Science: 600 Men & a Machine

To many a scientist the discouraging moment in life conies when his figures begin to run amok. Figures can bristle like barbed-wire barriers between his data and his conclusions. He finds that before he can get on with his work, he must multiply numbers as long as his middle finger, divide them, add them, square them, extract their roots. Sometimes a process involving a complicated equation with many variables must be repeated thousands or hundreds of thousands of times. Often the scientist gives up in despair. Many important lines of research have bogged down in a morass of figures.

Last week 600 scientists—from mathematicians to sociologists—gathered at Harvard to admire the latest of the great machines (large-scale computers) that eat their way through oceans of figures like whales grazing on plankton. At the invitation of Professor Howard H. Aiken, director of Harvard’s Computation Laboratory, the scientists arrived full of problems. Said Dr. Aiken: “We’ve built the machines. Now let’s start using them.”

Problems for All. The conference proved that hardly a science or branch of technology lacks problems for the computers.* Physicists, chemists, aircraft de signers had plenty of them to offer. So did psychologists and physiologists. Even sociologists wanted to use the machines, though they did not quite know how to go about it. All the scientists agreed that the large-scale calculators would encourage them to tackle many problems from which they had been scared away by computation difficulties.

A promising field is economics. Professor Wassily W. Leontief of Harvard explained that when economists try to figure out how the innumerable industries of a nation or continent affect one another, they run into a bramble-patch of interlaced figures. He hoped that the great calculators, by breaking this numerical barrier, might give nations a hint on how to keep their economies balanced.

Some of the speakers won applause, but the real hero of the conference, holding court in the Computation Laboratory up the street, was a machine: the Mark III Computer, built by Harvard for the Navy at a cost of $500,000. From the front, the Mark III looks like a giant radio panel, with a clean, serene dignity. But behind the panel hides a nightmare of pulsing, twitching, flashing complexity. Thousands of metal parts, big & little, all polished like costume jewelry, compete in frenetic activity. They hum and clack and chirp and roar like a hive of mechanical insects. Among them glow the filaments of 4,500 vacuum tubes, and between them run skeins of wire, 100 miles in all, with 400,000 soldered connections. The Mark III is so complicated that no one in the laboratory was willing to talk authoritatively about all of it.

Inner Memory. What can the Mark III do? For one thing, it can multiply two 16-digit numbers in a little more than twelve one-thousandths of a second. But this prodigious speed gives little idea of the machine’s talents. Its strong point is its “inner memory.” This “memory” consists of nine big aluminum cylinders revolving up to 7,200 r.p.m. Their surfaces are coated with black magnetic material. Huddled around them are staggered rows of little brass blocks enclosing electromagnets. When a brief electric impulse flashes through an electromagnet, it prints a dot of magnetism on the spinning cylinder’s surface. The dot stands for part of a coded number for the machine to store in its memory.

The nine cylinders can store 4,000 numbers of 16 digits each and 4,000 coded “commands.” In response to the proper command (either remembered or coming from outside), the numbers are “read off” electrically. They zip through the machine as coded electrical pulses. Basically the process is similar to a man’s pulling a telephone number out of his memory and spinning it on a dial.

With the aid of-its vast mechanical memory for numbers and commands, the machine can solve in a flash a complicated equation involving thousands of numbers and thousands of operations. It can do its trick tirelessly, over & over again, varying one or more of the factors in the equation. It prints the result (e.g. the range of a naval shell at different gun elevations) in the form of a neat table, as fast as electric typewriters can rattle the figures out. To do a comparable job by hand would take an army of trained mathematicians.

Most of last week’s conferees were down-to-earth men who flinch from sensationalism. They hate to hear the “Mark III and its fellows called “mechanical brains.” They insist that the machines have no intellect, but merely obey commands.

Other calculator experts (a minority) are not so sure. They admit that even the Mark III is crude compared to a human brain with its billions of cross-connected nerve cells. Yet the Mark III can already beat the human brain at certain tedious tasks. Imaginative scientists can only’ guess at what other mental marvels its more efficient descendants will be able to produce.

* This equation, introduced at Cambridge by a psychologist, is a typical problem for the machines.