They set the machine to play at 2-ply, i.e., it only looks two moves ahead. It's not so surprising even a good algorithm would lose in 15 moves at 2-ply.
The article mentions they also played at 5-ply and it lasted 27 and 30 moves against Kasparov, which is probably better than most casual players would do.
I do think this was a reasonable choice for the presentation because as I understand it, that's what Turing would have used when working it out with pencil and paper (and as they said, even that would take 15 minute between turns). 5-ply probably would have taken on the order of weeks or months to calculate the next move by hand.
I'm guessing it generates the tree (2 levels deep) and then uses an evaluation function to choose the best branch.
Not too far off from a modern algorithm, except that it lacks even basic optimizations like alpha-beta pruning, etc. and an incredibly fine-tuned evaluation function and a huge library of opening and closing moves.
humans can look 5-8 moves ahead with practice. GM's can probably look 15 or so moves ahead as well as study tactics of regular positions such as openings.
> Human grandmasters don’t work that way. They do not necessarily “see” the game several moves out. Indeed, they can’t — as Kasparov points out, chess is so complex that “a player looking eight moves ahead [faces] as many possible games as there are stars in the galaxy.”... “As for how many moves ahead a grandmaster sees,” as Kasparov concludes, the real answer is: “Just one, the best one.”
http://www.collisiondetection.net/mt/archives/2010/02/why_cy...
There's something I don't understand about this. My impression is that grandmasters don't play quickly (against each other). I'd be surprised if it was uncommon to spend more than five minutes on a move, for example.
So what takes so long? I assume their process is roughly "consider all possible moves, decide how good they are". So if they only look one move ahead, that sounds to me like when they evaluate a potential position, it can take more than a few seconds and they do it without reading possible future moves. Which seems unlikely.
I wrote a lesson plan that was used in a chess camp this summer, and one section covered various thought processes. Here is an excerpt from that section.
"As time goes on, your thought process will start to be based more and more on your positional intuition. Instead of explicitly weighing certain details against each other, you will instinctively be able to feel what direction the game is headed in. This intuition does not come naturally, but instead is the result of playing hundreds and thousands of chess games. Over the course of all those games, you will see positions similar to the one you are playing right now, and even though you may not explicitly internalize them, your subconsciousness will recognize the themes from before, and you will have some vague recollection or understanding of the position based on your prior experience."
The best answer to that that I know of is by http://en.wikipedia.org/wiki/Adriaan_de_Groot: "De Groot found that much of what is important in choosing a move occurs during the first few seconds of exposure to a new position. Four stages in the task of choosing the next move were noted. The first stage was the 'orientation phase', in which the subject assessed the situation and determined a very general idea of what to do next. The second stage, the 'exploration phase' was manifested by looking at some branches of the game tree. The third stage, or 'investigation phase' resulted in the subject choosing a probable best move. Finally, in the fourth stage, the 'proof phase', saw the subject confirming with him/herself that the results of the investigation were valid."
IIRC, in phase one, experts see wo or three moves to look deeper into. Non-experts see many more.
I recommend reading "het denken van den schaker" if this interests you. It is very readable by laypersons (and a translation in English exists)
I doubt this quote is to be taken literally. I rather think he means that good chess players are very, very good at deciding which moves to think deeper about based on intuition, hence reducing the need to visualize.
I think it was more of a poetic way of expressing it. That they consider many moves and their impacts, but it's based on experience, intuition, intelligence and knowledge, pruning out that bad branches quickly on, and identifying and focusing on promising ones.
He's seriously deluded. He believes that volumes of historical writings, C-14 dating, etc notwithstanding, that about a thousand years of recorded history never happened. Largely on the basis of the fact that the exponential population growth Europe has had since the British Agricultural revolution, if projected backwards in time, would say that there wouldn't have been as many people in Rome as there were.
Even a cursory understanding of, say, the trials our ancestors went through in the 1300s makes it obvious that he's wrong to project back his little exponential curve that far.
He's not the only grandmaster with some wacky ideas about history. Here's world #11, Alexander Grischuk, praising Joseph Stalin and discussing how 9/11 was a setup: http://www.whychess.org/node/514
Yep, he sure has some wacky ideas about history. Still, gotta give him credit for the fact that his wacky ideas are at least unusual -- he seems to have developed his stupid belief system for himself rather than acquiring it wholesale from someone else.
And yes, he's amazing at chess. However that is not necessarily an endorsement of his abilities at anything else. Paranoid delusions seem to be a significant occupational hazard at top levels of play. (See Bobby Fischer for a significant example.)
The brain is a pattern recognition engine well known to report many false positives. Most people are aware of that, be it often unconsciously. Chess masters learn to trust their intuition and pattern recognition, even when contrary signals are present, because, for some reason, that works in chess. It doesn't in the real world, where there are much more than 32 (the number of pieces) facts to keep track off.
The article mentions they also played at 5-ply and it lasted 27 and 30 moves against Kasparov, which is probably better than most casual players would do.