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.
2 The Promise of Technology In the mere half-century since their invention, computers have evolved from experimental curiosities to tools so widely used that today, the computer sector accounts for about 10 percent of the U.S. gross national product, and ahnost 10 percent of the nation's capital investment (CBEMA 1987~. Table 1 indicates the size and growth of the computer equipment industry, the core of the computer sector. Pitiably employed only for scientific and engineering calculations and later for certain business data processing calculations, computers are now used for innumerable practical applications of numerical and symbolic information processing in areas as diverse as manufactur- ing, education, communications, agriculture, medicine, and defense. In the world of business, machines that were once confined to payroll and accounting are now relied upon to help create documents, route messages, analyze financial data, conduct banking transactions, han- die airline reservations, run the telephone system, and gain access to the vast amounts of information stored in electronic databases. In the world of scientific calculation, machines once desired for calculat- ing numerical tables are now used to design transportation vehicles, guide satellites, predict the weather, explore for oil, increase food production, develop new pharmaceuticals, investigate the atom, and map the human genome. In the public sector, these machines have 7
v 1_ u) ~ - · - bO · - P4 o v c~ · - o ·. o ~2 0 ~ o ~ ~ x ~ 0 a, ~ ~ - o ~ u ~ ~o ~ o ~ - · - ~ ·= - ~ x l oD ~ l oN l oD a~ 1~) 1 1 1c°o oN 1 ~ l c~ 1 ~ ~ oo oD oN r~ cO a, oo a~ ~n cD oN oo o co oN H O ~ 00 0 . . . I - O ~ d. 1 0 O O ;` O O . . . . . 00 ~ 00 CO I O ~ . . O C~ N 01 . . CO O ~ \0 ~ ~ ~ ~ ~ ~D . . · · · · . N ~ Lt) 1 ' I 1 CO ~ ~ ~ m. . . . . Ln ~ 0 co ~r ~ ~ 1 ~1 d' a~ c~ ~ c~ ~ . . . . . ~ ~ LO ~ ~ CN ~ · . C~ ~ 1 . . 1 0\ 0 ln 1 ~ O OI mo I ~ r m kD ~ ON . . . . . C~ oo 0 0 I dl kD Ln O ~ t_ O ~ O · CO ON ~ ~ . . .. O 0N ~ O CO 0O 0 0 ~1 ~I C~ ~ N r o · CO ~ tD ~ · ~ . . .. ~ ~ oo ~ ~ o ~r oo ~ ~ r~ O 0\ CO kD s ~ oN · · · . · . O ~ o ~r co ~ ~n ~ 0 r ~ ~ ~ r~ . ~ oo ~ ~ · ON · . . . en a~ :~ ~ 0 Ln ~ In u~ OD CO 01 ~ ~ o, 0 ~ c, cn r~ 0 a, LO ~ ~ · \0 tD . . .. ~ ~ ~ ~ o c~ ~ a~ (D ~n ~ ko ~ a, ~ · C, ~ oo u, . ~ cn . . O 0 1 ~ O - 1 _1 1 N · ~r ~ ~ · tD ~ ~ 1 ' - 1 ~ O - o o - o ~ o o ~ ~r~ 0 ~ LC, ~ i'~ ~ ~ ~ ~ ~ ~ ~ ~ ot.~K ~o po~ o-: ~ ~ ~ ~ a) ~ rj ~ rj S. ~ ~ ~ ~ ~ _ ~ _ ~ 8 i~ .~ ~ ~ .m -¢ ·~) ~ L~ ·g o ~ ti g ~ ~ H ~ ~ 5~ ~ ~ .O o ~ .k ~ ~ ~ ~ _ ~ ~.5 83 !-~ S.i s~ -S l :r o iS t~ 75= o 06~ 3 0 ~ ~ ~ . §~= .. ~ '.~.~= ;~E ~ ; ~ ~'d' ~ ~ ~l ~ ~ ~S) ~ (h H
9 played a major role, not only in national defense, but also in the anal- ysis and management of the large amounts of information involved in such government programs as the decennial census and social se- curity. In short, they have brought about a revolution in the way we live in and think about the world and, in doing so, have become indispensable. A series of technological innovations rapidly changed computers and the way they are used. Time sharing, which distributed com- puter power from a single machine in a round-robin fashion among dozens of users in numerous locations, became commercially viable in the 1960s. In the 1970s very large scale integrated (VESI) circuits made possible the processor on a chip, which in turn made comput- ers ubiquitous, faster, cheaper, and more powerful while computer memories grew bigger, cheaper, and more reliable. In the 1980s, the personal computer has delivered cheaper computational and storage resources directly to the end user and captivated millions of people through easy-to-learn programs for spreadsheets, word processing, data bases, and business graphics. At the same time, computer networks became more widespread, interconnecting many personal and time-shared machines, thereby redefining the computing base together with a large array of sophisticated software. As this 30-year period draws to an end, processors have become thousands of times faster at constant cost, or thousands of times smaller and cheaper at constant performance, than when the period began. Current technological trends and pioneering research activities suggest con- tinuation, if not acceleration, of technological growth over the next decade accompanied by even more useful and powerful applications. The purpose of this chapter is to summarize briefly the techno- logical areas that, in the board's judgment, hold the greatest promise for influencing the field in the next decade and to speculate on the changes they may bring about in our world. The technologies iden- tified and their potential users are discussed in more detail in Part IT, along with some of the associated problems and prospects. In reading this chapter, the reader should keep In mind that we have selected key areas rather than attempting to provide a taxonomy of the field, and that we have excluded related technologies of commu- nications, semiconductors, packaging, and manufacturing, which are also necessary to meet the challenge facing the nation in computer science and technology.
10 MACHINES, SYSTEMS, AND SOFTWARE: Perhaps the greatest promise lies in the evolution of multi- processors systems that may harness hundreds, thousands, or po- tentially even millions of computers to work together on a single application (for example, transcription of human speech into text). Development of this technology is motivated by the fact that cur- rent computing power is wholly inadequate, by orders of magnitude, to perform most of the interesting applications of the future. The expectations from this technology are roughly analogous to those of harnessing several horses to a cart: they are more economical than one powerful horse; their number can be adjusted to match the load; and working together they can exceed by far the power of even the strongest animal, thereby making possible qualitatively different achievements. As with horses, effective ways must be developed to harness these machines in order to exploit their power. Multiprocessors over generic and broad potential utility. The technology involves new computational engines that are economical, scalable, and of potentially far greater power than those available today. That power should help multiprocessors achieve ambitious new applications of artificial intelligence, such as real-time speech understanding, machine vision, learning, natural language under- standing, and better machine reasoning. If successful, these applica- tions would open an entirely new world of computer uses. Finally, by linking together a large number of the most powerful superproces- sors, multiprocessor architectures might even lead to ultracomputers that could truly expand the capabilities of physical science through computer simulation of immense problems, thereby creating a new set of scientific tools, such as major computational observatories, computational microscopes, computational chemical or biochemical reactors, and computational wind tunnels. Beyond multiprocessors, another major direction in the systems area involves the interconnection via networks of geographically re- mote computers into distributed systems. A distributed system can consist of a handful of interconnected machines in a modest-sized office, of a few thousand machines in a large corporation, or even a few million computers belonging to individuals and organizations throughout the country. Unlike the processors in a multiprocessor, which work on the same task under central control, the computers of a distributed system work mainly on different tasks under the control of their different users. The power of a distributed system
11 of computers versus that of an equal number of independent nonin- teracting computers comes from its ability to intercommunicate in order to exchange the information needed for or supplied by the lo- cal computations. Distributed systems mirror human organizations and individuals, which, though largely autonomous in their work, communicate occasionally with one another toward achieving indi- vidual as well as common goals. Such systems therefore prorn~se to enhance information-related functions within and between organiza- tions, from routine office tasks to commercial transactions. Software is needed to realize and expand computer capabilities so that they are conceptually as well as physically accessible to end users. Software consists of the collection of computer instructions that specializes general purpose computers to their applications. Pro- gramming, the task of generating software, is difficult and expensive for a variety of reasons that we discuss in Chapter 6. The expected proliferation of multiprocessors and distributed systems will create further software and programming challenges. Nevertheless, devel- opment of software may achieve greater progress than in the past through development and use of tools and techniques for increasing software productivity, a crucial goal for researchers and industry. Reaping the benefits of computers will depend on improving the interface with the user. Graphics and visualization are one source of the necessary improvements. Until recently, computer graphics had been used primarily by scientists and engineers. With the large-scale introduction of personal computers and workstations with bit-map displays, graphics is fulfilling its promise. Direct manipulation of objects on the screen is replacing traditional, much less user-friendly interaction via typed command languages. As a result, sophisticated computer technology is becoming widely accessible to casual users, lay persons, and even young children.* Improvements in graphics, visualization, and the user interface in general will draw on advances in computer hardware and software and on inputs from cognitive psychologists and other experts on interactions between people and machines. Graphics and visualization are having a particularly strong im- pact on scientific research, especially when supercomputers, which *Much of this technology, popularly associated with the Apple Macintosh, emanates from developments at Xerox Palo Alto Research Center beginning in the early 1970~.
12 generate and process massive amounts of data, are involved (Mc- Corm~ck et al. 1987~. Fields such as molecular modeling and com- putational chemistry, solids modeling for mechanical engineering, computational fluicis dynamics, and computational astronomy re- quire visualizations of considerable complexity involving the use of color-shaded, (pseudo-) realistically portrayed objects and data. Ad- vanced applications are beginning to call for animated as well as static unages. ARTIIlICLAL INTELLIGENCE The expected growth of multiprocessors and continued research strides in speech understanding and machine vision are expected to advance sensory computing substantially during the next decade. To the extent that sensory systems become as successful as we expect, they will have a dramatic impact on the way people interact with computers, since speech and vision, unlike typing, are natural means of human communication. Thus, sensory systems can make comput- ers easier and faster to use and therefore accessible to a wider range of people than they are today. Another important aspect of artificial intelligence (Al) is ex- pert systems. These systems represent and use human knowledge for the solution of problems in specialized domains that are difficult enough to require significant human expertise for their solution. For example: in manufacturing, a well-known expert system is given a customer order for a computer installation and then designs a manufacturable configuration of the subsystems and schecluTes pro- duction; in finance, expert systems assist bank officers in deciding the credit worthiness of loan applications. Currently available expert systems present only the beginnings of what may someday be pos- sible. The evolution of more powerful multiprocessors, paradigms, and algorithms supporting ongoing research in the representation, acquisition, and utilization of the knowledge needed by expert shy tems is expected to increase their usefulness and ubiquity, blending them into the general stream of computer systems and applications. Machine intelligence ~ also advancing through deeper cognitive systems and, in particular, machine learning. Unlike expert systems, which are preprogrammed with expert knowledge, learning systems are capable of learning from the tasks they perform through practice, much as people do. Finally, robotics is another technology that involves machines
13 that are sufficiently intelligent to interact with the physical world to perform designated tasks. Robotics builds on sensory computing as well as mechanisms capable of subtle motions. The evolution of effective robotic systems could result in several benefits increased factory productivity and, perhaps as significantly, the ability to pro- duce individually tailored products at mass production costs. THEORETICAL COMPUTER SCIENCE As a young discipline, computer science is in the process of building up its theoretical base and will probably continue to do so for many years to come. Until that base of theory is more fully developed, we wiD be able to use computers to solve only a tiny fraction of known problems in theory and applications. The utility of theory in computer science, as in other more mature fields, is that it helps to order and explain complex phenomena through simple laws, it discovers limits on what is possible, and it guides the discovery of new principles and new possibilities for computers. To date, there have been important advances in the areas of computational complexity, which considers the intrinsic difficulty of solving a given problem; in algorithms and their analysis, whereby new procedures are sought to solve difficult problems; and In se- mantics and languages, whereby as a result of theoretical insights, certain unport ant system programs (compilers) are now routinely constructed. Theory has also contributed to cryptology, whereby methods have been developed for ensuring the privacy and authenti- cation of computer messages. CONCLUSION The technological developments highlighted above, along with the existing technological base of computers, paint a picture of formidable prospective tools for the Information Age. However, if these tools are to be widely available and truly effective, several steps must be taken to facilitate their development and use. Basic research and better, more widespread data networking are two of the most important such steps.