154

ing and decision making practices. The possibilities for constructing a mixed optimum model of policy making, superior to both the "muddling through" and "rational-comprehensive" ones also requires attention, especially because of the neglect of such a possibility in the professional literature.

Conditions Limiting the Validity of the "Science of ‘Muddling Through" "Thesis

Conceding the many penetrating insights in Lindblom's paper, there may nevertheless be a critical examination of two main elements of the "Science of 'Muddling Through'" thesis, namely the incremental nature of desired changes in policy, and agreement on policy as the criterion of its quality.

The basic strategy of incremental change, as stated by Lindblom, is one of maximizing security in making change. All reliable knowledge being based on the past, the only way to proceed without risk is by continuing in the same direction, limiting consideration of alternative policies "to those policies that differ in relatively small degrees from policies presently in effect." This is sound advice, provided certain conditions pertain—a requirement not adequately faced by Lindblom.

Unless three closely interrelated conditions are concurrently met, incremental change by "successive limited comparison" is not an adequate method for policy making. These three essential conditions are: (1) the results of present policies must be in the main satisfactory (to the policy makers and the social strata on which they depend), so that marginal changes are sufficient for achieving an acceptable rate of improvements in policy-results; (2) there must be a high degree of continuity in the nature of the problems; (3) there must be a high degree of continuity in the available means for dealing with problems.

When the results of past policies are undesirable, it is often preferable to take the risks involved in radical new departures. For instance, in newly developing states aspiring to accelerated socio-economic development, the policies followed by the former colonial policy makers clearly do not constitute an acceptable basis to be followed with only incremental

'Lindblom, "The Science of 'Muddling Through' ", p. 84.

change. Similarly, in modern countries whe changes in values make formerly accepte policy-results unacceptable, radical departure in policy are required despite the risk, for o stance in respect to the segregation proble in the United States.

When there are no past policies in respet to a discrete policy-issue, incremental chang is in fact impossible. For instance, many d the problems faced during the New Deal bat novel characteristics making most policie (other than doing nothing) a radical departur from the past.

Changes in knowledge-technological and behavioral-put at the disposal of policy makers new means of action, which, unless g nored, lead to radically new policies. The best illustrations are provided in military techno ogy, where "incremental change" results in the often noted tendency of a nation's armed forces to be excellently prepared for the las war. Similar illustrations can be cited in most spheres of social action where innovations in knowledge take place, such as medicine (po icy making in regard to smoking) and educ tion (utilization of programmed teaching chines).

The three conditions essential to the val ity of the "muddling through" thesis are mos likely to prevail where there is a high degre of social stability. Under conditions of stab ity, routine is often the best policy, and change being at a slow rate, incremental po icy-change is often optimal. But, even in the most stable societies, many of today's qual tatively most important problems are tied up with high speed changes in levels of aspire tions, the nature of issues, and the available means of action, and require therefore a pol icy making method different from "muddling through."4

A similar conclusion may be reached from examination of the reliance on agreement o policy as the criterion of the policy's quality, Under conditions of stability, when all rele vant parties have a more or less clear image of the expected results of a certain policy with a high correlation of subjective and ob

In his "Policy Analysis”, ibid., Lindblom explicith recognizes that his analysis applies to the United States and other "stable, well-established, deeply rooted democracies" (p. 30). But he fails to pursue this limitation and does not realize that there is to day no country, including the United States, "stable" enough to fit his analysis.

L.

rective probability, a policy agreed upon will rdinarily involve little risk of catastrophe; lso, under such conditions, it is in fact much asier to agree on a discrete policy than on bstract goals. In contrast, under conditions f high-rate change, ignorance can produce 2greement upon a catastrophic policy; under uch conditions, moreover, it is often much asier to agree on abstract or operational goals e.g., "raising the standard of living," "increase het per-capita product by 2% annually") than on policies, there being no background of shared experience to serve as a basis for consensus on policy. Lest the reader reach the conclusion that these comments apply only to new development countries, let him consider military policy, where decisions agreed upon by experienced military personnel were rejected, following their dissection by McNamara's "whiz kids."

The formula that "agreement" equals "high quality" is the more dangerous because of its appeal to a value highly regarded in democratic ideology, as attested to by the abundance of "administration by consent" literature. It is, therefore, highly necessary to emphasize that agreement should follow examination of the consequences of policy and not be substituted for it, in all save the most familiar and stable policy areas.

"

The conclusion is inescapable, therefore, that the "Science of 'Muddling Through' thesis has limited validity. It may be more valid for a larger number of policy areas in a relatively stable society, such as the United States, than in countries engaged in high-rate directed social change. But even in the United States many of the most critical policy problems involve factors changing at a high rate of speed.

The Impact of "Muddling Through" on Actual Policy Making Although Lindblom's thesis includes a number of reservations, these are insufficient to alter its main impact as an ideological reinforcement of the pro-inertia and anti-innovation forces prevalent in all human organizations, administrative and policy making. The

'Even more dangerous is acceptance of the "incremental change" attitude by social scientists, as it reduces their functions as an innovating social factor. By limiting their suggestions to incremental change, the social scientist also significantly reduces his utility

155

actual tendency of most organizations is to limit the search for alternatives to the minimum; there is little danger in real life then, that administration will become bogged down in exhaustive search for all alternatives and full enumeration of consequences, in order to achieve "rational-comprehensive" policy making. The "rational-comprehensive" model has at least the advantage of stimulating administrators to get a little outside their regular routine, while Lindblom's model justifies a policy of "no effort."

Taken together, the limited validity of the "muddling through" thesis and its inertiareinforcing implications constitute a very serious weakness. This criterion in no way diminishes Lindblom's pioneering role in pointing out the shortcomings of the "rational-comprehensive" policy making model. This may well prove to be one of his most important contributions, since the counter-model of "muddling through" is itself open to serious doubts. A choice between these two models would be difficult, but, luckily, may be avoided through reliance upon a third model for public policy making.

A Normative Optimum Model for
Policy Making

The bases for an optimum model for policy making are the following assumptions: (1) Optimum policy making involves an effort to increase rationality-content, through more explication of goals, extensive search for new alternatives, conscious attempts to elaborate expectations, with an explicit cut off point, and some formulation of decision-criteria. (2) Extrarational processes play a significant role in optimal policy making on complex issues. This is not only unavoidable because of lack of resources and capacity for complete rationality, but in fact, makes a positive contribution to better policy making. Intuitive judgment, holistic impressions derived from immersion in a situation, and creative invention of new alternatives are illustrations of extrarational phases of optimal policy making. The importance of such processes is not only ac

to policy makers, the latter being well aware of incremental alternatives and looking to the social scientist for new ideas. See Yehezkel Dror, "The Barriers Facing Policy Science," 7 The American Behavioral Scientist, 3-7 (January 1964), esp. pp. 4-5.

156

knowledged by all experienced policy makers but also by some modern research in psychology, for instance the works of Carl Rogers and Michael Polanyi discussing "experience" and "tacit knowledge" as sources of insight and understanding. (3) These extrarational policy making phases can be improved by various means, such as case discussions, sensitivity sessions and "brain storming." Similarly, the rational policy making phases can be improved, for instance, through increasing the input (especially time), increasing the knowledge and qualifications of policy-practitioners, and by setting up special "think" units for the improvement of conceptual analytical tools. (4) Actual policy making in modern states has a tendency to follow precedents, “incremental change," "muddling through," inertia and routine. Compared with the rate of change in the problems faced by policy, in the levels of aspiration, in the aspiration, in the available alternatives, and in knowledge on the policy making process itself (such as decision-techniques and group dynamics)—-most contemporary policy making practices lag behind. This can and should be improved.

A reprocessing of both the "comprehensiverationality" and the "successive limited comparison" models in light of these assumptions may result in a normative-optimum model for policy making which includes the following main features:"

-Some clarification of values, objectives, and decisioncriteria.

-Identification of alternatives, accompanied by a conscious effort to consider new alternatives (through survey of comparative literature, experience and available theories) and to stimulate creative alternative innovation.

-Preliminary estimation of expected pay-off of various alternatives and decisions whether a strategy of minimal risk or a strategy of innovation is preferable. -If the first, the "successive limited comparison" model should be followed. If the latter, the next element is establishment of a cut-off for considering possible results of alternative policies and identification of main expected results, relying on available knowledge and intuition.

-The test of the optimum policy is that it is agreed

upon by the various analysts after full and frank discussion of stages 1 to 4.

-A conscious effort is made to decide whether the

•For a detailed discussion of this model, see Yehezkel Dror, Public Policymaking Reexamined (Chandler Publishing Co., 1965). Part Two.

problem is important enough to make analysis : comprehensive.

-Theory and experience, rationality and ta tionality all are relied upon, the composition of se mix depending upon their availability and the ture of the problem.

-Explicit arrangements are made to improve the quality of policy making through systematic karming from experience, stimulation of initiative an creativity, staff development and encouragement intellectual effort.

The Need Redefined

To state the problem of policy making as a choice between the "rational-comprehensive and the "successive limited comparison" meth ods is misleading and dangerous. It is mislead ing because other policy making models ca be devised. It is dangerous because it leads either to an effort to achieve the impossible ar to an encouragement of inertia and a contin uation of the status quo.

What is needed is a model which fits reality while being directed toward its improvement, and which can in fact be applied to policy making while motivating a maximum effon to arrive at better policies. By all the criteria. the model presented in "The Science of 'Mud dling Through" is inadequate, having lim ited validity, and constituting a barrier to the improvement of policy making. Its favorable acceptance, the result in part of its many mer its, reflects the widespread disposition of ad ministrators and students of public adminis tration to accept the present as a guide to the future, and to regard contemporary practice as a norm for the future.

The broad acceptance of the "muddling through" thesis indicates that inertia and the tendency to "incremental change" are in fact widespread phenomena. This in itself serves to emphasize the need for models of policy making stressing the limits of such an ap proach and pointing out the needs and possi bilities for better policy making. The normative optimum model presented above is only one variation of such a model, many others being feasible.

It is time for administrative science to enter into the area of inquiry, illuminated by Lindblom, and take up the challenge of providing models for policy making instead of devoting time and resources to managerial techniques. on the one hand, and "Grand Theories" on the other. In doing so, the search should be di

HE two major objections that Yehezkel Dror brings against my model of decision aking through "muddling through" are to - taken very seriously. I grant the first objecon immediately; the method is unquestionbly one of less than universal usefulness. I elcome any attempt to specify more precisely an I have done the circumstances in which nuddling through" is a defensible strategy r decision making.

I have, however, further refined and deiled this model, referring to it as the strategy disjointed incrementalism, in A Strategy Decision. There it is argued that in a potical democracy like that of the United States nd probably also in a relatively stable dictorship like that of the Soviet Union, public olicy decisions are typically of the kind that ll for the disjointed incremental approach. ypically, policy decisions meet Dror's three sential conditions for the application of y model: "(1) the results of present policies ust be in the main satisfactory to the policy akers . . . ; (2) there must be a high degree f continuity in the nature of the problems; nd (3) there must be a high degree of connuity in the available means for dealing ith problems." Examples are policy making n taxation, social security, national security, form of the judiciary, foreign aid, adminisrative organization, antitrust, auto traffic, zonng, and agriculture price supports, among

thers.

Dror's second charge-that the strategy cannot but serve as an ideological reinforcement of the pro-inertia and anti-innovation orces prevalent in all human organizations"

'David Braybrooke and Charles E. Lindblom (The ree Press of Glencoe, 1963).

-I am skeptical about. To be sure, it is not a charge that can be conclusively refuted; but Braybrooke and I discuss it at some length. Logically speaking, one can make changes in the social structure as rapidly through a sequence of incremental steps as through drastic-hence less frequent-alterations. Psychologically and sociologically speaking, decision makers can sometimes bring themselves to make changes easily and quickly only because the changes are incremental and are not fraught with great risk of error or of political conflict. American society, for example, is a rapidly changing society. One can argue that it can change as fast as it does because it usually avoids big controversies over big changes. To be sure, anyone who proposes to scrap the world and start over again sounds more radical than one who proposes to start from the here-and-now, making only a sequence of modest changes in the social structure. But through which of these styles of thought does the continuing transformation of American society receive its impetus?

For administrative decisions, I do not agree that " 'the rational-comprehensive' model has at least the applied advantage of stimulating administrators to get a little outside their regular routine, while Lindblom's model justifies a policy of 'no effort' ". Nothing would be more paralyzing to an administrator than to take seriously the prescription of the rational comprehensive model that he make no decision until he canvas all possible alternative ways of reaching well formulated goals, making sure that he has investigated every possible major consequence of each possible alternative. He can only "get a little outside" his reg. ular routine by practicing some strategy that gives him some direction without asking for

158

the impossible. And this is what the strategy does. For the particular component features of the strategy constitute in effect answers to questions that an administrator must face as to what to do in the face of a decision that cannot be comprehensively analyzed or mastered. They constitute an alternative to mere guesswork, intuition, or random analysis.

Nevertheless, I am quite willing to entertain the possibility that in some subtle respects the strategy has effects on the decision maker's attitudes that might dispose him either more or less favorably toward preservation of the status quo than would some alternative strategy for decision making. It is a question, pertinent to any strategy for decision making, that bears even more investigation than Braybrooke and I have given it.

Dror goes on to ask what kind of model of decision making we need. We need a number

of them, I would say, because there are a m ber of decision making methods and strateg: in actual use, each of which needs clanic tion. We do not need "a model" as I unde stand Dror to suggest; and it is perhaps defect of his own proposed model that, tv, to encompass too many varied aspects of dec sion making, his propositions become simp a series of discrete observations and pres tions on decision making which, taken wi group, are not tightly interlocked and which taken one by one, are not each general valid or acceptable. Moreover, Dror asks tha a model: (a) "fits reality" while (b) "being directed toward its improvement" and "which can in fact be applied to policy mak ing" while (d) "motivating a maximum effor to arrive at better policies." I doubt whethe any reader, on reflecting on these four criter for a model, will accept them in combination