J. C. R. Licklider
IRE Transactions on Human Factors in=20
Electronics,
volume HFE-1, pages 4-11, March 1960
Man-computer symbiosis is an expected development in = cooperative=20 interaction between men and electronic computers. It will involve very = close=20 coupling between the human and the electronic members of the = partnership. The=20 main aims are 1) to let computers facilitate formulative thinking as = they now=20 facilitate the solution of formulated problems, and 2) to enable men = and=20 computers to cooperate in making decisions and controlling complex = situations=20 without inflexible dependence on predetermined programs. In the = anticipated=20 symbiotic partnership, men will set the goals, formulate the = hypotheses,=20 determine the criteria, and perform the evaluations. Computing = machines will=20 do the routinizable work that must be done to prepare the way for = insights and=20 decisions in technical and scientific thinking. Preliminary analyses = indicate=20 that the symbiotic partnership will perform intellectual operations = much more=20 effectively than man alone can perform them. Prerequisites for the = achievement=20 of the effective, cooperative association include developments in = computer=20 time sharing, in memory components, in memory organization, in = programming=20 languages, and in input and output equipment.
The fig tree is pollinated only by the insect Blastophaga = grossorun.=20 The larva of the insect lives in the ovary of the fig tree, and there it = gets=20 its food. The tree and the insect are thus heavily interdependent: the = tree=20 cannot reproduce wit bout the insect; the insect cannot eat wit bout the = tree;=20 together, they constitute not only a viable but a productive and = thriving=20 partnership. This cooperative "living together in intimate association, = or even=20 close union, of two dissimilar organisms" is called symbiosis [27].=20
"Man-computer symbiosis is a subclass of man-machine systems. There = are many=20 man-machine systems. At present, however, there are no man-computer = symbioses.=20 The purposes of this paper are to present the concept and, hopefully, to = foster=20 the development of man-computer symbiosis by analyzing some problems of=20 interaction between men and computing machines, calling attention to = applicable=20 principles of man-machine engineering, and pointing out a few questions = to which=20 research answers are needed. The hope is that, in not too many years, = human=20 brains and computing machines will be coupled together very tightly, and = that=20 the resulting partnership will think as no human brain has ever thought = and=20 process data in a way not approached by the information-handling = machines we=20 know today.=20
As a concept, man-computer symbiosis is different in an important way = from=20 what North [21] has called "mechanically extended man." In the = man-machine=20 systems of the past, the human operator supplied the initiative, the = direction,=20 the integration, and the criterion. The mechanical parts of the systems = were=20 mere extensions, first of the human arm, then of the human eye. These = systems=20 certainly did not consist of "dissimilar organisms living together..." = There was=20 only one kind of organism-man-and the rest was there only to help him.=20
In one sense of course, any man-made system is intended to help man, = to help=20 a man or men outside the system. If we focus upon the human operator = within the=20 system, however, we see that, in some areas of technology, a fantastic = change=20 has taken place during the last few years. "Mechanical extension" has = given way=20 to replacement of men, to automation, and the men who remain are there = more to=20 help than to be helped. In some instances, particularly in large=20 computer-centered information and control systems, the human operators = are=20 responsible mainly for functions that it proved infeasible to automate. = Such=20 systems ("humanly extended machines," North might call them) are not = symbiotic=20 systems. They are "semi-automatic" systems, systems that started out to = be fully=20 automatic but fell short of the goal.=20
Man-computer symbiosis is probably not the ultimate paradigm for = complex=20 technological systems. It seems entirely possible that, in due course,=20 electronic or chemical "machines" will outdo the human brain in most of = the=20 functions we now consider exclusively within its province. Even now, = Gelernter's=20 IBM-704 program for proving theorems in plane geometry proceeds at about = the=20 same pace as Brooklyn high school students, and makes similar = errors.[12] There=20 are, in fact, several theorem-proving, problem-solving, chess-playing, = and=20 pattern-recognizing programs (too many for complete reference [1, 2, 5, = 8, 11,=20 13, 17, 18, 19, 22, 23, 25]) capable of rivaling human intellectual = performance=20 in restricted areas; and Newell, Simon, and Shaw's [20] "general problem = solver"=20 may remove some of the restrictions. In short, it seems worthwhile to = avoid=20 argument with (other) enthusiasts for artificial intelligence by = conceding=20 dominance in the distant future of cerebration to machines alone. There = will=20 nevertheless be a fairly long interim during which the main intellectual = advances will be made by men and computers working together in intimate=20 association. A multidisciplinary study group, examining future research = and=20 development problems of the Air Force, estimated that it would be 1980 = before=20 developments in artificial intelligence make it possible for machines = alone to=20 do much thinking or problem solving of military significance. That would = leave,=20 say, five years to develop man-computer symbiosis and 15 years to use = it. The 15=20 may be 10 or 500, but those years should be intellectually the most = creative and=20 exciting in the history of mankind.=20
Present-day computers are designed primarily to solve preformulated = problems=20 or to process data according to predetermined procedures. The course of = the=20 computation may be conditional upon results obtained during the = computation, but=20 all the alternatives must be foreseen in advance. (If an unforeseen = alternative=20 arises, the whole process comes to a halt and awaits the necessary = extension of=20 the program.) The requirement for preformulation or predetermination is=20 sometimes no great disadvantage. It is often said that programming for a = computing machine forces one to think clearly, that it disciplines the = thought=20 process. If the user can think his problem through in advance, symbiotic = association with a computing machine is not necessary.=20
However, many problems that can be thought through in advance are = very=20 difficult to think through in advance. They would be easier to solve, = and they=20 could be solved faster, through an intuitively guided trial-and-error = procedure=20 in which the computer cooperated, turning up flaws in the reasoning or = revealing=20 unexpected turns in the solution. Other problems simply cannot be = formulated=20 without computing-machine aid. Poincare anticipated the frustration of = an=20 important group of would-be computer users when he said, "The question = is not,=20 'What is the answer?' The question is, 'What is the question?'" One of = the main=20 aims of man-computer symbiosis is to bring the computing machine = effectively=20 into the formulative parts of technical problems.=20
The other main aim is closely related. It is to bring computing = machines=20 effectively into processes of thinking that must go on in "real time," = time that=20 moves too fast to permit using computers in conventional ways. Imagine = trying,=20 for example, to direct a battle with the aid of a computer on such a = schedule as=20 this. You formulate your problem today. Tomorrow you spend with a = programmer.=20 Next week the computer devotes 5 minutes to assembling your program and = 47=20 seconds to calculating the answer to your problem. You get a sheet of = paper 20=20 feet long, full of numbers that, instead of providing a final solution, = only=20 suggest a tactic that should be explored by simulation. Obviously, the = battle=20 would be over before the second step in its planning was begun. To think = in=20 interaction with a computer in the same way that you think with a = colleague=20 whose competence supplements your own will require much tighter coupling = between=20 man and machine than is suggested by the example and than is possible = today.=20
The preceding paragraphs tacitly made the assumption that, if they = could be=20 introduced effectively into the thought process, the functions that can = be=20 performed by data-processing machines would improve or facilitate = thinking and=20 problem solving in an important way. That assumption may require = justification.=20
Despite the fact that there is a voluminous literature on thinking = and=20 problem solving, including intensive case-history studies of the process = of=20 invention, I could find nothing comparable to a time-and-motion-study = analysis=20 of the mental work of a person engaged in a scientific or technical = enterprise.=20 In the spring and summer of 1957, therefore, I tried to keep track of = what one=20 moderately technical person actually did during the hours he regarded as = devoted=20 to work. Although I was aware of the inadequacy of the sampling, I = served as my=20 own subject.=20
It soon became apparent that the main thing I did was to keep = records, and=20 the project would have become an infinite regress if the keeping of = records had=20 been carried through in the detail envisaged in the initial plan. It was = not.=20 Nevertheless, I obtained a picture of my activities that gave me pause. = Perhaps=20 my spectrum is not typical--I hope it is not, but I fear it is.=20
About 85 per cent of my "thinking" time was spent getting into a = position to=20 think, to make a decision, to learn something I needed to know. Much = more time=20 went into finding or obtaining information than into digesting it. Hours = went=20 into the plotting of graphs, and other hours into instructing an = assistant how=20 to plot. When the graphs were finished, the relations were obvious at = once, but=20 the plotting had to be done in order to make them so. At one point, it = was=20 necessary to compare six experimental determinations of a function = relating=20 speech-intelligibility to speech-to-noise ratio. No two experimenters = had used=20 the same definition or measure of speech-to-noise ratio. Several hours = of=20 calculating were required to get the data into comparable form. When = they were=20 in comparable form, it took only a few seconds to determine what I = needed to=20 know.=20
Throughout the period I examined, in short, my "thinking" time was = devoted=20 mainly to activities that were essentially clerical or mechanical: = searching,=20 calculating, plotting, transforming, determining the logical or dynamic=20 consequences of a set of assumptions or hypotheses, preparing the way = for a=20 decision or an insight. Moreover, my choices of what to attempt and what = not to=20 attempt were determined to an embarrassingly great extent by = considerations of=20 clerical feasibility, not intellectual capability.=20
The main suggestion conveyed by the findings just described is that = the=20 operations that fill most of the time allegedly devoted to technical = thinking=20 are operations that can be performed more effectively by machines than = by men.=20 Severe problems are posed by the fact that these operations have to be = performed=20 upon diverse variables and in unforeseen and continually changing = sequences. If=20 those problems can be solved in such a way as to create a symbiotic = relation=20 between a man and a fast information-retrieval and data-processing = machine,=20 however, it seems evident that the cooperative interaction would greatly = improve=20 the thinking process.=20
It may be appropriate to acknowledge, at this point, that we are = using the=20 term "computer" to cover a wide class of calculating, data-processing, = and=20 information-storage-and-retrieval machines. The capabilities of machines = in this=20 class are increasing almost daily. It is therefore hazardous to make = general=20 statements about capabilities of the class. Perhaps it is equally = hazardous to=20 make general statements about the capabilities of men. Nevertheless, = certain=20 genotypic differences in capability between men and computers do stand = out, and=20 they have a bearing on the nature of possible man-computer symbiosis and = the=20 potential value of achieving it.=20
As has been said in various ways, men are noisy, narrow-band devices, = but=20 their nervous systems have very many parallel and simultaneously active=20 channels. Relative to men, computing machines are very fast and very = accurate,=20 but they are constrained to perform only one or a few elementary = operations at a=20 time. Men are flexible, capable of "programming themselves contingently" = on the=20 basis of newly received information. Computing machines are = single-minded,=20 constrained by their " pre-programming." Men naturally speak redundant = languages=20 organized around unitary objects and coherent actions and employing 20 = to 60=20 elementary symbols. Computers "naturally" speak nonredundant languages, = usually=20 with only two elementary symbols and no inherent appreciation either of = unitary=20 objects or of coherent actions.=20
To be rigorously correct, those characterizations would have to = include many=20 qualifiers. Nevertheless, the picture of dissimilarity (and therefore = p0tential=20 supplementation) that they present is essentially valid. Computing = machines can=20 do readily, well, and rapidly many things that are difficult or = impossible for=20 man, and men can do readily and well, though not rapidly, many things = that are=20 difficult or impossible for computers. That suggests that a symbiotic=20 cooperation, if successful in integrating the positive characteristics = of men=20 and computers, would be of great value. The differences in speed and in=20 language, of course, pose difficulties that must be overcome.=20
It seems likely that the contributions of human operators and = equipment will=20 blend together so completely in many operations that it will be = difficult to=20 separate them neatly in analysis. That would be the case it; in = gathering data=20 on which to base a decision, for example, both the man and the computer = came up=20 with relevant precedents from experience and if the computer then = suggested a=20 course of action that agreed with the man's intuitive judgment. (In=20 theorem-proving programs, computers find precedents in experience, and = in the=20 SAGE System, they suggest courses of action. The foregoing is not a = far-fetched=20 example. ) In other operations, however, the contributions of men and = equipment=20 will be to some extent separable.=20
Men will set the goals and supply the motivations, of course, at = least in the=20 early years. They will formulate hypotheses. They will ask questions. = They will=20 think of mechanisms, procedures, and models. They will remember that=20 such-and-such a person did some possibly relevant work on a topic of = interest=20 back in 1947, or at any rate shortly after World War II, and they will = have an=20 idea in what journals it might have been published. In general, they = will make=20 approximate and fallible, but leading, contributions, and they will = define=20 criteria and serve as evaluators, judging the contributions of the = equipment and=20 guiding the general line of thought.=20
In addition, men will handle the very-low-probability situations when = such=20 situations do actually arise. (In current man-machine systems, that is = one of=20 the human operator's most important functions. The sum of the = probabilities of=20 very-low-probability alternatives is often much too large to neglect. ) = Men will=20 fill in the gaps, either in the problem solution or in the computer = program,=20 when the computer has no mode or routine that is applicable in a = particular=20 circumstance.=20
The information-processing equipment, for its part, will convert = hypotheses=20 into testable models and then test the models against data (which the = human=20 operator may designate roughly and identify as relevant when the = computer=20 presents them for his approval). The equipment will answer questions. It = will=20 simulate the mechanisms and models, carry out the procedures, and = display the=20 results to the operator. It will transform data, plot graphs ("cutting = the cake"=20 in whatever way the human operator specifies, or in several alternative = ways if=20 the human operator is not sure what he wants). The equipment will = interpolate,=20 extrapolate, and transform. It will convert static equations or logical=20 statements into dynamic models so the human operator can examine their = behavior.=20 In general, it will carry out the routinizable, clerical operations that = fill=20 the intervals between decisions.=20
In addition, the computer will serve as a statistical-inference,=20 decision-theory, or game-theory machine to make elementary evaluations = of=20 suggested courses of action whenever there is enough basis to support a = formal=20 statistical analysis. Finally, it will do as much diagnosis, = pattern-matching,=20 and relevance-recognizing as it profitably can, but it will accept a = clearly=20 secondary status in those areas.=20
The data-processing equipment tacitly postulated in the preceding = section is=20 not available. The computer programs have not been written. There are in = fact=20 several hurdles that stand between the nonsymbiotic present and the = anticipated=20 symbiotic future. Let us examine some of them to see more clearly what = is needed=20 and what the chances are of achieving it.=20
Any present-day large-scale computer is too fast and too costly for = real-time=20 cooperative thinking with one man. Clearly, for the sake of efficiency = and=20 economy, the computer must divide its time among many users. Timesharing = systems=20 are currently under active development. There are even arrangements to = keep=20 users from "clobbering" anything but their own personal programs.=20
It seems reasonable to envision, for a time 10 or 15 years hence, a = "thinking=20 center" that will incorporate the functions of present-day libraries = together=20 with anticipated advances in information storage and retrieval and the = symbiotic=20 functions suggested earlier in this paper. The picture readily enlarges = itself=20 into a network of such centers, connected to one another by wide-band=20 communication lines and to individual users by leased-wire services. In = such a=20 system, the speed of the computers would be balanced, and the cost of = the=20 gigantic memories and the sophisticated programs would be divided by the = number=20 of users.=20
When we start to think of storing any appreciable fraction of a = technical=20 literature in computer memory, we run into billions of bits and, unless = things=20 change markedly, billions of dollars.=20
The first thing to face is that we shall not store all the technical = and=20 scientific papers in computer memory. We may store the parts that can be = summarized most succinctly-the quantitative parts and the reference=20 citations-but not the whole. Books are among the most beautifully = engineered,=20 and human-engineered, components in existence, and they will continue to = be=20 functionally important within the context of man-computer symbiosis. = (Hopefully,=20 the computer will expedite the finding, delivering, and returning of = books.)=20
The second point is that a very important section of memory will be=20 permanent: part indelible memory and part published = memory. The=20 computer will be able to write once into indelible memory, and then read = back=20 indefinitely, but the computer will not be able to erase indelible = memory. (It=20 may also over-write, turning all the 0's into l's, as though marking = over what=20 was written earlier.) Published memory will be "read-only" memory. It = will be=20 introduced into the computer already structured. The computer will be = able to=20 refer to it repeatedly, but not to change it. These types of memory will = become=20 more and more important as computers grow larger. They can be made more = compact=20 than core, thin-film, or even tape memory, and they will be much less = expensive.=20 The main engineering problems will concern selection circuitry.=20
In so far as other aspects of memory requirement are concerned, we = may count=20 upon the continuing development of ordinary scientific and business = computing=20 machines There is some prospect that memory elements will become as fast = as=20 processing (logic) elements. That development would have a revolutionary = effect=20 upon the design of computers.=20
Implicit in the idea of man-computer symbiosis are the requirements = that=20 information be retrievable both by name and by pattern and that it be = accessible=20 through procedure much faster than serial search. At least half of the = problem=20 of memory organization appears to reside in the storage procedure. Most = of the=20 remainder seems to be wrapped up in the problem of pattern recognition = within=20 the storage mechanism or medium. Detailed discussion of these problems = is beyond=20 the present scope. However, a brief outline of one promising idea, "trie = memory," may serve to indicate the general nature of anticipated = developments.=20
Trie memory is so called by its originator, Fredkin [10], because it = is=20 designed to facilitate retrieval of information and because the = branching=20 storage structure, when developed, resembles a tree. Most common memory = systems=20 store functions of arguments at locations designated by the arguments. = (In one=20 sense, they do not store the arguments at all. In another and more = realistic=20 sense, they store all the possible arguments in the framework structure = of the=20 memory.) The trie memory system, on the other hand, stores both the = functions=20 and the arguments. The argument is introduced into the memory first, one = character at a time, starting at a standard initial register. Each = argument=20 register has one cell for each character of the ensemble (e.g., two for=20 information encoded in binary form) and each character cell has within = it=20 storage space for the address of the next register. The argument is = stored by=20 writing a series of addresses, each one of which tells where to find the = next.=20 At the end of the argument is a special "end-of-argument" marker. Then = follow=20 directions to the function, which is stored in one or another of several = ways,=20 either further trie structure or "list structure" often being most = effective.=20
The trie memory scheme is inefficient for small memories, but it = becomes=20 increasingly efficient in using available storage space as memory size=20 increases. The attractive features of the scheme are these: 1) The = retrieval=20 process is extremely simple. Given the argument, enter the standard = initial=20 register with the first character, and pick up the address of the = second. Then=20 go to the second register, and pick up the address of the third, etc. 2) = If two=20 arguments have initial characters in common, they use the same storage = space for=20 those characters. 3) The lengths of the arguments need not be the same, = and need=20 not be specified in advance. 4) No room in storage is reserved for or = used by=20 any argument until it is actually stored. The trie structure is created = as the=20 items are introduced into the memory. 5) A function can be used as an = argument=20 for another function, and that function as an argument for the next. = Thus, for=20 example, by entering with the argument, "matrix multiplication," one = might=20 retrieve the entire program for performing a matrix multiplication on = the=20 computer. 6) By examining the storage at a given level, one can = determine what=20 thus-far similar items have been stored. For example, if there is no = citation=20 for Egan, J. P., it is but a step or two backward to pick up the trail = of Egan,=20 James ... .=20
The properties just described do not include all the desired ones, = but they=20 bring computer storage into resonance with human operators and their=20 predilection to designate things by naming or pointing.=20
The basic dissimilarity between human languages and computer = languages may be=20 the most serious obstacle to true symbiosis. It is reassuring, however, = to note=20 what great strides have already been made, through interpretive programs = and=20 particularly through assembly or compiling programs such as FORTRAN, to = adapt=20 computers to human language forms. The "Information Processing Language" = of=20 Shaw, Newell, Simon, and Ellis [24] represents another line of = rapprochement.=20 And, in ALGOL and related systems, men are proving their flexibility by = adopting=20 standard formulas of representation and expression that are readily = translatable=20 into machine language.=20
For the purposes of real-time cooperation between men and computers, = it will=20 be necessary, however, to make use of an additional and rather different = principle of communication and control. The idea may be highlighted by = comparing=20 instructions ordinarily addressed to intelligent human beings with = instructions=20 ordinarily used with computers. The latter specify precisely the = individual=20 steps to take and the sequence in which to take them. The former present = or=20 imply something about incentive or motivation, and they supply a = criterion by=20 which the human executor of the instructions will know when he has = accomplished=20 his task. In short: instructions directed to computers specify courses;=20 instructions-directed to human beings specify goals.=20
Men appear to think more naturally and easily in terms of goals than = in terms=20 of courses. True, they usually know something about directions in which = to=20 travel or lines along which to work, but few start out with precisely = formulated=20 itineraries. Who, for example, would depart from Boston for Los Angeles = with a=20 detailed specification of the route? Instead, to paraphrase Wiener, men = bound=20 for Los Angeles try continually to decrease the amount by which they are = not yet=20 in the smog.=20
Computer instruction through specification of goals is being = approached along=20 two paths. The first involves problem-solving, hill-climbing, = self-organizing=20 programs. The second involves real-time concatenation of preprogrammed = segments=20 and closed subroutines which the human operator can designate and call = into=20 action simply by name.=20
Along the first of these paths, there has been promising exploratory = work. It=20 is clear that, working within the loose constraints of predetermined = strategies,=20 computers will in due course be able to devise and simplify their own = procedures=20 for achieving stated goals. Thus far, the achievements have not been=20 substantively important; they have constituted only "demonstration in=20 principle." Nevertheless, the implications are far-reaching.=20
Although the second path is simpler and apparently capable of earlier = realization, it has been relatively neglected. Fredkin's trie memory = provides a=20 promising paradigm. We may in due course see a serious effort to develop = computer programs that can be connected together like the words and = phrases of=20 speech to do whatever computation or control is required at the moment. = The=20 consideration that holds back such an effort, apparently, is that the = effort=20 would produce nothing that would be of great value in the context of = existing=20 computers. It would be unrewarding to develop the language before there = are any=20 computing machines capable of responding meaningfully to it.=20
The department of data processing that seems least advanced, in so = far as the=20 requirements of man-computer symbiosis are concerned, is the one that = deals with=20 input and output equipment or, as it is seen from the human operator's = point of=20 view, displays and controls. Immediately after saying that, it is = essential to=20 make qualifying comments, because the engineering of equipment for = high-speed=20 introduction and extraction of information has been excellent, and = because some=20 very sophisticated display and control techniques have been developed in = such=20 research laboratories as the Lincoln Laboratory. By and large, in = generally=20 available computers, however, there is almost no provision for any more=20 effective, immediate man-machine communication than can be achieved with = an=20 electric typewriter.=20
Displays seem to be in a somewhat better state than controls. Many = computers=20 plot graphs on oscilloscope screens, and a few take advantage of the = remarkable=20 capabilities, graphical and symbolic, of the charactron display tube. = Nowhere,=20 to my knowledge, however, is there anything approaching the flexibility = and=20 convenience of the pencil and doodle pad or the chalk and blackboard = used by men=20 in technical discussion.=20
1) Desk-Surface Display and Control: Certainly, for effective=20 man-computer interaction, it will be necessary for the man and the = computer to=20 draw graphs and pictures and to write notes and equations to each other = on the=20 same display surface. The man should be able to present a function to = the=20 computer, in a rough but rapid fashion, by drawing a graph. The computer = should=20 read the man's writing, perhaps on the condition that it be in clear = block=20 capitals, and it should immediately post, at the location of each = hand-drawn=20 symbol, the corresponding character as interpreted and put into precise=20 type-face. With such an input-output device, the operator would quickly = learn to=20 write or print in a manner legible to the machine. He could compose = instructions=20 and subroutines, set them into proper format, and check them over before = introducing them finally into the computer's main memory. He could even = define=20 new symbols, as Gilmore and Savell [14] have done at the Lincoln = Laboratory, and=20 present them directly to the computer. He could sketch out the format of = a table=20 roughly and let the computer shape it up with precision. He could = correct the=20 computer's data, instruct the machine via flow diagrams, and in general = interact=20 with it very much as he would with another engineer, except that the = "other=20 engineer" would be a precise draftsman, a lightning calculator, a = mnemonic=20 wizard, and many other valuable partners all in one.=20
2) Computer-Posted Wall Display: In some technological = systems,=20 several men share responsibility for controlling vehicles whose = behaviors=20 interact. Some information must be presented simultaneously to all the = men,=20 preferably on a common grid, to coordinate their actions. Other = information is=20 of relevance only to one or two operators. There would be only a = confusion of=20 uninterpretable clutter if all the information were presented on one = display to=20 all of them. The information must be posted by a computer, since manual = plotting=20 is too slow to keep it up to date.=20
The problem just outlined is even now a critical one, and it seems = certain to=20 become more and more critical as time goes by. Several designers are = convinced=20 that displays with the desired characteristics can be constructed with = the aid=20 of flashing lights and time-sharing viewing screens based on the = light-valve=20 principle.=20
The large display should be supplemented, according to most of those = who have=20 thought about the problem, by individual display-control units. The = latter would=20 permit the operators to modify the wall display without leaving their = locations.=20 For some purposes, it would be desirable for the operators to be able to = communicate with the computer through the supplementary displays and = perhaps=20 even through the wall display. At least one scheme for providing such=20 communication seems feasible.=20
The large wall display and its associated system are relevant, of = course, to=20 symbiotic cooperation between a computer and a team of men. Laboratory=20 experiments have indicated repeatedly that informal, parallel = arrangements of=20 operators, coordinating their activities through reference to a large = situation=20 display, have important advantages over the arrangement, more widely = used, that=20 locates the operators at individual consoles and attempts to correlate = their=20 actions through the agency of a computer. This is one of several = operator-team=20 problems in need of careful study.=20
3) Automatic Speech Production and Recognition: How desirable = and how=20 feasible is speech communication between human operators and computing = machines?=20 That compound question is asked whenever sophisticated data-processing = systems=20 are discussed. Engineers who work and live with computers take a = conservative=20 attitude toward the desirability. Engineers who have had experience in = the field=20 of automatic speech recognition take a conservative attitude toward the=20 feasibility. Yet there is continuing interest in the idea of talking = with=20 computing machines. In large part, the interest stems from realization = that one=20 can hardly take a military commander or a corporation president away = from his=20 work to teach him to type. If computing machines are ever to be used = directly by=20 top-level decision makers, it may be worthwhile to provide communication = via the=20 most natural means, even at considerable cost.=20
Preliminary analysis of his problems and time scales suggests that a=20 corporation president would be interested in a symbiotic association = with a=20 computer only as an avocation. Business situations usually move slowly = enough=20 that there is time for briefings and conferences. It seems reasonable,=20 therefore, for computer specialists to be the ones who interact directly = with=20 computers in business offices.=20
The military commander, on the other hand, faces a greater = probability of=20 having to make critical decisions in short intervals of time. It is easy = to=20 overdramatize the notion of the ten-minute war, but it would be = dangerous to=20 count on having more than ten minutes in which to make a critical = decision. As=20 military system ground environments and control centers grow in = capability and=20 complexity, therefore, a real requirement for automatic speech = production and=20 recognition in computers seems likely to develop. Certainly, if the = equipment=20 were already developed, reliable, and available, it would be used.=20
In so far as feasibility is concerned, speech production poses less = severe=20 problems of a technical nature than does automatic recognition of speech = sounds.=20 A commercial electronic digital voltmeter now reads aloud its = indications, digit=20 by digit. For eight or ten years, at the Bell Telephone Laboratories, = the Royal=20 Institute of Technology (Stockholm), the Signals Research and = Development=20 Establishment (Christchurch), the Haskins Laboratory, and the = Massachusetts=20 Institute of Technology, Dunn [6], Fant [7], Lawrence [15], Cooper [3], = Stevens=20 [26], and their co-workers, have demonstrated successive generations of=20 intelligible automatic talkers. Recent work at the Haskins Laboratory = has led to=20 the development of a digital code, suitable for use by computing = machines, that=20 makes an automatic voice utter intelligible connected discourse [16].=20
The feasibility of automatic speech recognition depends heavily upon = the size=20 of the vocabulary of words to be recognized and upon the diversity of = talkers=20 and accents with which it must work. Ninety-eight per cent correct = recognition=20 of naturally spoken decimal digits was demonstrated several years ago at = the=20 Bell Telephone Laboratories and at the Lincoln Laboratory [4], [9]. To = go a step=20 up the scale of vocabulary size, we may say that an automatic recognizer = of=20 clearly spoken alpha-numerical characters can almost surely be developed = now on=20 the basis of existing knowledge. Since untrained operators can read at = least as=20 rapidly as trained ones can type, such a device would be a convenient = tool in=20 almost any computer installation.=20
For real-time interaction on a truly symbiotic level, however, a = vocabulary=20 of about 2000 words, e.g., 1000 words of something like basic English = and 1000=20 technical terms, would probably be required. That constitutes a = challenging=20 problem. In the consensus of acousticians and linguists, construction of = a=20 recognizer of 2000 words cannot be accomplished now. However, there are = several=20 organizations that would happily undertake to develop an automatic = recognize for=20 such a vocabulary on a five-year basis. They would stipulate that the = speech be=20 clear speech, dictation style, without unusual accent.=20
Although detailed discussion of techniques of automatic speech = recognition is=20 beyond the present scope, it is fitting to note that computing machines = are=20 playing a dominant role in the development of automatic speech = recognizers. They=20 have contributed the impetus that accounts for the present optimism, or = rather=20 for the optimism presently found in some quarters. Two or three years = ago, it=20 appeared that automatic recognition of sizeable vocabularies would not = be=20 achieved for ten or fifteen years; that it would have to await much = further,=20 gradual accumulation of knowledge of acoustic, phonetic, linguistic, and = psychological processes in speech communication. Now, however, many see = a=20 prospect of accelerating the acquisition of that knowledge with the aid = of=20 computer processing of speech signals, and not a few workers have the = feeling=20 that sophisticated computer programs will be able to perform well as=20 speech-pattern recognizes even without the aid of much substantive = knowledge of=20 speech signals and processes. Putting those two considerations together = brings=20 the estimate of the time required to achieve practically significant = speech=20 recognition down to perhaps five years, the five years just mentioned.=20
[1] A. Bernstein and M. deV. Roberts, "Computer versus chess-player," = Scientific American, vol. 198, pp. 96-98; June, 1958.
[2] W. W. Bledsoe and I. Browning, "Pattern Recognition and Reading = by=20 Machine," presented at the Eastern Joint Computer Conf, Boston, Mass., = December,=20 1959.
[3] F. S. Cooper, et al., "Some experiments on the perception of = synthetic=20 speech sounds," J. Acoust Soc. Amer., vol.24, pp.597-606; = November,=20 1952.
[4] K. H. Davis, R. Biddulph, and S. Balashek, "Automatic recognition = of=20 spoken digits," in W. Jackson, Communication Theory, Butterworths = Scientific Publications, London, Eng., pp. 433-441; 1953.
[5] G. P. Dinneen, "Programming pattern recognition," Proc. = WJCC, pp.=20 94-100; March, 1955.
[6] H. K. Dunn, "The calculation of vowel resonances, and an = electrical vocal=20 tract," J. Acoust Soc. Amer., vol. 22, pp.740-753; November, = 1950.
[7] G. Fant, "On the Acoustics of Speech," paper presented at the = Third=20 Internatl. Congress on Acoustics, Stuttgart, Ger.; September, 1959.
[8] B. G. Farley and W. A. Clark, "Simulation of self-organizing = systems by=20 digital computers." IRE Trans. on Information Theory, vol. IT-4,=20 pp.76-84; September, 1954
[9] J. W. Forgie and C. D. Forgie, "Results obtained from a vowel = recognition=20 computer program," J. Acoust Soc. Amer., vol. 31, pp. 1480-1489;=20 November, 1959
[10] E. Fredkin, "Trie memory," Communications of the ACM, = Sept. 1960,=20 pp. 490-499
[11] R. M. Friedberg, "A learning machine: Part I," IBM J. Res. = &=20 Dev., vol.2, pp.2-13; January, 1958.
[12] H. Gelernter, "Realization of a Geometry Theorem Proving = Machine."=20 Unesco, NS, ICIP, 1.6.6, Internatl. Conf. on Information Processing, = Paris,=20 France; June, 1959.
[13] P. C. Gilmore, "A Program for the Production of Proofs for = Theorems=20 Derivable Within the First Order Predicate Calculus from Axioms," = Unesco, NS,=20 ICIP, 1.6.14, Internatl. Conf. on Information Processing, Paris, France; = June,=20 1959.
[14] J. T. Gilmore and R. E. Savell, "The Lincoln Writer," Lincoln=20 Laboratory, M. I. T., Lexington, Mass., Rept. 51-8; October, 1959.
[15] W. Lawrence, et al., "Methods and Purposes of Speech Synthesis," = Signals=20 Res. and Dev. Estab., Ministry of Supply, Christchurch, Hants, England, = Rept.=20 56/1457; March, 1956.
[16] A. M. Liberman, F. Ingemann, L. Lisker, P. Delattre, and F. S. = Cooper,=20 "Minimal rules for synthesizing speech," J. Acoust Soc. Amer., = vol. 31,=20 pp. 1490-1499; November, 1959.
[17] A. Newell, "The chess machine: an example of dealing with a = complex task=20 by adaptation," Proc. WJCC, pp. 101-108; March, 1955.
[18] A. Newell and J. C. Shaw, "Programming the logic theory = machine."=20 Proc. WJCC, pp. 230-240; March, 1957.
[19] A. Newell, J. C. Shaw, and H. A. Simon, "Chess-playing programs = and the=20 problem of complexity," IBM J. Res & Dev., vol.2, pp. = 320-33.5;=20 October, 1958.
[20] A. Newell, H. A. Simon, and J. C. Shaw, "Report on a general=20 problem-solving program," Unesco, NS, ICIP, 1.6.8, Internatl. Conf. on=20 Information Processing, Paris, France; June, 1959.
[21] J. D. North, "The rational behavior of mechanically extended = man",=20 Boulton Paul Aircraft Ltd., Wolverhampton, Eng.; September, 1954.
[22] 0. G. Selfridge, "Pandemonium, a paradigm for learning," = Proc. Symp.=20 Mechanisation of Thought Processes, Natl. Physical Lab., Teddington, = Eng.;=20 November, 1958.
[23] C. E. Shannon, "Programming a computer for playing chess," = Phil.=20 Mag., vol.41, pp.256-75; March, 1950.
[24] J. C. Shaw, A. Newell, H. A. Simon, and T. O. Ellis, "A command=20 structure for complex information processing," Proc. WJCC, pp. = 119-128;=20 May, 1958.
[25] H. Sherman, "A Quasi-Topological Method for Recognition of Line=20 Patterns," Unesco, NS, ICIP, H.L.5, Internatl. Conf. on Information = Processing,=20 Paris, France; June, 1959
[26] K. N. Stevens, S. Kasowski, and C. G. Fant, "Electric analog of = the=20 vocal tract," J. Acoust. Soc. Amer., vol. 25, pp. 734-742; July,=20 1953.
[27] Webster's New International Dictionary, 2nd e., G. and C. = Merriam=20 Co., Springfield, Mass., p. 2555; 1958.