in missile technology are continually threatening the whole offensive or defensive apparatus of one side or the other. Keeping ahead in the technological race is not in itself a guarantee of security in these circumstances; it remains essential to incorporate the technology in operational hardware ("forces in being") and to deploy them and use them with skill and intelligence. But no amount of production, skill, and intelligent use can compensate for significant technological inferiority. Our atomic monopoly in 1945 would have insured victory over Japan even if the ratio of our conventional forces to theirs had been reversed. Clever operations research can in exceptional cases improve some capability of a force by a factor of 3 to 10. Inventions can frequently change the same sort of capability by factors of 10 to 1,000.

Of course any particular nation will borrow much of its scientific knowledge and even its military technology from other countries. Science is international. Once a scientific discovery is made, it quickly becomes the property of scientists everywhere; frequently the same discovery is made nearly simultaneously in two or more widely separated research centers. Scientists in all leading countries (certainly including Russia) were familiar at the beginning of the Second World War with the basic scientific ideas that prompted us to initiate the Manhattan District Project. Few if any of these ideas were American in origin.

To a lesser extent the same is true of developments in military hardware. Despite military secrecy, the news of a new development leaks, and this leads to frantic copying on the part of potential enemies. The British "invented" the tank in World War I, but they disclosed it prematurely, and the most effective use of the tank in that war was made by the Germans. It is only a question of time before many additional countries perfect advanced nuclear weapons. Espionage is responsible for some of the transmission of new technological ideas from country to country, but espionage rarely does more than speed an inevitable process. New military hardware can usually be observed without too much difficulty when it becomes operational. In the United States and most Western countries full accounts are usually made available earlier than that to anyone who can read the public and technical press and the advertisements of contractors. Even if technical details are not revealed, the mere knowledge that a research objective has been achieved is an enormous help and stimulus. It means that the objective is achievable that there is a treasure to be found and that a search (for that is what research is) is therefore definitely worthwhile. Other countries might not be developing atomic weapons today if we had not demonstrated their feasibility.

Therefore the net gains from a successful military development can only be counted on for a few years at most. If a nation's civilian economy is made more productive by research, the nation continues to gain from

the increased productivity even if other countries do copy (as of course they will). But military strength is in its very essence relative. Accretions to our absolute strength are offset if potentially unfriendly nations make equivalent additions to theirs. And we must, in general, expect them to do so when a new scientific idea is responsible for our gain.2

Nevertheless, those few years of superiority may be decisive. Consider how our security would be prejudiced if the Russians had the technological initiative and we were reduced to copying their successes with a lag of several years (a by no means inconceivable circumstance since their launching of the first satellite). Suppose the Russians had had a four-year lead with the A-bomb. Imagine them with a similar lead in operational intercontinental ballistic missiles. Or in effective anti-ICBM's. It is misleading to say that primacy in military research and development can give us only lead time. This may be enough to prevent or "win" a war, and, for a nation on the strategic defensive, is essential to avoid defeat.

KINDS OF MILITARY RESEARCH AND DEVELOPMENT

Nomenclature in the field of research and development is thoroughly confused, partly because the processes by which new scientific and technical knowledge is wrested from nature are not well understood and are anything but neat and orderly. It is customary to use some such classification as the following in discussing problems of military research and development policy management :3

1. Weapon system development, directed toward the fabrication for testing of prototypes of operational weapons. In the case of missiles (and some other systems), this may mean the fabrication of rather large test inventories. The term "weapon system" in this context is used in a narrower sense than we have used it in discussing systems analysis. The system being developed typically includes the major equipment (say a bomber aircraft) and such auxiliary equipment as power plant, bombingnavigation system, other electronics gear, and armament, but would not necessarily include an operational concept, or base system, or personnel training program.

2 This by no means implies that all scientific ideas are “neutral” as between nations. The question whether the development of A- and H-bombs is on balance favorable to the West (because it enables us to offset the alleged Russian superiority in manpower) or to the Russians (because it robs us of our previously decisive advantage in mobilization potential), or unfavorable to both (because it threatens mutual destruction) is still being ardently debated. There may be aspects of national geographies, economies, or psychologies which make any given invention more useful to one country than to another.

8 This classification is not necessarily identical with that of the National Science Foundation or of other Agencies. Also, it might be noted that there are not sharp lines of demarcation between these categories. Nonetheless, the use of such a classification is essential to the understanding of R and D problems.

2. Component development, that is, the more or less independent development of items of military hardware which are not themselves "weapon systems" or "major equipment," but parts or sub-systems of such major equipment. The distinction between systems and sub-systems is, of course, arbitrary. The important thing is that the Services can and do spend money productively developing engines for aircraft, guidance systems for missiles, gyroscopes for guidance systems, communication systems for armored vehicles, gear for landing aircraft on carriers, and so on, without knowing precisely the type or model of the major equipment of which the component will be a part.

3. In exploratory development, or applied research the objective is advancing the state of the technological art rather than obtaining immediately operational hardware. Someone has had a promising idea for a new type of vacuum tube or a new fuel or a new design to reduce drag in an aircraft wing. The idea may have emerged from paper research or may simply have been someone's inspiration. Whatever its origin it requires some sort of testing or verifying before it can be accepted as valid and incorporated in the state of the art. This may involve a laboratory experiment, a wind tunnel test, or the construction of a working model, depending upon the nature of the idea. Test models are typically very much cheaper than operational models, although there are some exceptions.

4. Basic research is the deliberate search for knowledge. The military Services, AEC, NASA, and so on, support a good deal of basic research in areas of special interest to them, such as aerodynamics, atomic physics, and some branches of mathematics. Basic research is typically the cheapest type of research and development if we think of something like "cost per idea explored." It frequently requires no equipment except paper, and with some exceptions (betatrons, for example) even the scientific laboratories and computing equipment that are needed cost less than a single operational prototype of a new aircraft or missile.

We estimate that, using generally accepted definitions, military research and development in the late 1950's was costing about $6 billions per annum. This estimate is both arbitrary (the boundary between "development" and "production" is hard to draw) and misleading, since by far the largest part of this total is at the weapon system end of the research and development spectrum and includes large costs of fabricating operational prototypes. Expenditure really devoted to advancing science and technology in areas of special military interest is much less — perhaps $1 to $2 billions. This would include basic research (on the order of

'For example, when the Air Force (or Navy) contracts for a new model of aircraft, making the first "prototype" batch is regarded as development. But when the Navy contracts for a new model of submarine, constructing the very first is regarded as production.

$150 to $200 million), what we have called exploratory development, and some genuine state-of-the-art advancement undertaken in connection with component and weapon system development.

Most weapon system and component development is undertaken by industrial firms on contract with one of the Services, although the Navy and Army still do some in their yards and arsenals. All three Services support extensive test facilities as well as administrative staffs to determine requirements, to let contracts, and to supervise performance under contracts. Research and exploratory development is divided between internal and external research centers and laboratories. A major share of the basic research is performed on contract by laboratories and individual scientists in universities.

PROBLEMS OF RESEARCH AND DEVELOPMENT

It is apparent that many of the problems of managing a research and development program have important economic aspects.

For example, the most fundamental problem of all, how much to spend on the program, is an allocation problem. Economic theory tells us that we "should" spend on research and development until the marginal gain from the expenditures, properly discounted, is just equal to the gain from expenditures elsewhere. But in the case of research and development, whether it is performed for a military Service or a business firm, this formula is peculiarly difficult to interpret or apply. The gain is much more uncertain - much harder to predict with accuracy than the gain from, say, an additional fighter defense squadron with aircraft of known performance. This is particularly true of basic research and exploratory development, where the product, if any, will be knowledge- and knowledge usually far removed from any practical end use. Calculating gains from research and development is further complicated by the need for discounting. Typically the payoffs from research and development are expected in a more distant future than payoffs from procurement, since there is no enhancement of military capability until the results have been incorporated in operational hardware, procured, and deployed in operational units. This means that the discounted values of the payoff are highly sensitive to rates of discount - and as we have seen, there is great uncertainty about which discount rates are most appropriate.

These uncertainties make it difficult or, in many cases, impossible to use an explicit economic calculus to determine how much of total resources should be allocated to research and development; how these resources should be divided among the various kinds of research and development; and which specific projects should be selected. At the end of this chapter we will return to the question of what role economic calculations can play in answering such questions. Short of such calcula

tions, however, general economic reasoning and analysis of the institutional factors affecting the research and development process can throw some light on the way these problems should be approached, and on some common pitfalls and mistakes in research and development policies and management.

It is our opinion that the three main root causes of error in the management of military research and development are: (1) a failure to appreciate the degree of uncertainty involved in research and development or, what has the same effect, a tendency to ignore uncertainties and act as if they were certainties; (2) a pervasive tendency, discussed in Chapter 11, to undervalue future outputs relative to current ones; and (3) certain characteristic tendencies of bureaucracy, particularly the strength of centripetal (centralizing) forces.

Of these the most serious and destructive is the first. The important thing to appreciate in making good decisions with respect to research and development is the dominant role played by uncertainty. Research and development are uncertain by definition. Research is a search, and one rarely knows in advance whether the search will be successful at all, let alone how long it will take or which route will lead to the treasure. The military Services have all too frequently tried to command the research and development community to invent new weapons to specification, just as they would command a platoon of infantry to march by the right flank. One of the Service manuals attempts to prescribe research and development planning procedures as follows:

Planning for research and development is broadly similar to planning for a military operation. . . . The objective must be clearly identified, the operation or job must be divided into component parts, units or organizations must be assigned to accomplish each part, the time and resources required for each part must be estimated, and all efforts must be time-phased in proper relationship to each other.5

But the essence of research and development is that it is not like the military operations for which this kind of planning and scheduling is appropriate. Among research and development activities, this type of planning is least inappropriate for weapon system development — al

*ARDC Program Management Procedures, ARDC Manual No. 80-4, Headquarters, Air Research and Development Command, United States Air Force, Baltimore, Maryland, September 1, 1956, p. 11.

This is not to say that R and D planning is the same in 1960, or will be the same in the future, as in the past. In the U.S. Air Force, for example, some recent changes in procedures indicate a greater recognition of the crucial role of uncertainties in R and D planning.

This kind of planning is equally inappropriate for a good many classes of military operations, where uncertainties demand flexibility, probing techniques to gain information, improvisation on the battlefield, etc., rather than neat, orderly scheduling in advance to achieve a clearly identified objective.