The Census Taker
Moving from the sophistication of such logic devices, we find a tremendous advance in mechanical computers spurred by such a mundane chore as the census. The 1880 United States census required seven years for compiling; and that with only 50 million heads to reckon. It was plain to see that shortly a ten-year census would be impossible of completion unless something were done to cut the birth rate or speed the counting. Dr. Herman Hollerith was the man who did something about it, and as a result the 1890 census, with 62 million people counted, took only one-third the time of the previous tally.
Hollerith, a statistician living in Buffalo, New York, may or may not have heard the old saw about statistics being able to support anything—including the statisticians, but there was a challenge in the rapid growth of population that appealed to the inventor in him and he set to work. He came up with a card punched with coded holes, a card much like that used by Jacquard on his looms, and by Babbage on the dream computer that became a nightmare. But Hollerith did not meet the fate of his predecessors. Not stoned, or doomed to die a failure, Hollerith built his card machines and contracted with the government to do the census work. “It was a good paying business,” he said. It was indeed, and his early census cards would some day be known generically as “IBM cards.”
While Jacquard and Babbage of necessity used mechanical devices with their punched cards, Hollerith added the magic of electricity to his card machine, building in essence the first electrical computing machine. The punched cards were floated across a pool of mercury, and telescoping pins in the reading head dropped through the holes. As they contacted the mercury, an electrical circuit was made and another American counted. Hollerith did not stop with census work. Sagely he felt there must be commercial applications for his machines and sold two of the leading railroads on a punched-card accounting system. His firm merged with others to become the Computing-Tabulating-Recording Company, and finally International Business Machines. The term “Hollerith Coding” is still familiar today.
Edison was illuminating the world and the same electrical power was brightening the future of computing machines. As early as 1915 the Ford Instrument Company was producing in quantity a device known as “Range Keeper Mark I,” thought to be the first electrical-analog computer. In 1920, General Electric built a “short-circuit calculating board” that was an analog or model of the real circuits being tested. Westinghouse came up with an “alternating-current network analyzer” in 1933, and this analog computer was found to be a powerful tool for mathematics.
While scientists were putting the machines to work, writers continued to prophesy doom when the mechanical man took over. Mary W. Shelley’s Frankenstein created a monster from a human body; a monster that in time would take his master’s name and father a long horrid line of other fictional monsters. Ambrose G. Bierce wrote of a diabolical chess-playing machine that was human enough to throttle the man who beat him at a game. But it remained for the Czech playwright Karel Čapek to give the world the name that has stuck to the mechanical man. In Čapek’s 1921 play, R.U.R., for Rossum’s Universal Robots, we are introduced to humanlike workers grown in vats of synthetic protoplasm. Robota is a Czech word meaning compulsory service, and apparently these mechanical slaves did not take to servitude, turning on their masters and killing them. Robot is generally accepted now to mean a mobile thinking machine capable of action. Before the advent of the high-speed electronic computer it had little likelihood of stepping out of the pages of a novel or movie script.
As early as 1885, Allan Marquand had proposed an electrical logic machine as an improvement over his simple mechanically operated model, but it was 1936 before such a device was actually built. In that year Benjamin Burack, a member of Chicago’s Roosevelt College psychology department, built and demonstrated his “Electrical Logic Machine.” Able to test all syllogisms, the Burack machine was unique in another respect. It was the first of the portable electrical computers.
The compatibility of symbolic logic and electrical network theory was becoming evident at about this time. The idea that yes-no corresponded to on-off was beautifully simple, and in 1938 there appeared in one of the learned journals what may fairly be called a historic paper. Appearing in Transactions of the American Institute of Electrical Engineers, “A Symbolic Analysis of Relay and Switching Circuits,” was written by Claude Shannon and was based on his thesis for the M.S. degree at the Massachusetts Institute of Technology a year earlier. One of its important implications was that the programming of a computer was more a logical than an arithmetical operation. Shannon had laid the groundwork for logical computer design; his work made it possible to teach the machine not only to add but also to think. Another monumental piece of work by Shannon was that on information theory, which revolutionized the science of communications. The author is now on the staff of the electronics research laboratory at M.I.T.
Two enterprising Harvard undergraduates put Shannon’s ideas to work on their problems in the symbolic logic class they were taking. Called a Kalin-Burkhart machine for its builders, this electrical logic machine did indeed work, solving the students’ homework assignments and saving them much tedious paperwork. Interestingly, when certain logical questions were posed for the machine, its circuits went into oscillation, making “a hell of a racket” in its frustration. The builders called this an example of “Russell’s paradox.” A typical logical paradox is that of the barber who shaved all men who didn’t shave themselves—who shaves the barber? Or of the condemned man permitted to make a last statement. If the statement is true, he will be beheaded; if false, he will hang. The man says, “I shall be hanged,” and thus confounds his executioners as well as logic, since if he is hanged, the statement is indeed true, and he should have been beheaded. If he is beheaded, the statement is false, and he should have been hanged instead.
World War II, with its pressingly complex technological problems, spurred computer work mightily. Men like Vannevar Bush, then at Harvard, produced analog computers called “differential analyzers” which were useful in solving mathematics involved in design of aircraft and in ballistics problems.
A computer built by General Electric for the gunsights on the World War II B-29 bomber is typical of applications of analog devices for computing and predicting, and is also an example of early airborne use of computing devices. Most computers, however, were sizable affairs. One early General Electric analog machine, described as a hundred feet long, indicates the trend toward the “giant brain” concept.
Even with the sophistication attained, these computers were hardly more than extensions of mechanical forerunners. In other words, gears and cams properly proportioned and actuated gave the proper answers whether they were turned by a manual crank or an electrical motor. The digital computer, which had somehow been lost in the shuffle of interest in computers, was now appearing on the scientific horizon, however, and in this machine would flower all the gains in computers from the abacus to electrical logic machines.