"as schaeffer has done with checkers"
its kinda funny how schaeffer himself disagrees with you. tygxc.
Chess will never be solved, here's why
Sort:
Feb 4, 2023
0
#6243
"++ I do not assume, think, believe, or guess that, I know that. That is knowledge.
A very certain theorem. I even presented proof of it.
If you do not understand that, then that is your problem."
what proof lmao?
you just said a tempo isnt enough, and acted like you did something of value.
@8409
"what proof"
++ Take any strong tournament, e.g. Tata Steel Masters 2023.
There were 91 games: 60 draws and 31 decisive games.
Assume a Poisson distribution of the errors per game.
First assume Chess to be a white win or a black win.
Then the 60 draws must contain an odd number of errors (?).
Thus Poisson (1, lambda, 0) + Poisson (3, lambda, 0) + Poisson (5, lambda, 0) = 60 / 91
There is no lambda that satisfies that.
So chess cannot be a white or a black win.
So Chess must be a draw.
Now assume Chess to be a draw.
Then the 31 decisive games must contain an odd number of errors (?).
Thus Poisson (1, lambda, 0) + Poisson (3, lambda, 0) + Poisson (5, lambda, 0) = 31 / 91
Lambda = 0.57 errors per game average satisfies this.
Thus:
Drawn games with 0 errors: optimal play from both sides: 51
Decisive games with 1 error: 29
Drawn games with 2 errors: 9
Decisive games with 3 errors: 2
Thus Chess is a draw.
Applying the same reasoning to the ICCF World Championship Finals is even clearer.
It shows that drawn games there are > 99% certain to be perfect games with optimal play from both sides and there remains < 1% probability that such games contain 2 errors that undo each other. That means we have over 1000 perfect games: draws with optimal play from both sides.
Feb 5, 2023
0
#6246
Elroch wrote:
Of more interest is this more in depth look at ChatGPT and some of its idiosyncracies such as that it does not believe it can speak Danish (when in truth it is fluent).
Which should probably score highly on the Turing test.
When I worked in Holland my boss used to apologise for his poor English, which I would have judged to be better than the majority of colleagues I'd worked with in England. Then he'd break off to answer a telephone at length in Norwegian, where he probably said something similar.
Feb 5, 2023
0
#6247
MEGACHE3SE wrote:
tygxc wrote:
@8416
"Sveshnikov making those solvability claims"
++ Sveshnikov said in an interview on June 25, 2007 with Eldar Mukhametov:
'Give me five years, good assistants and the latest computers
- I will bring all openings to technical endgames'
OH MY GOD I WAS RIGHT YOU ARE JUST TAKING STUFF OUT IF CONTEXT
If that quote is accurate it entirely fails to support @tygxc's claims. It is nothing to do with formal solving of chess, but rather is about pragmatic, uncertain chess analysis. So it is not even that Sveshnikov was deluded.
Feb 5, 2023
0
#6248
mpaetz wrote:
Optimissed wrote:
Just grow up. 100,000 was excessively modest of me. It's a base-line figure, based on my proven IQ. But there are more human qualities than ability to complete numerical, verbal and visuo-spatial tests extremely quickly. I'm watching a film about how the Americans won WWII and also Agincourt so shut up.
What is the title of this film? It would be fascinating to learn how the Americans won the battle of Agincourt.
This would indeed be fascinating. The battle of Agincourt occurred 77 years before Columbus (an Italian) reached North America, and Native Americans rarely took their vacations in Europe.
Feb 5, 2023
0
#6249
Optimissed wrote:
Optimissed wrote:
Elroch wrote:
Is @Optimissed trying to develop a scale for narcissism?
As in water temperature, you could be the equivalent of boiling point if you like .... a standard which is impossible to surpass. It could be called "One Elroch" and zero narcissism is zero Elrochs. The scale could be divided into 99, which is a multiple of three primes, so you'd feel at home.
You'd feel great satisfaction in explaining that it is really a multiple of two primes. So you could call people foolish.
I should actually point out that you'd be wrong to insist that it was a multiple of two primes. That's a subjective judgement, at least to me. I would suggest that 99 is a product of three primes, most definitely: and not of two. But a multiple of either two or three, depending on which side you got out of bed this morning.
You don't say which three primes you're thinking of.
If we adopt your practice of using words to mean whatever we want, we could use "prime" to mean any digit and a "product" of primes to mean the primes written consecutively, in which case 99 would be the product of at most two primes (unless you've got a different meaning for "two" as well).
It may be you're really quite humble.
In @Optimissed speak, "I'm the most intelligent person who ever lived in the whole of history", may actually mean, "I'm trying to learn how to wipe my bum - can anybody help?".
@8414
"nothing to do with formal solving of chess" ++ It is formal solving of Chess. That is what Schaeffer did for Checkers: trace 19 of the 300 tournament openings to his endgame tablebase.
Once the initial position is traced to the 7-men endgame table base, a strategy is established to achieve the game-theoretic value against any opposition.
"pragmatic, uncertain chess analysis" ++ No. Analysis ends with some uncertain evaluation.
A solution ends with the exact evaluation draw / win / loss of the 7-men endgame tablebase.
Feb 5, 2023
0
#6251
They would most likely be disappointed. @tygxc probably doesn't regard awful positions as sensible, so his solution would be, "does not compute".
@8424
Only drawn positions are relevant. The strategy to achieve the draw against any opposition is then to follow the path from the initial position to other drawn positions until a 7-men endgame table base draw.
Here is a practical example: all theory moves memorised by both players
https://www.chessgames.com/perl/chessgame?gid=2457690
@8426
It is a perfect game with no errors and optimal play from both sides, but they did not play a single move of themselves, they just followed their memorised solution.
Chess
Feb 5, 2023
0
#6255
Bum wiping is important. The three least common denominator primes 3 3 11. Hope this helps.
Optimissed wrote:
It seems as though there's no way to follow the lines to show where they end up, except by holding them all in a read-only memory. But I'm not sure how such things would work in practice and it seems to me that a read-write memory would be necessary, to put markers in. Not that I thought about it for more than two seconds but that's how it seems. It's feasible but it could be slow to access. I don't know how slow but given that it can't be known that it will be perfectly accurate (the methodology you propose) I'd be extremely surprised if anyone wanted to do it except as a means to cheat at chess. That is, if you get an awful position, you feed it in to see if there's a theoretical draw hidden in it.
Let us assure everyone that read-only memory vs. read-write memory is not really a consideration. If you are going to use such terms, though, you should probably say ROM and RAM (read-only memory vs. random access memory).
Feb 5, 2023
0
#6257
BoardMonkey wrote:
Bum wiping is important. The three least common denominator primes 3 3 11. Hope this helps.
No doubt that's why the thread is so long. There are three trolls. @Optimissed @Optimissed @tygxc.
Feb 5, 2023
0
#6258
tygxc wrote:
@8409
"what proof"
++ Take any strong tournament, e.g. Tata Steel Masters 2023.
and already your "proof" is invalid. strong is a subjective term. its imperfect play. since your argument relies on a subjective evaluation it cannot be true.
There were 91 games: 60 draws and 31 decisive games.
you assume that the positions of the games were actually what the players agreed on.
Assume a Poisson distribution of the errors per game.
you cant make an assumption like that and consider an argument to be an objective proof.
First assume Chess to be a white win or a black win.
Then the 60 draws must contain an odd number of errors (?).
this is objectively false, the process of going to say, a white win from a black win, can be 1 or 2 errors.
Thus Poisson (1, lambda, 0) + Poisson (3, lambda, 0) + Poisson (5, lambda, 0) = 60 / 91
even if we assume the false poisson distribution, you falsely exclude games with even errors and games with more than 5 errors.
your basic assumptions are mathematically equivalent to saying that half of chess games played must be draws/wins.
and that isnt correct.
There is no lambda that satisfies that.
by definition there isnt. because of all of your false assumptions.
So chess cannot be a white or a black win.
So Chess must be a draw.
Now assume Chess to be a draw.
Then the 31 decisive games must contain an odd number of errors (?).
again, objectively incorrect.
Thus Poisson (1, lambda, 0) + Poisson (3, lambda, 0) + Poisson (5, lambda, 0) = 31 / 91
this is based off of multiple incorrect assumptions.
in addition, this assumes that drawn games could not have an odd number of errors, which, as before, is objectively false.
you also assume that there arent any with 7 or more errors.
Lambda = 0.57 errors per game average satisfies this.
Thus:
Drawn games with 0 errors: optimal play from both sides: 51
Decisive games with 1 error: 29
Drawn games with 2 errors: 9
Then how come the VERY FIRST tata steel draw that i looked at had 4+ errors? (aronian vs gukesh)
Decisive games with 3 errors: 2
Thus Chess is a draw.
tf u mean "thus", chess being a draw is your initial assumption.
Applying the same flawed reasoning to the ICCF World Championship Finals is even clearer.
It shows that drawn games there are > 99% certain to be perfect games with optimal play from both sides and there remains < 1% probability that such games contain 2 errors that undo each other. That means we have over 1000 perfect games: draws with optimal play from both sides.
even if we assume allllll of your objectively incorrect stuff before, that still doesnt prove we would have even a SINGLE perfect game. an expected value of something is not a proof that we would have the expected value.
none of that is proof lmao.
you prove my point of how you construct arguments.
you take obscure stuff, do obscure operations, and act like any of it has standing.
it is physically impossible to prove that chess must be a draw based on a statistical analysis of imperfect games. period.
You cant just assume a poisson distribution, and you cant just assume an odd number of errors for certain games, heck you literally have question marks within your own "proof", because you know that those are unfounded assumptions.
you provide no evidence that the iccf games are ">99% perfect"
Don't worry about engine play chess player on player as we are not calculating every possible position and are not perfect. Then again Google Go bot lost a game against human champion 4-1 . When we are transhumanism people then we mix it up like Bobby Fischer with something new (for the week) lulz . Bot v bot and human v human I am missing something if man/woman needs to be concerned. (?)
"You have refuted nothing at all. You just ridicule and then proceed with your erroneous assumption that weakly solving or strongly solving require the same number of positions."
nope
unless you take things out of context again.