How so?
Well the biggest problem was that the computer rang in with an actuator. So that in situations where more than one player had the correct answer before the question was done, the computer always had first crack. That's a huge advantage. And it's arguably munchkinism where the spirit of the game is to think quickly and accurately.
Of course, one might argue that ringing in quickly is part of the game and therefore part of the spirit of the game. My response to that is that other "parts of the game" were omitted for the benefit of the computer. For example, interpreting the questions by reading from the screen and listening to Alex is part of the game. But the computer had the questions fed to it in the form of a text file. The computer probably would not have been as effective if it had to get the questions by having a camera pan to the correct monitor, zoom in, and then do an OCR to interpret the questions. Or by doing voice recognition on Alex.
Another "part of the game" is that you normally have to play at the studio in Los Angeles. The computer (which was based in New York) would have had to deal with latency problems if the game had been played in the normal location.
Another "part of the game" is audio and visual clues -- as I recall there were no such clues.
On a slightly different topic, it was also a problem that there were two human players and one computer. So that on questions which favor humans over computers, the two humans would have had a tendency to split the points. Quite possibly the computer would have lost if the two humans had agreed in advance that one of them would always wait an extra 2 or 3 seconds before ringing in.
The bottom line is that the computer win was not satisfying. It reminds me of the annoying girl in your advanced math class who was always asking "Will this be on the test?"
The computer probably would not have been as effective if it had to get the questions by having a camera pan to the correct monitor, zoom in, and then do an OCR to interpret the questions.
The questions are always printed in the same font. Compared to the amount of processing power needed to answer them, the amount needed to do the OCR would be minimal.
Another "part of the game" is that you normally have to play at the studio in Los Angeles. The computer (which was based in New York) would have had to deal with latency problems if the game had been played in the normal location.
That's about 60 msec latency each way — not much.
Douglas Lenat's program EURISKO is legendary in the AI community for a distinct real-world achievement: allowing Lenat to win the the Traveller TCS roleplaying game tournament two years in a row (and then semi-voluntarily not competing subsequent years). Lenat never released EURISKO's source code, leaving how he managed to pull off this feat somewhat of a mystery. Yet Lenat's later work based on EURISKO does not seem to have yielded anything else in the way of practical benefits.
Some time ago on LessWrong, someone proposed trying to figure out what Lenat did and reimplementing EURISKO. But Eliezer is worried this could be dangerous. So I have another proposal: see if Lenat's accomplishment can be replicated using machine learning and genetic programming techniques that are already publicly known.
My suspicion is that Lenat's TCS win tells us more about TCS than about EURISKO, that TCS is likely a game that's inherently vulnerable to the "find winning strategies by simulating a lots of games on a computer" meta-strategy. I've heard, for example, that battles are often tactically trivial, with the outcome of battles effectively determined by the composition of the two fleets (and fleet composition is what Lenat used EURISKO for). If that hypothesis is correct, though, it suggests it shouldn't be necessary to reimplement EURISKO specifically to get a program that's good at designing TCS fleets. If that turns out not to be the case, it would be evidence that there really is something special about EURISKO after all.
Does anyone know if anyone has tried this? As a novice computer programmer, I think it might be a good project to hone my programming skills. Input on how to approach such a project would be appreciated.