It is only in retrospect that the HAL's breakdown can be seen as the inevitable result of a programming conflict. HAL is responsible for the safety of the space ship Discovery and its crew, and for the success of the mission. Through his pre-launch programming, HAL has been also been entrusted with knowledge of the true purpose of the mission to Jupiter (which the human crew thinks is merely to explore the Jupiter system, but is actually to follow the "burglar alarm" signal from the monolith on the moon). This continues the theme of information being withheld about this monolith's discovery; earlier in the movie, we see space bureaucrat Dr. Heywood Floyd traveling to the moon merely to orchestrate the secrecy campaign. But HAL has also been ordered not to reveal this purpose to the human crew -- to keep from adding to the already considerable psychological stress they will face during the months that they spend in isolation on their space mission. That these goals are logically irreconcilable and ultimately unachievable is gradually revealed by HAL's interaction with the crew during the course of the Jupiter mission. The plot of this part of the movie can be thought of as a game, similar to a game of chess, played out between HAL and the human crew: each trying to figure out how much the other knows, to anticipate and block their opponent's next move. In the dialog of a series of key scenes we see how this game, which depends on more than pure logic, leads HAL to irreconcilable conflicts and subsequent breakdown, with tragic results.
HAL: "No 9000 computer has ever made a mistake, or distorted information. We are all, by any practical definition of the words, foolproof and incapable of error."
HAL is already coming close to lying here, since his withholding the true purpose of the mission could be considered distorting information.
HAL: " ... my mission responsibilities range over the entire operation of the ship, so I am constantly occupied."
Actually, since the uneventful months of the mission, or even a second-long gap in a conversation with a human, are practically endless to a computer, HAL is doomed to spend what must seem like forever stuck in logical loops with no resolution. Boredom is as much as, or more of, a risk for HAL as it is for the human crew.
In this and subsequent scenes of daily life aboard the space ship, we see that indeed HAL acts more human than the crew members. Not only was HAL programmed to emulate human emotions, but the astronauts were no doubt selected for their ability to carry out months of tedious duty with machine-like reliability; they were chosen for this mission because they were especially boring people. Forcing both HAL and the humans to relate on the same intellectual level -- expecting a programmed machine to deal with the complexities and ambiguities of the human psyche -- contributes to HAL's breakdown.
HAL: "Queen to Bishop Three. Bishop takes Queen. Knight takes Bishop. Mate."
Here we see HAL playing an actual game of chess with Poole. If you look at the chess game on the board, you will see that even though he wins HAL has made a mistake: he incorrectly identifies one of the moves on the board (He should have said "Queen to Bishop Six.") This is the first clue, albeit a very subtle one which no one but a chess master like Kubrick would ever notice, that something is wrong with HAL.
HAL: "Well, forgive me for being so inquisitive, but during the past few weeks I've wondered whether you might be having some second thoughts about the mission ... I know I've never completely freed myself of the suspicion that there are some extremely odd things about this mission ... certainly no one could have been unaware of the very strange stories floating around before we left. Rumors about something being dug up on the moon ..."
Bowman: "You're working up your crew psychology report."
Bowman's mistake; actually HAL is trying to get enough information to know if he has succeeded in obeying his programming by keeping his secret, but recognizes that since he does not know everything that Bowman knows, he cannot say for sure. And he can't continue to question Bowman without giving away the secret. This is an endless loop that causes HAL's first serious breakdown.
HAL: [after one of those pauses which must seem an eternity to a computer] "Of course I am ... [The first of many lies; then he changes the subject] Just a moment ... just a moment ... I've just picked up a fault in the AE35 unit. It's going to go a hundred percent failure within seventy-two hours."
Why does HAL flag a non-existent failure in this piece of communications equipment? There are several possible reasons. Does he intend to sever communications with earth to keep the astronauts from prematurely learning the purpose of the mission? Is he just trying to get both astronauts outside the spaceship where he can easily kill them? Is he just sending them on a wild goose chase to buy time? Or is HAL's action truly irrational?
Bowman: "Well, HAL, I'm damned if I can find anything wrong with it."
He is in essence calling HAL a liar.
HAL: "Yes, it's puzzling. I don't think I've ever seen anything quite like this before. I would recommend that we put the unit back in operation, and let it fail ... we can certainly afford to be out of communication for the short time it will take to replace it."
HAL has bought more time but he most know that the unit will NOT fail, and his error will then be discovered.
Mission Control: "... our preliminary findings indicate that your on-board Niner-Triple-Zero computer is in error predicting the fault ... this conclusion is based on results from our twin Niner-Triple-Zero computer."
The twin computer can correctly analyze the AE35 unit because it does not have the secrecy programming dilemma.
Bowman [to HAL]: "How would you account for this discrepancy between you and the twin 9000?"
HAL: "... it can only be attributable to human error ..."
HAL is actually right here; it was humans that programmed him to lie. At this point in the novel by Arthur C. Clarke, mission control admits that they have identified the programming conflict, and recommend disconnecting HAL and operating the ship by remote control from the twin 9000 computer on Earth; in the movie HAL has to figure out that the humans will inevitably come to this conclusion.
Poole: "Listen, HAL -- there's never been any instance at all of a computer error occurring in a 9000 series, has there?"
HAL: "None whatsoever, Frank. The 9000 series has a perfect operational record."
Poole: "Well, or course I know all the wonderful achievements of the 9000 series, but are you certain there's never been any case of even the most insignificant computer error?"
In pressing his line of questioning , Poole makes two mistakes here that seal his fate: He words the question sarcastically; HAL has been programmed to respond as if he has human emotions. And he words it so vaguely (how do you define "insignificant"? Withholding the true purpose of the mission would have to meet that criterion). If HAL had been programmed with fuzzy logic, he might have made it past this dilemma. At this point HAL's nervous breakdown continues and he decides to protect his secret at all costs -- even if it conflicts with his programming to protect the crew.
HAL: "None whatsoever, Frank."
Poole: "Look, Dave -- let's say we put the unit back and it doesn't fail. That would pretty well wrap it up as far as HAL was concerned, wouldn't it? ... I don't think we'd have any alternatives ... I wouldn't see how we'd have any choice but disconnection."
Bowman: " ... no 9000 computer has ever been disconnected."
Poole: "Well, no 9000 computer has ever fouled up before."
Bowman: "That's not what I mean."
Bowman: "Well, I'm not so sure what he'd think about it ..."
Poole has misunderstood Bowman's point: they may be putting themselves in danger by attempting to shut down HAL. At this point we can't hear the rest of their conversation, but we see that HAL can: he is reading their lips. It is at this point that HAL's breakdown is complete and makes his final logical, but tragic, decision: to protect his secret, he must kill the astronauts, and we go into the movie's intermission with a feeling of dreadful expectation.
HAL [to Bowman, after killing the four other astronauts] : "This mission is too important for me to allow you to jeopardize it."
Bowman: "I don't know what you're talking about, HAL."
HAL: "I know that you and Frank were planning to disconnect me, and I'm afraid that's something I cannot allow to happen."
Bowman does not understand what HAL means, but this is as close as HAL himself will ever get to explaining his actions: to carry out his programming, he must also prevent anyone from turning him off, since the true purpose of the mission would then be lost. Logical on computer terms, but not on human terms: of course, if HAL kills all the astronauts, there will be no one left for him to reveal his secret to at the preprogrammed time, when the ship enters orbit around Jupiter. Bowman may not understand everything, but it is obvious to him that HAL must die.
This also brings up the question: did the murdered hibernating astronauts know the true purpose of the mission? One clue that they did know: HAL earlier made a comment that they were trained separately from Bowman and Poole. Again, the novel clarifies that they did know -- but if that is the case, why did HAL kill them? After all, he wouldn't have been programmed to keep the secret from them. He kills them simply because if they are revived from suspended animation, they will help Bowman disconnect HAL. In the novel, Bowman does attempt to revive one of the astronauts to replace Poole, but is foiled at the last moment by HAL.
Dr. Heywood Floyd [on computer monitor, after Bowman has turned off HAL's higher logic functions ... Check Mate]: "This is a pre-recorded briefing, made prior to your departure, and which, for security reasons of the highest importance, has been known on board during the mission only by your HAL 9000 computer. ... eighteen months ago, the first evidence of intelligent life off the Earth was discovered ... buried forty feet below the lunar surface ... a single, very powerful radio emission, aimed at Jupiter ..."
Again, this is as close as the movie comes to explaining everything that has gone before. The wording misleadingly implies that the astronauts in suspended animation were not in on the secret. But they were, and once we know that, we can see that the explanation given, "security reasons of the highest importance," is also misleading; the true purpose of the secrecy was psychological protection of the crew. Dr. Floyd alluded to this motivation during his earlier briefing at the moon base: "I'm sure you're all aware of the extremely grave potential for cultural shock and social disorientation contained in this present situation, if the facts were prematurely and suddenly made public without adequate preparation and conditioning." This bureaucratic policy, and its incompatibility with the limitations of technology, ultimately lead to HAL's breakdown.
So, in he end, we can't help but feel sorry for HAL. After all, he was only trying to do as he was told.
In the game of chess, no observer, nor even the players themselves, can understand all of the thought processes that transpire in the course of a match. Therefore, while my analysis is probably as complete as possible given the information in the movie, other questions and possibilities will always remain. Also, this explanation may not even agree with the one that was (unfortunately) spelled out in the sequel 2010: Odyssey Two -- a movie that did not realize that mystery is an essential part of the 2001 experience.
After writing the above, I read the book HAL's Legacy: 2001's Computer As Dream And Reality, edited by David G. Stork and published to coincide with the fictional date of HAL's birth as given in the novel: January 12, 1997. The book describes the state of the art in Artificial Intelligence in real computers, as compared with HAL's capabilities; there are chapters on fault tolerance, speech synthesis, speech recognition, image processing, speechreading, and computer ethics. These discussions include richly detailed analysis of HAL's behavior from many perspectives, including his programming and how it leads to his breakdown. One chapter reproduces the entire chess game between HAL and Poole -- actually a real match that was played in Hamburg in 1913!
Although the authors explore many possibilities for HAL's architecture and operation (including speculation that HAL's core may actually be a human brain!), they only briefly consider one possibility that has occurred to me: that HAL may actually be a lot less intelligent than he appears; many of his responses may be at least partially "canned," similar to the recorded voice on present-day automated telephone-answering systems. This seems particularly evident during the BBC interview; HAL's answers seem curiously disconnected from the questions, especially the response that begins "Let me put it this way ..." One author admits that HAL's refusal "I'm sorry, Dave, I'm afraid I can't do that" could be an automatic response if one assumes a "storytelling" programming model, and that such a computer would be even possible with today's technology. But then the author dismisses the notion; why would HAL's creators have programmed him with paranoid responses? That line of reasoning is belied by the frustratingly generic and misleadingly personal error messages of today's PC's; one should not attribute to deliberate malicious intent what could be the inadvertent result of technological limitations. Indeed, many modern PC owners (Arthur C. Clarke included!) recognize the humorous irony inherent in this situation by setting up their computer to deliver that very line of HAL's dialog when instructed to do anything contrary to programming! Likewise, in the chapter on speechreading it is never considered that HAL didn't actually read the astronauts' lips, but, through a mix of extrapolation and paranoia, "guessed" that they had to be discussing his disconnection. After all, HAL merely says "I could see your lips move" -- perhaps intentionally misleading Bowman (and the audience) to infer that he was truly speechreading.
The authors point out many other inconsistencies in HAL's operation, but also recognize that much of this can be attributed to dramatic license rather than actual failure to understanding the technology. One outright error occurs in the chapter on speech recognition by computers, when a phrase of dialog from the movie ("Pod 3BA") is used as an example. But in this case the author himself has misheard the line -- Bowman actually says "[Prepare] B Pod for E.V.A. [Extra-Vehicular Activity]" -- negating the whole point of how easy it is for humans to understand the phrase in context!
Overall, this book is filled with interesting information and intelligent analysis, and anyone who wants to know how close we are to building an actual HAL will find it fascinating.