The Science of Design: Creating the Artificial


110 Remembering and Learning Human beings, viewed as behaving systems' are quite simple' The apparent com- plexity of our behavior over time is largely a reflection of the complexity of the environment in which we find ourselves . . . provided that we include in what we call the human environment the co- coon of information, stored in books and in long-term memory' that we spin about ourselves. That information, stored both as data and as procedures and richly indexed for access in the presence of appropriate stimuli, enables the simple basic information processes to draw upon a very large repertory of information and strategies, and accounts for the apPearance of com- plexity in their behavior. The inner environment, the hardware, is simple. Complexity emerges from the richness of the outer environment, both the world apprehended through the senses and the information about the world stored in long-term memory. A scientific account of human cognition describes it in terms of several sets of invariants. First, there are the parameters of the inner environment. Then, there are the general control and search-guiding mechanisms that are used over and over again in all task domains. Finally, there are the learning and discovery mechanisms that permit the system to adapt with gradually increasing effectiveness to the particular environment in which it finds itself. The adaptiveness of the human organism, the facility with which it acquires new representations and strategies and becomes adept in dealing with highly specialized environments, makes it an elusive and fascinating target of our scientific inquiries-and the very prototype of the artificial. 5 The Science of Design: Creating the Artificial Historically and traditionallS it has been the task of the science disci- plines to teach about natural things: how they are and how they work. It has been the task of engineering schools to teach about artificial things: how to make artifacts that have desired properties and how to design. Engineers are not the only professional designers. Everyone designs who devises courses of action aimed at changing existing situations into preferred ones. The intellectual activity that produces material artifacts is no different fundamentally from the one that prescribes remedies for a sick patient or the one that devises a new sales plan for a company or a social welfare policy for a state. Design, so construed, is the core of all professional training; it is the principal mark that distinguishes the pro- fessions from the sciences. Schools of engineering, as well as schools of architecture, business, education, law, and medicine, are all centrally con- cerned with the process of design. In view of the key role of design in professional activiry it is ironic that in this century the natural sciences almost drove the sciences of the artifi- cial from professional school curricula, a development that peaked about two or three decades after the Second World War. Engineering schools gradually became schools of physics and mathematics; medical schools became schools of biological science; business schools became schools of finite mathematics. The use of adjectives like "applied" concealed, but did not change, the fact. It simply meant that in the professional schools those topics were selected from mathematics and the natural sciences for emphasis which were thought to be most nearly relevant to professional practice. It did not mean that design continued to be taught, as distin- guished from analysis. 112 The Science of Design The movement toward natural science and away from the sciences of the artificial proceeded further and faster in engineering, business, and medicine than in the other professional fields I have mentioned, though it has by no means been absent from schools of law, journalism, and library science. The stronger universities were more deeply affected than the weaker, and the graduate programs more than the undergraduate. During that time few doctoral dissertations in first-rate professional schools dealt with genuine design problems, as distinguished from problems in solid- state physics or stochastic processes. I have to make partial exceptions- for reasons I shall mention-of dissertations in computer science and management science, and there were undoubtedly some others, for ex- ample, in chemical engineering. Such a universal phenomenon musr have had a basic cause. It did have a very obvious one. As professional schools, including the independent engineering schools, were more and more absorbed into the general cul- ture of the university, they hankered after academic respectability. In terms of the prevailing norms, academic respectabirity calls for subject matter that is intellectually tough, analytic, formalizable, and teachable. In the past much, if not most, of what we knew about design and about the artificial sciences was intellectually soft, intuitive, informal, and cook- booky. why would anyone in a universify stoop to teach or learn about designing machines or planning market strategies when he could concern himself with solid-state physics? The answer has been clear: he usually wouldn't. The damage to pro the loss of design from professional cur on in engineering and medicine and ro e schools did not think it a problem (and a few still do not), because they regarded schools of applied science as a superior alternative to the trade schools of the past. If that were the choice, we could agree.r But neither alternative is I lop: . . . rhat increasing attention in the Institute may be given ro the fundamen-tal sciences; that they may achieve as nevcr before the siirit end reruJts of ,e. The Science of Design 113 satisfactory. The older kind of professional school did not know how to educate for professional design at an intellectual level appropriate to a university; the newer kind of school nearly abdicated responsibility for training in the core professional skill. Thus we were faced with a problem of devising a professional school that could attain two objectives simulta- neously: education in both artificial and natural science at a high intellec- tual level. This too is a problem of design-organizational design. The kernel of the problem lies in the phrase "artificial science." The previous chapters have shown that a science of artificial phenomena is always in imminent danger of dissolving and vanishing. The peculiar properties of the artifact lie on the thin interface between the natural laws within it and the natural laws without. rVhat can we say about it? rUfhat is there to study besides the boundary sciences-those that govern the means and the task environment? The artificial world is centered precisely on this interface between the inner and outer environments; it is concerned with attaining goals by adapting the former to the latter. The proper study of those who are con- cerned with the artifcial is the way in which that adaptation of means to environments is brought about-and central to that is the process of de- sign itself. The professional schools can reassume their professional re- sponsibilities just to the degree that they discover and teach a science of design, a body of intellectually tough, analytic, partly formalizable, partly empirical, teachable doctrine about the design process. It is the thesis of this chapter that such a science of design not only is possible but also has been emerging since the mid-1970s. In fact, it is fair to say that the 6rst edition of this book, published in 1969,was influential in its development, serving as a call to action and outlining the form that the action could take. At Carnegie Mellon University, one of the first engi- neering schools to move toward research on the process of design, the search; that all courses of instruction may be examined carefully to see where training in details has been unduly emphasized at the expense of the more power- ful training in all-embracing fundamental principles. Notice that President Compton's emphasis was on 'fundamental," an emphasis as sound today as it was in 1930. What is called for is not a departure from the fundamental but an inclusion in the curriculum of the fundamental in engineering along with the fundamental in natural science. That was not possible in 1930; but it is possible today. 112 The Science of Design The movement toward natural science and away from the sciences of the artificial proceeded further and faster in engineering, business, and medicine than in the other professional fields I have menrioned, though it has by no means been absent from schools of law, journalism, and library science. The stronger universities were more deeply affected than the weaker, and the graduate programs more than the undergraduate. During that time few doctoral dissertations in first-rate professional schools dealt with genuine design problems, as distinguished from problems in solid- state physics or stochastic processes. I have to make partial exceptions- for reasons I shall mention-of dissertations in computer science and management science, and there were undoubtedly some others, for ex- ample, in chemical engineering. Such a universal phenomenon must have had a basic cause. It did have a very obvious one. As professional schools, including the independent engineering schools, were more and more absorbed into the general cul- ture of the university, they hankered after academic respectability. In terms of the prevailing norms, academic respectability calls for subject matter that is intellectually tough, analytic, formalizable, and teachable. In the past much, if not most, of what we knew about design and about the artificial sciences was intellectually soft, intuitive, informal, and cook- booky' rfhy would anyone in a universiry stoop to teach or learn about designing machines or planning market strategies when he could concern himself with solid-state physics? The answer has been clear: he usually wouldn't. The damage to professional competence caused by the loss of design from professional curricula gradually gained recognition in engineering and medicine and to a lesser extent in business. some schools did not think it a problem (and a few still do not), because they regarded schools of applied science as a superior alternative to the trade schools of the past. If that were the choice, we could agree.l But neither alternative is 1. That was in fac schools needed to did not exist even more fundamental Taylor Compton's presidential inaugural address at MIT in 1930: I hope , . . that increasing attention in the Institute may be given to the fundamen- tal sciences; that they may achieve as never before the cpirit and results of rc. The Science of Design 113 satisfactory. The older kind of professional school did not know how to educate for professional design at an intellectual level appropriate to a university; the newer kind of school nearly abdicated responsibility for training in the core professional skill. Thus we were faced with a problem of devising a professional school that could attain two objectives simulta- neously: education in both artificial and natural science at a high intellec- tual level. This too is a problem of design-organizational design. The kernel of the problem lies in the phrase "artificial science." The previous chapters have shown that a science of artificial phenomena is always in imminent danger of dissolving and vanishing. The peculiar properties of the artifact lie on the thin interface between the natural laws within it and the natural laws without. lVhat can we say about it? \(hat is there to study besides the boundary sciences-those that govern the means and the task environment? The artificial world is centered precisely on this interface between the inner and outer environments; it is concerned with attaining goals by adapting the former to the latter. The proper study of those who are con- cerned with the artificial is the way in which that adaptation of means to environments is brought about-and central to that is the process of de- sign itself. The professional schools can reassume their professional re- sponsibilities just to the degree that they discover and teach a science of design, a body of intellectually tough, analytic, partly formalizable, partly empirical, teachable doctrine about the design process. It is the thesis of this chapter that such a science of design not only is possible but also has been emerging since the mid-1970s. In fact, it is fair to say that the first edition of this book, published in 1.969,was influential in its development, serving as a call to action and outlining the form that the action could take. At Carnegie Mellon University, one of the first engi- neering schools to move toward research on the process of design, the search; that all courses of instruction may be examined carefully to see where training in details has been unduly emphasized at the expense of the more power- ful training in all-embracing fundamental principles. Notice that President Compton's emphasis was on "fundamental," an emphasis as sound today as it was in 1930. \07hat is called for is not a departure from the fundamental but an inclusion in the curriculum of the fundamental in engineering along with the fundamental in natural science. That was not possible in 1930; but it is possible today. 114 The Science of Design initial step was to form a Design Research Center, about 1975. The Cen- ter (since 1985 called the "Engineering Design Research center") facili- tated collaboration among the faculty and students undertaking research on the science and practice of design and developed elements of a theory of design that found their way back into the undergraduare and graduate curricula. The center continues to play an important role in the modern- ization and strengthening of education and research in design at carnegie Mellon and elsewhere in the United States. In substantial parr, design theory is aimed at broadening the capabili- ties of computers to aid design, drawing upon the tools of artificial intel- ligence and operations research. Hence, research on many aspects of computer-aided design is being pursued with growing intensity in com- puter science, engineering and architecture departments, and in opera- tions research groups in business schools. The need to make design theory explicit and precise in order to introduce computers into the process has been the key to establishing its academic acceptability-its appropriate- ness for a university. In the remainder of this chapter I will take up some of the topics that need to be incorporated in a theory of design and in instruction in design. The Logic of Design: Fixed Alternatives 'we must start with some questions of logic.2 The natural sciences are concerned with how things are. ordinary systems of logic-the standard propositional and predicate calculi, say-serve these sciences well. since the concern of standard logic is with declarative statements, it is well suited for assertions about the world and for inferences from those assertions. Design, on the other hand, is concerned with how things ought to be, with devising artifacts to attain goals. we might question whether the greater length i#:LI:i::: Action (Pitts- burghr University of Pittsburgh Press, 19671, pp, l-35. The presenr discussion is based on these two papers, which have been reprinred as chapters 3.1 and 3,2 in my Models of Discouery (Dordrechtr D. Reidel Pub. Co,, 1977). The Science of Design 115 forms of reasoning that are appropriate to natural science are suitable also for design. One might well suppose that introduction of the verb "should" may require additional rules of inference' or modification of the rules already imbedded in declarative logic. Paradoxes of lmperative Logic Various "paradoxes" have been constructed to demonstrate the need for a distinct logic of imperatives, or a normative, deontic logic. In ordinary logic from "Dogs are pets" and "Cats are petsr" one can infer "Dogs and cats are pets." But from "Dogs are petsr" "Cats are petsr" and "You should keep pets," can one infer "You should keep cats and dogs"? And from "Give me needle and thread!" can one deduce, in analogy with de- clarative logic, "Give me needle or thread!"? Easily frustrated people would perhaps rather have neither needle nor thread than one without the other, and peaceJoving people, neither cats nor dogs, rather than both. As a response to these challenges of apparent paradox, there have been developed a number of constructions of modal logic for handling "shoulds," "shalts," and "oughts" of various kinds. I qhink it is fair to say that none of these systems has been sufficiently developed or sufficiently widely applied to demonstrate that it is adequate to handle the logical requirements of the process of design. Fortunately, such a demonstration is really not essential, for it can be shown that the requirements of design can be met fully by a modest adap- tation of ordinary declarative logic. Thus a special logic of imperatives is unnecessary. I should like to underline the word "unnecessarS" which does not mean "impossible." Modal logics can be shown to exist in the same way that giraffes can-namely by exhibiting some of them. The question is not whether they exist, but whether they are needed for, or even useful for, design. Reduction to Declarative Logic The easiest way to discover what kinds of logic are needed for design is to examine what kinds of logic designers use when they are being careful about their reasoning. Now there would be no point in doing this if de- signers were always sloppy fellows who reasoned loosely, vaguel5 and 116 The Science of Design intuitively. Then we might say that whatever logic they used was not the logic they should use. However, there exists a considerable area of design practice where stan- dards of rigor in inference are as high as one could wish. I refer to the domain of so-called "optimization methods," most highly developed in statistical decision theory and management science but acquiring growing importance also in engineering design theory. The theories of probability and utility, and their intersection, have received the painstaking attention not only of practical designers and decision makers but also of a consider- able number of the most distinguished logicians and mathematicians of recent generations. F. P. Ramsey, B. de Finetti, A. Wald, J. von Neumann, J. Neyman, K. Arrow, and L. J. Savage are examples. The logic of optimization methods can be sketched as follows: The "in- ner environment" of the design problem is represented by a set of given alternatives of action. The alternatives may be given in extenso: more commonly they are specified in terms of command uariables that have defined domains. The "outer environment" is represented by a set of pa- rameters, which may be known with certainty or only in terms of a proba- bility distribution. The goals for adaptation of inner to outer environment are defined by a utility function-a function, usually scalar, of the com- mand variables and environmental parameters-perhaps supplemented by a number of constraints (inequalities, say, between functions of the command variables and environmental parameters). The optimization problem is to find an admissible set of values of the command variables, compatible with the constraints, that maximize the utility function for the given values of the environmental parameters. (In the probabilistic case we might say, "maximize the expected value of the utility function," for instance, instead of "maximize the utility function.") A stock application of this paradigm is the so-called "diet problem" shown in figure 6. A list of foods is provided, the command variables being quantities of the various foods to be included in the diet. The envi- ronmental parameters are the prices and nutritional contents (calories, vitamins, minerals, and so on) of each of the foods. The utility function is the cost (with a minus sign attached) of the diet, subject to the con- straints, say, that it not contain more than 2,000 calories per day that it ("Means") (.'taws") fEnds") The Science of Design Emmplc: Thc diet prcblem . Quantitiesoffoods f ki."s of foods I 1 I Nutritional contents I Nutritionatrequirements ) [-c*orai", Constraints characterizc the inner environment; Parameters charadtetizf, the outcr envimnment. Prcblem: Given the constraints and fxed parameters, find values of the comrnand variables t}at maximize utility. Figure 5 The paradigm for imPerative logic meet specified minimum needs for vitamins and minerals, and that ruta- baga not be eaten more than once a week. The constraints may be viewed as characterizing the inner environment. The problem is to select the quantities of foods that will meet the nutritional requirements and side conditions at the given prices for the lowest cost' The diet problem is a simple example of a class of problems that are readily handled, even when the number of variables is exceedingly large, by the mathematical formalism known as linear programming' I shall come back to the technique a little later. My present concern is with the logic of the matter. Since the optimization problem, once formalized, is a standard mathe- matical problem-to maximize a function subiect to constraints-it is evident that the logic used to deduce the answer is the standard logic of the predicate calculus on which mathbmatics rests. How does the formal- ism avoid making use of a special logic of imperatives? It does so by deal- ing with sets of possib le worlds: First consider all the possible worlds that meet the constraints of the outer environment; then find the particular world in the set that meets the remaininS constraints of the goal and 117 118 The Science of Design maximizes the utility function. The logic is exactly the same as if we were to adjoin the goal constraints and the maximization requirement, as new "natural laws," to the existing natural laws embodied in the environmen- tal conditions.3 We simply ask what values the command variables would have in a world meeting all these conditions and conclude that these are the values the command variables shouldhave, Computing the Optimum Our discussion thus far has akeady provided us with two central topics for the curriculum in the science of design: l. Utility tbeory and statistical decision theory as a logical framework for rational choice among giuen ahernatiues. 2. The body of tecbniques for actually deducing wbicb of the auailable alternatiues is the optimum. Only in trivial cases is the computation of the optimum alternative an easy matter (Recall Chapter 2). If utility theory is to have application to real-life design problems, it must be accompanied by tools for actu- ally making the computations. The dilemma of the rational chess player is familiar to all. The optimal strategy in chess is easily demonstrated: simply assign a value of *1 to a win,0 to a draw, -1 to a loss; consider all possible courses of play; minimax backward from the outcome of each, assuming each player will take the most favorable move at any given point. This procedure will determine what move to make now. The only trouble is that the computations required are astronomical (the number 10120 is often mentioned in this context) and hence cannot be carried out-not by humans, not by existing computers, not by prospective computers. A theory of design as applied to the game of chess would encompass not only the utopian minimax principle but also some practicable pro- 3. The use of the notion of "possible worlds" to embed the logic of imperatives in declarative logic goes back at least to JorgenJorgensen, "Imperatives and Logic," Er k ennt ni s, 7 (19 37 -19 3 8l :28 8-29 6. See a lso my A dm in i s*at i u e B e h au i or (N ew York: Macmillan, 19471, chapter 3. Typed logics can be used to distinguish, as belonging to different types, statements that arc true under different conditions (i,e,, in different pomible worlda), but, se my example ahowa, even thic device is not u3u€lly ncedcd, Eaeh new €quation or eonEtlelnt we lntroduee into a eyetem !€du€Gr thc rct ef poillbl€ rteter t€ e rubi€t cf thme p!€vl€uily pomibl€, The Science of Design 119 cedures for finding good moves in actual board positions in real time' within the computational capacities of real human beings or real comput- ers. The best procedures of this kind that exist today are still those stored in the memories of grandmasters, having the characteristics I described in chapters 3 and 4. But there are now s an rather regularly defeat all but a few of rs' Even these programs do not possess of human masters' but succeed by a combination of brute-force computa- tion (sometimes hundreds of millions of variations are analysed) with a good deal of "book" knowledge of opening variations and a reasonably Iophisticated criterion function for er aluating positions' ih. ,..orrd topic then for the curriculum in the science of design con- sistsintheefficientcomputationaltechniquesthatareavailable'foractu- ally finding optimum too"t' of action in real situations' or reasonable approximations to real situations' As I mentioned in chapter 2' that topic has a number of important components today' most of them developed- at reast to the level of practical apprication-within the past years' These include linear programming theory, dynamic programming' geometrlc programming, queuing theory, and control theory' Finding Satisfactory Actions . , The suLiect of computational techniques need not be limited to optimlza- tion. Traditional engineering design methods make much more use of ine- qualities-specifications of satisfactory performance-than of maxima and minima. So-called "figures of merit" permit comparison beween de- signs in terms of "better" and "worse" but seldom provide a iudgment of "best." For example, I may cite the root-locus methods employed in the design of control systems' Since there did not seem to be any word in English for decision methods thatlookforgoodorsatisfactorysolutionsinsteadofoptimalones,some years ago I iniroduced the term "satisficing" to refer to such procedures' Nownooneinhisrightmindwillsatisficeifhecanequallywelloptimize; no one will settle foi good or be$er if he can have best' But that is not the way the problem usually poses itself in actual design situations' In chepter 2 I argued that in the real world we usually do not have a ehoiee between ratiifiory and optimal colutiong, for we only rarely have 120 Tbe Science of Design a method of finding the optimum. consider, for example, the well-known combinatorial problem called the traveling salesman problem: given the geographical locations of a set of cities, find the routing th"t -ill take a salesman to all the cities with the shortest mileage.a For this problem there is a straightforward optimizing algorithm (analogous to the mini- max algorithm for chess): try all possibre routings, and pick the shortest. But for any considerable number of cities, the algorithm is compuration- ally infeasible (the number of routes through N cities will be N/). Al- though some ways have been found for cutting down the length of the search, no algorithm has been discovered sufficiently po*.rful ro solve the traveling salesman problem with a tolerable "-ount of .omputing for a set of, say, fifty cities. Rather than keep our salesman at home, we shall prefer of course tofind a satisfactory if not optimal, routing for him. Under most circum- stances' common sense will probabry arrive at a fairly good route, but an even better one can often be found by one or another of several heuristic methods. An earmark of all these situations where we satisfice for inability to optimize is that, although the set of available alternatives is .,given,, in a certain abstract sense (we can define a generator guaranteed to genefate all of them eventually), it is not "given" in the only sense that is practicaily relevant. 'we cannor within practicable compurational limits generate all the admissible alternatives and compare their respective merits. Nor can we recognize the best alternative, even if we are fortunate enough to gen- erate it early, until we have seen all of them. we satisfice by lotking for alternatives in such a way that we can generaily find an acceptable one after only moderate search. Now in many satisficing situations, the expected length of search foran alternative meeting specified standards of acceptability depends on how high the standards are set, but it depends hardly at all o'the total size of the universe to be searched. The time required for a search through a haystack for a needle sharp enough to sew with depends on the density of distribution of sharp needles but not on rhe total size of the stack. 4. "Thg traveling salesman probrem" and a number of crosery anarogous c.mbi-natorial problems-such as the "warehouse location problem"-hr"u. .-,ri.ia.r-able practical .import'nce, for instance, in siring c.nirar powc-totiu,ir-'f,,, uninterconnecred grid, The Science of Design 12L Hence, when we use satisficing methods, it often does not matter whether or not the total set of admissible alternatives is "given" by a formal but impracticable algorithm. It often does not even matter how big that set is. For this reason satisficing methods may be extendable to design problems in that broad range where the set of alternatives is not "given" even in the quixotic sense that it is "given" for the traveling sales- man problem. Our next task is to examine this possibility. The Logic of Design: Finding Alternatives 'lfhen we take up the case where the design alternatives are not given in any constructive sense but must be synthesized, we must ask once more whether any new forms of reasoning are involved in the synthesis, or whether again the standard logic of declarative statements is all we need. In the case of optimization we asked: "Of all possible worlds (those attainable for some admissible values of the action variables), which is the best (yields the highest value of the criterion function)?" As we saw, this is " p,treiy empirical question, calling only for facts and ordinary declarative reasoning to answer it. In this case, where we are seeking a satisfactory alternative' once we have found a candidate we can ask: "Does this alternative satisfy all the design criteria?" Clearly this is also a factual question and raises no new issues of logic. But how about the process of searching for candidates? rU7hat kind of logic is needed for the search? Means-Ends Analysis The condition of any goal-seeking system is that it is connected to the out- side environment through two kinds of channels: the afferent, or sensory' channels through which it receives information about the environment and the efferent, or motor, channels through which it acts on the environment.s The system must have some means of storing in its memory informa- tion about states of the world-afferent, or sensory, information- 5. Notice that we are not saying that the two kinds of channels operate indepen- dently of each other, since they surely do not in living organisms, but that we can distinguish conceptually, and to some extent neurologically, between the incoming and outgoing flows. 122 The Science of Design and information about actions-efferent, or motor, information. Ability to attain goals depends on building up associations, which may be simple or very complex, befween particular changes in states of the world and particular actions that will (reliably or not) bring these changes about. In chapter 4 we described these associations as productions. Except for a few built-in re is for correlating its sensory information with i t part of its early learning is that particular acti will bring about particular changes in the state of the sensed world. until the infant builds up this knowledge, the world of sense and the motor world are two en- tirely separate, entirely unrelated worlds. only as it begins to acquire ex- perience as to how elements of the one relate to elements of the other can it act purposefully on the world. The computer problem-solving program called Gps, designed to model some of the main features of human problem solving, exhibits in stark action depends on building this kind of bridge the efferent worlds. On the afferent, of sensory, represent desired situations or desired objects as well as the present situation. It must be able also to represent differences between the desired and the presenr. on the efferent side, Gps must be able to represent actions that change obiects or situations. To behave pur- posefully GPS must be able to select from time to time those particular actions that are likely to remove the particular differences between desired and present states that the system detects. In the machinery of Gps, this selection is achieved through a table of connections, which associates with each kind of detectable difference those actions that are relevanr to reducing that difference. These are its associations, in the form of produc- world. Since reaching a goal since some attempts may be ecting the progress it is mak- ing (the changes in the differences berween the actual and the desired) and for trying alternate paths. The Logic of Search GPS then is a sysrem thar searches selectively through a (possibly large) environment in order to discover and assemble sequences of actions that The Science of Design 123 will lead it from a given situation to a desired situation. 'I7hat are the rules of logic that govern such a search? Is anything more than standard logic involved? Do we require a modal logic to rationalize the process? Standard logic would seem to suffice. To represent the relation between the afferent and the efferent worlds, we conceive GPS as moving through a large maze. The nodes of the maze represent situations, described affer- ently; the paths joining one node to another are the actions, described as motor sequences, that will transform the one situation into the other. At any given moment GPS is always faced with a single question: "'Sfhat action shall t try next?" Since GPS has some imperfect knowledge about the relations of actions to changes in the situation, this becomes a ques- tion of choice under uncertainty of a kind already discussed in a previ- ous section. It is characteristic of the search for alternatives that the solution, the complete action that constitutes the final design, is built from a sequence of component actions. The enormous size of the space of alternatives arises out of the innumerable ways in which the component actions, which need not be very numerous, can be combined into sequences. Much is gained by considering the component actions in place of the sequences that constitute complete actions, because the situation when viewed afferently usually factors into components that match at least ap- proximately the component actions derived from an efferent factoriza- tion. The reasoning implicit in GPS is that, if a desired situation differs from a present situation by differences D, D2, ...,Do, and if action A, removes differences of type D,, action A, removes differences of type D, and so on, then the present situation can be transformed into the desired situation by performing the sequence of actions AtA2 , , , A,, This reasoning is by no means valid in terms of the rules of standard logic in all possible worlds. Its validity requires some rather strong as- sumptions about the independence of the effects of the several actions on the several differences. One might say that the reasoning is valid in worlds that are "additive" or "factorable" in a certain sense. (The air of paradox about the cat-dog and needle-thread examples cited earlier arises pre- cisely from the nonadditivity of the actions in these two cases. The 6rst is, in economists' language, a case of decreasing returnsi the second, a case of increasing returns.) 124 The Science of Design Now the real worlds to which probrem solvers and designers address themselves are seldom completely additive in this sense. Actions have side consequences (may create new differences) and sometimes can only be taken when certain side conditions are satisfied (call for removal of other differences before they become applicable). Under these circumstances one can never be certain that a partial sequence of actions that accom- plishes certain goals can be augmented to provide a solution that satisfies all the conditions and a*ains ail thegoals (even though they be satisficing goals) of the problem. For this reason problem-solving systems and design procedures in the real world do not merely assembre problem sorutions from components but must searcb for appropriate assemblies. In carrying out such " s.arch, it is often efficient to divide one's eggs among a number of baskets-that is, not to follow out one line until it succeeds completely or fails definitely but to begin to explore several tentative paths, continuing to pursue a few that look most promising at a given moment. If one of the active paths begins to look less promising, it may be replaced by another that had previously been assigned a lower priority. our discussion of design when the akernatives are not given has yielded at least three additional topics for instruction in the science of design: 3. Adaptation of standard logic to tbe search Design so_lutions are sequences of acdons that lead to satisfyingspecified constraints. With satisficing goals the ble worldsare seldom unique; the search is foi iufficient, not necessary, actions forattaining goals. 4. The exploitation of parailer, or near-parailer, factorizations of differ-ences. Means-end analysis is an exa Lple of a broadly applicable problem-solving technique that exploits this factorization. 5' lhe allocation of resources for searcb to arternatiue, partry exproredaction sequences. I should like to elaborate somewhat ort. this last_mentioned topic. Design as Resource Allocation There are two ways in which design processes are concerned with the allocation of resources. First, conservation of scarce resources may be one of the criteria for a sarisfactory design. second, the design process itself The Science of Design 125 involves management of the resources of the designer, so that his efforts will not be dissipated unnecessarily in following lines of inquiry that prove fruitless. There is nothing special that needs to be said here about resource con- servation-cost minimization, for example, as a design criterion. Cost minimization has always been an implicit consideration in the design of engineering structures, but until a few years ago it generally was only implicit, rather than explicit. More and more cost calculations have been brought explicitly into the design procedure, and a strong case can be made today for training design engineers in that body of technique and theory that economists know as "cost-bene6t analysis." An Example from Highway Design The notion that the costs of designing must themselves be considered in guiding the design process began to take root only as formal design proce- dures have developed, and it still is not universally applied. An early ex- ample, but still a very good one, of incorporating design costs in the design process is the procedure, developed by Marvin L. Manheim as a doctoral thesis at MII for solving highway location problems.6 Manheim's procedure incorporates two main notions: first, the idea of specifying a design progressively from the level of very general plans down to determining the actual constructionl second, the idea of attaching val- ues to plans at the higher levels as a basis for deciding which plans to pursue to levels of greater specificity. In the case of highway design the higher-level search is directed toward discovering "bands of interest" within which the prospects of finding a good specific route are promising. Within each band of interest one or more locations is selected for closer examination. Specific designs are then developed for particular locations. The scheme is not limited of course to this specific three-level division, but it can be generalized as appropnate. Manheim's scheme for deciding which alternatives to pursue from one level to the next is based on assigning costs to each of the design activities and estimating highway costs for each of the higher-level plans. The 5. Marvin L. Manheim, Hierarchical Structure: A Model of Design and Planning Processes (Cembridge: The MIT Presc, 1966), 126 The Science of Design highway cost associated with a plan is a prediction of what the cost would be for the actual route if that plan were particu larizedthrough subsequent design activity. In other words, it is a measure of how ,,promising,, a plan is. Those plans are then pursued to completion that look most p-romising after the prospective design costs have been offset against them. In the particular method that Manheim describes, the ,.promise,, of a plan is represented by a probability distribution of outcomes that would ensue if it were pursued to completion. The distribution must be esti- mated by the engineer-a serious weakness of the method_but, once estimated, it can be used within the framework of Bayesian decision the-ory' The particular probability moder used is not the important thing about the method; other methods of varuation without the Bayesian su- perstructure might be just as satisfactory. In the highway location procedure the evaluation of higher-level plans performs two functions. First, it answers the question, :.wh... shall I search next?" second, it answers the question, ..'when shall I stop the search and accept a solution as satisfactory?" Thus it is both a steering mechanism for the search and a satisficing criterion for terminating the search. Schemes for Guiding Search Let us generalize the notion of schemes for guiding search activity beyond Manheimt specific application to a highway location problem and be-yond his specific guidance scheme based on Bayesian decision theory. Consider the typical structure of a problem_solving program. The pro_ gram begins to search along possible paths, storing in memory a ,.tree,, of the paths it has explored. Attached to the end of each branchjeach partialpath-is a number that is supposed to express the "value,' of that path. But the term "value" is really a misnomer. A partial path is ,rot " ,olu-tion of the problem, and a path has a ..true', value of zero unless it leads toward a solution' Hence it is more useful to think of the values as esti- mates of the gain to be expected from further search along the path thanto think of them as ..values" in any more direct ,.nr.. io, example, itmay be desirable to attach a rerativery high varue to a partial exploration that may lead to a very good sorution but with a low probability. If theprospect fades on further exploration, onty the cost of the search has been lost, The disappoirrting outconle need not be accepted, but nn rrltcrrrativc The Science of Design 127 path may be taken instead. Thus the scheme for attaching values to partial paths may be quite different from the evaluation function for proposed complete solutions.T 'When we recognize that the purpose of assigning values to incomplete paths is to guide the choice of the next point for exploration, it is natural to generalize even further. All kinds of information gathered in the course of search may be of value in selecting the next step in search. We need not limit ourselves to valuations of partial search paths. For example, in a chess-playing program an exploration may generate a continuation move different from any that was proposed by the initial move generator. lThatever the context-the branch of the search tree- on which the move was actually generated, it can now be removed from the context and considered in the context of other move sequences. Such a scheme was added on a limited basis by Baylor to MAIER, a program for discovering check-mating combinations in chess, and it proved to en- hance the program's power significantly.8 Thus search processes may be viewed-as they have been in most dis- cussions of problem solving-as processes for seeking a problem solu- tion. But they can be viewed more generally as processes for gathering information about problem structure that will ultimately be valuable in discovering a problem solution. The latter viewpoint is more general than the former in a significant sense, in that it suggests that information ob- tained along any particular branch of a search tree may be used in many contexts besides the one in which it was generated. Only a few problem- solving programs exist today that can be regarded as moving even a modest distance from the earlier, more limited viewpoint to the newer one.e 7. That this point is not obvious can be seen from the fact that most chess-playing programs have used similar or identical evaluation procedures both to guide search and to evaluate the positions reached at the ends of paths. 8. George \0. Baylor and Herbert A. Simon, "A Chess Mating Combinations Pro- graml' Proceedings of the Spring Joint Computer Conference, Boston, April 26- 28, (79551:431-447 (Washington: Spartan Books, 1965), reprinted tn Models of Thought, chapter 4.3, 9, A formal theory of the optimal choice of search paths can be found in H. A, Simon and J. B. Kadnne, "Optimal Prohlem-Solving Search: All-or-nonc Solu- ti trn s," A rti;f c i a I I n t e I I i ge n <' e, 6( I 97 S | 23 5-247 . 128 The Science of Design The Shape of the Design: Hierarchy The Generator-Test Cycle One way of consider interrelations among cknowledging that the think or the design! ::::1":Hr:.Jlll:ll,l: 10' I have developed this argument at greater length in my essay ..The Architec-rure of Complexiry" chapter g. v ---i 11, For a recent discussion of ,fu1c-1i9n1l analysis in design, see Clive L. Dym, iii::r,:;* Design (New york, Nt; ail;i;. unive.Iity p,.,,,-'rsear, pp. The Science of Design 129 tives and, then, the testing of these alternatives against a whole array of requirements and constraints. There need not be merely a single generate- test cycle, but there can be a whole nested series of such cycles. The gener- ators implicitly define the decomposition of the design problem, and the tests guarantee that important indirect consequences will be noticed and weighed. Alternative decompositions correspond to different ways of dividing the responsibilities for the final design between generarors and tests. To take a greatly oversimplified example, a series of generators may generate one or more possible outlines and schemes of fenestration for a building, while tests may be applied to determine whether needs for par- ticular kinds of rooms can be met within the outlines generated. Alterna- tively the generators may be used to evolve the structure of rooms, while tests are applied to see whether they are consistent with an acceptable over-all shape and design. The house can be designed from the outside in or from the inside out.t2 Alternatives are also open, in organizing the design process, as to how far development of possible subsystems will be carried before the over-all coordinating design is developed in detail, or vice-versa, how far the over- all design should be carried before various components, or possible components, are developed. These alternatives of design are familiar to architects. They are familiar also to composers, who must decide how far the architectonics of a musical structure will be evolved before some of the component musical themes and other elements have been invented. Computer programmers face the same choices, between working down- ward from executive routines to subroutines or upward from component subroutines to a coordinating executive. A theory of design will include principles for deciding such questions of precedence and sequence in the design process. As examples, the ap- proach to designing computer programs called structured programming is concerned in considerable part with attending to design subproblems 12. I am indebted to John Grason for many ideas on the topic of this section. J. Grason, "Fundamental Description of a Floor Plan Design Program," EDRAl, Proceedings of the First Enuironmcntal Design Association Conference, H, Sa- noff and S, Cohn (eds.), North Cnrolinn State Univeraity, 1970. 130 Tbe Science of Design in the proper order (usually top-down); and much instruction in schools of architecture focuses on the same concerns. Process as a Determinant of Style '\tr7hen we recall that the process will generally be concerned with finding a satisfactory design, rather than an optimum design, we see that se- quence and the division of labor between generators and tests can affect not only the efficiency with which resources for designing are used but also the nature of the final design as well. rJilhat we ordinarily call "style" may stem iust as much from these decisions about the design process as from alternative emphases on the goals to be realized through the final design.t3 An architect who designs buildings from the outside in will arrive at quite different buildings from one who designs from the inside out, even though both of them might agree on the characteristics that a satisfactory building should possess. 'When we come to the design of systems as complex as cities, or build- ings, or economies, we must give up the aim of creating systems that will optimize some hypothesized utility function, and we must consider whether differences in style of the sort I have just been describing do not represent highly desirable variants in the design process rather than alter- natives to be evaluated as "better" or "worse." Variety, within the limits of satisfactory constraints, may be a desirable end in itself, among other reasons, because it permits us to attach value to the search as well as its outcome-to regard the design process as itself a valued activity for those who participate in it. 'We have usually thought of city planning as a means whereby the plan- ner's creative activity could build a system that would satisfy the needs of a populace. Perhaps we should think of city planning as a valuable cre- ative activity in which many members of a community can have the op- portunity of participating-if we have wits to organize the process that way. I shall have more to say on these topics in the next chapter. However that may be, I hope I have illustrated sufficiently that both the shape of the design and the shape and organization of the design process 13. H. A. Simon, "Style in Design," Proceedings of the 2nd Annual Confercnce of the Environmental Design Research Association, Pittsburgh' PA: Carnegie Mellon University (1971\, pp. l-10. The Science of Design 131 are essential components of a theory of design' These topics constitute the sixth item in my proposed curriculum in design: 6. The organization of complex structures and its implication for the ot- ganization of design Processes' Representation of the Design I have by no means surveyed all facets of^the emerging science of design' In particular I have saidiil;;t"t the influence of problem representation on design. elthot'gt' ti;trnf """"tt of the question is recognized today' we are still far from " 'r';;;;ittheory of the subiect-in particular' a theorv that would tell us h;;';;;t"trate effective problem rePresentations''a I shall cite o,tt t*"-plt' to "i"kt clear what I mean by *representatton' Here are 'r" "'rt' oi"";;"' which t shall call number scrabble' The game is played Uy r*o pto-ple with nine cards-let us say the ace through the nine of hearts' T;::;; are placed in a row' face up' between the two players. Tht p;;;'"* "l""'"'elY' one at a time' electing any one of the cards tr'"t *"1"t;;;;t Tht "i- of the game is for a plaver to make op " "booki; lt'"' i" a set of t*"ttt'three cards whose spots add to 15, before hi' ";";;; t"t a" so' The fi"t pl"yt' who makes a book wins; if all nine t""d' h"ut been drawn without either player making a t"Siint"?:;;'1il; in this game? How would vou so about finding one? If the reader ;;;;;. aheady discovered it for himself, let me show how a change in representation will lake it easy to play the game well' The magic 'Ott""-i*-o-*n"f' I introduced in the third chapter' is made "n "f rft". numerals from 1 through 9' 492 357 816 14. As examples of current thinking ' ;;;;6;;6ed Artiracts") and 6 ( DYm' For a Re Pre 132 The Science of Design Each row, column, or diagonal adds to 15, and every triple of these nu_merals that add to 15 rs a row, column, or diagonal of the magic square.From this, it is obvious that "making a book,, in number scrabble isequivalent to getting "three in a row" in the game of tic-tac-toe. But mostpeople know how to pray dc-tac-toe welr, hence can simply transfer theirusual strategy to number scrabble.rj Problem Solving as Change in Representation That represenradon makes a difference is a rong-familiar point. ve alrbelieve that arithmetic has become easier since Arabic nu-.r"I, and placenotation replaced Roman numerals, arthough I know of no theoretictreatment that explains why. That representation makes a difference is evident for a different reason.AII mathematics exhibits in its concrusions only what is arreadyimplicitin its premises, as I mentioned in a previous chapter. Hence all mathemati_cal derivation can be viewed ri-pry as change in representation, makingevident what was previously true but obscure. This view can be extended to all of problem solving_solving a prob_lem simply means representing it so as to make the sol,itio'trarrs'parent.,.If the problem solving courd actually be organized in these terms, the issueof representation would indeed b..o-. Jrrt "l. But even if it cannot-if this is too exaggerated a view_a deeper understanding of how re_presentations are created and how they contribute to the solution ofproblems will become an essential component in the future theory ofdesign. Spatial Representation Since much of design, particularly architecturar and engineering design,is concerned with objects or arrangements in real Euclidean rwo- 15' Number scrabble is no1 gh.e only isomorph of tic-tac-toe. John A. Michon hasdescribed another, JAM, which it rrrrar"i"i'iili".*o. in the sense of projective, the rows, columns, a _toe become pointssquares of the former .,^-,_,-_1-1.'* bv "jamming" a[ the ]:]:i"8 the points' j amming a singre segmenr. other isomoron:t:i;,T.";:..:l; i ft ,il"l iffl'l)-?,X:t' Mech a n i za tion o f crea ti ve Processes," IEEE sp e ct r u m The Science of Design 133 dimensional or three-dimensional space, the representation of space and of things in space will necessarily be a central topic in a science of design. From our previous discussion of visual perception, it should be clear that "space" inside the head of the designer or the memory of a computer may have very different properties from a picture on paper or a three- dimensional model. These representational issues have already attracted the attention of those concerned with computer-aided design-the cooperation of human and computer in the design process. As a single example, I may mention Ivan Sutherland's pioneering SKETCHPAD program which allowed geo- metric shapes to be represented and conditions to be placed on these shapes in terms of constraints, to which they then conformed.lT Geometric considerations are also prominent in the attempts to auto- mate completely the design, say of printed or etched circuits, or of build- ings. Grason, for example, in a system for designing house floor plans, constructs an internal representation of the layout that helps one decide whether a proposed set of connections among rooms, selected to meet design criteria for communication, and so on, can be realized in a plane.ls The Taxonomy of Representation An early step toward understanding any set of phenomena is to learn what kinds of things there are in the set-to develop a taxonomy. This step has not yet been taken with respect to representations. We have only a sketchy and incomplete knowledge of the different ways in which prob- lems can be represented and much less knowledge of the significance of the differences. In a completely pragmatic vein we know that problems can be de- scribed verbally in natural language. They often can be described mathe- matically using standard formalisms of algebra, geometry, set theory, analysis, or topology. If the problems relate to physical objects, they (or their solutions) can be represented by floor plans, engineering drawings, 17, l,E. Sutherland, "SKETCHPAD, A Man-Machine Graphical Communicn- tion System," Proceedings, AFIPS Spring loint Computer Conferencq 196J (Bel- timore: Spartan Books), pp, 329-346, 18. See also C, E. Pfefferkorn, "The Design Problem Solverr A Syatem for Derlgn- ing Equipment or Furniture Leyouta," in G, M, Eestmen (ed.), Spat/a/ Synthetlt in Computer-Aidcd BuildingDciltn (Londonr Applied Seienee Publiahers, 1975), renderings, or three-dimensionar models. problems that have to do withactions can be attacked with flow charts and programs.other items most rikery wilr need to be added to the rist, and there mayexist more fundamentar and significant ways of classifying its members.But even though our classific"tion is incomplete, we are beginning tobuild a theory of the properties "f .h.r;;;;.;;;^"'" ;:"_:theories or .o-p*.r archire*ures and- ;rjJ:rff;TJllrt'r'rffexample' the work on functional ranguages and object-oriented ran-guages-iilustrate some of the direcdon, th"t a theory of representationscan take ' There has also been closery paralrel progress, some of it reviewedin chapters 3 and 4, toward u.ra.rrt".rairrg the hum"r, use of representa_tions in thinking. These topics begin to frovide substance for the finalsubiect in our program on the th.ory of design, 7. Abernatiue representations for design problems. Summary-Topics in The Theory of Design My main goal in this chapter has been to show that there already existtoday a number of components of a theory of design and a substantialbody of knowledge, theoretical "rrd ._p-iri.al, relating to each. As wedraw up our curriculum in design-ir th. science of the artificiar-totake its place by the side of .,"t.rrl ..i.rr.. i., the whole engineering cur_riculum, it includes at least the followrn* ,ofr.r, THE EVALUATION OF DESIGNS 1' Theory of evaluation: utility theory, statisticar decision theory2. Computational methods: 134 The Science of Design "' lj?il,1T:j:*::::: :!::::,t arternadves such as rinear pro- il:T.TlT :::f ::i:1i :o*':r'h;il ;;;;il'a"ffi iJb. Algorithms and heuristics for .hoorirrg ;oi ;'rf**;':;::;':;;::3. THE FoRMAL Locrc oF DESTcN: imperative and declarati".;;;;THE SEARCH FOR AITERNATIVES 4. Heuristic search: factorization and means_ends analysis5. Allocation of resources for search 5. TTTSONY OF' STRUCTURE AND DESIGN ORGANIZATION: hicrArChiCsystems 7. nspnesrurATloN oF DEsrcN pRoBLEMs The Science of Design 135 In small segments of the curriculum-the theory of evaluation, for ex- ample, and the formal logic of design-it is already possible to organize the instruction within a framework of systematic, formal theory. In many other segments the treatment would be more pragmatic, more empirical. But nowhere do we need to return or retreat to the methods of the. cookbook that originally put design into disrepute and drove it from the engineering curriculum. For there exist today a considerable number of examples of actual design processes, of many different kinds, that have been defined fully and cast in the metal, so to speak, in the form of run- ning computer programs: optimizing algorithms, search procedures, and special-purpose programs for designing motors, balancing assembly lines, selecting investment portfolios, locating warehouses, designing highways, diagnosing and treating diseases, and so forth.re Because these computer programs describe complex design processes in complete, painstaking detail, they are open to full inspection and an. alysis, or to trial by simulation. They constiture a body of empirical phe. nomena to which the srudent of design can address himself and which hc can seek to understand. There is no question, since these programs exist, of the design process hiding behind the cloak of "judgment" or ,,experi. ence." 'Whatever judgment or experience was used in creating the pro- grams must now be incorporated in them and hence be observable. The programs are the tangible record of the variety of schemes that'man has devised to explore his complex outer environment and to discover in that environment the paths to his goals. Role of Design in the Life of the Mind I have called my topic "the theory of design" and my curriculum a .,pro- gram in design." I have emphasized its role as complement to the natural 19. A number of these programs are described in Dym, op. cit,, and others ore ok on Engineering Design in the Large, written Engineering Design Research Center at Carnegic des each chapter of his book with a commentary . Dym's book has n hibliography of more than 200 items, a majority of them referring to specific design projects and eyEtemrl its extent gives some indication of the rare at which the ee ience of derign ir now progreeaing, 135 The Science of Design scrence curriculum in the totar training of a professional engineer_or ofany professional whose task is to sorve probrems, to.h""r.;;;:;thesize,to decide. But there is another way in which the theory of design may be viewedin relation ro orher knowredge. My third and fourth .h""pt r, *.r. .h"p-ters on psychology-specificalry on man's relation to his biological innerenvironment' The present chapter may arso be construed ", " Jh"pr.. onpsychology: on man's relation to the complex outer environment in whichhe seeks to survive and achieve. All three chaprers, so construed, have import that goes beyond theprofessional work of the person we have called the ..designer.,, Many ofus-have been unhappy about the fragmentation of our society into twocultures' Some of us even think there are not iust two curtures but alarge number of curtures' If we regret that fragmentation, then we mustlook for a common core of knowrJge th"t can be shared by the membersof all cultures-a core that incrudes more significant topics than the of children, or perhaps lation to the inner and we live and choose can This may seem an extravagant claim. Let me use the rearm of music toillustrate what I mean. Music is one of th. most]illil'; ,.i.n.., ed by the Greeks. Anything I have as well to music, its composition ing topics I have used for most of Music involves a formar paftern. It has few (but important) contactswith the inner environment; that is, it is capable of .uoking ,rron, ._o_tions' its patterns are detectabre by human ri.t.rr.rr, and some of its har-monic relations can be given physicar and physiological interfretations(though the aesthetic import oi ,h.r. is debatable). As for the outer envi-ronment' when we view composition as a problem in design, *..rr.oun-ter just the same tasks of evaruation, of search fo. "lt.rl"iiues, arrd ofrepresentarion that we do in any other design probrem. If it pleases us,we can even apply to music some of the same techniques of automaticdesign by computer that have been ur.d rn other fierds of design. If The Science of Design 137 computer-composed music has not yet reached notable heights of aes- thetic excellence, it deserves, and has already received, serious attention from professional composers and analysts, who do not find it written in tongues alien to them.2o Undoubtedly there are tone-deaf engineers, just as there are mathemati- cally ignorant composers. Few engineers and composers, whether deaf, ignorant, or not, can carry on a mutually rewarding conversation about the content of each othefs professional work. What I am suggesting is that they can certy on such a conversation about design, can begin to perceive the common creative activity in which they are both engaged, can begin to share their experiences of the creative, professional design process. Those of us who have lived close to the development of the modern computer through gestation and infancy have been drawn from a wide variety of professional fields, music being one of them. We have noticed the growing communication among intellectual disciplines that takes place around the computer. We have welcomed it, because it has brought us into contact with new worlds of knowledge-has helped us combat our own multiple-cultures isolation. This breakdown of old disciplinary boundaries has been much commented upon, and its connection with computers and the information sciences often noted. But surely the computer, as a piece of hardware, or even as a piece of programmed software, has nothing to do directly with the matter. I have already suggested a different explanation. The ability to communicate across fields-the common ground-comes from the fact that.all who use computers in complex ways are using computers to design or to par- ticipate in the process of design. Consequently we as designers, or as de- signers of design processes, have had to be explicit as never before about what is involved in creating a design and what takes place while the cre- ation is going on. The real subjects of the new intellectual free trade among the many cul- tures are our own thought processes, our processes of judging, deciding, 20. L. A. Hillier and L. M, lsaacson's Experimental Music (New York: McGraw- Hill, 1959), reporting experiments lregun more than four decades ago, still pro- vides a good introduction to the subject of musicnl composition, viewed as design. See elco Walter R. Reitman, Cogttition and Thought (New York: Wiley, 1955), chapter 6, "Crcative Problem Solvlngr Noter from the Autobiography of n Fuguc," 138 Tbe Science of Design ,."T,:i5;"::ljl.ljtlg:y. "..: importing and exporting from one intet_ *:'* -*::liT . "", *er i de a s ; b" * il; ;ffi i;';H;T,I::i Social Planning: Designing the Evolvingmatioa-processing system rike a human oai - ""o"t urgantzect comnlev nf _o- ^-t ____ ng_or a computer, ;",T"r1.:,:lf:i"11-:T*andcompute*,"".""",*o'." jo#,.l,ij Artifact ::*].fi:'ems and achieves go"rr' i; ou;,"."#;ffi:::?T:;complexity. 6 The proper study of mankind ha gued that people-or at least their itively simple, that most of the comp from their environment, from their s my case, then we can cgn;l1de rhat, in large part, the proper study ofmankind is the science of.design, ,,* orrt/", ,fre professional component :fffi:.}|Hl education rut ",; ."* ;J;rine for *.,ri,u-.","rir.a,- In chapter 5 I surveyed some of the modern tools of design that are used by planners and artificers. Even before most of these tools were available to them, ambitious planners often took whole societies and their environ- ments as systems to be refashioned. Some recorded their utopias in books-Plato, Sir Thomas More, Marx. Others sought to realize their plans by social revolution in America, France, Russia, China. Many or most of the large-scale designs have centered on political and economic arrangements, but others have focused on the physical snyilqnrnsnl- river development plans, for example, reaching from ancient Egypt to the Tennessee Valley to the Indus and back to today's Nile. As we look back on such design efforts and their implementation, and as we contemplate the tasks of design that are posed in the world today, our feelings are very mixed. We are energized by the great power our technological knowledge bestows on us.'We are intimidated by the magni- tude of the problems it creates or alerts us to. We are sobered by the very limited success-and sometimes disastrous failure-of past efforts to de- sign on the scale of whole societies. S(e ask, "If we can go to the Moon, why can't we . . . )"-ns1 expecting an answer, for we know that going to the Moon was a simple task indeed, compared with some others we have set for ourselves, such as creating a humane society or a peaceful world. tU7herein lies the difference? Going to the Moon was a complex matter along only one dimension: it challenged our technological capabilities. Though it was no mean ac- complishment, it was achieved in an exceedingly cooperative environ- ment, employing a single new organization, NASA, that was charged with n single, highly operatiorral g
还剩19页未读

继续阅读

下载pdf到电脑,查找使用更方便

pdf的实际排版效果,会与网站的显示效果略有不同!!

需要 5 金币 [ 分享pdf获得金币 ] 0 人已下载

下载pdf

pdf贡献者

HudsonJoe

贡献于2015-06-20

下载需要 5 金币 [金币充值 ]
亲,您也可以通过 分享原创pdf 来获得金币奖励!
下载pdf