MEGA - do you think its possible that tygxc is trying to attack mathematics the way the flat-earthers try to attack geography?
Flat-earthers apparently resent and distrust and hate scientists.
Could tygxc have a similiar motivation?
More likely he just wants attention and this is his way of getting it.
Chess will never be solved, here's why
Sort:
Jul 18, 2024
0
#10922
Jul 18, 2024
0
#10923
playerafar wrote:
MEGA - do you think its possible that tygxc is trying to attack mathematics the way the flat-earthers try to attack geography?
Flat-earthers apparently resent and distrust and hate scientists.
Could tygxc have a similiar motivation?
More likely he just wants attention and this is his way of getting it.
no, i think that tygxc simply has no real conception of the existence of mathematical rigor.
this is drawn from tygxc at times claiming his lunatic "definitions" as actual mathematical terms.
hmm
Jul 18, 2024
0
#10926
@tygxc has agreed that if there were more ICCF games, some of them would be decisive. So he claims that the draws are perfect because they are draws, not perfect because the players are reliable. It is an adequate explanation that the players are perfect at failing to win, rather than that they are perfect at preventing the opponent at winning.
IMHO, a future generation of engines played against the current ICCF players would humiliate them and show their lack of perfection. This would show that a key part of the current blanket drawing is how unchallenging the current players are (by future standards).
This has always been the way. Last I heard, more processing power was still providing rating increases (just not enough to make a difference at several days a move, which is where the increase is smallest).
@14100
"if there were more ICCF games, some of them would be decisive" ++ Because of random factors: clerical error, illness, computer crash, power blackout, fire, flood, war, revolution...
"the draws are perfect" ++ because all 113 games are draws.
"the players are perfect at failing to win"
++ They cannot win unless their opponent makes an error.
"they are perfect at preventing the opponent at winning"
++ Yes, they now reached 0 error/game.
"IMHO, a future generation of engines played against the current ICCF players would humiliate them and show their lack of perfection" ++ That is your humble opinion.
The fact that there were decisive games in previous ICCF WC Finals, every year fewer, and now none indicates at 5 days/move perfection has been reached.
"more processing power was still providing rating increases"
++ But at 5 days/move you cannot get below 0 error/game.
@14097
"one side may conceivably sacrifice three minor pieces for a pawn and the opportunity to promote a pawn" ++ OK, no problem, that is no underpromotion.
"if they can only promote to a previously captured piece" ++ They can promote to a queen, but not underpromote to a rook, bishop, or knight not previously captured.
"I didn't/can't even see a reason why such positions should be vetoed"
++ Promoting to a queen is OK, only underpromoting to a rook, bishop, or knight not previously captured is removed from the Tromp count of 4.82 * 10^44 to arrive at 3.3 * 10^38.
A typical position as counted by Tromp has 9 promotions to pieces not previously captured.
A position as counted by Gourion (3 * 10^37) has 0 promotions to pieces not previously captured. That is too restrictive, that is why I multiply by 10.9456 to include positions with 3 or 4 queens to reach 3.3 * 10^38.
"swap of a queen for three minor pieces can suit either side depending on position"
++ No problem, though Q < B + N + N + P, also sacrificing rook + bishop to promote to a queen is fine: Q = R + B + P
@14104
"tygxc is implying, by the way he argues, that this isn't the place for mathematical rigour"
++ On the contrary I am a fan of mathematical rirour, but not of purism.
Combinatorial game theory is a branch of mathematics, but game knowledge can and should play a major role in solving chess. Allis weakly solved Connect Four using a set of 9 strategic rules.
That is why 1 e4 e5 2 Ba6? can and should be dismissed without any game tree.
'Chess is the art which expresses the science of logic' - Botvinnik, World Champion and PhD. in engineering.
That is why 1 a4 cannot be superior to 1 e4.
That is why 1 Nf3 d5 2 Ng1? cannot be superior and can be dismissed.
That is also rigour, but a different kind of rigour, more game rigour than combinatorial rigour.
Elroch wrote:
@tygxc has agreed that if there were more ICCF games, some of them would be decisive. So he claims that the draws are perfect because they are draws, not perfect because the players are reliable. It is an adequate explanation that the players are perfect at failing to win, rather than that they are perfect at preventing the opponent at winning.
IMHO, a future generation of engines played against the current ICCF players would humiliate them and show their lack of perfection. This would show that a key part of the current blanket drawing is how unchallenging the current players are (by future standards).
This has always been the way. Last I heard, more processing power was still providing rating increases (just not enough to make a difference at several days a move, which is where the increase is smallest).
I'm not sure the first part is exactly aligned with Tygxc, but it might be likely that these ICCF games are mostly played without mistakes simply because the engines are very equal in strength and can't consistently find find the most challenging (alltho still drawn) lines against each other. I would agree that that theory is more likely than there being many errors per game (switching between losing/drawing evaluations).
MARattigan wrote:
tygxc wrote:
...
#14
Chess even stays a draw if stalemate = win.
The paper shows that the draw rate increases with more time.
Compare figure 2 (a) and (b).
The time spent by AlphaZero also means nothing in terms of theoretically accurate play.
The paper doesn't show that chess stays a draw even if stalemate = win. It shows that Alpha Zero can't see a forced win in either case.
But AlphaZero or any current engine has a very limited look ahead capability.
I can't run AlphaZero but here is LC0 based on the same kind of software playing a drawn position with only five men on the board. It blunders into a loss on it's second move. It's not even a very accurate blunder. The longest forced mate with the black pawn on h3 is 71 moves but it blunders into a mate in 56 (according to Nalimov) and collapses in 21 instead.
If Haworth's law (http://centaur.reading.ac.uk/36276/3/HaworthLaw.pdf) continues to hold up to 32 men there would be winning positions where the forced mates need at least tens of trillions of moves against accurate defence. The performance of anything that plays chess in much simpler positions such as the one above hardly inspires confidence in their assessments.
but max number of moves is 5488
Kotshmot wrote:
... it might be likely that these ICCF games are mostly played without mistakes simply because the engines are very equal in strength ...
Agreed. But even more than equal in strength - they're undoubtedly the same pool of engines.
It's almost certain that none of the ICCF WC finalists have access to some mysterious, superior engine that the others don't.
So the engines can't find wins because they're playing against themselves. This doesn't suggest infallibility - it suggests that engines aren't capable of outplaying their own evals.
For example, if you pit Chessmaster's "The King" engine against itself - it draws repeatedly. It can't find a win against itself, despite the fact that it's only about 2650 in strength ...
Jul 18, 2024
0
#10933
Excellent point. It's worth emphasizing that you are referring to an engine that might lose almost 100% of the time versus the latest Stockfish, despite being of GM standard.
Disputations to the contrary are not really valid - every single participant will be using the latest top engine, with other engines being used less uniformly.
MEGACHE3SE wrote:
playerafar wrote:
Optimissed will always pretend that his trolling is about whoever else.
Whether its a particular individual - or groups of people.
On occasion some persons actually fall for that crass illogic of Optimissed's.
But most don't.
Usually - whatever Optimissed is claiming or asserting - its the reverse that is the reality. Constantly. Year in year out.
you know, optimissed recently has been making actual falsifiable arguments against elroch. idk what made him change but his some of his statemetns actually are startng to have some substance.
I did trash talk him recently when it started but idk if I played a part in it someone else came along and teased him pretty hard (maybe the guy that said fake 250 iq boy?)
tygxc wrote:
@14104
"daily chess is a joke to you"
++ chess.com daily without engines, not competitive, played by weak players is a joke
"but 5 days/move is proven perfect chess"
++ ICCF World Championship Finals very competitive to become the World Champion played by ICCF (grand)masters who qualified through Preliminaries, Semifinals, Candidates and with twin engines each 90 million positions/s produces 113 draws out of 113 games and is thus 99.992% sure to be perfect chess with optimal play by both sides.
ICCF, played by weak human players, is a joke. This only follows since the ratings of the top ICCF players is right in line with top daily chess players. More of a joke, since the humans playing ICCF are not really playing chess, just secretary for their engines.
Optimissed wrote:
Your reaction is visible and obviously identifiable as real. It's just obviously real and not a pretence. What is also noteable is that you don't realise it comes across as real. You perhaps don't recognise how close your character is to that of playerafar, either. Your basic personalities are identical and only the rhetoric and strategies are sometimes different. If you did recognise it, you'd take pains to alter your behaviour so it would seem less obvious. Not even you should wish to appear like him.
You lump me in with Playerafar based on your perception of how you are treated, not based on any real resemblance in general. This is because you live in your Optimissed-centered universe and cannot perceive or sympathize with other perspectives.
My reactions are uniformly calm and measured. What you read into them is your problem.
Jul 18, 2024
0
#10937
Optimissed wrote:
tygxc is implying, by the way he argues, that this isn't the place for mathematical rigour and I think, if I'm right, that he's right, since the whole thing is a bunch of guesswork which some people try to disguise as mathematical rigour but it's a pretence imo.
I'm not saying the implication is deliberate but it makes sense to accept it, since that is obviously what it DOES imply.
if tygxc was trying to argue that this isnt the place for mathematical rigor... he would be making that claim. Even the bare minimum, acknowledging that there is a difference, has never even once been done by him.
in addition, tygxc tries to claim mathematical definitions for the statements he makes, and cites journal articles that deal in mathematical rigor to try to claim that his methods are valid (of course, the contents of the articles do not match what he claims).
this makes me believe that tygxc cannot even recognize the difference to begin with.
MaetsNori wrote:
Kotshmot wrote:
... it might be likely that these ICCF games are mostly played without mistakes simply because the engines are very equal in strength ...
Agreed. But even more than equal in strength - they're undoubtedly the same pool of engines.
It's almost certain that none of the ICCF WC finalists have access to some mysterious, superior engine that the others don't.
So the engines can't find wins because they're playing against themselves. This doesn't suggest infallibility - it suggests that engines aren't capable of outplaying their own evals.
For example, if you pit Chessmaster's "The King" engine against itself - it draws repeatedly. It can't find a win against itself, despite the fact that it's only about 2650 in strength ...
Two of the same engines evaluate a shared position the same way and see the same continuations. Id imagine a 2650 engine would stumble into a decisive game against itself more often than stronger engines.
Optimissed wrote:
DiogenesDue wrote:
Optimissed wrote:
Dioinnnnng.
And now we've reached kindergarten level retorts...regression, even for you.
This I think signifies the two faces of Dio. This is a kind of veiled psychopathy. Trying to convince others that I'm subject to his own failings and he himself isn't subject to them at all is regressive, childish and rather normal: the sort of thing attempted by the naughty but mentally unengaging boy in the class. The one who isn't benefitting at all from the attempt at educating him.
He objected to a one word post where he had called me Sherlock, being sarcastic, so I called him Watson, since Watson was portrayed by Conan-Doyle as solid but dim. So then I wrote Dioinng or whatever. No reason but just poking a bit of fun. The guy is apparently dim but not dangerous: completely mad but mad within normal, ineffectual bounds.
It's the other one who SEEMS the more outlandish of the two but then occasionally, there'll be a lapse, where there'll be a language switch between the two of them or it will simply become identical for a few minutes. By now I'm convinced of it. Don't feel as if I'm guessing any more. It isn't a proof but there's a 99.992% chance (approximately).
The fact that you insist posters that are obviously different people are alts of each other just proves how suspect your observations and logic are in general, you know.
This was the most long winded "no, you" reply I have ever seen, by the way.
@14135
"I'm not a fan of this number"
++ 113 games out of 113 in the ICCF WC Finals ended in draws.
Suppose game 114 ends in a clear win: no clerical error or ill player or crashed computer...
Then the error rate is 1/114.
Thus the probability of one of the 113 games having 2 errors is (1/114)² = 0.0077%
Thus the 113 draws are 99.992% sure to be perfect games with optimal play from both sides.
The data at hand is not enough to calculate probability of an error, hence you have to make many unjustified assumptions in your calculation to get this result.
The 113 draws could potentially have one or more errors in them. Same goes with the 114th decisive game.
You make a confident assumption this is not the case based on probability distribution, which assumes all potential errors have the same probability and that they're independent events. This is not the case.
This is why your confidence in these games being optimal play as well as chess being a draw is based on a misunderstanding.