POLL: What is more important?

[Marek]

Engines with perfect play will always play perfect no matter what gambit or move it is because perfect engines will have solved chess.

What is perfect play? Possession of 32TBs won't result in perfect play. (If chess is a draw) the possessor won't lose. But that's not the entire problem. Effective play means drawing least too. The 32TBs won't discriminate between moves that draw. But practical play sometimes will. After 1.d4 both 1...d5 and 1...Nf6 draw (probably, maybe other moves too). Which move does the "perfect" engine or the 32TBs-reader choose? And why?

[Dr. Ivannik]

Engines with perfect play will always play perfect no matter what gambit or move it is because perfect engines will have solved chess.

What is perfect play? Possession of 32TBs won't result in perfect play. (If chess is a draw) the possessor won't lose. But that's not the entire problem. Effective play means drawing least too. The 32TBs won't discriminate between moves that draw. But practical play sometimes will. After 1.d4 both 1...d5 and 1...Nf6 draw (probably, maybe other moves too). Which move does the "perfect" engine or the 32TBs-reader choose? And why?

Gentlemen

It is quite simple enough. The engine will choose the move that gets the result be it a win or draw the fastest. Ex. After 1.d4 1...d5 (draw in 65) or 1...Nf6 (draw in 75) the engine plays the move that gets result fastest.The 32TBs WILL discriminate between moves that draw. If two moves will have same result in same amount of moves a annotation window will note this information. Not really to hard.

Thank you

Dr. Ivannik

[Marek]

Engines with perfect play will always play perfect no matter what gambit or move it is because perfect engines will have solved chess.

What is perfect play? Possession of 32TBs won't result in perfect play. (If chess is a draw) the possessor won't lose. But that's not the entire problem. Effective play means drawing least too. The 32TBs won't discriminate between moves that draw. But practical play sometimes will. After 1.d4 both 1...d5 and 1...Nf6 draw (probably, maybe other moves too). Which move does the "perfect" engine or the 32TBs-reader choose? And why?

Gentlemen

It is quite simple enough. The engine will choose the move that gets the result be it a win or draw the fastest. Ex. After 1.d4 1...d5 (draw in 65) or 1...Nf6 (draw in 75) the engine plays the move that gets result fastest.The 32TBs WILL discriminate between moves that draw. If two moves will have same result in same amount of moves a annotation window will note this information. Not really to hard.

Thank you

Dr. Ivannik

So, the engine will play the "shortest" draw (in a drawn position). But that won't necessarily give the opposition the greatest opportunity to go wrong. In fact, it would make more sense to choose the "longest" draw to test the opposition's standard of play (which is unknown). But that's still not certainly the most effective policy, is it? Presumably a "complex" or unbalanced position would be more testing than a simple and symmetrical one. But that's not revealed by tablebases alone.

[Uly]

Wow Marek, I admire your patience.

Anyway, I came with a new argument for xshat: Imagine that chess is won for white with perfect play, that is, 32men Tablebases say that e4 wins the game against all defenses.

Now, we put this perfect engine to play from the black side against me, and I play e4. Does this perfect engine resign? Because your ideas of all engines with 32men TB playing the same moves lead to them also resigning on move one in this situation.

If they don't resign, then, what do they play? The path maximizing the way to mate may not be best, say, that 1...f6 is the longest mate so the engine plays it, but the longest mate is also the most clear one to victory, so playing it is wrong. The new task of the engine is to try to win from this lost position (which is possible, and likely since in the example the perfect engine is playing a patzer). There are many ways of doing it, so all perfect engines would play differently.

It's the same if the game is drawn, the perfect engine knows the opponent could play a perfect game and go for a draw, but if it's playing me that's not true, so it would try to defeat me from a drawn position.

If your perfect engine would resign at move 1 because it all sees is losing moves, and because it knows another perfect engine would beat it 100% of the time, then your perfect engine would play weaker against me than a nonperfect engine, which makes no sense.

[Marek]

Wow Marek, I admire your patience.

Anyway, I came with a new argument for xshat: Imagine that chess is won for white with perfect play, that is, 32men Tablebases say that e4 wins the game against all defenses.

Now, we put this perfect engine to play from the black side against me, and I play e4. Does this perfect engine resign? Because your ideas of all engines with 32men TB playing the same moves lead to them also resigning on move one in this situation.

If they don't resign, then, what do they play? The path maximizing the way to mate may not be best, say, that 1...f6 is the longest mate so the engine plays it, but the longest mate is also the most clear one to victory, so playing it is wrong. The new task of the engine is to try to win from this lost position (which is possible, and likely since in the example the perfect engine is playing a patzer). There are many ways of doing it, so all perfect engines would play differently.

It's the same if the game is drawn, the perfect engine knows the opponent could play a perfect game and go for a draw, but if it's playing me that's not true, so it would try to defeat me from a drawn position.

If your perfect engine would resign at move 1 because it all sees is losing moves, and because it knows another perfect engine would beat it 100% of the time, then your perfect engine would play weaker against me than a nonperfect engine, which makes no sense.

Slowly we're getting somewhere, but I need to recap some things. It was never my idea that all engines with 32TBs will play the same moves (unless there's only one winning move); it was never my idea that a perfect engine would immediately resign a lost position either.

The perfect engine cannot know the standard of opposition. How could it? If 1.e4 wins (or draws) according to 32TBs it should still come up with a reply as Black, and hope that White goes wrong. It need not play "weaker" against one opponent rather than another. It doesn't know who it's playing, remember. Nevertheless, the clock is ticking and the perfect engine still has to come up with moves. But access to 32TBs won't completely solve that problem. Is 1...c5 more likely to succeed against imperfect opposition than 1...e5 or 1...c6, etc.? The tablebases won't say. In other words, the "perfect" engine has to be more than a mere tablebase-reader, as one poster erroneously suggested. And if in your words, "all perfect engines would play differently" in the same certain circumstances, then they may score differently too. They will win the won positions of course, but in a lost or drawn position against an imperfect but strongish opponent (e.g. Rybka on fast hardware) one perfect engine could be outperformed by another. I don't know what the best policy is to turn around a game against Rybka. In a drawn or lost position typically several moves will be of equal value. (Tablebases don't confer degrees of advantage.) Should the engine head for a particular endgame or for complications or what exactly? I don't know. That's why 32TBs, which are practically impossible anyway, won't completely solve the eternal riddle of chess - what move to play next.

[xshat]

Engines with perfect play will always play perfect no matter what gambit or move it is because perfect engines will have solved chess.

What is perfect play? Possession of 32TBs won't result in perfect play. (If chess is a draw) the possessor won't lose. But that's not the entire problem. Effective play means drawing least too. The 32TBs won't discriminate between moves that draw. But practical play sometimes will. After 1.d4 both 1...d5 and 1...Nf6 draw (probably, maybe other moves too). Which move does the "perfect" engine or the 32TBs-reader choose? And why?

Perfect play is how an engine will play once chess is solved and it always makes the right move.
http://en.wikipedia.org/wiki/Solved_game

[xshat]

Wow Marek, I admire your patience.

Anyway, I came with a new argument for xshat: Imagine that chess is won for white with perfect play, that is, 32men Tablebases say that e4 wins the game against all defenses.

Now, we put this perfect engine to play from the black side against me, and I play e4. Does this perfect engine resign? Because your ideas of all engines with 32men TB playing the same moves lead to them also resigning on move one in this situation.

If they don't resign, then, what do they play? The path maximizing the way to mate may not be best, say, that 1...f6 is the longest mate so the engine plays it, but the longest mate is also the most clear one to victory, so playing it is wrong. The new task of the engine is to try to win from this lost position (which is possible, and likely since in the example the perfect engine is playing a patzer). There are many ways of doing it, so all perfect engines would play differently.

It's the same if the game is drawn, the perfect engine knows the opponent could play a perfect game and go for a draw, but if it's playing me that's not true, so it would try to defeat me from a drawn position.

If your perfect engine would resign at move 1 because it all sees is losing moves, and because it knows another perfect engine would beat it 100% of the time, then your perfect engine would play weaker against me than a nonperfect engine, which makes no sense.

If the engine is programmed to resign after the first move then I imagine it would. If it is programmed to fight to the end, as it should be, then it will. And it also means you're only going to win if you play perfect. Tehy will play what they are programmed to. The engine will attempt to turn a losing position into either a draw or win and if the non-perfect humans is playing then the odds could change before the middle game is even reached. I disable my chess engines from resigning and make the game play all the way to checkmate, it's more fun that way.

[xshat]

Wow Marek, I admire your patience.

Anyway, I came with a new argument for xshat: Imagine that chess is won for white with perfect play, that is, 32men Tablebases say that e4 wins the game against all defenses.

Now, we put this perfect engine to play from the black side against me, and I play e4. Does this perfect engine resign? Because your ideas of all engines with 32men TB playing the same moves lead to them also resigning on move one in this situation.

If they don't resign, then, what do they play? The path maximizing the way to mate may not be best, say, that 1...f6 is the longest mate so the engine plays it, but the longest mate is also the most clear one to victory, so playing it is wrong. The new task of the engine is to try to win from this lost position (which is possible, and likely since in the example the perfect engine is playing a patzer). There are many ways of doing it, so all perfect engines would play differently.

It's the same if the game is drawn, the perfect engine knows the opponent could play a perfect game and go for a draw, but if it's playing me that's not true, so it would try to defeat me from a drawn position.

If your perfect engine would resign at move 1 because it all sees is losing moves, and because it knows another perfect engine would beat it 100% of the time, then your perfect engine would play weaker against me than a nonperfect engine, which makes no sense.

Slowly we're getting somewhere, but I need to recap some things. It was never my idea that all engines with 32TBs will play the same moves (unless there's only one winning move); it was never my idea that a perfect engine would immediately resign a lost position either.

The perfect engine cannot know the standard of opposition. How could it? If 1.e4 wins (or draws) according to 32TBs it should still come up with a reply as Black, and hope that White goes wrong. It need not play "weaker" against one opponent rather than another. It doesn't know who it's playing, remember. Nevertheless, the clock is ticking and the perfect engine still has to come up with moves. But access to 32TBs won't completely solve that problem. Is 1...c5 more likely to succeed against imperfect opposition than 1...e5 or 1...c6, etc.? The tablebases won't say. In other words, the "perfect" engine has to be more than a mere tablebase-reader, as one poster erroneously suggested. And if in your words, "all perfect engines would play differently" in the same certain circumstances, then they may score differently too. They will win the won positions of course, but in a lost or drawn position against an imperfect but strongish opponent (e.g. Rybka on fast hardware) one perfect engine could be outperformed by another. I don't know what the best policy is to turn around a game against Rybka. In a drawn or lost position typically several moves will be of equal value. (Tablebases don't confer degrees of advantage.) Should the engine head for a particular endgame or for complications or what exactly? I don't know. That's why 32TBs, which are practically impossible anyway, won't completely solve the eternal riddle of chess - what move to play next.

32TB's will tell it what to do and it will not be asking human questions such s whether to play c5 or e5. The tablebases will say. If programmed correctly, it will have solved chess from move 1.

Chess is not impossible to solve.

[Uly]

It was never my idea that all engines with 32TBs will play the same moves (unless there's only one winning move); it was never my idea that a perfect engine would immediately resign a lost position either.

Yes, they were xshat's ideas (the former, the latter is just an extension), I agree with your arguments, it's him who seems to be stuck on a loop. Like in his last post, where he says the TB's are going to tell the engine whether to play e5 or c5, which is impossible if both moves lead to draw, the engine has to decide which one to play in some other manner.

[xshat]

It was never my idea that all engines with 32TBs will play the same moves (unless there's only one winning move); it was never my idea that a perfect engine would immediately resign a lost position either.

Yes, they were xshat's ideas (the former, the latter is just an extension), I agree with your arguments, it's him who seems to be stuck on a loop. Like in his last post, where he says the TB's are going to tell the engine whether to play e5 or c5, which is impossible if both moves lead to draw, the engine has to decide which one to play in some other manner.

You have once again twisted my words around. I said when chess is solved the engines will make perfect play, and they will have 32 men tablebases. The "resigning on first move" has nothing to do with what I was talking about.