What do you think?
Rate this book
Behind Deep Blue: Building the Computer That Defeated the World Chess Champion
On May 11, 1997, as millions worldwide watched a stunning victory unfold on television, a machine shocked the chess world by defeating the defending world champion, Garry Kasparov. Written by the man who started the adventure, Behind Deep Blue reveals the inside story of what happened behind the scenes at the two historic Deep Blue vs. Kasparov matches. This is also the story behind the quest to create the mother of all chess machines. The book unveils how a modest student project eventually produced a multimillion dollar supercomputer, from the development of the scientific ideas through technical setbacks, rivalry in the race to develop the ultimate chess machine, and wild controversies to the final triumph over the world's greatest human player.
In nontechnical, conversational prose, Feng-hsiung Hsu, the system architect of Deep Blue, tells us how he and a small team of fellow researchers forged ahead at IBM with a project they'd begun as students at Carnegie Mellon in the the search for one of the oldest holy grails in artificial intelligence--a machine that could beat any human chess player in a bona fide match. Back in 1949 science had conceived the foundations of modern chess computers but not until almost fifty years later--until Deep Blue--would the quest be realized.
Hsu refutes Kasparov's controversial claim that only human intervention could have allowed Deep Blue to make its decisive, "uncomputerlike" moves. In riveting detail he describes the heightening tension in this war of brains and nerves, the "smoldering fire" in Kasparov's eyes. Behind Deep Blue is not just another tale of man versus machine. This fascinating book tells us how man as genius was given an ultimate, unforgettable run for his mind, no, not by the genius of a computer, but of man as toolmaker.
In nontechnical, conversational prose, Feng-hsiung Hsu, the system architect of Deep Blue, tells us how he and a small team of fellow researchers forged ahead at IBM with a project they'd begun as students at Carnegie Mellon in the the search for one of the oldest holy grails in artificial intelligence--a machine that could beat any human chess player in a bona fide match. Back in 1949 science had conceived the foundations of modern chess computers but not until almost fifty years later--until Deep Blue--would the quest be realized.
Hsu refutes Kasparov's controversial claim that only human intervention could have allowed Deep Blue to make its decisive, "uncomputerlike" moves. In riveting detail he describes the heightening tension in this war of brains and nerves, the "smoldering fire" in Kasparov's eyes. Behind Deep Blue is not just another tale of man versus machine. This fascinating book tells us how man as genius was given an ultimate, unforgettable run for his mind, no, not by the genius of a computer, but of man as toolmaker.
320 pages, Paperback
First published January 1, 2002
8 people are currently reading
281 people want to read
Ratings & Reviews
Friends & Following
Create a free account to discover what your friends think of this book!
Community Reviews
Displaying 1 - 24 of 24 reviews
October 6, 2009
If this book had been written forty years earlier, it would have been a typical piece of hard science-fiction. In fact, there is an SF short story by Fritz Leiber, The Sixty-Four Square Madhouse, which is quite similar. Like everyone else, I knew the bare bones of the real life version. A group of smart computer scientists at IBM built an extremely strong chess computer called Deep Blue, with the express aim of beating World Champion Garry Kasparov in a match. They played two matches, each of six games. In the first match, the machine won Game 1, but was then comprehensively outplayed by Kasparov. In the second match, the score was tied at 2.5 - 2.5, when Kasparov played a risky opening in the last game, and was crushed.
But I hadn't heard any of the details, and they were fascinating. Hsu, the author, was an extremely gifted and hard-working chip designer. In the late 80s, he was a graduate student at Carnegie Mellon University (CMU), where Hans Berliner, a former World Correspondence Chess Champion, led a team which had pioneered the idea of building custom chess chips to get extra-fast search. Berliner's machine was indeed chalking up results which were impressive for the time, but Hsu was sure he could do better. I don't know VLSI, so I couldn't follow all the details, but one important point was that he thought he could get all the important stuff on a single chip, speeding things up a great deal.
Some of the early stages provided good examples of the old saw about genius being 1% inspiration and 99% perspiration. Chips these days are almost always designed at an abstract level, with a compiler handling the concrete layout of the components. Hsu's design was very close to succeeding, but the compiler couldn't quite squeeze it all on to the surface of the chip. Most people would have given up, but Hsu spent several days working nonstop to do the layout by hand; he had to place about 40,000 components. Sort of the hardware architecture equivalent of cleaning your whole house with a toothbrush.
After a while, Hsu's chips were in business, and a team of people had come together around them. The prototype machine, dubbed Deep Thought, started getting good results. But now, at least according to Hsu, they ran into a lot of trouble with Berliner's team. First, Berliner's people rubbished Hsu's ideas; then, when it became obvious that things were going in a good direction, they tried to make out that Deep Thought was really part of their project. I am not sure whether this is the whole truth. I have read most of Berliner's papers, and greatly admire him. I met him once and talked for a couple of hours, and he seemed like a very nice guy; it doesn't sound like the kind of thing he would do. But what do I know.
At any rate, the Deep Thought gang made steady progress. Hsu kept refining the chips, working day and night; he gives you a lot of technical detail. There are some striking anecdotes. The best and scariest one is about the day he was coming back from the chip plant with a new batch. A car suddenly pulled out of a side street and drove right towards him. He had a second or two to see that a crash was inevitable. While he was recovering in hospital, a colleague asked him what his thoughts were during those two seconds. "I thought I was going to die," said Hsu, "and I was sad, because that would be the end of the project." His friend was shocked. But Hsu said that, at that point, the project was his whole life, and he really couldn't have thought anything else.
Finally, after many adventures, the team ended up playing the fateful second match against Kasparov. They were told over and over again by senior management that they had to treat their great opponent with the utmost respect. Above all, there was to be no crowing if they were fortunate enough to win. When they did indeed win the sixth game, they followed orders, though Hsu says that it was of course extremely difficult. They had worked their guts out for seven or eight years, and they couldn't even show how pleased they were to have succeeded. IBM called a huge press conference, and the CEO turned up to make a speech. Well, he said, he couldn't see what the fuss was about. This was just a chess match between the best chess player in the world and... Garry Kasparov. Evidently, the "no crowing" rule didn't apply to people high enough up the ladder.
If you're a software engineer and dream of making history, Hsu's book is obligatory reading. Highly recommended.
But I hadn't heard any of the details, and they were fascinating. Hsu, the author, was an extremely gifted and hard-working chip designer. In the late 80s, he was a graduate student at Carnegie Mellon University (CMU), where Hans Berliner, a former World Correspondence Chess Champion, led a team which had pioneered the idea of building custom chess chips to get extra-fast search. Berliner's machine was indeed chalking up results which were impressive for the time, but Hsu was sure he could do better. I don't know VLSI, so I couldn't follow all the details, but one important point was that he thought he could get all the important stuff on a single chip, speeding things up a great deal.
Some of the early stages provided good examples of the old saw about genius being 1% inspiration and 99% perspiration. Chips these days are almost always designed at an abstract level, with a compiler handling the concrete layout of the components. Hsu's design was very close to succeeding, but the compiler couldn't quite squeeze it all on to the surface of the chip. Most people would have given up, but Hsu spent several days working nonstop to do the layout by hand; he had to place about 40,000 components. Sort of the hardware architecture equivalent of cleaning your whole house with a toothbrush.
After a while, Hsu's chips were in business, and a team of people had come together around them. The prototype machine, dubbed Deep Thought, started getting good results. But now, at least according to Hsu, they ran into a lot of trouble with Berliner's team. First, Berliner's people rubbished Hsu's ideas; then, when it became obvious that things were going in a good direction, they tried to make out that Deep Thought was really part of their project. I am not sure whether this is the whole truth. I have read most of Berliner's papers, and greatly admire him. I met him once and talked for a couple of hours, and he seemed like a very nice guy; it doesn't sound like the kind of thing he would do. But what do I know.
At any rate, the Deep Thought gang made steady progress. Hsu kept refining the chips, working day and night; he gives you a lot of technical detail. There are some striking anecdotes. The best and scariest one is about the day he was coming back from the chip plant with a new batch. A car suddenly pulled out of a side street and drove right towards him. He had a second or two to see that a crash was inevitable. While he was recovering in hospital, a colleague asked him what his thoughts were during those two seconds. "I thought I was going to die," said Hsu, "and I was sad, because that would be the end of the project." His friend was shocked. But Hsu said that, at that point, the project was his whole life, and he really couldn't have thought anything else.
Finally, after many adventures, the team ended up playing the fateful second match against Kasparov. They were told over and over again by senior management that they had to treat their great opponent with the utmost respect. Above all, there was to be no crowing if they were fortunate enough to win. When they did indeed win the sixth game, they followed orders, though Hsu says that it was of course extremely difficult. They had worked their guts out for seven or eight years, and they couldn't even show how pleased they were to have succeeded. IBM called a huge press conference, and the CEO turned up to make a speech. Well, he said, he couldn't see what the fuss was about. This was just a chess match between the best chess player in the world and... Garry Kasparov. Evidently, the "no crowing" rule didn't apply to people high enough up the ladder.
If you're a software engineer and dream of making history, Hsu's book is obligatory reading. Highly recommended.
Read
June 27, 2011Wow. I think Hofstadter's thoughts (from NYT, 1996) are so way, way off on this:
May I counter with a column I wrote on my bridge blog a couple of years ago:
On the role of beauty and regret in bridge.
Does bridge meet art, and if so where?
I was recently reading a review of a book of chess games which was rather dismissive of The Evergreen Game, one of the classically famous games of the Romantic period.
Bridge doesn’t have periods, it doesn’t have famous hands or themes in the way chess does. Aesthetics don’t have the same weight in the bridge player’s mind, as far as I can tell. Coming to bridge from chess where beauty counted for so much – where one had an obligation to be aesthetically pleasing – this came to me as something of a surprise.
One of the things the reviewer had against this particular game (and presumably, therefore, the period) it was its simplicity. It has a famous sacrificial theme with attendant variations. Yet do not simplicity and beauty go hand in hand and do not we bridge players get our pleasure more of the simple moments than the complex ones? If you perform a backwash squeeze with a triple somersault and a forward tuck landing does that really stay with you the way a simple flip of a card onto the table might which is diabolically deceptive and subsequently wins you the hand?
I was wondering if the difference in part between these two scenarios that makes the latter aesthetically important and the former trivial is that the trivial, but complex one, is there for the taking. The cards lie a particular way, therefore…The simple act of deception, however, requires a creative interaction between you and the cards. Even if you don’t think of it the backwash squeeze is still there. But the one card deceptive play is only there because you create it in your mind.
That made me wonder if it affects the nature of regret in bridge. Because if you have an aesthetic duty to the game, then the regret that follows the errant path is really quite profound. You have disappointed the game as you might disappoint God. You have let down so much more than a mere partner or teammate, or a result.
Just so we have a clear understanding of what is meant by regret, I have in mind Gerald Abraham’s distinction between it and sorrow:
“Sorrow emphasises, by reluctant acceptance, the goneness of what has gone. Regret dwells on the persistent reality of what might have been…..The thought that what might have been has, in fact, not been, does not deprive the might have been of reality, in the way that the sorrowful acceptance of some present anguish exorcises the spectres of the past. Regret stays sadly and quietly with the mind; dwelling on unactual realities….
“To many chess players this experience is real – is at once acute and chronic. When the old campaigner Morry was asked by a spectator: ‘What is that player thinking about?’ he replied: ‘He’s not thinking, he’s regretting.’ But the person referred to was living in a real world: the world in which he moved the other Rook; the world in which he did not foolishly capture the Pawn, or foolishly refuse the Pawn, as the case may be….
“[The] chess player does not accept as valid a rigid distinction between the ‘actually is’ and the ‘never was.’ He does not accept, in other words, the unreality of possibility, including the possibilities of the past. To the logician the whole issue is too easy: true or false. It is just untrue to say: ‘I moved the King’s Rook,’ when in fact you moved the Queen’s Rook. And if you say: ‘But what if I feel as if I had moved the King’s Rook?’ he will reply, if he is a modern: ‘From a false proposition, all absurdities follow.’ But the chess player will not be convinced, for he has lived in a dimension of reality which professional philosophers ignore or pretend to ignore. His regrets accompany him in a dimension of thought which even the idealists find hard to recognise….
“Most taking of all words: it might have been. I do not know whether the philosopher Bergson was a chess player…..But if I ever meet Bergson in the shades, I’ll tell him that the chess players of the world know the real meaning of The Reality of Time. I do not refer to Zietnot which is time mechanised and formalised, and only a clog to the creative spirit. I refer to the richness of time, with all its possible dimensions which are the dimensions of possibility.”
To this I add: only a weak and ignorant player could think that chess ‘doesn’t have deep emotional qualities to it, mortality, resignation, joy, all the things that music deals with.’ Whether or not a computer can beat a human at chess will never detract from that. And maybe, if we want to define the intelligent quality which a human has that a machine will never ever have, it is the one Abrahams so poignantly captures of ‘regret.’
Several cognitive scientists said Deep Blue's victory in the opening game of the recent match told more about chess than about intelligence.
"It was a watershed event, but it doesn't have to do with computers becoming intelligent," said Douglas Hofstadter, a professor of computer science at Indiana University and author of several books about human intelligence, including "Godel, Escher, Bach," which won a Pulitzer Prize in 1980, with its witty argument about the connecting threads of intellect in various fields of expression. "They're just overtaking humans in certain intellectual activities that we thought required intelligence. My God, I used to think chess required thought. Now, I realize it doesn't. It doesn't mean Kasparov isn't a deep thinker, just that you can bypass deep thinking in playing chess, the way you can fly without flapping your wings."
Those who ascribe to the theory that machines are just machines that will always be apprenticed to human masters tend to view the hoopla over the chess match and the worry over the ascension of the machine as, well, "crazy," to use the word of Berkeley's Searle.
"It's just a hunk of junk that somebody's designed," he said of Deep Blue.
Paul Saffo, a technology expert at the Institute for the Future, a think tank in Menlo Park, Calif., more or less agreed. "People who fear machines don't need to lose any sleep just yet," he said. "To me, the match was interesting as a cultural event. Chess, whereas it's a difficult problem to solve for computer scientists, is just a constrained formal problem. O.K., a computer beat a grandmaster, but computers aren't any smarter than they were the day before. The question I'd ask, now that this Rubicon has been passed, is What's the new arbitrary measure? Maybe it's a computer that plays go," -- the ancient Japanese board game that has yet to be conquered by a machine -- "or a computer that can fill in an IRS form without getting an audit."
Hofstadter has another idea. He says it is not impossible that a machine might one day learn from experiences and feel things as humans do. But in the meantime, that is still what separates us and them.
In "Godel, Escher, Bach" he held chess-playing to be a creative endeavor with the unrestrained threshold of excellence that pertains to arts like musical composition or literature. Now, he says, the computer gains of the last decade have persuaded him that chess is not as lofty an intellectual endeavor as music and writing; they require a soul.
"I think chess is cerebral and intellectual," he said, "but it doesn't have deep emotional qualities to it, mortality, resignation, joy, all the things that music deals with. I'd put poetry and literature up there, too. If music or literature were created at an artistic level by a computer, I would feel this is a terrible thing."
May I counter with a column I wrote on my bridge blog a couple of years ago:
On the role of beauty and regret in bridge.
Does bridge meet art, and if so where?
I was recently reading a review of a book of chess games which was rather dismissive of The Evergreen Game, one of the classically famous games of the Romantic period.
Bridge doesn’t have periods, it doesn’t have famous hands or themes in the way chess does. Aesthetics don’t have the same weight in the bridge player’s mind, as far as I can tell. Coming to bridge from chess where beauty counted for so much – where one had an obligation to be aesthetically pleasing – this came to me as something of a surprise.
One of the things the reviewer had against this particular game (and presumably, therefore, the period) it was its simplicity. It has a famous sacrificial theme with attendant variations. Yet do not simplicity and beauty go hand in hand and do not we bridge players get our pleasure more of the simple moments than the complex ones? If you perform a backwash squeeze with a triple somersault and a forward tuck landing does that really stay with you the way a simple flip of a card onto the table might which is diabolically deceptive and subsequently wins you the hand?
I was wondering if the difference in part between these two scenarios that makes the latter aesthetically important and the former trivial is that the trivial, but complex one, is there for the taking. The cards lie a particular way, therefore…The simple act of deception, however, requires a creative interaction between you and the cards. Even if you don’t think of it the backwash squeeze is still there. But the one card deceptive play is only there because you create it in your mind.
That made me wonder if it affects the nature of regret in bridge. Because if you have an aesthetic duty to the game, then the regret that follows the errant path is really quite profound. You have disappointed the game as you might disappoint God. You have let down so much more than a mere partner or teammate, or a result.
Just so we have a clear understanding of what is meant by regret, I have in mind Gerald Abraham’s distinction between it and sorrow:
“Sorrow emphasises, by reluctant acceptance, the goneness of what has gone. Regret dwells on the persistent reality of what might have been…..The thought that what might have been has, in fact, not been, does not deprive the might have been of reality, in the way that the sorrowful acceptance of some present anguish exorcises the spectres of the past. Regret stays sadly and quietly with the mind; dwelling on unactual realities….
“To many chess players this experience is real – is at once acute and chronic. When the old campaigner Morry was asked by a spectator: ‘What is that player thinking about?’ he replied: ‘He’s not thinking, he’s regretting.’ But the person referred to was living in a real world: the world in which he moved the other Rook; the world in which he did not foolishly capture the Pawn, or foolishly refuse the Pawn, as the case may be….
“[The] chess player does not accept as valid a rigid distinction between the ‘actually is’ and the ‘never was.’ He does not accept, in other words, the unreality of possibility, including the possibilities of the past. To the logician the whole issue is too easy: true or false. It is just untrue to say: ‘I moved the King’s Rook,’ when in fact you moved the Queen’s Rook. And if you say: ‘But what if I feel as if I had moved the King’s Rook?’ he will reply, if he is a modern: ‘From a false proposition, all absurdities follow.’ But the chess player will not be convinced, for he has lived in a dimension of reality which professional philosophers ignore or pretend to ignore. His regrets accompany him in a dimension of thought which even the idealists find hard to recognise….
“Most taking of all words: it might have been. I do not know whether the philosopher Bergson was a chess player…..But if I ever meet Bergson in the shades, I’ll tell him that the chess players of the world know the real meaning of The Reality of Time. I do not refer to Zietnot which is time mechanised and formalised, and only a clog to the creative spirit. I refer to the richness of time, with all its possible dimensions which are the dimensions of possibility.”
To this I add: only a weak and ignorant player could think that chess ‘doesn’t have deep emotional qualities to it, mortality, resignation, joy, all the things that music deals with.’ Whether or not a computer can beat a human at chess will never detract from that. And maybe, if we want to define the intelligent quality which a human has that a machine will never ever have, it is the one Abrahams so poignantly captures of ‘regret.’
August 18, 2023
This is a personal account of what a chance encounter within the University and approaching a problem without encroached views can lead to. A very interesting narrative of how in the end the World Champion was beaten by a computer program that provides, even to the computer uninitiated a view of what it takes. Read this if you are interested to dispell in your mind theories and stupid assumptions of what Deep Blue was and understand what it actually was, and why there was no need for a newer one.
October 4, 2015
While this is an interesting book on the 'match of the century' and the development of the Deep Blue computer, it doesn't really address the problems with the match itself. There is a documentary, "Game Over - Kasparov and the Machine", that discusses the "questionable gamesmanship" of the IBM team (to put it nicely), namely, having approximately a dozen grandmasters in attendance in the locked and secured computer room where the real work of the computer hardware was happening. The team member and the monitor out front on stage across from Kasparov was only a "dummy" terminal so to speak. The hardware was in a locked and guarded back room and this is where the moves were processed and made. (The IBM representative on stage only moved the physical chess piece according to what appeared on the monitor next to him.)
So why were there so many grandmasters hanging out in the machine-room? The only humans who need be in attendance there were IBM technicians. This is where the major problem begins with this, the second match.
All that was needed was someone to make the move for the computer and then the computer would automatically "think" and indicate its next move. The attendance of a group of grandmasters in that room can only lead one to suspect that a discussion was going on about the move and upcoming moves that Deep Blue proposed. A grandmaster, or even an international master or national master for that matter, with the help of Deep Blue (or even a very good commercial chess computer program such as Fritz) could beat a top player when working together.
The first game of the second match saw Kasparov easily defeat Deep Blue. The machine made many poor moves, what Kasparov described as "computer moves". Then suddenly in the second game, Deep Blue's expertise increased appreciably and it won the game. The next three games were draws and then Deep Blue won the final and deciding game.
How could the computer increase in skill so dramatically and almost literally overnight? And why were printouts of Deep Blue's move analysis never provided (as promised) following the match? IBM simply dismantled the machine, sending half of it to the Smithsonian and the other half being disassembled back at IBM.
Why dismantle the first machine to beat a grandmaster? IBM spent millions (I assume ... it was their "own" development costs but it couldn't have been cheap) on the development and seemed to think of it as nothing more than disposable, outdated digital hardware.
The reason that makes sense is that Deep Blue did its analysis and the group of grandmasters present in the secluded back room then analyzed Deep Blue's suggestions from there. If they saw a better move or line among the ones proposed by Deep Blue as it was "thinking", then they simply made that move rather than the move "honestly" compiled by Deep Blue.
The fact that Deep Blue played like a "patzer" in the first game and then overnight, with some sort of "tweak" by the IBM technicians, it suddenly started to play at a world class level just does not make sense. The technicians had every game Kasparov ever played (as well as millions of other games) to uae as a database while "teaching" Deep Blue as well as for an onboard database (or "book" as it is usually called for a computer chess program) for the machine to use during the match. Then why wasn't Deep Blue ready right off the mark in the first game?
A group of grandmasters assessing and assisting the computer is the most obvious answer.
In other words, IBM cheated and then immediately dismantled the machine so that there would be no future matches by which someone might figure out just how it was that they beat the highest-rated chess player (at the time) in the history of the game.
While computers now are be able to play at the top level this was nearly twenty years ago and that is a lifetime as far at digital technology development is concerned. Deep Blue's amazing jump in ability during this match was far too big in far too short a time.
Of course, these questions are not going to be discussed in a book which is basically one of those "look how smart I am" PR scripts and tells only IBM's side of the story.
My apologies if I am a bit terse but cheating is why I stopped watching sports also (steroids, human growth hormone, performance-enhancing drugs of all sorts). I don't know how these people live with themselves and how they can stand up in front of millions of people and lie and lie and lie about not cheating (think of Lance Armstrong). The human race would be a far better species if the natural ability of lying could be taken out of our DNA.
So why were there so many grandmasters hanging out in the machine-room? The only humans who need be in attendance there were IBM technicians. This is where the major problem begins with this, the second match.
All that was needed was someone to make the move for the computer and then the computer would automatically "think" and indicate its next move. The attendance of a group of grandmasters in that room can only lead one to suspect that a discussion was going on about the move and upcoming moves that Deep Blue proposed. A grandmaster, or even an international master or national master for that matter, with the help of Deep Blue (or even a very good commercial chess computer program such as Fritz) could beat a top player when working together.
The first game of the second match saw Kasparov easily defeat Deep Blue. The machine made many poor moves, what Kasparov described as "computer moves". Then suddenly in the second game, Deep Blue's expertise increased appreciably and it won the game. The next three games were draws and then Deep Blue won the final and deciding game.
How could the computer increase in skill so dramatically and almost literally overnight? And why were printouts of Deep Blue's move analysis never provided (as promised) following the match? IBM simply dismantled the machine, sending half of it to the Smithsonian and the other half being disassembled back at IBM.
Why dismantle the first machine to beat a grandmaster? IBM spent millions (I assume ... it was their "own" development costs but it couldn't have been cheap) on the development and seemed to think of it as nothing more than disposable, outdated digital hardware.
The reason that makes sense is that Deep Blue did its analysis and the group of grandmasters present in the secluded back room then analyzed Deep Blue's suggestions from there. If they saw a better move or line among the ones proposed by Deep Blue as it was "thinking", then they simply made that move rather than the move "honestly" compiled by Deep Blue.
The fact that Deep Blue played like a "patzer" in the first game and then overnight, with some sort of "tweak" by the IBM technicians, it suddenly started to play at a world class level just does not make sense. The technicians had every game Kasparov ever played (as well as millions of other games) to uae as a database while "teaching" Deep Blue as well as for an onboard database (or "book" as it is usually called for a computer chess program) for the machine to use during the match. Then why wasn't Deep Blue ready right off the mark in the first game?
A group of grandmasters assessing and assisting the computer is the most obvious answer.
In other words, IBM cheated and then immediately dismantled the machine so that there would be no future matches by which someone might figure out just how it was that they beat the highest-rated chess player (at the time) in the history of the game.
While computers now are be able to play at the top level this was nearly twenty years ago and that is a lifetime as far at digital technology development is concerned. Deep Blue's amazing jump in ability during this match was far too big in far too short a time.
Of course, these questions are not going to be discussed in a book which is basically one of those "look how smart I am" PR scripts and tells only IBM's side of the story.
My apologies if I am a bit terse but cheating is why I stopped watching sports also (steroids, human growth hormone, performance-enhancing drugs of all sorts). I don't know how these people live with themselves and how they can stand up in front of millions of people and lie and lie and lie about not cheating (think of Lance Armstrong). The human race would be a far better species if the natural ability of lying could be taken out of our DNA.
January 22, 2012
Feng-hsiung Hsu was born in 1959 in Taiwan. After graduating from National Taiwan University with a B.S. in Electrical Engineering, he served in the ROC Army for two years, and then was sent to the United States for graduate studies at Carnegie-Mellon University. For his thesis, Hsu was going to design a chip for improved printing of Chinese characters by laser printers; however, he went to a seminar about a specialized 64-chip circuit board for a chess-playing computer.and was approached by another professor who asked him to improve the design. Hsu realized that you don't need so many chips: you can simplify the board drastically and only have one chip! In 1985 the new chip was made; a program built around it became one of the world's best computer chess players. After graduating, Hsu was hired by IBM Research, where he built a series of successively faster specialized-hardware chess computers that beat human chess players of successively higher ranking. In 1996 one of these computers lost a match to Garry Kasparov, then the World Chess Champion, winning one game, drawing in two, and losing three. A computer called Deep Blue won a 1997 rematch, winning two, drawing in three, and losing one. This was the culmination of 12 years of Hsu's life, 7 years of one of his coworkers', and 5 years of another's. After that, games between human chess champions and computers would have made as much sense as strength competitions between human weightlifters and excavators. If Deep Blue were recreated today, it would have been even more powerful, given that its specialized chips were built using 0.6 micron CMOS technology, 20-30 times the feature size of today's microprocessors; Hsu is now building a similar machine for Go. I am always happy to read about people who pursued their dreams and succeeded.
December 9, 2015
Beyond Deep Blue is an interesting book with a first-hand look at the trials and tribulations of Deep Blue. The main appeal of this book is its author, Feng-Hsiung Hsu, who was of the lead developers of the Deep Blue computer. Hsu tells of Deep Blue's genesis as the computer Deep Thought, and the long road to beating the world chess champion. The author maintains an air of humor that is a breath of fresh air next to the sometimes stifling chess lingo. However, the author is a self-admitted amateur at chess, and sometimes stumbles over his commentary for the matches Deep Blue played. Despite these minor problems, this book is quite engrossing, and I would recommend it to all who are interested in Chess, Computers, or both.
August 10, 2019
A thoroughly engaging and candid account
Taiwanese-born Feng-Hsiung Hsu has written a most engaging and readable account of how Deep Blue came to be, and how it defeated world chess champion Garry Kasparov in perhaps the greatest chess match of all time. I say "perhaps" because there are many who still consider the 1972 encounter at Reykjavik, Iceland between Bobby Fischer and Boris Spassky to be the greatest match ever. One thing both matches had in common, in addition to a worldwide audience, is two deeply suspicious and idiosyncratic geniuses, Kasparov and Fischer.
However, while Fischer's triumph rejuvenated interest in chess, especially in the US, Kasparov's defeat, many fear, may have rung the death knell for the ancient game. Before Deep Blue's victory, it was easy to imagine that the human mind was light-years ahead of any artificial intelligence. After Kasparov slunk off mumbling vague charges of human intervention ("cheating"), it became necessary to face the possibility that machine intelligence was on its way to exceeding that of humans.
But what did the match really prove? According to Hsu himself, the triumph of Deep Blue "might be the more important human achievement when all was said and done." (p. 256) By a "more important human achievement," he means, more important than the one that would have been Kasparov's had he won.
This I think is the crux of the matter. Deep Blue, an IBM computer of enormous power, is the product of human minds and human engineering. Look at it this way: as computers become more and more powerful and their algorithms become more and more sophisticated, there will no thought at all that a human might compete with them at chess. It would be like expecting the world's fastest human to beat a motor car in a race. Or for the world's best human calculator to add numbers faster than a personal computer.
In a deeper sense what was destroyed by this match was not human intellectual superiority but the delusion that somehow a board game--even the greatest board game ever invented--is a true measure of human intelligence. Quite simply, the ability to play chess at the highest level is only one talent, similar to (but different from) the ability to play the violin or to run fast. More significant is the greater human ability to conceive and build a machine that does something better than humans can do themselves.
Hsu's account includes a lot of information about his personal adventures in academia and the corporate structure, including rivalries with others in the race to build the ultimate chess-playing computer. He is candid, and self-revelatory to a surprising degree, and it is this candor that helps to make this a fascinating read, not only for computer specialists and chess players, but for anyone interested in how the human competitive spirit works. His portrait of Garry Kasparov--perhaps the strongest chess player of all time--captures the arrogant, suspicious genius at his most human and makes it clear how he came to lose a match he fully expected to win.
Ah, the match itself! The book includes the moves of the games in an appendix, but one can readily see that the match turned on two very strange decisions by the hitherto nearly invincible Kasparov. Strange to say, it appears that Kasparov lost the match mainly because of poor psychological decisions. In game two, believing that he was lost, mainly because he believed that the computer would not have made the move it had made had there been a perpetual check available to the human player that would have drawn the game, Kasparov resigned. However, the machine had erred, and there was a way to draw the game. Against a human opponent, I believe that Kasparov would have closely investigated that line and found the drawing resource.
In the final game again Kasparov made a decision based on what he thought was the nature of the way computers play chess. He allowed a sacrificial line as Black in the Caro-Kahn Defense, a line that he believed Deep Blue would never play since computers are notoriously bad at figuring out how to conduct a complicated attack. Indeed, commercial chess software for PCs typically exclude this line from their opening repertoire so as not to burden the program! So Kasparov thought in playing 7... h6 that Deep Blue would retreat its knight giving Kasparov easy equality. Instead Deep Blue plunged in with 8. Nxe6! Eleven moves later Kasparov resigned--easily one of the quickest defeats of his career.
So, with better decisions, based on sound chess and NOT on mistaken preconceptions about Deep Blue's prowess, Kasparov might have won the match. However, the irony is that it is unlikely that there ever will be another match between the world chess champion and a machine simply because Kasparov and the whole chess world know that the ultimate victory of machine over man, in the arcane test of will and calculation that is chess, is inevitable. But what we also know is that it doesn't matter. We still hold races between humans even though our machines can easily out distant them. And humans will continue to play chess even though they would have no chance against a computer because chess is first and foremost a human sporting event, a test of mental strength and skill much as a boxing match is a test of physical strength and skill.
--Dennis Littrell, author of “The World Is Not as We Think It Is”
Taiwanese-born Feng-Hsiung Hsu has written a most engaging and readable account of how Deep Blue came to be, and how it defeated world chess champion Garry Kasparov in perhaps the greatest chess match of all time. I say "perhaps" because there are many who still consider the 1972 encounter at Reykjavik, Iceland between Bobby Fischer and Boris Spassky to be the greatest match ever. One thing both matches had in common, in addition to a worldwide audience, is two deeply suspicious and idiosyncratic geniuses, Kasparov and Fischer.
However, while Fischer's triumph rejuvenated interest in chess, especially in the US, Kasparov's defeat, many fear, may have rung the death knell for the ancient game. Before Deep Blue's victory, it was easy to imagine that the human mind was light-years ahead of any artificial intelligence. After Kasparov slunk off mumbling vague charges of human intervention ("cheating"), it became necessary to face the possibility that machine intelligence was on its way to exceeding that of humans.
But what did the match really prove? According to Hsu himself, the triumph of Deep Blue "might be the more important human achievement when all was said and done." (p. 256) By a "more important human achievement," he means, more important than the one that would have been Kasparov's had he won.
This I think is the crux of the matter. Deep Blue, an IBM computer of enormous power, is the product of human minds and human engineering. Look at it this way: as computers become more and more powerful and their algorithms become more and more sophisticated, there will no thought at all that a human might compete with them at chess. It would be like expecting the world's fastest human to beat a motor car in a race. Or for the world's best human calculator to add numbers faster than a personal computer.
In a deeper sense what was destroyed by this match was not human intellectual superiority but the delusion that somehow a board game--even the greatest board game ever invented--is a true measure of human intelligence. Quite simply, the ability to play chess at the highest level is only one talent, similar to (but different from) the ability to play the violin or to run fast. More significant is the greater human ability to conceive and build a machine that does something better than humans can do themselves.
Hsu's account includes a lot of information about his personal adventures in academia and the corporate structure, including rivalries with others in the race to build the ultimate chess-playing computer. He is candid, and self-revelatory to a surprising degree, and it is this candor that helps to make this a fascinating read, not only for computer specialists and chess players, but for anyone interested in how the human competitive spirit works. His portrait of Garry Kasparov--perhaps the strongest chess player of all time--captures the arrogant, suspicious genius at his most human and makes it clear how he came to lose a match he fully expected to win.
Ah, the match itself! The book includes the moves of the games in an appendix, but one can readily see that the match turned on two very strange decisions by the hitherto nearly invincible Kasparov. Strange to say, it appears that Kasparov lost the match mainly because of poor psychological decisions. In game two, believing that he was lost, mainly because he believed that the computer would not have made the move it had made had there been a perpetual check available to the human player that would have drawn the game, Kasparov resigned. However, the machine had erred, and there was a way to draw the game. Against a human opponent, I believe that Kasparov would have closely investigated that line and found the drawing resource.
In the final game again Kasparov made a decision based on what he thought was the nature of the way computers play chess. He allowed a sacrificial line as Black in the Caro-Kahn Defense, a line that he believed Deep Blue would never play since computers are notoriously bad at figuring out how to conduct a complicated attack. Indeed, commercial chess software for PCs typically exclude this line from their opening repertoire so as not to burden the program! So Kasparov thought in playing 7... h6 that Deep Blue would retreat its knight giving Kasparov easy equality. Instead Deep Blue plunged in with 8. Nxe6! Eleven moves later Kasparov resigned--easily one of the quickest defeats of his career.
So, with better decisions, based on sound chess and NOT on mistaken preconceptions about Deep Blue's prowess, Kasparov might have won the match. However, the irony is that it is unlikely that there ever will be another match between the world chess champion and a machine simply because Kasparov and the whole chess world know that the ultimate victory of machine over man, in the arcane test of will and calculation that is chess, is inevitable. But what we also know is that it doesn't matter. We still hold races between humans even though our machines can easily out distant them. And humans will continue to play chess even though they would have no chance against a computer because chess is first and foremost a human sporting event, a test of mental strength and skill much as a boxing match is a test of physical strength and skill.
--Dennis Littrell, author of “The World Is Not as We Think It Is”
October 19, 2009
Amazon third party, 2009-10-08. Suffers from Smoot's Disease (see: Wrinkles in Time), but has its strong moments.
July 10, 2023
Behind Deep Blue tells the story of the team that built the first chess computer that managed to beat the world chess champion. The book recounts the various chess AI competitions and encounters with strong human chess players the team went through and how they learned from their experiences. Central themes of the book are hardware and software development, chip manufacturing, artificial intelligence, team work, personal sacrifices and last, but not least, chess.
While Behind Deep Blue is not a chess book, it does cover chess extensively. To cater for chess beginners, the book does explain everything that is needed to follow the events. However, a reader who is already familiar with concepts such as rooks on open files and bishop endgames will have an easier time to just enjoy the excitement of the games.
Apart from chess, knowledge about VLSI design and manufacturing will make Behind Deep Blue more intriguing. The author, Feng-Hsiung Hsu, was responsible of the VLSI design of the chess chips from the earliest prototypes all the way until the end. In the book, the parts about VLSI design can definitely be skipped by a reader who either has no interest or no expertise in the subject. While the team did the VLSI design themselves, the team used contractors to manufacture the chips so there's not much more to it than reports of the yields.
Most computer scientists will find it interesting that Ken Thompson had a non-trivial role in the development of chess AI's. Mr Thompson's was part of a team that built a chess computer called Belle, which was strong contender in chess AI competitions in late 1970's and early 1980's. The design of Belle was also the starting point that Mr Hsu used when he started working on chess computers.
Creating the first chess computer capable of beating the world chess champion was something many scientists aspired to do. It is natural that with success and glory, you will also manage to create a few adversaries. The book covers also this aspect.
Finally, in Behind Deep Blue we learn about long working hours Mr Hsu and others had to do. Also, skilled hardware and software developers were in high demand. Many of the early contributors in the team took up on the lucrative offers and left for greener pastures. As such, one of the many themes in the book is about weighing money against glory.
The book is well-written and an exciting read for anyone who is interested in the various themes presented. Personally, I'm a huge fan of spectator sports and chess. Reading about all the competitions was interesting and I felt like I could relive the events as an observer.
On a very personal note, some of the most interesting content for me was the VLSI design work. When I studied at the University, I focused somewhat extensively on embedded systems and VLSI design. However, the global financial crisis of 2008 hit the market in Finland quite hard and I was unable to find a job in the field. Therefore, although I have a lot of theoretical expertise, I have never been able to work with VLSI chips myself. Hence, the VLSI content in Behind Deep Blue was exceptionally interesting to me.
While Behind Deep Blue is not a chess book, it does cover chess extensively. To cater for chess beginners, the book does explain everything that is needed to follow the events. However, a reader who is already familiar with concepts such as rooks on open files and bishop endgames will have an easier time to just enjoy the excitement of the games.
Apart from chess, knowledge about VLSI design and manufacturing will make Behind Deep Blue more intriguing. The author, Feng-Hsiung Hsu, was responsible of the VLSI design of the chess chips from the earliest prototypes all the way until the end. In the book, the parts about VLSI design can definitely be skipped by a reader who either has no interest or no expertise in the subject. While the team did the VLSI design themselves, the team used contractors to manufacture the chips so there's not much more to it than reports of the yields.
Most computer scientists will find it interesting that Ken Thompson had a non-trivial role in the development of chess AI's. Mr Thompson's was part of a team that built a chess computer called Belle, which was strong contender in chess AI competitions in late 1970's and early 1980's. The design of Belle was also the starting point that Mr Hsu used when he started working on chess computers.
Creating the first chess computer capable of beating the world chess champion was something many scientists aspired to do. It is natural that with success and glory, you will also manage to create a few adversaries. The book covers also this aspect.
Finally, in Behind Deep Blue we learn about long working hours Mr Hsu and others had to do. Also, skilled hardware and software developers were in high demand. Many of the early contributors in the team took up on the lucrative offers and left for greener pastures. As such, one of the many themes in the book is about weighing money against glory.
The book is well-written and an exciting read for anyone who is interested in the various themes presented. Personally, I'm a huge fan of spectator sports and chess. Reading about all the competitions was interesting and I felt like I could relive the events as an observer.
On a very personal note, some of the most interesting content for me was the VLSI design work. When I studied at the University, I focused somewhat extensively on embedded systems and VLSI design. However, the global financial crisis of 2008 hit the market in Finland quite hard and I was unable to find a job in the field. Therefore, although I have a lot of theoretical expertise, I have never been able to work with VLSI chips myself. Hence, the VLSI content in Behind Deep Blue was exceptionally interesting to me.
June 24, 2020
Very interesting. A lot of details on the projects that from 1985 to 1997 (and some times later) made deep blue possible. Many times they were short of time (as in every non trivial projects) and a lot of myths about deep blue and the possible third match with Kasparov are solved.
Also incredible how players playing against computers barely understand how some chess engines works, although the people are very good in chess. Impressive how one can be very good in a field, while misunderstanding another field that is very similar.
One thing is that the author, he himself should have known a bit better, dismisses the improvements of the evaluation function of commercial programs after deep blue. Thinking that the evaluation function of deep blue was the best. Surely it was very complicated (8000 terms), but not necessarily the most effective.
Moreover if one searches online there are more indepth technical articles about deep blue chips. In one is stated that on average deep blue searched on averare 220 million positions per second, a peak of 300 million and a theoretical peak (never reached on sustained computation) of 1 billion.
Nowadays chess engines are several classes of strength stronger than deep blue, evaluate way less positions, and thus they have a much better evaluation function than the one available for Deep Blue.
Oh another thing "in swiss tournaments the strongest player has only 50% chances to win it", no, it doesn't work in this way (easy example: a strong master vs a field of novices. The probability is close to 100%). I wonder whether the author put this statement, several times, to increase the dramatic effect, or whether he never paid attention to the mathematics behind it (because the author is clearly mathematically apt).
Really nice, it is a shame that humans attain a certain knowledge, that then is locked behind secrecy and has to be rediscovered. I wonder how much are we slowed down by this, what a pity. Maybe we will learn one day.
Also incredible how players playing against computers barely understand how some chess engines works, although the people are very good in chess. Impressive how one can be very good in a field, while misunderstanding another field that is very similar.
One thing is that the author, he himself should have known a bit better, dismisses the improvements of the evaluation function of commercial programs after deep blue. Thinking that the evaluation function of deep blue was the best. Surely it was very complicated (8000 terms), but not necessarily the most effective.
Moreover if one searches online there are more indepth technical articles about deep blue chips. In one is stated that on average deep blue searched on averare 220 million positions per second, a peak of 300 million and a theoretical peak (never reached on sustained computation) of 1 billion.
Nowadays chess engines are several classes of strength stronger than deep blue, evaluate way less positions, and thus they have a much better evaluation function than the one available for Deep Blue.
Oh another thing "in swiss tournaments the strongest player has only 50% chances to win it", no, it doesn't work in this way (easy example: a strong master vs a field of novices. The probability is close to 100%). I wonder whether the author put this statement, several times, to increase the dramatic effect, or whether he never paid attention to the mathematics behind it (because the author is clearly mathematically apt).
Really nice, it is a shame that humans attain a certain knowledge, that then is locked behind secrecy and has to be rediscovered. I wonder how much are we slowed down by this, what a pity. Maybe we will learn one day.
December 22, 2025
Can’t stress in enough how much I loved this book - it’s a blend of chess, human emotions, experimentation and sheer will.
It is my favourite genre of chess book - it has barely any diagrams or calculated variations; just the raw story around it.
The entire story is not man vs machine but instead it’s humanity through natural intelligence vs humanity through tools.
It’s very heavy focused on the physical hardware of the chips themselves as well as the computational limitations which I do not care about (EEE people would love it) but the pure determination to achieve his goal is so impressive and inspiring.
The author himself is very interesting with most of his personal life and stories relegated to the appendix and his navigation of the human elements of this journey is real. He’s clearly a genius but without any pompousness or arrogance - he’s truly honest and his disappointment at how Kasparov reacts + his comments after aren’t fake humble or directly aggressive (even though he’d have the right to be).
It’s got the perfect middle ground of technical detail, chess information and human character that made it a delicious read, I truly munched this book in a way I haven’t for years.
Reading it in 2025 is also interesting - more than 20 years on and also in an era where the concept of a machine beating a human in a match is stupidly obvious. With AlphaGo and the neural network revolution in chess, this book is like reading about Diners club -> it won’t offer any practical information for how to build a powerful chess engine. However it sheds so much light on what it takes to be able to build it from a human perspective and at the end of the day that’s the most important thing.
It is my favourite genre of chess book - it has barely any diagrams or calculated variations; just the raw story around it.
The entire story is not man vs machine but instead it’s humanity through natural intelligence vs humanity through tools.
It’s very heavy focused on the physical hardware of the chips themselves as well as the computational limitations which I do not care about (EEE people would love it) but the pure determination to achieve his goal is so impressive and inspiring.
The author himself is very interesting with most of his personal life and stories relegated to the appendix and his navigation of the human elements of this journey is real. He’s clearly a genius but without any pompousness or arrogance - he’s truly honest and his disappointment at how Kasparov reacts + his comments after aren’t fake humble or directly aggressive (even though he’d have the right to be).
It’s got the perfect middle ground of technical detail, chess information and human character that made it a delicious read, I truly munched this book in a way I haven’t for years.
Reading it in 2025 is also interesting - more than 20 years on and also in an era where the concept of a machine beating a human in a match is stupidly obvious. With AlphaGo and the neural network revolution in chess, this book is like reading about Diners club -> it won’t offer any practical information for how to build a powerful chess engine. However it sheds so much light on what it takes to be able to build it from a human perspective and at the end of the day that’s the most important thing.
April 19, 2023
This book was really good, it was about the guy and the team who made the chess computer, deep blue, which beat Garry Kasparov in a chess match. They went to college over in Pittsburg at Carnegie Mellon, which is cool. They really went through a lot with this computer and this book details everything really well and makes it readable to someone who does not know a lot about computers.
4.5/5
4.5/5
August 18, 2025
Leí esto como parte de una investigación. A pesar de ser muy técnico en múltiples aspectos, reconozco la habilidad de Fen-Hsiung Hsu para mantener al lector en tensión de lo que sucederá a continuación. Es curioso leerlo en una época en la que la IA se ha vuelto parte de la vida diaria, sin embargo, queda la reflexión de Hsu sobre el humano como intérprete vs el humano como creador.
July 29, 2022
Would’ve been able to follow better if I was a little bit smarter
September 30, 2025
2002年，「深藍」團隊領導者許峰雄執筆的深藍發展史。當年的許峰雄只追求硬體速度，且人工智能的前景渺茫。於今想來，不知AI的突破點為何？
December 18, 2025
Disappointing. No index, no scholarship, no tools to discern the "how." No defense nor absolution for IBM against the charge of cheating. The topic is great; timely as heck with AI bearing down, but the work is not. It is a puff piece.
August 10, 2016
This is the story of how humans built a machine that could outthink humans, at least in a narrow field.
The author was one of the creators of Deep Blue, the chess machine that beat World Champion Grandmaster Garry Kasparov in 1997. This book is the story of how the machine came to be.
One of the themes of the book is the struggle between man in his own abilities and man as tool maker. The insecurity of man when he considers his tools is evident in a lot of literature -- particularly science fiction -- and movies. Time after time we are confronted with stories of robots run amok, often threatening to kill men or mankind.
In contrast, this book is a celebration of man's ability to create. How can we create great things when we are afraid of our creations?
The author was one of the creators of Deep Blue, the chess machine that beat World Champion Grandmaster Garry Kasparov in 1997. This book is the story of how the machine came to be.
One of the themes of the book is the struggle between man in his own abilities and man as tool maker. The insecurity of man when he considers his tools is evident in a lot of literature -- particularly science fiction -- and movies. Time after time we are confronted with stories of robots run amok, often threatening to kill men or mankind.
In contrast, this book is a celebration of man's ability to create. How can we create great things when we are afraid of our creations?
April 23, 2012
Fantastic story about the underlying history culminating in the first machine able to defeat the world champion Gary Kasparov in tournament-style chess. Although I have some familiarity with computer science and chess, neither is required to enjoy this book. I also had no idea that Joel Benjamin assisted the IBM team with the project. Lots of details, and really interesting to note that the principal architect 許峰雄 is Taiwanese.
March 17, 2013
Hsu isn't exactly a gifted wordsmith, and it would be nice to hear a little more about his ethos as a toolmaker and his thoughts on what was happening, and a little less about each and every single move from a particular chess game, but this remains a fascinating, enjoyably unvarnished look at the day-to-day machinations of pushing technology forward.
April 9, 2007
The author is much too self-congratulatory for my taste, and the creation of a world-class chess-playing computer is less interesting than one might think. I finished the book with a feeling of "so what?".
Read
June 24, 2007 Just started reading it -- looks interesting.
June 30, 2008
Makes you think that you migth actually undertsand how geniuses create a computer strong enough to beat the Chess World Champion. I liked reading about the battle of egos most of all.
May 4, 2011
This was very well-written. I was introduced to the world of computer chess. Gary Kasparov behaved like a bad sport in the end. I think he should have lost with some class.
February 1, 2021
Great book
Displaying 1 - 24 of 24 reviews