- Blind mode tutorial
lichess.org
Donate

Why can't humans beat computers in chess?

#61

Draw any conclusion you like.
The given "test" was 5 minute blitz matches of old vs. new.

Outcome of games
The first game was drawn. Kramnik won games 2 and 3 by "conventional" anti-computer tactics—play conservatively for a long-term advantage the computer is not able to see in its game tree search. After a draw in game 4, Kramnik lost game 5 due to a blunder. Game 6 was described by commentators as "spectacular". Kramnik, in a better position in the early middlegame, sacrificed a piece to launch an attack—a strategy known to be highly risky against computers, which are at their strongest when defending such attacks. True to form, Fritz found a watertight defense and Kramnik was left in a bad position. Kramnik resigned the game, believing his position to be lost. However, post-game analysis has shown that Fritz was unlikely to have been able to force a win—Kramnik gave up a drawn position.[1] The final two games were draws.[2]
Wiki

Fifteen years ago, in October of 2002, Vladimir Kramnik and Deep Fritz were locked in battle in the Brains in Bahrain match. If Kasparov vs. Deep Blue was the beginning of the end for humans in Chess, then the Brains in Bahrain match was the middle of the end. It marked the first match between a world champion and a chess engine running on consumer-grade hardware, although its eight-processor machine was fairly exotic at the time.
Ultimately, Kramnik and Fritz played to a 4-4 tie in the eight-game match. Of course, we know that today the world champion would be crushed in a similar match against a modern computer. But how much of that is superior algorithms, and how much is due to hardware advances?
Wiki

Draw any conclusion you like. The given "test" was 5 minute blitz matches of old vs. new. Outcome of games The first game was drawn. Kramnik won games 2 and 3 by "conventional" anti-computer tactics—play conservatively for a long-term advantage the computer is not able to see in its game tree search. After a draw in game 4, Kramnik lost game 5 due to a blunder. Game 6 was described by commentators as "spectacular". Kramnik, in a better position in the early middlegame, sacrificed a piece to launch an attack—a strategy known to be highly risky against computers, which are at their strongest when defending such attacks. True to form, Fritz found a watertight defense and Kramnik was left in a bad position. Kramnik resigned the game, believing his position to be lost. However, post-game analysis has shown that Fritz was unlikely to have been able to force a win—Kramnik gave up a drawn position.[1] The final two games were draws.[2] Wiki Fifteen years ago, in October of 2002, Vladimir Kramnik and Deep Fritz were locked in battle in the Brains in Bahrain match. If Kasparov vs. Deep Blue was the beginning of the end for humans in Chess, then the Brains in Bahrain match was the middle of the end. It marked the first match between a world champion and a chess engine running on consumer-grade hardware, although its eight-processor machine was fairly exotic at the time. Ultimately, Kramnik and Fritz played to a 4-4 tie in the eight-game match. Of course, we know that today the world champion would be crushed in a similar match against a modern computer. But how much of that is superior algorithms, and how much is due to hardware advances? Wiki
#62

Brains in Bargain.

I remember.

Brains in Bargain. I remember.
#63

CM Sarg0n said:
Brains in Bargain.

LOL

mdinnerspace said:
But how much of that is superior algorithms, and how much is due to hardware advances?

This is quite easy to answer: about 80% (my estimate) is hardware advances, 17-18% is making use of these hardware advances and about 2-3% is optimizing the implementation of the same algorithms used 15 years ago.

The algorithmic side of chess programming has not changed over the last years: you construct a search tree of all possible positions and work on that with a minimax algorithm by using alpha-beta pruning to intelligently limit the necessary work. This is the algorithmic side of the problem.

Real-world chess programs (experimental programming concepts aside) are implementations of this concept and the difference in playing strength is just reflecting the different optimisation strategies used by the programmers. Chess programming (along with signal processing) is probably the last stand of assembler programming in a world where languages like C are already considered low-level. *)

What makes todays chess programs better is mostly the raising of the calculating horizon and the fact that endgame databases are getting always bigger AND it is possible to store them on current storage. With more pieces covered the number of positions rises dramatically, as Kyrill Kryukov once pointed out:

2: 462
3: 368,079
4: 125,246,598
5: 25,912,594,054
6: 3,787,154,440,416
7: 423,836,835,667,331
8: 38,176,306,877,748,245
(from: http://kirill-kryukov.com/chess/nulp/results.html )

Assuming the relationship of positions and necessary space to store them to be O(n) it is clear that one needs the increased size of todays long-term storage media to make practical use of these databases in a program.

This and the feasibility of storing ever-growing opening books all contributes to the programs increasing strength.

krasnaya

*) Compare the time to start M$$ Winword 6 on a 386 and the time the latest Word version takes to start on your current i7-system. It takes about the same time even though you have hardware a thousand times faster. This is what that "high-level integrated intelligent visual optimized 3D-pointy-clicky-blah-blah" development environments do to the resluting code. A couple of gigabytes just for starting a hello-world-program.

CM Sarg0n said: Brains in Bargain. LOL mdinnerspace said: But how much of that is superior algorithms, and how much is due to hardware advances? This is quite easy to answer: about 80% (my estimate) is hardware advances, 17-18% is making use of these hardware advances and about 2-3% is optimizing the implementation of the same algorithms used 15 years ago. The algorithmic side of chess programming has not changed over the last years: you construct a search tree of all possible positions and work on that with a minimax algorithm by using alpha-beta pruning to intelligently limit the necessary work. This is the algorithmic side of the problem. Real-world chess programs (experimental programming concepts aside) are implementations of this concept and the difference in playing strength is just reflecting the different optimisation strategies used by the programmers. Chess programming (along with signal processing) is probably the last stand of assembler programming in a world where languages like C are already considered low-level. *) What makes todays chess programs better is mostly the raising of the calculating horizon and the fact that endgame databases are getting always bigger AND it is possible to store them on current storage. With more pieces covered the number of positions rises dramatically, as Kyrill Kryukov once pointed out: 2: 462 3: 368,079 4: 125,246,598 5: 25,912,594,054 6: 3,787,154,440,416 7: 423,836,835,667,331 8: 38,176,306,877,748,245 (from: http://kirill-kryukov.com/chess/nulp/results.html ) Assuming the relationship of positions and necessary space to store them to be O(n) it is clear that one needs the increased size of todays long-term storage media to make practical use of these databases in a program. This and the feasibility of storing ever-growing opening books all contributes to the programs increasing strength. krasnaya ___________ *) Compare the time to start M$$ Winword 6 on a 386 and the time the latest Word version takes to start on your current i7-system. It takes about the same time even though you have hardware a thousand times faster. This is what that "high-level integrated intelligent visual optimized 3D-pointy-clicky-blah-blah" development environments do to the resluting code. A couple of gigabytes just for starting a hello-world-program.
#64

@krasnaya
"This is quite easy to answer: about 80% (my estimate) is hardware advances, 17-18% is making use of these hardware advances and about 2-3% is optimizing the implementation of the same algorithms used 15 years ago."
2-3%? The top programs have improved at a rate of about 50 Elo per year over the last couple of years. The basic algorithms are the same, but many heuristics have been improved and nowadays the tree is pruned very aggressively, especially in Stockfish, so the engines can search much deeper even on the same hardware.

@krasnaya "This is quite easy to answer: about 80% (my estimate) is hardware advances, 17-18% is making use of these hardware advances and about 2-3% is optimizing the implementation of the same algorithms used 15 years ago." 2-3%? The top programs have improved at a rate of about 50 Elo per year over the last couple of years. The basic algorithms are the same, but many heuristics have been improved and nowadays the tree is pruned very aggressively, especially in Stockfish, so the engines can search much deeper even on the same hardware.
#65

IR So confused why this thread is 7 pages long.

Wasnt this basically a silly question?

To me it was in the same boat as:

Why cant humans run as fast as cars?

lol.

IR So confused why this thread is 7 pages long. Wasnt this basically a silly question? To me it was in the same boat as: Why cant humans run as fast as cars? lol.
#66

The verb in the question was Beat.
A human cannot beat a computer, because a chess engine cannot be hurt or injured. Humans can get completely exhausted trying to find the main path to victory, but computers do not get exhausted.
Chess is all 1's and 0's, which could be simply dark and light squares. Maybe if we used some sort of binary colour complex system, we could force the computer to move pieces to our advantage. The chess battle is human strategies vs engine tactics. Looks like tactics is dominating the chessboard. Maybe a chess fortress is the solution. The Hedgehog Defence is not a fortress but it is an interesting strategy. Another strategy could be to force the engines to not pick the main strategic line. Chess most likely has a main strategic winning line and a main tactical winning line. The tactical main line will probable be solved before the strategic one. If the strategic main line is solved first, it is because chess players worked on finding a solution.

The verb in the question was Beat. A human cannot beat a computer, because a chess engine cannot be hurt or injured. Humans can get completely exhausted trying to find the main path to victory, but computers do not get exhausted. Chess is all 1's and 0's, which could be simply dark and light squares. Maybe if we used some sort of binary colour complex system, we could force the computer to move pieces to our advantage. The chess battle is human strategies vs engine tactics. Looks like tactics is dominating the chessboard. Maybe a chess fortress is the solution. The Hedgehog Defence is not a fortress but it is an interesting strategy. Another strategy could be to force the engines to not pick the main strategic line. Chess most likely has a main strategic winning line and a main tactical winning line. The tactical main line will probable be solved before the strategic one. If the strategic main line is solved first, it is because chess players worked on finding a solution.
#67

There has been loads lot misconception in this threas (as well as very accurate information)

  1. Evaluation function has positional understanding on ELo 1300. Well perhaps in seventies. Have someone elo 1500 play against a computer and allow it only Q-search on top 1-ply (ie lookjonly on move but check for obvious forces checks and captures) an he/she will be beaten

  2. Opening book in computer a) are unfair b) play a part in their strength. for a) no, if you have a perfect memory ofcourse it fair to use if. b) there has been challenge match between a GM and top enginen without opein move ad result was as expected. Also 2015 humans got crushed in odds matches I do doubt if opening book helps one bit if you a re missing a pawn or two.

Chess is game where humans have lost all the way. Analysis power rules. In correspondence games a human assisted computer may be alittle bit better than pure computer. At least this is what correspondence players claim. And they still do have tournaments

There has been loads lot misconception in this threas (as well as very accurate information) 1. Evaluation function has positional understanding on ELo 1300. Well perhaps in seventies. Have someone elo 1500 play against a computer and allow it only Q-search on top 1-ply (ie lookjonly on move but check for obvious forces checks and captures) an he/she will be beaten 2. Opening book in computer a) are unfair b) play a part in their strength. for a) no, if you have a perfect memory ofcourse it fair to use if. b) there has been challenge match between a GM and top enginen without opein move ad result was as expected. Also 2015 humans got crushed in odds matches I do doubt if opening book helps one bit if you a re missing a pawn or two. Chess is game where humans have lost all the way. Analysis power rules. In correspondence games a human assisted computer may be alittle bit better than pure computer. At least this is what correspondence players claim. And they still do have tournaments
#68

@breakreign We all know that computers are better but some (like me) don't really understand the mechanics of it. This thread has some people who provided really good answers.

Well it also kick-started the whole discussion of whether humans can still beat chess computers. While it is foolish to discuss if humans can outrun cars, perhaps it's not so clear cut here.

@breakreign We all know that computers are better but some (like me) don't really understand the mechanics of it. This thread has some people who provided really good answers. Well it also kick-started the whole discussion of whether humans can still beat chess computers. While it is foolish to discuss if humans can outrun cars, perhaps it's not so clear cut here.
#69

GMs lose against engines nowadays either

-with development advantage, given a couple of extra-moves
-given extra material, some pawns
-don‘t have to defend tb draws
-infinite time
-correspondence chess
-are allowed to use additional help like a „weak“ engine, books or dbs
-and so on

This is utterly humiliating, isn‘t it? The case was closed long ago; sticklers, trolls and partypoopers may find a point in searching exceptions frantically as well as correspondence chess players are denying that they are slaves to the engines.

Nakamura: „We are all clowns.“

PS: see number #2, it says it all: we only see a tiny part of the exponential jungle, that‘s the reason we don‘t stand a chance against brute-force. Those up to 100.000 acquired chess chunks/patterns (GM) as „understanding“ is nothing.

GMs lose against engines nowadays either -with development advantage, given a couple of extra-moves -given extra material, some pawns -don‘t have to defend tb draws -infinite time -correspondence chess -are allowed to use additional help like a „weak“ engine, books or dbs -and so on This is utterly humiliating, isn‘t it? The case was closed long ago; sticklers, trolls and partypoopers may find a point in searching exceptions frantically as well as correspondence chess players are denying that they are slaves to the engines. Nakamura: „We are all clowns.“ PS: see number #2, it says it all: we only see a tiny part of the exponential jungle, that‘s the reason we don‘t stand a chance against brute-force. Those up to 100.000 acquired chess chunks/patterns (GM) as „understanding“ is nothing.
#70

While engines have improved a lot (and not because of hardware, because of software; Stockfish on your cellphone would easily beat an old Fritz version running on a 1000 core cluster) humans + engines combo is still stronger in Correspondence Chess. However the roles have changed a bit, nowadays obviously the computer does the hard calculating work however humans are still useful to decide how much time to use on which variation, which moves are critical and deserve a lot of calculation, how is the style of the position, which engine should I use for this specific position etc etc. (engines are really strong, however they aren't perfect and also they have different strengths; so by using different engines depending on the position humans can draw from the strength of each engine while avoiding their weak spots)

While engines have improved a lot (and not because of hardware, because of software; Stockfish on your cellphone would easily beat an old Fritz version running on a 1000 core cluster) humans + engines combo is still stronger in Correspondence Chess. However the roles have changed a bit, nowadays obviously the computer does the hard calculating work however humans are still useful to decide how much time to use on which variation, which moves are critical and deserve a lot of calculation, how is the style of the position, which engine should I use for this specific position etc etc. (engines are really strong, however they aren't perfect and also they have different strengths; so by using different engines depending on the position humans can draw from the strength of each engine while avoiding their weak spots)

This topic has been archived and can no longer be replied to.