Clarence Philbrook1, “‘Realism’ in Policy Espousal,” The American Economic Review, vol. 43, no. 5 (December, 1953), pp. 846-59.


Economists in general have learned to live under the recurring charge that they are “unrealistic.” The term may, of course, mean many things. Where the meaning applies to what they do strictly as “scientists,” the barbs need not go deep. Against such slings and arrows there hangs ready a familiar armor — namely, the prestige of scientific method, made magically tough by having clothed also those heroes of the battle for knowledge, the natural scientists. Not always are economists content, however, to remain “scientists” describing relationships and eschewing reference to the desirability of things, or even to remain “engineers” suggesting devices to achieve ends laid before them entirely by others. They do step out of these rôles and give their society, or some sector of it, advice which commits the adviser to responsibility for value judgments.2 They do “take positions” — act as advisers in a full sense. In this rôle too they are often faced with the charge of being “unrealistic.”

Used in this connection the epithet becomes a different missile and will not be turned aside by the same armor. What range of meanings may possibly attach to the term need not for present purposes be explored. One particular connotation which it frequently carries will be singled out and attributed to the word. This meaning is, that the economist charged with “unrealism” has made recommendations requiring for their fulfillment changes in things which must, for the purpose at hand, be treated as unchangeable. It would be brash to deny that such a charge may often be true. A basic responsibility implied in it has, at any rate, a valid claim to attention: of course the economist, to give useful advice to society, must regard various things as in a significant sense beyond our power to alter. Failure to do this seems virtually a definition of idle dreaming. Wherever he has failed to be guided properly by this precept, the economist must indeed confess that he has sinned.

Yet there is a more cankerous sin which economists have all too often committed in order to escape indictment. So great has become the scramble to turn state’s evidence, that one who suggests a solution not treating as “given” (“here to stay”) any particular popular social practice is merely fortunate if no colleague misquotes him Carlyle’s response to Margaret Fuller’s decision to accept the reality of the universe: “Egad, you’d better!” The charge of “unrealism” is used with telling effect to discredit policy recommendations without adequate consideration. It probably affects in no small degree the determination of what types of work are respected and supported in universities. Science has rightly been said to aim at stability of belief by cultivating doubt; but the spirit behind the familiar charge, by restricting the range of questioning deemed worth while, limits the use of science for this fundamental purpose. In so doing that spirit, on the negative side, creates a presumption that the time of scholars is readily available for other activity, and, on the positive side, implies that worthy uses for time must be sought elsewhere. It therefore fosters a test of scholars which, at a moderate stage of development, asks whether their activity is “constructive” (Does it build, not tear down? Does it look forward, not backward?) and, at a now common extreme, asks whether their activity immediately and discernibly influences practical affairs. Of course the man least demonstrably ineffectual is he who advises others to do what he knows they will do without his advice. Indeed, the competition for reputation as “realists” works toward a condition in which students of society are loath to take a minority position. Society tends as a result to lose the benefit of that disinterested, fundamental, continuous criticism which, unless provided by persons made independent of practical affairs for that very purpose, is unlikely to be forthcoming. There has grown a widespread practice of cooperation with “things as they are,” without explicit criticism of them, which is bound to have the effect of active approval regardless of whether such is intended. Thus, the spirit of “Egad, you’d better!” bids fair to render the field of political economy not merely useless but actually damaging to the social welfare.

However, I wish to invite contemplation and discussion; for the practical consequences hinging upon the treatment of this issue are enormous, and yet the matter seldom crosses the threshold into systematic consensus-seeking, but rather remains a little-explored source of recurring irritation and ad hoc dicta. I should like, therefore, to avoid resting upon the controversial conclusion stated above, and find a more neutral point of departure.

The methodology of this essay

Perhaps this can be found along the following lines. In deciding what advice to give, one clearly must regard some conditions as given. On the other hand, the very fact that some one is to be advised to do something which presumably he might not otherwise do means that not everything in the situation is being treated as given. Evidently what is at issue is “correctness” in the selection of things to be so treated. If we can find literally no criteria, consensus on policy is indeed hard to come by. It must then be impossible to force the “unrealist” to regard the necessary things as fixed; and, on the other hand, his critic can always, by “realistically” accepting the universe of his choice, find sanctuary against the necessity of defending his own positive policies. Is it possible to say anything helpful toward consensus about what actually constitutes “realism,” in some defensible sense, in the selection of proposals on policy? It may turn out so, if we explore a path through a series of intellectual positions, asking how far it may reasonably be supposed that plaintiff and defendant can walk together and where they might feel forced to part company. The method will be to start with a position which a “realist” surely will not think labels alterable any conditions which are not so, and progressively introduce elements which he might consider less “realistic,” at each stage inquiring whether he really would be likely to reject the new element. Finally the question will be raised whether the “realist” can reasonably insist upon any principle which would enable him to use meaningfully the familiar charge. An effort will be made to maximize the ease with which any telling rebuttal that may be available against the argument of this paper can be made, so that if the conclusions reached here are wrong the fact may be made clear and many of us may mend our ways in keeping with a sound “realism.”

The starting “realistic” position: Determinism

The position most safely “realistic” would seem to be that which took everything in the situation as given, including both the physical universe and the attitudes of all persons. One who adopted it would presumably interest himself exclusively in the discovery of regularities and the forecasting of events. The fact that this view by no means corresponds to the scientific outlook in total does not assure that the two are never thought to be identical. Such identification probably accounts in no small measure for the high prestige of simply any “empirical” research as opposed to “arm-chair theorizing”: almost any conceivable collection of figures will be regarded in important quarters as representing a certain hard-headed practicality, while any serious discussion of what is the real significance of the collection is considered either jealous carping or mere mental gymnastics by minds too lazy to do “constructive” work. Finding “what the facts are” is the great task according to this view; for, if they were once discovered, we should then be in position to make forecasts.

This position has great appeal to men of scientific bent, and indeed to all of us. It is “obvious” to common sense that every effect must have a cause and that any state of mind must be an effect of preceding states of mind, body, and environment. To be sure, some embarrassment may arise in trying to find a function and purpose for the scientist himself; for, although the surrounding stimuli may cause him to emit sounds and cause us to call the sounds a forecast, he cannot in this view “influence” anything, and the notion of function, including that of purpose, is itself meaningless. Nevertheless, the mere fact that an analyst will typically use words which imply that his conclusions are yet to be determined by the effort of investigation or discussion does not prove the deterministic position false. Activity that is apparently purposeful may be viewed either as the result of a free will operating with its available means toward some goal, or alternatively, as the movement of a set of pre-established forces working toward equilibrium. In the determinist view the pseudofunction of the scientist is that of a cog in a machine. The tempting query why, if determinism is true, the scientist would struggle as he does, or society pay him to do so, is itself meaningless, for “would” no longer can connote volition and “struggle” has no meaning; he and society do not do, in any significant sense, what they think best — they simply do.

The well-known and fatal difficulty with the deterministic attitude that we must treat everything as given is that the truth of its opposite also is obvious and is inevitably dominant in our procedure. It is certainly impossible for the mind to act regularly as if it regarded itself as an epiphenomenon accompanying the burning out of a physicochemical reaction. Although nuclear physicists could not alter the fact of the fissionability of the atom, it is said that some of them believe they could (as well as should) have refused to discover the fact or, at least, to put it to its most famous use. No analyst will, where it counts, cling to a purely deterministic position.  All will “play like” their “decisions” are decisions in the common-sense meaning of the word and capable of affecting reality.

The second “realistic” position: Discretionary edict

Social analysts as advisers will, then, take at least one step away from regarding everything as given and still feel themselves to be “realists.” The most modest step is exemplified where the analyst occupies an administrative post with some discretionary power. A man in that position would believe that the word “discretion” had some meaning — that the external situation could be influenced by a decision, such as, for example, to allow a certain wage rate to be raised. The step is modest in that it leaves the scientist in the same instrumentalistic relation to the situation as he might bear to a group of chemical materials to be manipulated. Further, nearly all social analysts would take another step without at all losing a sense of “realism,” in believing that they might actually influence the world by giving advice to crucially placed administrators or legislators — on occasion simply by making evident to them the truth of a cause-and-effect sequence or the validity of a value. Indeed, adding together these two steps, we have the basis for what appears to be regarded as the most valuable activity of analysts as social advisers, although a relatively high respectability seems to attach also to the act of advising organized pressure groups.

The range of possible change, and of influence of the social adviser, is relatively narrow if channels of influence are restricted to those just mentioned. For most important changes (usually legal) require a change of attitudes on the part of a large number of persons. This need not appear true, to be sure, to one willing to take a purely instrumentalistic view. In that outlook, the range of action would be considerably broader, since it would be possible to advise holders of power to use their power “for the good of the people” with small regard for what the people thought they wanted. Indeed, much recent history of government might be accounted for in terms of a general belief that “realism” is violated by those who hope social advisers may have influence otherwise than directly. This instrumentalistic position need not disavow the possibility of change in mass attitudes: a benevolent government may “educate” the people to agreement with its measures. This is a clear-cut, possible intellectual outlook; but if such neo-Machiavellianism is to be espoused, the social scientists adopting it should make as clear as they can that they have given up hope of that consensus-created, not merely consensus-creating, government which is usually called democracy. Those who believe that consensus should precede political action must, it is merely tautological to say, believe the archetype of their “action” as advisers to consist of effort to help others discover correct attitudes; so we may suppose that, in considering attitudes to be among the alterable conditions, we still have the company of “realists” whose ultimate values include a rôle for democracy.

The third “realistic” position: Deciding what alterations of attitude to attempt

Since important changes in society do require alterations in the attitudes of many persons, social analysts, in order to form opinions on what changes would be worth while if they could be brought about, must often consider at great length what processes would go forward under assumed conditions which are often at variance with the facts. Strangely, by carrying on such considerations carefully rather than carelessly, men call down upon their heads some of the most scornful charges: “unrealistic,” “too theoretical,” “purist,” and the like. Yet it is difficult to see what could be more “unrealistic” than saying that a change would be worth while (unless mere change for its own sake is the sole issue) without first methodically considering what results might reasonably be expected. So surely it may be assumed that the school of “realism” would accept the study of “models,” or “theories,” as indispensable.

It is in deciding what alterations of attitude to attempt, that the point arises which presumably most persons stressing “realism” have in mind. Should we not, they might ask, distinguish among conceivable changes according to whether we stand some reasonable chance of actually effecting the necessary shift of attitude? Why waste effort by making suggestions which we cannot hope will be accepted? These queries appear to have much justice. The idea obviously is that one ought to list the actions open to him and attach to each a weight determined by the probability that it would in fact bring about the attitude change intended.

If we described a “pure realist” position as calling for decision by simply selecting the advice with the highest “probability” weight, we ought to be able to treat it as only a straw man. To be sure, the mere fact that most “realists” would reject an approach so stated does not mean it really differs from what they have effectively in mind; a major part of the thesis of this paper is that numerous economists are in effect espousing positions which, if stated clearly, would be repudiated by them. Nevertheless, one fatal criticism of any such bald “probability” approach must surely be treated as noncontroversial. No one would explicitly maintain that a high relative probability of having some effect should give priority to an act completely without regard to the relative desirability of the outcome it would tend to have. It seems reasonable to suppose, therefore, that the “realist” would accept as a part of “realism” the necessity of consulting end values as determinants of advice.

It becomes important, then, to recall what is required in due allowance for this element of “realism”; for, although few would deny its claims in the abstract, many become quite impatient when called upon in specific cases to give full weight to it. (Who having experienced the earlier years of the New Deal can forget how frequently the conclusive defense of almost any policy was that the government was doing something!)

The fourth “realistic” position: Weighing the desirability of the change promoted

Evidently the adviser, having arrayed the lines of advice he might give on a particular issue, must assign a weight to each according to the desirability of the change it would tend to promote. A change can be desirable otherwise than for the sake of novelty only as it furthers some end. But since there are conflicts of ends, an action may further one end and lessen the realization of another, and the net significance cannot be known without first evaluating the various ends (or degrees of realization of them). Relative evaluation requires a common measure, hence can be accomplished only if the “ends” are but intermediate, serving in turn as means to, and deriving their value from, a more general end (which itself is merely one of a number of ends all serving as means to a yet more nearly ultimate goal.) Assigning the correct desirability weight to a change means judging its net influence upon the realization of some end sufficiently general (“high”) to include all the intermediate ends touched by the change; and this judgment involves tracing the effects of the change upon each intermediate end and evaluating those effects in each case by reference to the inclusive end. Thus, literally full justification of the whole set of value weights one attaches to a set of policy suggestions would require laying bare the relations between the proposed changes and some all-subsuming, ultimate end of social organization. Only in this way could the mutual consistency of all the various means and subsidiary ends which constitute a total program for society be established.

In discussion aimed at consensus, the problem is one of showing a disputed policy to be a means to some intermediate end and then inquiring, of parties to the discussion, whether this “end” is in their view necessary as a means to what they regard as some more fundamental goal. If not, agreement is not logically called for. The proponent must then try to show that the disputed “end” is a means to something he himself regards as a more fundamental end, and if successful repeat the query. It is for present purposes unnecessary to dwell upon the complexities hidden under the light reference to showing a disputed end to be a means; what is needed is merely a suggestion of the extensiveness of the obligation taken on by acceptance of the necessity of value weights and consensus. If consensus is to be soundly reached, a process of the sort indicated must go forward until the derivation of importance is traced back to some end the authenticity of which is accepted by the parties to the discussion. Fortunately, it is usually not necessary to go far toward ultimates: but the ideal student of policy, far from refusing to talk about the sheer desirability of a proposal opposed to his own, would stand ready and eager to reduce the two, no matter how different, to commensurability in the manner suggested.

There is good reason to be appalled by the difficulty of the task cited, especially in a world become self-conscious and discordant about its values. It is of no avail, however, to assail as “purists,” “absolutists,” or “unrealists” those who call attention to the task. The only alternative to readiness to defend a position in the sense indicated is to grant that at least some of one’s means and intermediate ends have not been tested for mutual consistency, and that one’s policy proposals have therefore no more authority than some diametrically opposite set. These considerations surely place a demand upon an economist giving advice on social policy that he shall have thought his way through, so far as the state of knowledge makes possible, to an internally consistent model of economic organization on which he bases his advice. The gravity of the potential effect of error in policy, and the immensity of the scope for growth and ripening of intellect toward power to grapple with the problems involved in discerning good policy, render merely ludicrous the widespread impatience toward “arm-chair theorizing,” which, unless the integrity of the men involved is in question, must be regarded as an arduous process of critical contemplation and integration aimed at the development of wisdom.

Of course nothing in the foregoing discussion denies the notion that the end must be dealt with in the light of the availability of means. Neither is it affirmed that “the end justifies the means” in the sense that there is any end for the furtherance of which no device is too bad: but the ultimate end itself must contain criteria of acceptability of means, so that the means, if justified, are indeed justified by the end. Even though strong doses of starvation might have shortened immensely the employment depression of the ‘thirties, the ends of the society justified the relief of distress. These things are probably all that need saying about the perennial argument over the short-run versus the long-run view in the selection of policy; yet one’s faith that the full import of attributing value weights is noncontroversial is bound to be shaken by memory of how often any manifestation of the long-run view is answered with some form of the completely irrelevant “realistic” proposition, “In the long run we are all dead.” Although this reaction has done much damage to the search for sound policy, by limiting the consideration of ends; it may nevertheless reasonably be interpreted as something other than repudiation (in the abstract) of thinking in terms of ends and means. Indeed, it is probabiy as a rule a confused, distorted, and misapplied criticism, not of means-ends thinking, but of presumed failure to apply fully such a calculus. More often than not, the thought reflected is simply a generalization of the example given above: the idea that the time dimension must be allowed for — that radical change takes time and meanwhile we must make the best of the situation, perhaps by means which hamper realization of some accepted goal. To be sure, a proposal involving thought of the long run might often be criticized because it seemed to violate the true “probability” approach, which we shall in a moment try to state. But, it may be taken for granted, relatively few “realists” would question that in such an approach value considerations must have a place.

Summary of most likely meaning of a “probability” approach

It seems possible now to sum up the most likely meaning of a “probability” approach — a position which must be the one that would be held by any who, in the light of considerable discussion, still were willing to make use of the charge, “unrealistic.” Although in some cases only under pressure, they surely would accept these propositions: that many elements in a social situation are capable of change, including the attitudes which determine policy; that actions of individuals in their rôle of “social advisers” could have some influence on those attitudes; and that the advice to be given by them cannot be chosen without the aid of weights attached to different lines of advice according to the relative desirability of the change which these respectively would tend to further. They would insist, however, that a choice can be made only after superimposing another weight on each line of advice according to the probability of its taking effect if offered. Men called “unrealistic” are those who disregard this principle and presumably those who assign probability weights incorrectly.

There is a considerable appearance of reason about this method of thought. Still, it requires the most careful inspection.

“Realism” and “probability” can be mutually exclusive

It is important not to slight the fact that two lines of advice such as typically give rise to controversy over “realism” are not necessarily so related as to make “probability” considerations relevant at all. A believer in states’ rights, for example, presumably may make clear his belief while yet offering good counsel on how best to operate a program which disregards such a conception. The relation required if “probability” is to have any place is that the two lines of advice be truly alternative — that is, mutually exclusive.

Two possible causes of mutual exclusiveness are suggested. First, sheer limitation on time, energy, and intellect requires specialization; and some decision must be made as to the field in which to develop special competence. May one not make an ample contribution by advising the Federal Reserve authorities how best to use their powers, without questioning whether the present banking system is the best conceivable? Actually, specialization raises no problem. The question is whether, not having studied the possibilities of alternative systems, the specialist causes or allows it to be supposed that his activities constitute support of this system against alternatives, or whether, having decided another system would be better, he conceals the fact. If one applies “probability” thinking and recommends that approach for others, it must be from sources other than specialization that he derives the implied mutual exclusiveness of different pieces of advice.

The second cause of mutual exclusiveness which the “realist” presumably has in mind is that the very act of approving a different general course may destroy power to guide action along the best path in an established, although less desirable, direction. There is no denying that suggestions from one listed among the faithful with respect to a program are, as “self-criticism,” more likely to wield influence than is advice from “outsiders.” Thus we can identify conceptually a type of case in which the lines of advice are alternative in a sense sufficient to justify at least entertaining the question of whether it is desirable to be guided by “probability.”

It goes without saying, that the mere identification of such a class of cases gives no license to beg the question of whether any particular case belongs to the class. If indeed “probabilities” are sometimes a proper guide, the epithet “unrealistic” can at best be justified only upon an explicit showing that giving one of the lines of advice does preclude the usefulness of giving the other. But even for use in cases where “probability” is formally relevant the “probability” approach is open to grave doubt on a number of scores.

Problems with the “probability” approach

In the first place, what is the criterion for the relative weighting of a “degree of probability” and a “degree of desirability”? There is always the devastating possibility that the total weight of an action in keeping with a completely evil outcome might be greater than that of one tending to further a desirable end. Suppose that in Germany under Hitler it appeared that one hundred thousand Jews would probably be bayoneted to death. Should I, if a German, have suggested merely a less brutal execution by gas — thus quite possibly doing some good — or should I have cried out against the whole idea, quite probably having no discernible influence? And, if I decided on the former, should I have claimed credit as a “realist” and decried as “unrealistic” any who advised the society to let the Jews live?

In the second place, the problem of actually assigning “probabilities” must be faced — some aspects of which are relegated to our third and fourth points. At present we may consider the matter from a relatively uncomplicated point of view. The task of assignment involves knowing attitudes, including in some sense their “strength,” and the force of the impact of each suggestion under consideration as tending to change them. Diffidence about analyzing matters of “attitude” would become us, for there are indications that merely agreeing upon a scientific content for the term might keep sociologists busy for some time. However, it appears that an attitude rests upon at least these underlying factors: a value system, the “brute” elements provided by the senses, and theories relating these elements. Thus, basic to assigning “probabilities” would be a knowledge, with respect to each person affecting the adoption of a policy, of how these factors combine in him to form attitudes, including what is the “marginal influence” of changes in the respective factors. Given such information, successful forecasting would require discerning also exactly what persons the advice would reach and in what form and, for each person reached, just what changes would occur in each of his attitude-making factors. (Since forecasts are, for the purposes in question, to run in terms of probability distributions, the knowledge required — whatever its nature — would be more rather than less complex than these statements suggest.)

In the case of an official economic adviser to one or a few immediate determiners of policy, presumably sufficient knowledge for a forecast might be possessed — although it would be naïve to take for granted that the enigma which man presents to man regularly yields to the degree of intimacy involved in such a relationship. However, to assay how appropriately the cited requirements might in general give pause to a forecaster, it would be well to recall what must be meant by “lines of action” open to economists as social advisers. These vary greatly as to immediacy of impact upon policy, ranging from the report of an official adviser, through systematic promotion (honest propaganda), down to a mere planting of seeds of thought whether in the classroom or in casual conversation. Evidently the overwhelming majority of “actions” exercise their effects quite indirectly. What is the “probability” that a policy-influencing marginal change in attitudes will result from planting an idea, or the means of arriving at good ideas, in the classroom or in an address or in social intercourse? With what degree of immediacy, both in point of time and in point of the number of intermediate impacts? A mere showing of difficulty would not invalidate the conception of a “probability” approach. But it would scarcely be extreme to suspect that estimates of “probability” would in the case of most expressions of espousal have a probable error so large as to render them “statistically” insignificant.

The knowledge stipulated as necessary was not characterized as sufficient for the appraisal of “probabilities.” Indeed, contemplation suggests some mystery about the very nature of such a process in the kind of universe in question. These facts suggest a problem which for emphasis we may single out as our third point: a fundamental difficulty in the very notion of “probability” as a property of the force of a piece of advice. Unless the intent is simply to lead others to act under false impression of the results to be expected, advice consists of making evident the soundness of an idea and relying upon the resulting recognition of that soundness to cause action based upon it. The force at work changing attitudes is, then, “idea force.” “Probability” implies a mechanical relation, some kind of quantitative continuity. But correct conclusions cannot flow from treating “idea force” as if the law of conservation of energy applied to it; for, however an idea may get into a mind, it is capable of dying there or of gathering immense force. Moreover, a number of minds can be seeded with one expression of the idea. Potentially, then, the force may grow at an astronomical rate. All the determinants of whether it dies or gathers strength in a particular mind, we simply do not know. We do know, however, that that which is believed to be true has an appeal to the mind believing it, over and above the attraction it may have on other grounds: truth has a positive appeal in its own right. Although other appeals may swamp this one, there is no limit to the possibilities of its breaking through and causing action. The degree of apparent influence of the person holding an idea is, therefore, no measure of the potential effect of his giving utterance to it. If there is one belief fundamental to, and universal in, our culture, it is this notion that truth as such has power. Such justification as may exist for the hope of a better world lies here and here only. In the face of these considerations it seems doubtful, not merely whether we can effectively measure “probabilities” for different “actions” upon attitudes, but indeed whether any real meaning attaches to the conception.

In the fourth place, to guide oneself by a “probability” calculus is one thing, but to say then that all persons should do likewise is to fall little if at all short of nonsense. There might be some sort of rationality to urging that others espouse according to value weights only, so as to leave one or a few free to add the “probability” factor. The latter procedure implies estimating the current division of the forces of espousal and deciding whether one’s own influence would suffice to change an inferior force into a superior. Evidently, then, much of the force exercised by others must have been somehow committed. If all, however, follow the “probability” principle, no one can commit himself until many others (nearly all?) have committed themselves. Thus, for keeping public affairs relatively free from the influence of the disinterested, the “probability” principle would be a reasonably effective device.

Since it would seem harsh to attribute to the “realist” a meaning so patently foolish, an attempt at a different interpretation may be appropriate. Although the inference does not appear obviously logical, could the actual meaning involve a process in which the many individual advisers, each with the “probability” factor in mind, somehow simultaneously determine their respective positions? If so, to be sure, the implications are still not attractive. Major economic policy, in so far as it is influenced at all by economists, apparently ought to be the product of infinite involutions of guesses by each about what others are guessing about what he is guessing about what they will advocate! However, for example, having advocated farm-price supports on grounds of their “probability,” it appears that one may take a fearless stand in favor of ninety per cent of parity as opposed to ninety-five per cent — unless, indeed, the probabilities are too much against that. It is a depressing picture of the rôle of representatives of that higher learning upon which much of the hope of social improvement has sometimes been supposed to depend.

In view of the preceding point, a fifth one is evident: the “probability” approach rests upon an ethical atomism which might reasonably be supposed long ago to have been barred by a well-established principle of ethics, that a precept is unacceptable for one man unless it would have an acceptable outcome when followed by all. In the face of this principle, how can the degree of immediate influence of a scholar upon practical affairs be deemed a matter of greater concern than the desirability of the direction of such influence as he does wield?

A sixth and major indictment of the “probability” approach, although contained in much of the preceding discussion, should be made explicit. Fundamental to such a procedure is the fact that, while believing one course to be the best, I say or allow it to be thought that I believe a different one to be the best. That is to say, the approach explicitly calls for the active or passive concealment of truth. It should scarcely be necessary to argue against an idea which so patently strikes at the foundations of scientific integrity.


If the thoughts assembled here are sound, men will best serve the cause of social welfare if they refrain both from disposing of opponents as “unrealistic,” in our defined sense, and from bowing to fear of that charge in selecting their own positions on policy. Degrees of “probability” are incommensurable with degrees of value, and ethical considerations do not permit the former to outweigh the latter. The assignment of “probabilities,” even with a highly oversimplified conception of their meaning, would in the great majority of cases be so difficult as to become absurd. The nature of “idea force,” moreover, makes it doubtful whether any real meaning attaches to the very notion of “probabilities of success” of different lines of advice and whether, therefore, there can be any such weights to superimpose upon value weights. In any case, to call the “probability” approach proper for all who take a position on policy would be to advocate a practice which would remove any rational basis which “probability” estimates might be imagined to have, by reducing all policy position-taking to a logically impossible universal mutual anticipation ending only in universal support of the status quo (in so far as change depends on deliberate human planning), and which would of course divorce policy selection from reference to value. Finally, the required concealment of truth is intolerable to the most fundamental conception of the scholar’s reason for being. The “probability” approach must surely be rejected, to say nothing of its being considered prerequisite to respectability. Only one type of serious defense of a policy is open to an economist or anyone else: he must maintain that the policy is good. True “realism” is the same thing men have always meant by wisdom: to decide the immediate in the light of the ultimate. The economist must follow this ideal as best he can — in humility and in readiness to compare notions both of technical relations and of ultimate values.


  1. The author is associate professor of economics at the University of North Carolina. He expresses thanks for criticism and stimulating discussion to Mr. George M. Woodward, Major Robert L. Bunting, Dr. Edwin J. Stringham, and Professor Dudley J. Cowden.
  2. Throughout this paper the word “advice” and terms used synonomously are to be interpreted in the broadest possible sense — to include any recommendation, inside or outside the classroom, made without hypothetical purpose clauses more restrictive than the necessarily implied aim to achieve the desirable.