In more conventional politics, the Soviet Union is gone, which was a big surprise when it happened. There were some who had expected the Soviet empire to gradually decline and perhaps ultimately collapse, but the quick demise of the Soviet Union's Eastern European empire in 1989 and the breakup of the Soviet Union itself in 1991 were totally unexpected.
And I'm not sure where this comment ought to go, but unlike in the movie, many women nowadays wear pants full-time.
I suspect that this is partly technology and partly politics. On the technology side, space exploration continues to be expensive and difficult, and the most successful spacecraft have tended to be the automated ones, because they do not have a human crew that must be kept alive and safely brought home, thus allowing them to be relatively small and cheap. And on the political side, there has not been much motivation for really massive outer-space projects; the closest to that was the space race of the 1960's, which had been "won" by the US with its Moon landings. The Russians had wanted to do that also, but despite their initial outer-space successes, they could not develop their capabilities fast enough, and they eventually quit and affected a never-in-the-race pose.
There are also some oddities which may either be a side effect of trying to make a one-g environment seem like a zero-g one, or else a lack of imagination. In the movie, all the flight attendants are careful to keep their feet in contact with some surface as if they were walking; while in real life, the inhabitants of sufficiently-voluminous spacecraft and space stations have preferred to float through the air. Also, women have generally not worn big swimming caps, but have often let their hair go free.
However, during the 1950's, 1960's, and 1970's, optimistic predictions about AI had been commonplace. One example is in Turing's paper; he predicted that typical RAM capability would increase from 165,000 bits (20 kilobytes) to 10^9 bits (125 megabytes), and also that computers would be able to pass his Turing Test at least 30% of the time. His RAM prediction has come to pass; the computer I had owned back then had 160 megabytes, and it was a typical model that was 3 years old. But his Turing-Test prediction has not. The classic Turing Test can be performed with chatterbot software, as it is called; in my experience, such software utterly fails that test. Other AI software will also fail various generalizations of the Turing Test.
Successful AI would be extremely valuable, with numerous applications easily imaginable; there have been numerous researchers who have devoted much of their careers to AI research, and if it was easily feasible, investors would be willing to spend large amounts of money on developing commercial AI apps. But these researches have had only very limited success, despite valiant efforts, and most commercial software has had only very limited AI, despite investors' willingness to spend large amounts of money on far-out business ventures during the recent dot-com bubble.
This includes games, where the greatest AI success has been in games with very abstract game worlds, like chess and backgammon. However, AI for games with more real-world-like game worlds, like most real-time-3D games, has been much more difficult; the most success has been with non-player characters that do not have to much planning or strategy.
Despite the great AI letdown, real-life user interfaces are *much* improved over the movie's user interfaces, as user-interface guru Donald Norman has noted. The movie features numerous individually-labeled buttons and screens that show only one sort of thing at a time; in real life, a much more effective approach has been to create lots of virtual screens in a physical screen, and to use typewriter-style keyboards as generic sets of buttons. This even includes the creation of virtual buttons and other controls on a screen; these can easily be created, modified, and destroyed, unlike physical-object controls.
There are some other engineering issues that I think are worth discussing. The movie pictures HAL 9000 as an integrated hardware/software combination. This shows that the makers of the movie may have failed to appreciate the separability of hardware and software designs; there is no reference to uploading or downloading control instructions, as opposed to data. This separability was apparent as early as Alan Turing's proof of universal computability -- that any computer that can perform certain simple sorts of operations can act like any other computer that can perform these operations. But that separability has not been widely appreciated as one might hope. Such separability, however, is what is behind emulators of other computer architectures; such emulators are almost too numerous to mention.
Furthermore, HAL 9000 is apparently constructed on the old mainframe model -- a single big standalone computer. However, in recent decades, alternative models for computing have emerged; the main ones are large numbers of smaller computers networked together a computer (or more than one!) for each user, and computers small enough to be easily portable. These models not only coexist, but have even assimilated mainframes, turning them into super servers in networks. However, there is none of this in the movie; we even see one of Discovery's crew doing a pencil-and-paper checklist instead of using some Personal Digital Assistant.
Also, HAL's aberrant performance was regarded as strikingly abnormal; that is a very different perception than what much present-day software produces, especially that produced by a certain well-known software company, one well-known for monopolistic business practices. Also, cutting-edge software being very well-behaved is a totally unrealistic expectation; such software often tends to be flaky, because almost by definition it is often a work-in-progress.
Finally, the line that "No member of the HAL 9000 line has ever made a mistake or distorted data" is also a bizarre one -- whether something is a mistake is often a matter of context, and distortion suggests analog instead of digital processing. Now to the various kinds of errors, and when they may not be "real" errors. Hardware errors can often be detected and sometimes corrected; that is standard procedure for storage systems and network links. Software errors are essentially errors in algorithms, and some such errors are inevitable, such as roundoff errors in floating-point quantities, which necessarily have finite precision. Other such errors are accepted because trying to get a highly precise result would be a waste of computing resources in the contexts of some problems. Furthermore, some algorithms start with erroneous but easily-guessed values and then improve those values until the desired level of accuracy is achieved. A variant on that is generating errors on purpose. One example is reading some file with unknown or unpredictable length; a program will read it until an attempt to read it signals an end-of-file error, which the program then treats as an expected occurrence.
However, the most interesting sort of error is errors in HAL's AI routines, which are what make HAL what HAL is. Was it those kinds of errors that HAL was referring to? With that thought, I close this essay.
To my Notebook-Binder Page