Of Machines and Men
As warmer weather advanced over the Northern Hemisphere, a certain dreaminess and languor became evident in the pronouncements of eminent scientists: a young student was said to have solved a complex mathematical equation deemed unsolvable for generations; a Soviet thinker named Dr. I. Shklovsky announced that after prolonged study he has concluded that the two tiny moons circulating around Mars are really Martian-made satellites, put into orbit by highly intellectual beings, now extinct, who inhabited that planet two or three billion years ago; and an astronomer from Wisconsin told a scientific gathering in Washington that after extensive study he had decided there might be diamonds in the craters of the moon. A kind of scientific wool-gathering, speculations well-nigh poetic in their extravagance, suffused the scientific imagination, and it hardly seemed surprising, therefore, when Univac began to play the races and tried to pick the winner of the Kentucky Derby.
The experiment took place in a big, bare, scrubbed and dustless room on the second floor of the Remington Rand Building in downtown Manhattan, where Univac disgorged its conclusions, the day before the Derby was run, after five months of study. The atmosphere of the occasion was so calm that it seemed well-nigh irresponsible. Obviously much, including perhaps the survival of horse racing, hinged on which horse Univac picked, but only a few unhurried, deliberately voiced, neatly dressed technicians and engineers were on hand, and the machine itself, humming and buzzing in the middle of the room, tapes spinning merrily in a row of glass containers, and lights glowing in the bank-vaultlike structure of the brain, gave an impression of utter indifference. One odd circumstance of the Univac- Kentucky Derby experiment was that most of the people connected with it knew nothing about horse racing. Rom Slimak, a Polish-born, English-educated mathematical genius, manager of Univac operations, never saw a horse race in his life. William Mickelfelder, a New York newspaperman who handled the operation for the New York World-Telegram and Sun, was once taken to Aqueduct when he was a boy but had never bet on a race; his specialty has been medicine. Last December Richard Starnes, the managing editor of the World-Telegram, put Mickelfelder on the Univac experiment for just those reasons.
The exception among these racing innocents was Bill Bloome, the newspaper's specialist. He prepared the material on past performances that was codified, put on tape and stored in Univac's memory: the distances run by all the entries in all their previous races, how they finished and against what competition; their speed ratings (best time against track records) and all such variables as track conditions and jockeys' performances as could be translated into the symbols used by the computer.
Now, it would be gratifying to be able to relate that all this professional background, freshness of approach and scientific acumen resulted in a clear-cut finding on the degree to which science can determine in advance the outcome of a horse race. What really happened was that the Remington Rand scientists worked on the material Handicapper Bloome provided, and evolved a weighted point-score system that came out right much of the time. That is, they took old races run in the long ago and, without knowing the outcomes (not being race fans), fed the relevant material into Univac, and compared Univac's decision with the real outcome. When Univac's choice wasn't on the nose, the weighing of factors started over. When Univac began to win these ancient races, the operation was returned to the Derby experiment, together with a few warm-up races in the days before the Derby.
Maybe there are diamonds in the craters of the moon. Perhaps Dr. I. Shklovsky is on solid ground when he says the moons of Mars are really hollow metal spheres which the ancient Martians sent aloft a few billion years ago. These are still agreeable scientific speculations. It appears, however, that it is now a good hypothesis that a machine is no better than the generality of mankind in predicting the outcome of a horse race. Human brains generally picked First Landing, Tomy Lee and Sword Dancer in the Derby, in that order. Univac picked First Landing, Tomy Lee and Royal Orbit, and put Sword Dancer well down the list, below Silver Spoon and Easy Spur. And yet, at this season of the year, it is difficult to evade a suspicion of a bit of springtime carelessness and irresponsibility in Univac; perhaps a devil-may-care spirit entirely different from what the machine might have felt if it had had some of its own money riding.
Words of the Week
The executive director of the United States Golf Association is called Joe Dey. He is also called taciturn, button-lipped and laconic. He was, at least, up until last week when he opened his mouth on two separate occasions to say something. And when he was done, everybody else was talking at once.
Dey first broke his customary silence on Monday for Oscar Fraley, a sportswriter for United Press International. "What percentage of golfers play by the rules?" asked Fraley in a story-fishing telephone call to Dey. "Ohhh," said Dey guardedly, "probably not more than 2%." Fraley pressed on: "Does that mean 98% do not play by the rules?" Dey, sensing the logic of that, admitted: "Ahhh...ummmm...yes." And before Dey could hang up, he had allowed that indeed 98% might be low, that 98 out of 100 golfers will somehow break a rule before reaching the sixth tee, and that, contrary to common practice, old balls may not be substituted for water-hazard shots and winter rules do not apply on the Fourth of July. Next day Oscar Fraley's story whistled out over the UPI system, and 98 out of 100 readers sat up, stiffened, and said, "He's right, of course, but he can't be talking about me."
That same day the Associated Press called Joe Dey. "What," asked a reporter, "do you think of Calcutta tournaments?" The Calcutta pools, such as the recent Tournament of Champions at Las Vegas, answered Dey, were alarming to the golf association. And he said a suggestion had been made to bar Calcutta players from the National Open championship, sponsored by the USGA. "I don't know what action will be taken at our meeting next January," said Dey, "but golf is strongly opposed to any tournaments having the slightest connection with organized gambling. We are presented with a situation in which a golfer [who is auctioned off to gamblers and usually gets 10% of any prize his syndicate wins] might collect more money by throwing a match than by winning it.... Under our present rules we can ban anyone guilty of conduct detrimental to the game. Consorting with known gamblers could fall under this category."