the rebuts to his claims for YEARS.

@11872

"Engines are flawed and are not evaluating for perfect play"
++ That is what I say the whole time.
The provisional, heuristic engine evaluations like +0.33 play NO ROLE.
Only the end result matters: a draw by either the 7-men endgame table base,
or a prior 3-fold repetition, or a known drawn position.
We have 108 sequences of average 39 moves ending in draws.

Weakly solving Chess requires 1 way to draw after 1 e4 and 1 way after 1 d4.
We have not 1, but 4 ways to draw after 1 e4:

1. 1 e4 e6
2. 1 e4 c5
3. 1 e4 e5 2 Nf3 Nf6
4. 1 e4 e5 2 Nf3 Nc6

We have not 1, but 5 ways to draw after 1 d4:

1. 1 d4 Nf6 2 c4 e6
2. 1 d4 Nf6 3 c4 g6
3. 1 d4 d5 2 c4 c6
4. 1 d4 d5 2 c4 dxc4
5. 1 d4 d5 2 c4 e6

This redundancy makes it fail safe.
Even if a pair of errors were found in one line of defense, then still 3-4 backup lines are available.

Yes, obviously: some more than others.