[Elster, Jon (1993), Why things don't happen as planned, in N.┼kerman, ed., The Necessity of Friction, Heidelberg: Physica Verlag, pp. 248-56.]
[ Books | Articles | Reviews | Index | The main menu ]
Why things don't happen as planned
[start of page 248]
This is an essay about thwarted plans and frustrated expectations. The experience that thing's don't always work out as planned is a common one. However, instead of just citing Murphy's law or the inherent malevolence of the universe, we need to understand the structure of frustration. Sometimes, things go wrong simply because we make a mistake, such as stepping on the accelerator instead of the brake. I am concerned here with more systematic, recurrent sources of frustration. Some of these are located within the individual, some of them in the interaction among different individuals.
I begin with two complernentary phenomena, weakness of will and excess of will. The former and better-known of these mechanisms has the following structure. We want to do x, for good reasons. We also want to do y, for good reasons. x and y are mutually exclusive. All things considered, we believe the reasons for doing x are stronger than the reasons for doing y. And yet we do y. Weight-watchers and Alcoholics Anonymous owe their existence to this mechanism. It is also illustrated by cases in which x could be to quit gambling, stop smoking, start jogging, or being saving for one's old age. These are all instances that oppose short-term interest as y and long-term interest as x. One could also have y embody self-interest and x some more altruistic concerns.
But x and y may also be turned around. Rigid and compulsive personalities are often unable to give themselves a break. They may well
[end of page 248, start of page 249]
believe that on a given occasion, all things considered, short-term self-indulgence is justified and even recommended, and yet be unable to deviate from the rules they have set for themselves. This is not the phenomenon of excess of will that I consider below. Rather, these rigid characters are weak-willed, prisoners of their self-control techniques. The best perspective on such phenomena is, I believe, a neo-Freudian one. The ego is constantly engaged in a two-front war, between the impulses of the id on the one hand and the impulse-control mechanisms of the superego on the other. It has to fight both against the primary problem and against the secondary problems that arise in trying to cope with the primary one.
To connect this analysis to the idea of frustrated plans I shall focus on the special case in which x and y represent, respectively, long-term and short-term interest. Typically, three temporal moments are involved. At time t0, we know that at time t1 we shall have the choice between getting a small reward at time t1 and a larger reward at time t2. The weakness of will scenario then is as follows. At t0, we decide that we would rather have the later and larger reward than the early, smaller one. As t1 approaches, however, there comes a moment when preference reversal occurs. The imminent availability of the earlier reward clouds our judgment, and prevents us from sticking to the decision we made in a cool and considered moment. What happens in a case like this is that our reasons for choosing y become stronger than the reasons for choosing x - not stronger qua reasons, but stronger qua sheer psychic turbulence.
The idea of excessive will encompasses a very different set of phenomena, which can be summarized as "willing what cannot be willed". It is notorious that certain states are very recalcitrant to attempts to bring them about intentionally. They are essentially by-products of actions undertaken for other ends. Paradigmatic cases are the efforts to overcome impotence, insomnia, calculatedness, unpleasant memories or stuttering by concentrating hard on the desired state - lust, sleep, spontaneity, forgetfulness or uncluttered speech. The very concentration interferes with the state one is trying to bring about and which essentially requires a relaxed, unselfconscious attitude.
On reflection, one can perceive this mechanism everywhere. It explains, I think, the notorious failure of self-help books. (Of course, many such books succeed in realizing the goal of the author, which is to bring in money. I have in mind, though, the goals of the readers.) Self-esteem is essentially a by-product: it is supervenient on other activities that are pursued for reasons other than that of providing self-esteem. It explains, or so I believe, the failure of certain social movements, whose members join them less for the sake of the cause than for the impact that joining will have on their own character formation, neglecting that that impact has to
[end of page 249, start of page 250]
be mediated by a belief in the cause. Character, in fact, is essentially a byproduct.
Many failures of rationality are due to the choice of inappropriate means to a given end. With excess of will, the problem runs deeper. The mistake here is the belief that the desired state is such that it can be realized by means-ends rationality. Instead of talking of excess of will, one might well talk about hyperrationality. But that phrase could also be taken in a different sense, to refer to the desire to find reasons or give reasons for decisions that essentially cannot be rationalized. An example will indicate what I have in mind. Suppose a surgeon is brought to a disaster area in which there are many people in urgent need of surgery. Assume, for simplicity, that he has to decide which of two patients to operate on first. To make a rational decision, he has to assess the relative urgency of the two cases, which be does by means of the appropriate examinations and tests. While he is doing so, however, first one and then the other patient dies on him. He could have saved one of them had he operated without going through the ritual of rationality.
This example is probably farfetched. But there are other, very real cases in which rationality prove to be self-defeating. In child custody and adoption decisions, the rule nowadays is that the custodial or the adopting parent ought to be chosen to as to promote the interests of the child. Courts and adoption agencies must, therefore, compare the parents or adopting couples with respect to their ability to raise the child. In disputed custody cases, especially, this can be a very time-consuming process. Typically, the child suffers a lot, not least because he or she is used as a pawn in the quarrels between the parents. On average, it would have been better for the child if the courts bad not sought to promote the interest of the child, but instead used some swift, mechanical decision rule such as the principle of maternal presumption or even a lottery. In adoption cases, the problem is compounded by the fact that as children grow older, they become less adoptable. When parents rather than children are the bottleneck (as with the adoption of black children in the United States), the insistence on waiting until the optimal parents come up may lead to no parents being found at all. The best, in such cases, is the enemy of the good. To insist on finding the decision that would have been optimal if found instantaneously and costlessly, is pointless when the search for this solution is in fact both time-consuming and costly.
Both varieties of hyperrationality - excess of will, and the search for the optimum that neglects the costs of searching - are instances of self-defeating plans. But they also differ. The latter problem can be solved, once it is recognized. However, as further explained below, recognition is often made difficult by the deep need human beings seem to have for being able to
[end of page 250, start of page 251]
give sufficient reasons for what they do. The former seems more untractable. In some cases, to be sure, one can use indirect strategies. Instead of trying to will sleep to come, one can take a sleeping pill. Instead of trying to induce religious faith by sheer will, one can engage in religious acts that will have the predictable result of inducing faith, while simultaneously inducing forgetfulness about the process itself. This was the strategy proposed in Pascal's wager. But not all the states that I characterized as essentially by-products can be attained by indirect strategies. Once one has eaten of the apple, there is no way back to innocence and spontaneity.
Our minds play many tricks on us. In some cases, the result is to give us some peace of mind in what would otherwise have been an uncomfortable situation. This outcome is illustrated in the fable about the fox and the sour grapes. For the fox, wanting the grapes and knowing that be couldn't get them would have induced an unpleasant state of "cognitive dissonance". By being able to tell himself that they weren't worth having anyway, be made his life easier. But consider another case. I want promotion, but there are few chances that I will get it. In that case, I can imitate the fox and tell myself that the higher position is really too burdensome to be worth the extra pay. But I can also refuse to believe the evidence, and persist in an unjustified belief that promotion is just around the corner. This will, for a while, also bring my beliefs and desires in harmony with one another. However, the mistaken factual belief can also get me into trouble, for instance if I buy a new house on the strength of my expectations.
Beliefs born of passion serve passion badly.
This is not to say, however, that we would be better off without the passions. Without them, we might as well be dead. We would be better off if we could have the motivating power of the passions without their distorting power, but that may not be a feasible combination. The motivation to make plans and to follow them up is positively correlated with the strength of our passions. The ability to follow them up by choosing the appropriate means is negatively correlated. In particular, the passions tend to make us overestimate our own abilities. The only individuals, by and large, with a correct appreciation of their own capacities are the clinically depressed. Imagine that achievements are ranked on a scale from 0 to 10. If I believe I will achieve 9, I shall in fact only achieve 6. If I aim at 7, I shall realize 5. It is only by limiting my aim to 3 that I can in fact achieve what I set out to do. My plan, in the last case,
[end of page 251, start of page 252]
suffers no frustration, but at a price that we might not want to pay (assuming we have the choice, which we don't).
When preferences adjust to circumstances, the result is, as I said, to give us some peace of mind. But the opposite rnechanisrn is also observed. Sometimes, we want what we cannot have simply because we cannot have it, and the moment we get it we no longer want it. Forbidden fruit tastes best, and the grass is always greener on the other side of the fence. When we live in New York, we dream only of living in Paris, and vice versa. Or consider two modern and pre-modern conceptions of love. The pre-modern conception is encapsulated in Hermione's question in Andromaque. "Je t'aimais inconstant, qu'aurais-je fait fidÚle?"' - I loved you while you were inconstant, what would I not have done bad you been faithful? The implicit answer is clearly that she would then have loved him even more. The modern conception is described by Stendhal and Proust. Love relationships are like a see-saw: when one is up the other is down. There is no such thing as happy, consummated love.
Nor is there such a thing as successfully consummated hate. For the person who defines himself by his opposition to a person or a doctrine - the fanatical anti-royalist, anti-communist or atheist - the worst that could happen would be if his destructive efforts actually succeeded. John Donne (The Prohibition) said it this way:
Take heed of hating me,
Or too much triumph in the victory
Not that I shall be mine own officer,
And hate with hate again retaliate;
But thou wilt lose the style of conqueror,
if I, thy conquest, perish by thy hate.
Then, lest my being nothing lessen thee,
If thou hate me, take heed of hating me.
* * *
So far I have looked at some sources of frustrated plans that are internal to the individual. In the second half of the essay I shall consider frustration due to the structure of interaction among individuals. To some extent, as we shall see, these frustrations can also be described as caused by individual failures to form rational beliefs, viz. beliefs about what others are likely to do.
Sartre coined the term counterfinality for a certain type of unintended consequence. In this example, we are asked to consider the process of erosion in the Chinese countryside. For each individual peasant,
[end of page 252, start of page 253]
deforestation is a rational strategy, given his pressing need for more land. However, when all peasants behave in this way, erosion sets in, and they all end up with less land than they bad initially. For each peasant, deforestation is optimal, regardless of what others do. If others let their trees stand, he will get the extra land be hoped for. If they don't, be will at last get wood for burning, since the land would go in any case. The causal mechanism underlying this story is one of externalities. By cutting down the threes on his plot, each peasant raises the probability of erosion on all plots by a small amount. His utility calculus tells him that the risk is worth it. However, universal deforestation raises the probability of erosion to a certainty. In the collective utility calculus, it was not worth it.
I have stated the peasant's dilemma as a Prisoner's Dilemma. But in a way this is misleading. In standard Prisoner-Dilemma situations such as pollution or congested roads, all actors are fully aware of the causal structure of the situation, and know that others are so aware. Nobody is surprised when they find the park full of litter or the roads blocked by cars, for this is only the predictable outcome of others behaving as they do. But when peasants lose their land, they are surprised. Initially at least, they will tend to ascribe the outcome to bad weather or the anger of the gods. Only after a while, if ever, will they come to suspect that the deforestation is the result of their own behaviour. And even then they may not understand the causal structures involved. Perhaps each peasant comes to believe that the erosion on his plot is fully explained by the deforestation on his plot. He may then try to remedy the situation by stopping the deforestation or planting new trees, and be frustrated again when be he finds that it doesn't help.
Rather than pursuing this speculation, let me turn to the better understood and better-documented mechanism underlying the cobweb cycle (so called because if illustrated in a demand-and-supply diagram the mechanism looks somewhat like a cobweb). I shall illustrate the mechanism by an example from farming, but it is much more widespread. In Norway, for instance, overinvestment in the shipbuilding industries in the 1970s and in the fisheries in the 1980s follow exactly the same logic. And innumerable examples from other countries could be cited.
Suppose that in year 1 farmers (in a competitive market) are making up their mind about how much to produce for year 2. To do so, they must estimate what prices will be in year 2. A natural assumption is that prices will remain constant. Planning on that assumption, the farmer will market a certain volume in year 2. Now suppose that year1 prices were unusually low. With capitalist farmers, this will induce a low level of production for year 2. With lower prices, the fact that capital and labour have decreasing marginal productivity makes it necessary to produce less in order to
[end of page 253, start of page 254]
equalize price and marginal cost. With family farms, we will observe a high level of production. Since the family cannot fire any of its members, they now have to produce more to earn a decent income.
In both cases, the farmers will be surprised. In an economy of capitalist farmers, they will be pleasantly surprised when they observe they they [sic] get very high prices at the low volumes produced by all. In an economy of family farmers, the surprise will be an unwelcome one, as they observe that at the large volumes produced by all, prices have fallen to even lower levels. This mechanism was observed in the Nordic countries during the Great Depression, and ultimately led to regulation of agriculture. In the following I shall leave it aside, and follow only the development of the capitalist case.
In year 2, then, capitalist farmers will produce little and get a good price. When planning for year 3, I assume that they assume that prices will remain high. The logic of equalizing price and marginal cost will then induce them to plan a large volume for year 3. However, when all act on that plan, the large volume thrown on the market will drive prices down. In theory, this cycle can go on forever. In one possible scenario, there will be explosive cycles, in which the below-average prices that obtain in alternate years will fall more and more, and the above-average prices that obtain in intermediate years will rise more and more. In another scenario, the cycles will be damped, so that the difference between below-average and above-average prices is steadily reduced. In the latter case, prices will eventually converge to the equilibrium level defined by the intersection of supply and demand curves. There will be no more surprises, no more frustrated plans.
But consider the former case. As prices oscillate in ever-widening cycles, it is not unreasonable to assume that the farmers eventually come to understand that this is their own doing. Each farmer will come to see that a year of low prices tends to be followed by a year of high prices. He will, therefore, adopt a new principle to generate his expectations: instead of assuming that next year will be as this year, he will come to think that next year will show the opposite pattern of this year. However, this will not be the end of his frustration. For when all farmers form their expectations on this new pattern, they will all be proven wrong. In a year of low prices, each farmer expects that prices will be high next year. As a consequence, he plans for a big volume. Next year, however, when all farmers dump their large crops on the market, prices will not reach the expected levels. Even if the farmers eventually understand what is happening and adjust accordingly, this will only reproduce the problem at a new level.
The root of the problem is that each farmer believes himself to be one head ahead of everyone else. The initial assumption, that prices will remain
[end of page 254, start of page 255]
constant, amounts implicitly to an assumption that all other farmers are like automata that mechanically do the same thing year in year out, so that he is the only one to adjust to market conditions. After he (together with the others) learns that others are in fact adjusting to the market, he now (along with all the others) acts on the tacit assumption that he is the only one to have understood this fact. But each farmer cannot be the only one to have seen through the mechanism.
Similar phenomena can arise in politics. Often, voters make their decisions on the basis of previous opinion polls. Suppose that the polls predict that the Labour Party will win by an absolute majority of 55%. Some voters at the left wing of the Labour Party will then switch their vote to the party immediately at the left, to send a message to the Labour Party to change its polices in that direction. But if more than 5 % of the voters act in this way, the Labour Party might lose the election, contrary to the expectations of these voters. Each of them had tacitly believed that all other voters would behave as they had told the pollsters they could do, so that the Labour majority would be safe from defections.
In these cases the frustrated expectations are due to the failure of the farmers or voters to understand the strategic nature of the situation. Once they come to understand that it would be irrational to believe that others are less rational then oneself, they can form rational expectations that will have the self-fulfilling property that when all act on them they will be exactly fulfilled. The equilibrium prices defined by these expectations have the property that when all but one of the farmers make their plans on the basis of them, the remaining farmer has no incentive to do otherwise. This is in fact the general definition of what is called Nash equilibrium in economics: a set of strategies, one for each actor, with the property that each strategy is the optimal response to the others. In equilibrium, nobody has an incentive to deviate.
However, the existence of an equilibrium (and one usually exists) is not enough to ensure the non-frustration of expectations and plans. In the first place, the actors may not know enough about each other to converge to the equilibrium (assuming there is only one). Only if the actors have a dominant strategy, as in the Prisoner's Dilemma, will they be able to figure out what to do without figuring out first what others are going to do. In the absence of a dominant strategy, they cannot determine the equilibrium without knowledge of the preferences of the other actors, and even with this knowledge they may hesitate to use their equilibrium strategy if they are unsure whether others have enough knowledge to do the same.
But there is an even more serious problem. Sometimes, there is more than one equilibrium. This need not create any coordination difficulties, but it often does. If one equilibrium is better for everybody than all the others,
[end of page 255, start of page 256]
it will obviously be chosen (if the knowledge requirements are fulfilled). But often one equilibrium is better for some, and another better for others. If only two person are involved, considerations of bargaining power may help us to predict the outcome. But in large market situations with many equilibria, the outcome is inherently indeterminate.
Now, indeterminacy need not yield frustration. If the actors understand that the equilibrium outcome cannot be uniquely determined even under assumptions of full rationality and perfect knowledge, they will have to adopt an imperfect decision heuristic, fully knowing that after the fact they may come to see that they could have done better. Investment decisions by firms are, for instance, very largely indeterminate in the sense just indicated A given firm will not be able to calculate how much other firms will invest on the basis of mutual rationality and self-fulfilling expectations. Instead of optimizing, the firm can then try do so as well as it can under the circumstances, for instance by following a rule of thumb such as "invest the same proportion of profit as last year, unless the rate of return on capital falls below 6%", with some additional rule about what to do in the latter eventuality. Although there is no reason to believe this to be an optimal response to what other firms will do, there is no other policy that is demonstrably better either. After the fact, the managers may be disappointed when they see that they could have done better, but as they will know that they could not have known they will not be surprised.
Such sagacity is probably rare. More frequently, managers persuade themselves that their policy is not simply no worse than any other, but is actually optimal in the light of what others can be expected to do. Human beings, as I said, have a very strong need to act in terms of sufficient reasons. This deep-seated need for rationality is in fact an important source of irrational behaviour. It is also an important source of frustrated plans. Any mechanism that makes us form expectations when there are no proper grounds for doing so will also induce frustration.
[end of page 256, start of page 257]
Without citing them, this essay has drawn extensively on some of my earlier writings. Her [sic] are some brief references. For counterfinality, see Ch, 5 og [sic] Logic and Society, Chichester: Wiley 1978: For weakness of will, se Ch.II of Ulysses and the Sirens, rev.ed., Cambridge University Press 1984. For states that are essentially by-products, se Ch.II of Sour Grapes, Cambridge University Press 1983. For the indeterminacy of rational action, se Ch.I of Solomonic Judgements, Cambridge University Press 1989. Child custody dilemmas are discussed in Ch.III of the same book. For whishful thinking and its consequences, see Ch.IV of Sour Grapes. For the idea of adaptive and counteradaptive preferences, see Ch.III of Sour Grapes. For the idea of inherently contradictory goals, see Ch.4 of Logic and Society and Ch.IV of Ulysses and the Sirens. Needless to say, many of the ideas developed in these books owe much to the work of other authors. A compact set of references will be found in the bibliographicla [sic] essay at the end of my Nuts and Bolts for the Social Sciences, Cambridge University Press.
[end of page 257]
[ Books | Articles | Reviews | Index | The main menu ]
[Elster, Jon (1993), Why things don't happen as planned, in N.┼kerman, ed., The Necessity of Friction, Heidelberg: Physica Verlag, pp. 248-56.]