HAL or the digits of PI

By Martin Conrads, 10 September 1997

Would you take seriously the idea that, thirty years from now, Roddenberry's creation Data will not only be seen as a shaky technological prediction but as reliable proof of a misplaced trust in fault tolerance in the culture of the 1990s? As usual with the mysteries of space, there are other, equally - if not more - significant histories to be discovered and stories to be told.

Especially when there are computers involved: Think of Godard's Alpha 60 or of the fatal bomb in Carpenter's Dark Star or this thread's possible anticlimax: HAL 9000. Making a book on HAL 9000, the murderous mainframe computer of 2001 - A Space Odyssey can be difficult when you take it all too literally. Or in this case, too scientifically. HAL's Legacy - 2001's Computer as Dream and Reality, edited by David G. Stork, chief scientist at the Ricoh California Research Centre, includes essays by computer scientists, programmers and psychologists such as Raymond Kurzweil, Murray S. Campbell, Marvin Minsky, Daniel C. Dennett and Roger C. Schank and starts with an enlightening foreword by Arthur C. Clarke himself.

The book, intended to mark HAL's birthday in keeping with the chronology of Clarke's book ("I became operational at the HAL plant in Urbana, Illinois on January 12, 1997"), nevertheless has a strong focus on Kubrick's masterpiece (in which his date of birth was changed to January 12, 1992). Despite the book's aim to be fully contemporary, it seems in keeping with its main subject that it couldn't 'escape velocity': After the '97 Kasparow versus Deep Blue match, many of the speculations on chess computers seem outdated. Even if HAL's promise wasn't realised in Deep Blue, Big Blue "now seems quite proud of the link and no longer fears guilt by association", as Clarke puts it when referring to the idea that the name of H A L was chosen with the aim of being one step ahead of I B M.

As for the movie, Murray S. Campbell, well known for being Kasparow's opponent in the '97 match, deconstructs a seemingly casual detail like the chess game between Frank and HAL (which appears in the movie for only 30 seconds and isn't mentioned in Clarke's book). Here, Campbell - programmer of Deep Blue - discovers that HAL's game exactly replicates a match between two undistinguished players in Hamburg in 1913 - one that wasn't even included in Deep Blue's database of 600,000 games. Only if you know that specific game (and Campbell suspects that Kubrick must have), would you recognise HAL's first mistake - long before he claims the AE 35 unit to be faulty - his play is too human. In a similar vein, computer scientist Rosalind W. Picard [sic!], whose essay "Does HAL Cry Digital Tears? Emotions and Computers" is the only text by a female contributor in the book, emphasises that HAL displays more emotions than any other crew member. Surprisingly, she also declares HAL to be a kind of digital art critic due to his decision that Dave's drawings have improved over time.

In an interview with the editor, Marvin Minsky, himself mentioned in 2001 ("In the 1980s, Minsky and Good had shown how neural networks could be generated automatically"), clings to his visionary identity by deciding that "if we work really hard - and smart - we can have something like HAL in between four and four hundred years." In contrast, Roger C. Schank ("That is the bad news. HAL could never exist") realises that HAL not only fails as a model for AI but also as a model for human intelligence. His observation that it is a misconception to posit problem solving as the heart of intelligence is a convincing one. Schank suggests that giving HAL more experience could help and suggests that sending him on a Carribean Cruise or giving him a sweetheart would more likely develop (and test) his intelligence. Daniel A. Norman of Apple ("Living in Space: Working with the Machines of the Future") is posing as the Best Boy of the book by counting out telling but silently accepted details in a manner Poe's Auguste Dupin would be proud of. Did you ever think, for example, how unlikely it is to have heavy cashmere sweaters in a space station, considering how much energy it takes to get them there? Ending with a statement on the lack of hand held computers on Discovery, Norman amusingly wonders why there are also no computer games to play on board. HAL's Legacy traces the path along which fiction turns to science - which turns out to be one of its most problematic traits. Most authors see HAL as a more or less accurate 60s attempt to predict scientific research and progress, rather than understanding Clarke's/Kubrick's vision as a story interlinked with a whole genre of movies and myths.

Nevertheless HAL's Legacy offers simple, yet intriguing observations: according to Norman, Kubrick and Clarke failed to imagine the importance of networks aswell as the fact that computers get smaller (and must do so to become more effective). Also, as Stork observes, both failed to understand the role of complex software systems in combination with minimalised hardware: think of how many displays there are on Discovery and how many displays your PC has on its way to becoming your personal I'm-Sorry-I'm-afraid-I-can't-do-that. Detailed analyses on speech and lipreading (Stork, Ariel Rosenfeld) emphasise the role of silence in the movie, maybe one of its most sublime and subliminal qualities. Of course, HAL's Legacy is not an interpretation of the whole 2001 plot, but only of the HAL 9000 and his function, mostly isolated from the realms of the Alien motif with which the book and movie are also concerned. Forget about hermeneutics. No wonder then that none of the authors reflect on the role of the black monolith, or on the scary conspiracy theme, which might be the HAL real estate. In the end HAL's Legacy turns out to be a cultural studies reader written by computer scientists, which casts light more on the influx of computer science on today's cultural imagination than it helps us understand whether we could actually build HAL - one of the major motifs of this book and a purely technical one. But why should we build HAL or why should HAL be built? "As we suspected, the fault does not lie in the AE 35 unit, and there is no need to replace it again. The trouble lies in the prediction circuits..."

HAL' Legacy. 2001's Computer as Dream and Reality. Edited by David G. Stork. Cambridge, Massachusetts; London, England: MIT Press, 1997.

Martin Conrads <luxus AT> is co-editor of contd. and a Berlin-based freelance author in the fields of arts, music and media