The two best-known models for systematic personal decision-making are the maximization of utility and satisficing. The first was invented by Jevons and paradigmatically formulated by Wicksteed, with many formal refinements by succeeding authors; the second, which exemplifies a ‘bounded rationality’ approach, has been advocated by Herbert A. Simon. Neither translates easily into a model for systematic collective decision-making. Even as a personal model, maximization strictly conceived makes impossible demands for information. One would not expect it to fare any better collectively. Indeed it is hard to find a collective analogy even for its personal use in adjusting expenditures in any two categories so as to get the same marginal utility in each. It offers at best an idealized hypothetical basis for criticizing omissions in deliberation. But the ideal survives on the collective level in the confusion between being able to point out painfully omitted considerations after the event and being inclined to demand that all considerations be attended to beforehand, which is impossible.
Satisficing does not make this demand on either level; it asks only that a few threshold criteria (say, location, price, floor space) be laid down and that the first option that meets the criteria be chosen, though if this requires only a brief search through options, the next time the aspiration level of the decision-maker will rise, the thresholds will be set higher, and more criteria may be added. Unfortunately, at the collective level, unless an autocrat or a perfect bureaucracy (with a hierarchy that divides up all the issues that come up and assigns responsibility for deciding each of them under a coordinated plan) is authorized to act for the collectivity, a problem arises about different criteria with different thresholds being invoked at the same time by different people. Which are to guide the decisions of the collectivity? This is a problem for social choice, and if the orderings of personal preferences for social criteria vary every whichway, aggregating the orderings runs into the Arrow theorem, before orderings for policy options themselves are considered. The theorem also applies to any attempt to decide directly on the options themselves. No system of aggregating personal orderings into social choices will meet at once all of Arrow’s apparently very weak conditions about having the choices reflect the orderings. One of Arrow’s conditions, typical in its weakness, is that if an option rises in one or more personal orderings without falling in any, then the option should not fall in the aggregated social ordering. The aggregated orderings can meet this condition and some others only at the cost of always being liable to fail because some three alternatives exhibit a cycle rather than a true ordering: for example, a majority prefer A to B, and a majority prefer B to C, but a majority prefer C to A. (An analogue of the theorem appears at the personal level, if the criteria in the set of criteria give very different orderings for the options [Thomas Schwartz 1970].)
One way of dealing with the Arrow theorem is to rely on the enterprise of politicians to invent a further option when attempts to aggregate personal orderings break down in cyclical voting. However, social choice theory raises an even worse problem (laid out in papers by R. D. McKelvey (1976) and Norman Schofield (1978), respectively) that a majority can be contrived for any option whatsoever if options are paired for decision, one pair after another, across the whole field of options. One way of dealing with this is to rely on the inertia of political institutions and on their divided jurisdictions to make it difficult to force immediate reconsideration of policies once chosen. Another way is to hope for widespread political agreement on criteria, taken together with a very limited scope for politics. A society of like-minded subsistence farmers (which was perhaps approximated in parts of New England in the eighteenth century) might get along very well without transacting much political business. The business need not be more than plausible departures from the status quo, satisficing departures that are also incremental, in important features of institutions no more than small changes; and the departures might fall into recurrent patterns answering to variations in the environment. This arrangement could combine with a limited use of the market, either a perfectly competitive market or a market with a robust place for bargaining, allowing for trade in the products of crafts and small industry, operating with only a negligible draft on nonrenewable resources. It could even combine with a market in which modest technological innovations continually arise.
In either case, the market could serve, as Hayek (1948 following Adam Smith) maintained, to elicit details of information too multifarious to be easily dealt with by a deliberated plan, and give effects to these details in subtly coordinated features of the outcomes of economic activities. The market could serve in the same way in much more complex societies, and in many connections unobjectionably. It is thus itself a collective decision-making device furnishing from moment to moment a great variety of instructive information for producers and consumers. A purely competitive market does this without offering any room for bargaining; but a market in which bargaining is widespread can serve, too, and may likewise reach acceptable economic results that could not easily have been reached by any program for deliberate planning. The market with bargaining, moreover, is a device whose virtues can be carried over to some extent to political bargaining, in which contending interest groups reach one way or another, not always in ways that involve threats and counterthreats, what Lindblom has called ‘partisan mutual adjustment,’ in respect, say, to a presence of religion in the schools or on abortion policies. Bargaining will often, of course, reflect inequalities of power. The inequalities may be unjustified, and exploited in unconscionable ways. But Lindblom is supposing that partisan mutual adjustment, including hard bargaining, takes place in a context in which other systems or processes of decision, among them polyarchy, control through at least minimally democratic procedures, will operate to check oppression. Indeed, for him and Dahl, systems of decision-making come in variable combinations of four processes: the market; bargaining; polyarchy; and hierarchy (as in bureaucracies). Each does some things well that the others cannot, and all have a substantial presence in current democratic societies.
That is no ground for complacency. The present position of collective decision-making in the advanced industrial countries is one in which those taking part are compelled to make incremental decisions both because of inertia (institutional resistance to change) and because the risks of going wrong are so high. Incremental decisions, since they put only so much in present institutions at risk as can be repaired, are appropriate ones for agencies with very imperfect information, and that is the situation for all agencies during rapid social change. Prudence would not license doing more. Nevertheless, it may be suggested, things are going wrong in a very big way. Technological advances that combine with ever-increasing population to use up nonrenewable resources by leaps and bounds may be outracing attempts to combine limited scope for politics with the operation of the market. Think of deforestation; global warming; the collapse of commercial fisheries; the proliferation of the automobile; the marketing in the United States of firearms. In all these cases, collective decision-making seems to many knowledgable people to have faltered because of inattention in the market, inherently, with its shortrun preoccupations; and in politics, because attention there has been intermittent and inconclusive. (This is not to say that government can be counted on to choose better).
Moreover, with powerful political forces arrayed against taking due account of the troubles (which would mean, among other things, big interferences with the market, private profits, and personal incomes), it is difficult to see how the circumstances of decision-making can be made safe for satisficing and incrementalism. Things would not necessarily be better if corruption and special interests miraculously vanished from the system. The quickly multiplying evils cited are all public evils, and remedies for them would be public goods, from which no one could be excluded with respect to benefits and to which no one (taken by himself?) would make an indispensable contribution. So, as Mancur Olson showed in a basic application of rational choice theory (the utility-maximization theory of consumer behavior generalized to personal political choices), everyone would have a motive to escape making a contribution and to free-ride on the contributions of other people. Everyone could see as much by considering just what benefit he would get from the public good in question and how that benefit would not depend on his making a contribution.
This difficulty arises before any policy is adopted to produce the specific public goods that would remedy the evils in question. It arises about taking one’s part in the system of collective decision-making. Why should a rational agent vote? There is a public good in having a polyarchal system in which the votes of the multitude exert some control over the politicians and thus check their greed and their hunger for power. It is desirable that the control extends to correcting for inequitable or shortsighted results from political bargaining and the operation of the market. Indeed a polyarchal system of decision-making will work as intended by its advocates only if this control in all its aspects is effective (an empirical condition) and a large proportion of the population participates (a definitional condition, often linked with an empirical hypothesis about the characterological benefits of participating). But no system for decision-making will work unless the people who have parts to play in the system actually take part; and polyarchy is peculiarly vulnerable on this point. Here, besides a troubling question about whether people will play their parts well, there is an equally troubling question about whether people will play their parts at all. Every voter has a motive to avoid the bother of voting herself, since her contribution to maintaining the system will be effort thrown away either if a sufficient number of other people vote (which would make her vote redundant) or a sufficient number do not vote (which would make her vote quixotic). The same thing is true of voting for a specific policy or candidate.
Nevertheless, people do vote. In spite of complaints about turnout, they vote in large numbers even when there are no ward-heelers acting for political machines to turn them out. A common explanation is that they get entertainment value from voting. But ‘entertainment’ as invoked in this connection must, awkwardly, stretch to embrace an unexpected assortment of things, including the conscientious performance of duty, something like rooting for a horse to win, and perhaps something like prayer. Here the media, much berated for treating election campaigns as horse races, play a vital part. Even when they trivialize the issues, they make politics entertaining. At their best they induce people to take sides on important topics and consequently become anxious about the outcome. Voting is a ritual that discharges the anxiety and brings about closure for weeks, maybe months, of obsession with campaigns. Perhaps it is paradoxical that the maintenance of a system of decision-making should depend on such considerations, extraneous, it would seem, to the mechanics of recording and aggregating votes, and outside the program of narrow self-interest that rational choice theory starts with. But maybe the paradox lies with being attached to that program regardless of the wayward complications of human motivations.
It is true, rational choice theory can accommodate almost any considerations by accepting the considerations as having utility, even decisive utility, for the agents in question. If it keeps doing this, adding considerations after the event, it falls into tautology as an explanation of human decisions: humans decide on the basis of greater utility; but what has greater utility is to be inferred from what they decide. The tautology is easily escaped, however, even with the most diluted notions of utility: at one stage, consulting the agent and presenting her with experimental choices, drawing up the agent’s utility map in advance. Then predicting her decision on the basis of the map is an empirical hypothesis, true or false as the facts turn out.
Moreover, the case of voting gives at least as much comfort to the theorists of narrow self-interest as disappointment. Large numbers of people eligible to vote do vote; but then large numbers do not, in spite of having the same stake in keeping up a democratic system and polyarchal control. In other connections, the free-riders form an even greater proportion of the public, sometimes enough to jeopardize the public good’s being produced at all. Only something under 10 percent of its listeners make even small contributions to keeping the classical music radio station in Austin,Texas, on the air. Some of them are enticed (as Olson would expect) by private goods offered by the station as premiums to those who pay members’ dues. Are the rest of those who decide to contribute just dupes? They may belong to that minority of the human race ‘too rarely found to be presumed on’ who out of ‘generosity’ act conscientiously and disdain to act otherwise (Leviathan, Chapter XIV, at the end).
In special respect to voting, turn outs in the millions, even if that is only half, maybe only a third, of the people eligible, suggest that Hobbes underestimated the proportion of conscientious people in the multitude. But even if they number enough to keep a democratic (or polyarchical) system going, will they be able to induce the system to make good decisions? However conscientious they may be, can they know enough about the complex issues of modern politics to tell good decisions from bad ones? There is a style of theorizing about decisions that throws great doubt upon the voters’ capacities in this respect. It is just the style that treats social decisions as maximizing or satisficing and in either case having to strike a subtle balance between many different considerations. Doubt that the run of voters can join to do anything of the sort is reinforced by repeated empirical findings that most voters lack even what seems the most elementary political information (like which party is most eager to curtail welfare expenditures).
Yet some issues, even important ones, may lend themselves to forms of presentation that make the choices for voters easy. Among these may figure the grand absurdities of US politics, including the lack of gun control and the soaring rate of imprisonment; here a stark contrast between the status quo and a status quo ante where the problem did not exist, could elicit widespread agreement on the criterion of restoring the status quo ante so far as possible. A similarly stark and useful contrast may be drawn on occasion between the status quo in one’s own country and the status quo in another, more successful in dealing with crimes or drugs or health insurance. In both sorts of cases sometimes the contrast can be brought before the public with the question, ‘Shall we move to the better status quo?’; and if a simple remedy (which costs less and clearly promises to reduce or eliminate the trouble) is available, the public might reasonably be expected to make the appropriate decision in a referendum on the remedy.
In the absence of a referendum, the public might keep up pressure on suitably representative institutions for the decision that a referendum would lead to; but such pressure is difficult to sustain, and cross-purposes among the institutions, aggravated by corruption, may dissipate the impulse to reform. Tom Paine thought that it was desirable to have arrangements that made it possible to appeal from the people drunk to the people sober, and this may be applied as a caution about the unhappiness that may come from resorting to a referendum; but there may be occasion for equal unhappiness with the decisions emerging from a complex of representative institutions. In any case, omitting to present the easy cases in the forms that make them easy, omitting to make them salient, and omitting to keep them salient until they have been resolved is not an argument against the capacities of voters or the use of polyarchy as a system of decision-making.