The Designing Computer
It seems strange that the computer was a bookkeeper and clerk for years before anyone seriously considered that it might be an engineer as well, yet the men who themselves designed the computer were loath to use it in their other work. Part of this resistance stems from the high premium placed on the creativity of research and design work. The engineer uses science in his work, to be sure, but he professes to use it as an artist, or with the personal touch of, say, a brewmaster. There is another possible reason for the lag in computer use by the men who should appreciate its ability the most. In the early days of the computer, it clacked away all week figuring payrolls, and perhaps writing checks. That’s what it was ordered for, and that’s where the money was—in the businessman’s application of the computer.
To be sure, the military was using the computer for other purposes, but the average scientist or engineer not employed by Uncle Sam had access to an electronic computer only on Sunday, if at all, when the big machine had done its primary work and could take a breathing spell. To further compound excuses for the foot-dragging engineer, there was a difference in needs in payroll computation and scientific mathematical calculation. Commercial computers are designed for a high rate of input and output, with a relatively slow arithmetic going on inside. The engineer, on the other hand, might need only several minutes of computer time, but it could take him a couple of days to put the problem into a form the machine could digest.
Slowly, however, enough engineers fought the battle of translation and forewent Sunday pursuits like church, picnics, and golf to learn haltingly how to use the electronic monster. It took courage, in addition to sacrifice, because the computer was pooh-poohed by some sharp scientific brains as an idiot savant at best. Behind the inertia there could have been a touch of concern too—concern that the machine just might not be as stupid as everybody kept saying it was.
Heavy industry made use of the machines. The steel plants, petroleum and chemical plants, and even the designers of highways were among the early users of computer techniques. There was of course good reason for this phenomenon. Faced with problems involving many variables and requiring statistical and probabilistic approaches, these people could make the best use of machines designed for repetitive computations. The refiner with a new plant in mind could simulate it in the computer and get an idea of how, or if, it would work before building his pilot plant. Today the notion of dispensing with even the pilot plant is getting serious consideration.
One program used by a gasoline producer analyzed thirty-seven variables and thirty-seven restrictions, a matrix that could never be evaluated by ordinary methods. Textile fiber research is another example, with thread tests run on dozens of samples and averaged statistically for valid conclusions. B. F. Goodrich put the computer to work in its laboratories at such tasks as multiple-regression studies of past production of processes like polymerization and the running of a batch of new material on the computer.
These applications were accomplishing a two-fold benefit. First, years were being telescoped into weeks or even days; second, complete investigation rather than sketchy sampling was possible. Optimum solutions took the place of the guesswork once necessary because of the lack of sufficient brainpower to run down all the possibilities. Still there were scientists and designers in other fields who shook their heads loftily and said, “Not for me, thanks.” The computer was but a diligent clerk, they held, relieving the engineer of some onerous chores. It could do nothing really creative; that must be left to man and his brain.
By now many industrial firms had purchased or rented computers for the technical people so that they would not have to fight for a place in line at the payroll computer. Civil engineering agencies, perhaps a hundred strong, used computers to design bridges and plan and lay out highways. Designers at the Tudor Engineering Company of San Francisco put its Bendix G-15D to work planning the highway that Contra Costa County will need in 1980. Almost all of our fifty states now use computers in their highway departments. In 1960, Georgia solved more than a thousand highway bridge design problems in its computers. Besides doing the work faster and cheaper, the computer produces a safer product. For example, if substandard materials are programmed in, the computer will print out a warning or even stop working altogether so that the error can be corrected.
Steel companies, like Jones & Laughlin, use computers not only to run production mills, but also as research tools. Three hours of operation of a new furnace can be simulated in the computer in thirty seconds. Tracing the steel back to its ore, the computer is used again. The Bureau of Mines has used the machines for several years; they are helpful in problems ranging from open-pit operation, grades of ore, drill-core data logging, reserve calculations, and process control.
Gradually, then, the resistance was worn down. Grudgingly at first, and accepting the computer only as an assiduous moron, engineers in other fields put it to work. Complex machine operations like gear-shaping were planned and carried out by computers that even punched out tapes for controlling the production tools. Optics designers switched from desk calculators to electronic computers. Mechanical engineers in jobs from ultrasonic vibrators to tractor design became users of computers. Mass spectrometry, heat-exchanger design, and waterworks design joined the jobs the computer could do.
The computer had figured in plotting trajectories for missiles, and in the production of aircraft; engineers found it could design them too. Back in 1945, an analysis of twenty-one different flight conditions at each of twelve stations of an airplane fuselage took 33 days and cost more than $17,000. Today, by using a high-speed computer instead of a desk calculator, the analysis is completed in a day and a half, at a cost of $200!
The last of the diehards seemed to be the electronics people themselves. A survey conducted by a technical journal in the field showed that in 1960 many designers were not using computers in their work. Admitting that the computer was a whiz just about everywhere else, the electronics engineer still could say, “The machine is great on paperwork, but I do creative work. The computer can’t help me.” Other reasons were that computers were expensive, took much time to program, and were helpful only with major design problems. Fortunately, all designers do not feel that way, and progress is being made to put the computer to work in the electronics field. It is helping in the design of components (Bendix saves ten man-hours in computing a tenth-order polynomial and associated data) and of networks (Lenkurt Electric saves close to 250 engineering hours a week in filter network design). Bell Telephone uses the computer approach in circuit analysis, and Westinghouse in the design of radar circuitry. It is interesting that as we move up the design scale, closer to what the engineer once considered the domain of human creativity, the computer still is of great value. In systems design it is harder at the outset to pin down the saving in time and the improvement in the system (the latter is perhaps hard to admit!) but firms using computers report savings in this field too.
One interesting job given the computer was that of designing the magnetic ink characters to be used in its own “reading” applications. This project, conducted by Stanford Research Institute, is typical of the questions we have begun to ask the computer about its needs and ways to improve it. A larger scale application of this idea is that of letting the computer design itself. Bell Telephone Laboratories developed such a system, called BLADES, for Bell Laboratories Automatic Design System, to design a computer used in the Nike-Zeus antimissile defense system.
A wag once noted that the computer would one day give birth to an electronic baby. His prophecy came true perhaps quicker than he anticipated, but there is one basic difference in that the progeny is not necessarily a smaller machine. The giant LARC, for instance, was designed by lesser computers. As A. M. Turing has pointed out, it is theoretically possible for a simple computer to produce a more complex one. This idea is borne out in nature, of course, and man is somewhat advanced over the amoeba. Thus the implication in the computer-designed computer is far more than merely the time and money saved, although this was certainly a considerable amount. The BLADES system in twenty-five minutes produced information for building a subassembly, a job that required four weeks of manual computation.
Notable improvements in the general-purpose computer are doing much to further its use as a technical tool. Present machines do jobs as varied as the following: personnel records, inventorying, pattern determination, missile system checkout, power-plant control, system simulation, navigation, ballistic trajectory computations, and so on. Special computers are also provided now for the engineer; and among these is the Stromberg-Carlson S-C 4020 microfilm recorder. Engineering specifications are put into the computer and the machine can then produce on request mechanical drawings as required by the engineer. Data stored in the memory is displayed on a Charactron tube. There is little resistance to this type of computer, since the engineer can say it is doing work below his level of ability! Of course, the draftsman may take a dim view of computers that can do mechanical drawing.
After a rather hard to explain slow start, then, the computer is now well established as a scientific and engineering tool. Blue-sky schemes describe systems in which the engineer simply discusses his problem with the machine, giving specifications and the desired piece of equipment. The machine talks back, rejecting certain proposed inputs and suggesting alternatives, and finally comes up with the finished design for the engineer’s approval. If he laughs overly loud at this possibility, the engineer may be trying to cover up his real feelings. At any rate the computer has added a thinking cap to its wardrobe of eyeshade and work gloves.