That is all no evidence, i dont even mention the pawnless queen endgames were first mover wins.
Seriously Ponz, the knowledge of GMs only applies to the known. If only 100 lines Lead to forced win the knowledge is absolutly not relevant for the topic.
The likeniness of Chess being a draw or a win isnt even in anyway related to the so called evidence
Yekatrinas - I observed this thread for 1000 posts and didn't see even so much as a brief glimpse of an intelligent thought from you. But periodically you step in and say that the conversation is beyond you thus you think it is stupid. If you have something to add, please do.
I think sapientdust made a fine post above summarizing some of the points now lost. A few points:
a) I'm surprised he is not banned after that outrageous remark " no mathematically literate person could ever make such an idiotic argument as the bolded sentence" about ponz. Are you saying ponz isn't mathematically literate? Really?
The bolded sentence was one that smyslovfan said was Jetson's position (nothing to do with ponz). Since the sentence was about Jetson, who is in fact a mathematician and thus certainly mathematically literate, I was effectively saying that Jetson didn't put forward such a silly position, because no mathematician would ever make such a ludicrous argument. It's the sort of fallacy that even a first-year philosophy student would notice.
"Perhaps I mischaracterized Jetson's position.
The evidence given by Jetson in mathematical problems that turned out contrary to empirical evidence are not relevant to chess. For example, the example about some theorem concenring prime numbers being false despite expectations happened because the space of integers is vastly bigger than the space of chess games. That theorem wasn't true at some number like 10*10^10^100 or something that utterly dwarfs the number of possible chess games. I would contend that we have explored the finite space of chess games much more thoroughly than we have explored the infinite space of integers.
I think the point of this analogy is rather to show how hard it can be to find a conter-example, and that some results turn out to be wrong even though everyone expected them to be right. The idea is that the number of possible chess games and the size of the counter exampled shown by george are both way beyond human reach. This example suggests that in the same way we shouldn't have trusted the mathematician's intuition on that conjecture on prime, it is not reasonable to trust human's intuition on wether chess is a draw or not. This question is too hard for human alone. That is why we try to build tablebases.
The example of the Q+P vs Q endgame was interesting and when those results came out of tablebases they were surprising. But Q vs Q+P endings being wins for the superior side just expand the set of winning positions without changing the fundamental argument that chess is a draw. N v N+P is unquestionably winning almost all (all?) the time so adding more winning positions doesn't change much and tablebases haven't changed our understanding of this position. Chess is a draw not because people can't figure out how to convert wins, but because forcing as much as a single pawn advantage seems impossible. Tablebases have increased our understanding of chess by showing us how to convert wins in superior positions. My contention is that these positions are largely irrelevant because they can't be achieved to begin with.
I think the most important point is not to know if these position will occur in a perfect game, but to acknoledge they exist. The immediate consequence of such positions existing is that human analysis (using heuristics) can't exhaust the game of chess. In other words, there are positions that resist human analysis. What this means is that we should be extremely careful when we say a position is drawn, because we know there are very simplified positions that we used to assess as drawn and that turned out to be wins. So it is possible that a lot of much more complex middlegame positions that we think are drawn turn out to be wins as well.
To be very clear, the most important point is the following :
We know that playing with heuristics (i.e. experience based methods) fails to achieve perfect play in some situations, so we should be very carefull when using experience-based "evidence" (like grandmaster games or centaur chess) to conclude chess is a draw.
I think sapientdust made a fine post above summarizing some of the points now lost. A few points:
a) I'm surprised he is not banned after that outrageous remark " no mathematically literate person could ever make such an idiotic argument as the bolded sentence" about ponz. Are you saying ponz isn't mathematically literate? Really?
The bolded sentence was one that smyslovfan said was Jetson's position (nothing to do with ponz). Since the sentence was about Jetson, who is in fact a mathematician and thus certainly mathematically literate, I was effectively saying that Jetson didn't put forward such a silly position, because no mathematician would ever make such a ludicrous argument. It's the sort of fallacy that even a first-year philosophy student would notice.
ah. misunderstood that. No wonder you aren't banned.
Thanks Tronchen for your intelligent and thoughtful post. I don't believe any of that stuff I posted; I just posted it as an example of how conversations like this can be productive (i.e., they stimulate thoughtful replies like yours). I think it would probably just cause a mess if I continued on defending some position I don't believe.
Centaur chess is not just experience based. It is experienced based along with brute force chess engines. That is why it is stronger than chess engines alone.
Yes, it is possible that a lot of the more complex middlegame positions that we think are drawn turn out to be wins. It is just not likely. But regardless why even use such a complex middle game positions when there are hundreds of games and positions which have no errors and end up in a draw? [this is a question]
I think what you are doing is to pick out the one in more than 100 games which are unclear and saying because we do not know the result of this one game [out of more than 100] that we cannot look at the other 99 plus games to see what we can learn from them.
@chiaro :
Ok I didn't get that.
This thread is most likely going to die anyway, since most people, for various reasons, are discouraged to argue here. It was an interesting thread though.
ponz I already explained several times (and george did it aswell) that we do not have "hundreds of game that have no errors and end up in a draw". We are unable to tell in great number of positions (inculding the starting position) if a move is or not a mistake. If you think you can seriously convince anyone that you know the list of first moves that are not mistakes, you should publish a paper because the whole math community is waiting for that answer.
I'm not trying to pick rare games or whatever. What I am saying is that as long as there are more than 7 pieces on the board, our means of analysis (human + engines) are by nature unable to make the distinction between perfect play and good practical play. We cannot rely on them, engines are not programmed to find perfect moves and humans are way too limited to probably understand why a perfect move is a perfect move in a lot of situations (including, again, very simplified positions like Q+P vs Q endgames).
Don't expect to get more answers from me, for reasons we both know. I came back essentialy to answer chiaro's post, which looked like an invitation for a constructive debate. Turns out he wasn't believing what he said, so I guess I'm out again.
This has been a fascinating debate. Filled with all sorts of emotions, opinions, theories, etc. I hope it doesn't end....
Centaur chess is not just experience based. It is experienced based along with brute force chess engines. That is why it is stronger than chess engines alone.
Nevertheless, Centaur chess is heuristic.
Philidor predicted that the King's Gambit and this position were draws:
As for the King's Gambit this is a forced win for black as discovered by Bobby Fischer, though many GMs throughout the early 20th century held the gambit in contempt, such as Capablanca:
Yes, Philador made a mistake. But your contention is that the King's Gambit is busted because Fischer said it was busted?
So you are showing that players can make mistakes?
Centaur Chess has an element of heuristics in it but it is much more than that. That is why it is better than just heuristics.
I think there is confusion about finding perfect moves in a possible perfect game. I may be wrong but believe some think that there is only one perfect move in a position? this is not true at all. Most opening positions have several moves which will not change the eventual outcome of a game. To give an example. 1. e4 now all of these moves will not change the outcome of the game. 1. ...e5 1. ...c5 1. ...c6
Thus it is rather easy to find examples of perfect games as we have defined it here. Remember it takes at least a pawn advantage to change the eventual outcome of a game.
Trying to solve chess seems too ambitious with today's technology, but certain openings have been refuted to either a loss for one side or a draw. The most obvious places to look for refutations are gambits: 1.e4,c5 2.b4,cxb4 3.a3,d5 for example is at the very least a headache for white.
1.e4,c5 2.f4,d5 I think might be refuted to a draw but aren't 100% on it.
The term "refuted" has several different meanings depending on the person using it. There is no consensus..
Centaur Chess is getting to the point that Black or White can draw vs any opening.
I would be 99% sure solved chess is a draw. Tic tac toe is a draw and the advantage of first move is much higher. Checkers has been solved and is also a draw with perfect play.
Checkers, BTW doesn't even need perfect play to be a forced draw:
http://chinook.cs.ualberta.ca/users/chinook/index.html
Centaur Chess is getting to the point that Black or White can draw vs any opening.
This is likely fruitless but hope springs eternal..
It is not enough to continually claim this as evidence of anything. There is no argument attached to this stylized fact that you repeatedly post on this thread. I'd like to come up with an example argument for you but the are all so dumb, I just can't do it. But think about what this fact might mean - Anand armed with Houdini draws Topalov armed with Houdini 100% of the time. Let's just say that we completely accept that as fact (I wouldn't but am willing to go with it). Does that mean anything about the ultimate state of chess? It means:
a) Houdini has made it so that neither of these great players make human tactical errors.
b) Without these tactical errors neither of them can gain enough advantage to win ever.
From that you can probably conclude that super-GM wins are due to tactical errors that machines wouldn't make. That's not a million miles close to being able to show that chess is ultimately a draw. It doesn't even address the question.
It is hard to get around problems like:
a) The state of chess knowledge for both Topalov and Anand is about the same since the have access to the same public body of knowledge.
b) There is no way at all of knowing how much this knowledge will change going forward. Capablanca and Fischer both felt that the state of knowledge was nearly complete toward the ends of their career. Both were deeply insightful and both were wrong.
c) Comparing the tactical abilities of the same engine is useless. If they were armed with different engines, the claim about centaur chess always being a draw is looking ridiculous.
d) I can take any chess engine run by either of these guys and throw more money than either of them has at the engine and make it instantly better. If Anand wants to play me at centaur chess, I would give him draw odds. I would simply go outspend Anand on my chess computer simply so I could beat the world champion (the ego value of that to me is worth way more than it would be to Anand and he is paid like a chessplayer). How we can think that centaur chess is a draw when we are certain Anand would come back afterwards and say it wasn't fair isn't clear at all.
Ponz should try hard to address all that. Claiming that this is an attack on you or repeating silly stuff like "You don't think centaur chess is a draw?" is not helpful. Try hard to make an argument about why centaur chess has something to do with proving chess is a draw. You should try to address at least some of the points above (which may mean that you give some reason to believe they aren't true).
Not the Latvian Gambit or Englund Gambit, those are forced wins for white. I think you mean draw vs. any reasonable opening. Have to agree there.
Oh ... and more importantly... Ponz should address this:
"ponz I already explained several times (and george did it aswell) that we do not have "hundreds of game that have no errors and end up in a draw". We are unable to tell in great number of positions (inculding the starting position) if a move is or not a mistake. If you think you can seriously convince anyone that you know the list of first moves that are not mistakes, you should publish a paper because the whole math community is waiting for that answer."
I don't know if Ponz doesn't understand that or what the problem is. It seems like the response is always the non-response of something like "We know what mistakes are in many settings". Maybe Tronchenblais's point needs to be explained more slowly?
Terribly boring discussion now.
This is the 1593rd post