< Earlier Kibitzing · PAGE 14 OF 14 ·
|May-11-13|| ||offramp: May 11th, 1997. Sixteen years ago today. Kasparov lost 2-1, but since then we've seen Mickey Adams lose 5-0.|
But we've also seen Nakamura giving computers some major kick-boxing lessons. So it's not all one way traffic.
|May-11-13|| ||AylerKupp: <copablanco> I do not know what you mean by "foresight". Foresight is usually defined as the ability to look forward and the chess engine does this by creating a tree of possible moves at ever increasing depth, evaluating the position after each of those moves, and selecting as its next move the move whose line provides the highest evaluation. This is what most chess engines have done since their inception, as describe by Shannon in his 1949 paper, "Programming a Computer for Playing Chess". See http://vision.unipv.it/IA1/Programm.... So increasing "foresight" is strictly a matter of searching deeper in the available time to make a move, something that can be accomplished by more capable hardware (brute forth) or more sophisticated (computer jargon for "clever") algorithms, or both.|
And evaluation functions have also been around since the beginning, see Shannon's paper also. The reason that computers played so poorly initially was that their evaluation functions were very primitive, often consisting only of evaluating material, because of the low speed of computers at the time, as well as the limited searching possible for the same reason. Today's much more powerful computers can afford much more sophisticated evaluation functions as well as much deeper searches, so they play better.
This is usually considered different than heuristics, although you could say that evaluation functions are a form of heuristics (computer jargon for "guesses"). More typical are the heuristics used to prune the search tree, such as first evaluating moves that have led to higher position evaluations in previous searches first. There is no guarantee that a heuristic will yield the best result, as contrast to an algorithm such as the minimax or alpha-beta pruning which will give the best result each time by a specific set of calculations.
All this has nothing to do with memory. Computer chess playing algorithms are very efficient in their use of memory so very good play can be achieved with relatively little memory. Even the use of opening books does not require much memory; they are typically stored on disk and searched when needed, and the disk storage required is not all that great. The main exception is hash tables which allow the chess engine to reduce the number of position evaluations that it needs to do by storing the evaluation of previously encountered positions. If the same position occurs later in the search, it does not need to be re-evaluated.
But you're right that a chess engine will not necessarily evaluate better than a human chessmaster, again depending on what you mean by "better". It can certainly evaluate faster and consider more move possibilities, and this allows it to often find moves that a human will not find because, with our limited evaluation capabilities, we are unable to look for them. And, of course, a lot depends on the quality of the chess engine's evaluation function.
|May-11-13|| ||AylerKupp: <offramp> True, Nakamura has been able to give computer some major kick-boxing lessons by exploiting some well-known limitations of computer chess engines. But these have occurred at blitz when the computer did not have much time per move and so its ability to search was limited to very low ply. I doubt that Nakamura, or any other grandmaster or team of grandmasters could defeat today's more powerful engines at regular time controls running on reasonably powerful hardware.|
And, of course, the Nakamura vs. computer games that are publicized are usually the games that Nakamura (or any other grandmaster) wins. Kind of the "man bites dog" situation. Seldom seen are the 99% of games that Nakamura lost. And I picked that number arbitrarily, I don't know the actual percentage of games that Nakamura lost when playing against a computer, but it was the greater majority.
|May-29-13|| ||Guitar1352: I agree with KittySafe. Garry deliberately wanted to lose this game. Besides, who knows? Maybe his nerves got the better of him? |
Anyway, Deep Blue cheated to win him because it had to be upgraded for this match, so technically IBM had to cheat Garry in order to win him!
And whoever says Garry shouldn't have taken the baited Bishop, you are all wrong. According to Critter AND Stockfish, it was the correct move, but he should have taken the rook with the D5 Knight instead of the F8 Bishop.
I don't care what anybody says, Garry Kasparov will forever(!!) be one of the greatest chess players ever!
|May-29-13|| ||perfidious: <Guitar1352>: That's a powerful set of accusations there: first your hero Kasparov 'wanted to lose', then Deep Blue et al 'cheated'.|
|May-29-13|| ||Just Another Master: one game...lets not end the chess world|
|May-29-13|| ||KlingonBorgTatar: And the chess world is poorer since the rematch Garry K asked for somewhat did not materialize.|
|Jun-30-13|| ||HeadCrunch: I was beaten by my chess computer once as it was no match for me at kick boxing.|
|Jul-10-13|| ||Wyatt Gwyon: Amazing that so many think Kasparov threw this game.|
|Jul-11-13|| ||RookFile: He gambled on the computer not knowing what to do in the opening. Who knows, maybe the wrong moves would have been in it's opening database? If he had to do it all over again, he might have preferred the Petrov, and offered a draw around move 25 or so. The IBM team probably would have accepted it.|
|Aug-31-13|| ||mmousez: Let's not be harsh on Garry. In a recent interview with Sir David Frost, he said about this game:
1. Both he and Deep Blue missed a way for Black to save the game. He didn't go into the specifics, so perhaps a chess expert can elaborate on this one. (I am thinking maybe after 8. Nxe6 fxe6, 9. Bg6+ Ke7. Even though the Black King is in the middle, there's no way of getting to it.)
2. This was a rematch, and people forgot he won the first match. So technically, the score is 1-1.
The interview is available on Chessbase.|
|Sep-01-13|| ||JoergWalter: Well, instead being the last chess champion who could beat a computer and thus defending the human race Kasparov became the first champion that got beaten. Maybe carrying the burden of the human race was too much. Even against bugs in the program|
|Sep-01-13|| ||JoergWalter: kasparov was able to swing a dollar out of that
|Apr-04-14|| ||tranquilsimplicity: <Serenpidity.ejd> "..or else I will switch loyalty to Karpov!"|
I honestly do not mean to be insulting or patronising nor give a lesson in philosophy but I could not resist commenting on the above words. What you have for Kasparov is not loyalty. The term loyalty implies being supportive of somebody or a group of persons you like for whatever reason, especially when they are facing difficulties; that is when the word loyalty is used with any significance.#
|Apr-13-14|| ||The Rocket: <"He gambled on the computer not knowing what to do in the opening.">|
No, he did not. Kasparov claims to have simply forgot. The theory of gambling doesn't make a hole lot of sense anyway, in this particular case, since the computers opening book missing the knight sacrifice, would not lead to any advantage at all for black. So what would Kasparov gambling about, when there's nothing to gain. Nonsense.
|Apr-15-15|| ||Paraconti: If only Kasparov hadn't lost faith in his Sicilian after his previous night's crush against Morphy!|
|Apr-15-15|| ||offramp: Playing through this game is obvious to me that Kasparov took a bung to lose. 11....b5?? That's not even funny!!|
|Mar-09-16|| ||whiteshark: Mankind is doomed:
Today the DeepMind engine <AlphaGo> beat the currently best (9 dan) Go player, Korean <Lee Se-dol> in their 1st game (of a 5 game match).
|Mar-09-16|| ||Shams: Here is a writeup of Game 1:
Sodol better load for bear next time.
|Mar-09-16|| ||Shams: From my link. Hard to read this and not think of the famous Caro-Kann debacle that cost Garry the 1997 match.|
<Everything was fairly standard until Lee played Black 7, which was quite unusual. It appeared that Lee had prepared a strange move to test AlphaGo’s strength in the opening.
Perhaps he also wanted to play an opening that AlphaGo was unlikely to have seen before, while crunching millions of positions as part of its training process.
Regardless, it wasn’t a good idea, because it didn’t lead to a good result...>
|Mar-10-16|| ||Shams: I don't speak German but it looks like 2-0 for AlphaGo.
|Mar-10-16|| ||Shams: "Yesterday, I was surprised, but today I am quite speechless," Lee said. "If you look at the way the game was played, it was a very clear loss on my part. Yesterday I felt like AlphaGo played certain problematic positions, but today I felt that AlphaGo played a near perfect game. There was not a moment I felt like its moves were unreasonable."|
|Mar-11-16|| ||whiteshark: DeepMind beats Lee Se-dol again to go 2-0 up in historic Go match: Full game http://www.youtube.com/watch?v=l-Gs...|
|Aug-16-16|| ||WorstPlayerEver: Not a clear loss. Not at all.|
|Aug-17-16|| ||WorstPlayerEver: PS I am referring to this game, not a go game.|
< Earlier Kibitzing · PAGE 14 OF 14 ·