Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
402 THE LIFE SCIENCES TABLE 62 Computing Costs Funded by Different Sources for Selected Research Areas COMPUTING COSTS, BY SOURCE (PERCENTAGE OF B-HOURS) RESEARCH AREA Federal Funds Own Life Sciences Funds of All Research Computing Non-Life- Other Unknown Sources Grant Center Sciences Funds Funds Source AVERAGE ALL BIOLOGY 100 42 29 9 11 Behavioral Biology 100 49 39 5 6 1 Cellular Biology 100 72 13 2 11 2 Developmental Biology 100 28 13 1 57 1 Disease Mechanisms 100 31 58 2 6 3 Ecology 100 40 19 11 3 26 Evolutionary and Systematic Biology 100 34 4 42 6 14 Genetics 100 50 26 1 12 8 Molecular Biology and Biochemistry 100 38 30 7 14 8 Morphology 100 51 14 15 16 4 Nutrition 100 32 12 12 19 24 Pharmacology 100 68 11 1 11 9 Physiology 100 44 24 5 9 13 Source: Survey of Individual Life Scientists, National Academy of Sciences Committee on Research in the Life Sciences. ~.xamnle almost all the computing of the cellular biologists is supported r 7 ~ ~ _ ~ A from research grants, while a much smaller fraction is thus funded In the areas of ecology, and evolutionary and systematic biology. Biologists in the latter two areas reported a large percentage of unknown support, which probably implies the receipt of "free" computation at a university computa- tion center. CONCLUSIONS AND RECOMMENDATIONS When this picture of computing in the life sciences in 1967 is combined with the basic lessons about the evolution of the computer industry and of computer use in any field, the following conclusions and recommenda- tions emerge. 1. Basic continuing support. Computers are now an integral part of the life sciences, through all its subdomains. Computing costs must be
DIGITAL COMPUTERS IN THE LIFE SCIF,NCES added to space, personnel, and instruments as basic continuing support items. As with these other resources, computing time should be awarded to the individual investigator or research project according to the merits of the research, through the regular funding channels. If funds permit, computing will continue to increase during the coming years as a fairly predictable increased percentage of life scientists (now probably of the order of 10-15 percent new scientists per year) come to use computers, and there is an unknown increase in the number of the already heavy users. The increased use by the new scientists will slow toward the base growth rate of the field in about five years, but the proportion of medium and heavy users will continue to increase for a long time. Offsetting the expense of this growth, while at the same time tending to increase its rate, is the decreasing cost of computation. The emphasis on funding computation in relation to the quality and significance of the research to which it will contribute, is not meant to ignore the need for stability that large facilities require. Indeed the need for stability constitutes the main pressure for block funding of facilities. However, the experience with block funding of computer facilities (e.g., at the National Institutes of Health) makes it clear that one must move as rapidly as possible to associate with each research effort the cost of its computing and then assign to investigators both the freedom and the xe- sponsibility to obtain computing funds appropriate to their research. 2. Multiplicity of facilities and decentralization of control. The extreme diversity of uses of computers implies need for an equal diversity of computing facilities. As we have said, the generality of the computer is that it can be shaped (configured and programmed) for almost any type of information-processing task, but it cannot simultaneously be all things to all users. In fact, all computer facilities become highly specialized, the large university computing center being no exception. A part of the history of computing in the life sciences is written in the struggle of research groups to obtain computers of their own, which can be shaped to their own uses- as laboratory instruments or as data-retrieval and display systems, for example. A second reason for decentralization of the selection, development, and control of computers is that only through the parallel attempt of many life scientists to adapt the computer to their needs will the computer play its appropriate role in life science research. If the technical development is isolated in a relatively few centralized centers, this assimilation will be substantially retarded. 3. Development. The life sciences can rely on the computer industry to continue to produce cheaper, more powerful technology. They cannot, however, look to anyone else for their development-that is, to assimilate 403
404 THE LIFE SCIENCES the computer into work on life science problems. Development is expensive and unsatisfactory, in that initial goals are seldom met. It is frustrating, in that frontier projects always have troubles that later projects (often ex- ploiting better and cheaper technology) seem to avoid, making it appear that originally the wrong approach was taken. But this is characteristic of development, and is the price of getting the computer fully assimilated into particular fields. There are no special "computer areas" in the life sciences. All subareas are assimilating the computer, though in somewhat different ways and, currently, at different rates. Every subarea has its unique forms of symbolic processing, which, as it is successfully developed, can make large differences in the progress of research in the subarea. For example: image processing. generally in microscopy, but also in ecology; large-scale simulations in ecology; laboratory computerized instruments in physiology and biochem- istry; taxonomic retrieval systems in systematics. Almost all the heavy users in our census would reveal somewhat special developments. The projects in the life sciences noted above have analogues in other fields, which may short-circuit the research effort but not the development effort. However, life scientists have some symbolic functions in common with all other scientists and all other technologically oriented professionals. The best example is small numerical calculations analogous to engineering calculations. The computer field can be relied on for development relating to these for instance, in the multitude of time-consuming mathematical calculations necessary to untangle and to understand complex biochemical reactions. Another example is the development of time-sharing, which is of immense importance in getting the computer widely assimilated. Massive development efforts repeatedly produce wedges that open up new technology. The life sciences must support such efforts for themselves. The amounts of money spent on such projects may often seem out of proportion to what the same amounts, distributed otherwise, could yield, but this is an illusion, for there is no other way to gain entry into new areas other than by paying the apparently "excessive" costs of large development efforts.