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.
CHAPTER SEVEN DIGITAL COMPUTERS IN THE LIFE SCIENCES Throughout organized society we are engaged in a massive experiment to learn how to integrate the technology of information processing by com- puter into the working fabric of each area of endeavor. This experiment started in some areas two dozen years ago; in some it is just starting, and in others it is still largely in the planning stage. These experiences have begun to provide a picture of the process through which a given field assimilates the computer. No field is much beyond the beginnings of this experiment, as a retrospective study 20 years from now will undoubtedly demonstrate. To assess the role of the computer in a particular field in this case, the life sciences-some general lessons must be understood and some questions examined. After presenting the general lessons rather briefly and dogmatically, we shall ask some appropriate questions concern- ing the life sciences. Together, these two approaches lead to a number of recommendations concerning management of computers for the life sciences in the immediate future. GENERAL FACTS ABOUT COMPUTER USAGE 1. The computer is a general processor of symbolic information. Most importantly, the computer is a general device for processing symbolic in 385
386 THE LIFE SCIENCES formation. It is not just a device for doing arithmetic fast, and information need not be in numerical form for the computer to "understand" it. Because all tasks employing symbolized information are candidates for performance by the computer, there is a great diversity of uses of computers. 2. Limits on computer application In a given field. Though all symbolic tasks are candidates, not all such tasks are appropriately performed by the computer at any moment. The most important determinants are: The power of existing computers: speed, memory, file space, and re- liability. The ways existing computers can interact with the outside world: printers, visual displays, direct coupling to instruments, typewriters, and so on. The cost of existing computers: dollars per million operations, or per thousand bits of memory. Our ability to make the computer perform, viz., programming techniques; of all the multitude of things computers can potentially do, we have suc- ceeded in getting them to do very few. The financial resources available to the field. The sophistication of practitioners in the field with respect to computers, especially their programming and operation. 3. Cost effectiveness continuously and radically increases. More than any other device man has ever built, the power of computers continuously increases from year to year, while costs per computation decrease. The yearly increment of these changes is huge, compared with such changes in other technological products ( e.g., the airplane or metal products ~ . Figure 39 shows the inverse nature of these changes. Estimates of costs and computer capability must be radically revised every five years. The inter- faces between the computer and the world are also changing, but much more slowly and conventionally. 4. Learning to accomplish a new type of task with a computer is analo- gous to "development" in other fields. The term "development" is used in its standard industrial meaning, i.e., development as opposed to re- search, prototype construction, and production. When initiating research it is not certain the job is possible, for essential knowledge and techniques are missing and must be discovered or invented. At the development stage the job can be done, but since it has never been done before, a multitude of unknown difficulties are certain to emerge, each of which can generally be overcome by further application of manpower and time. When proto- type construction is undertaken, all the elements of the process are under- stood, but the total process must still be organized into a smoothly working
DIGITAL COMPUTERS IN THE LIFE SCIENCES 387 109 108 o a, ~ 107 Q ·- 106 Q o 105 / 104 , 1955 1960 1965 1970 1975 A. Computer speed In o ·- .10 a) Q o _ .01 . _ $10 .001 .0001 _ .00001 1 1 1 1 1955 1960 1965 1970 1975 1 1 1 B. Computer cost FIGURE 39 A. Computer speed. Storage speed is expressed in thousands of addi- tions per second. B. Computer cost. Storage cost is expressed in dollars per million additions. (Adapted from P. Armer, "Computer aspects of technological change, automation, and economic progress," in The Outlook for Technological Change and Employment, Appendix Volume 1, Technology and the American Economy, The Report of the Commission, Studies Prepared for the National Commission on Tech- nology, Automation, and Economic Progress, U.S. Government Printing Office, Washington, D.C., February 1966.) whole, until, at the production stage, only maintenance activity is required to keep things working efficiently. In many instances, the application of computers still requires research, e.g., to achieve automatic processing of microscopic images for chromosome analysis. But the bulk of computer applications begin at the development stage; development is expensive and always takes longer than anticipated. Once we understand what we require the computer to do, we cannot simply tell it to do so, viz., write down instructions. Each time, a new, relatively large program has to be written. Hence, getting the computer to achieve the new task successfully is an expensive, time-consuming development job. 5. Assimilation of computers by a field must be accomplished by the practitioners themselves. It has often seemed that successful practice in the use of computers in one field can be transferred to another (since they seem to be "doing the same thing," regardless of the field) or that the computer industry can develop the application systems and provide them to new fields as a marketing service. Both assumptions are false, though both