Calculating Machines Can Yield National Industrial Production Goals, Expert Says

Large scale calculating machines may make it feasible for the government to set production goals for major industries, Frederick V. Waugh of President Truman's Council of Economic Advisers told a conference at Harvard last week.

The conference also heard Frederick Mosteller, lecturer on Mechanical Statistics, predict that new Large-scale Digital Calculating machinery may soon enable scientists to predict mathematically the chances of success of a given marriage or the degree of readjustment to the community that a given parolee will make when he is released from prison.

Waugh and Mosteller spoke at the Symposum on the new Mark III Calculator which Harvard built for the Navy. The Symposium was hold from September 13 to 16.

The implications of the new machine to the field of economics were discussed by Waugh and by Wassily W. Leontief, professor of Economics.

"There is a crying need for a few basic studies dealing in detail with the whole economy," Waugh stated. The machine, he continued, will enable us to make such extensive analysis because for the first time it will be possible to take into account the thousands of variables that influence the economy.

"We need to know how much of each kind of employment and production is needed, and how expenditures need to be allocated among various groups of consumer goods and investment categories," he said.

Referring to post-war studies of agricultural needs made by the department of Agriculture, Waugh urged the study of national clothing, housing, coal, steel, automobile, road, and school needs and the setting of production goals in these and other fields.

Statistics Needed

The advanced data which the machine can provide us with will help to enable us to arrive at goals which are "feasible to reach," he said. The economist emphasized that accurate and detailed statistics are not enough to determine a good set of goals for the economy. In a democracy of free people the citizens determine such goals. In so doing they taken into account factors such as the amount and kind of work they prefer and standards of health and diet prevalent among different working groups.

But to make intelligent decisions about their economic future, citizens must know what the alternatives are, and the large-scale computer, Waugh states, can give us useful information about the alternatives.

Economic goals cannot be "static," the economist continued. The policy of one period naturally affects the following period. A high degree of investment now will bolster futur productive capacity. Temporary deficit financing may forestall a depression but will create unique problems, Waugh said.

Forecasts Boom and Bust

What about timing? Can the mathematician, with the help of large-scale computers, help us to make intelligent choices and avoid alternate periods of boom and bust? Computers can, Waugh said, to the extent that they will help us to foresee the probable results of the various economic policies up for consideration.

Emphasizing the interdependence of economic factors and the effect that manpulation in one field can have on other fields, Waugh pointed to the repercussions on the overall economy that a policy of over support or under support of farm prices might have.

Excessive taxation or over-extended credit could also be disastrous to the country. Studies of monetary and credit controls, taxes, public works programs and social security were urged by the economist.

Aid to ERI'

Leontief illustrated a situation in which the new device could have been of great service. If the necessary information had been available, he said, one could have determined the rates of surplus imports of various commodities which would have been required to raise the domestic output of Marshall Plan countries from given original levels to prescribed higher levels over a stated period of time.

There are so many variables that must be taken into account in a given problem in the Social Sciences, Mosteller stated, that before the advent of the Mark III it was impossible to predict statistically the chances for marital success.

More Complex Problems

Until now, he added, little has been done with discriminate functions when it is desired to split the population into three or four parts. The lack of progress is mainly due to the very heavy computational work that would be associated with proper formulation of such a problem. As modern computing machinery becomes available to scientists, Mosteller stated, they will no longer be so reluctant to formulate problems which require heavy computation.

A typical problem in education would appear like this: We are given letter grades in four coursed for a certain number of individuals. We would like to pool these letter grades to form a scale of scholastic achievement. Since the distribution of grades in the several courses are quite different, even more variable must be considered in the problem.

In a speech entitled Physiology and Computational Devices, William J. Crosier, professor of General Physiology, cautions that so-called "thinking machines" such as the Mark III can never take the place of the human mind. No machine, he points out, could over not up a problem or analyze the results obtained through its own calculations.

It is, in fact, the setting up of the problem which poses the greatest problem for researchers now.

The value of calculators lies in their speed in handling difficult problems, Crozier said.

"This . . . point of interest finds parallels in, for instance, weather prediction or even in cosmology, where the basic difficulty is rather a mechanical one--arising from the degree of appropriateness of the mathematical system applied and the onerousness of using it, granted data which are sufficient. Beyond this is the second focus of interest, essentially more attractive and more stimulating. It centers upon the possibility of creating models of individual biological processes. Such models, both mathematical and material, have of course played a considerable role in general physiology,--as aids to the clarification and the concreteness of thinking, and as springboards for new experiments. Somewhat crudely put, but not unfairly the question has arisen: Do complex, fast, computation devices provide an effective model of mental processes, or even of one general class of human cerebral operations?--even to the extent that such machines, with developments of kinds now foreseeable, may be used to serve as surrogate for human decisions or actions.

Two Questions

"Note that there are here two distinct questions, Crozier continued. "The adequate imitation of a given kind of end result as achieved by an organism does not at all imply that the mechanism whereby the organism acts or decides has been duplicated. For engineering purposes, as in the "no hands" operation of a production line, this may be quite immaterial (so long as men keep the surrogate in good working order). But the physiologists's job is different. What he seeks is not merely an overall model. He really looks for an understanding of the actual mechanisms whereby the organic, biological machine operates. An even partially successful model may be enormously helpful in furthering this quest, in a particular case; but its character as an analogical crutch must not be lost sight of.

The biologist is not necessarily too grumbly about this--or at least I should not like to have you think that he is. He is perfectly able to accept, for a time being, a system of kinetc equations embracing the data of photosynthesis as picturing the known essence of the mechanism of this process. He also wants to learn the molecular inwardness of the matter, the kinetic mechanism with all its defects serving as a ladder toward specific experimental inquiry. He is not so crude as to look for a nexus of springs pulling dashpots through baths of hydraulic oil when he peers at muscle fibers in electron microscope pictures, though this kind of model has had brilliant uses for some purposes. On the other hand he knows in a practical way, as the logician knows, that reasoning by analogy from dynamical properties of the model is liable to stumble over imperfections in the analogy.