The Consequences for Consequentialism of Utility Monsters and Barely Happy Multitudes

The fundamental definition of maximizing consequentialism is that the objectively right action is the action, among available alternatives, that maximizes net good consequences. Do barely happy multitudes and utility monsters mean that consequentialism so understood cannot provide the fundamental guiding principle of morality?

1 Maximizing the average or median. 

Happiness is the most common candidate for the good. For simplicity, I am going follow the majority here and take happiness to be what the consequentialist wants to maximize. Having particularized that much, I will feel free to use “utility” and “utilitarianism,” these terms often being employed in place of “the good” and “consequentialism” especially by those who take happiness or pleasure to be the consequential good. For a little additional simplicity, I am going to assume that the subjects in whose happiness we are interested are human beings.  

Next we need to specify the math of the maximization of the happiness of human beings. Is it total happiness over the whole population, or average happiness, or median happiness or one of many other alternatives?

At this step the problems arise. Consider the maximization of average happiness. Suppose there is someone in the population capable of experiencing very great happiness – much more happiness than we ordinary folks. In the literature, such persons are called “utility monsters.”   Average consequentialism, then, would apparently recommend pruning all the other members of the population whose existence does not contribute materially to the happiness of the monster. After all, the others just bring down the average.

Although total and average good are the usual candidates for what consequentialism is to maximize, I am going to take a quick look at a couple of other functions on the good, which can stand proxy for many more. Even without a utility monster, person-pruning would typically be recommended for median as well as average happiness. Eliminating those less happy than the median person, assuming they do not contribute materially to either that person’s happiness or those who would potentially become median, moves the median up to happier and happier people. 

Maximizing the happiness of the least happy person, distantly related to a principle of Rawls, would hardly be a candidate for the expression of consequentialist insight. It is interesting to note, however, that it runs into the same problem. If we want a maximally happy least happy person, homicide is again an obvious expedient. Subject to certain details about interpersonal relationships and retaining enough people to keep the society functioning, eliminating a large chunk of the population may well result in a least happy person who is quite happy. Similar reasoning applies to any percentile on the happiness ladder.

 2. Read this section only if you have consequentialist concerns about consequentialist genocide; otherwise skip to 3. 

If you are reading this section, you may have doubts that consequentialism could really promote the “pruning” of living persons in support of a better average or median or minimum.  Killings, after all, are generally un-utilitarian. They erase all the future happiness that the victim would have had, and, especially if there are as many of them as average happiness consequentialism and its cousins might require, they could fill society in general with terror and loathing.

However, erasure of the future happiness of the happiness-deficient is exactly the point. Once that happiness quantity is off the books, the average or median happiness goes up.

 

As for terror and loathing, they would have to be taken into account in a real world exercise. One way of minimizing these dis-utilities would be propaganda celebrating the superiority of consequentialism in the form being enforced. Those who are less happy, it could be inculcated, ought to be eliminated. They deserve it for bringing down the average or median happiness. 

Repugnant as these rationalizations for murder are, single-minded real world consequentialists are not going to toss them from their toolbox. We have seen horrible, but quite effective, racist and nationalist propaganda not very different from this. 

A second way of dealing with eliminations for the greater average or median good combines some mix of secrecy and fraud.  If a loner is not seen again or if everyone but a few insiders thinks a death is the result of an aneurysm or a freak accident, terror and loathing can be minimized, at least until the numbers of disappearances, aneurysms, and freak accidents become too large.

These practical patches would, I think it is safe to say, only be partially successful, and hence the eliminations would, for consequential reasons, have to be somewhat less sweeping than the worst imaginable case.

In any event, however, this is all an unnecessary detour away from our purely theoretical investigation. A consequentialism of averages can license murder in the service of a utility monster, or a new median, by virtue of the convention, long recognized in philosophy, that philosophical hypotheticals are frictionless. If the existence of A is not logically inconsistent with the existence of B, then I am allowed put them together in a philosophical counterexample. It is a thought experiment, after all, not an engineering drawing. Maybe a utility monster is not a biological possibility, and a society that puts up with person-pruning is not anthropologically, sociologically, or politically possible. The rocket scientists cannot produce spaceships capable of closing towards each other at nearly the speed of light, but that does not stop us from equipping the twins of their famous “paradox” with such vehicles.

Moreover,  our real worry about average happiness consequentialism nags forcefully before we even start thinking about such details as terror and loathing.

Suppose that an average good month gives you and me 30 utiles of happiness. Anthropologists find an isolated mountain village of 50 persons. In a typical month, they enjoy about 100 utiles each. On another side of the mountain lives a hermit, U. M. who is deliriously happy at all times. The utility meter only goes up to 10,000, but the loud click with which it slams into the top pin suggests that U.M.’s monthly happiness far exceeds 10,000.

On yet a third side of the mountain is a community whose members are not very happy. Their economists tell them that if they follow one set of policies, they will evolve, over several generations, towards the happiness level of the first community with roughly their own existing level of population. Other policies, including birth control, but no killings, will eventually lead them to a U.M. like personage, self-perpetuated by cloning. Consistent average utilitarians should argue that the villagers should follow the second set of policies. 

 

If you disagree, then we do not have to worry about the more distressing sorts of average utility pruning to conclude against the average good consequentialists and their cousins. 

3. Maximizing the total.

Planet A has a billion (10⁹) inhabitants, each of whom enjoys roughly 100 utiles per month.  Planet B has a trillion (10¹² ) inhabitants , each of whom enjoys 1 utile per month. It is just barely a life worth living.  B has ten times more total happiness than A.  Has there ever been a consequentialist who would think B truly has the better of it? 

4. Can consequentialism be correct?

So the most obvious candidates for what it is to maximize the good can come into severe conflict with our intuitions, even when we focus our intuitions on the good a la consequentialism.  Can consequentialism yet claim to provide the principle of right action?  Here are some conceivable arguments in support of an affirmative answer.

5. An anti-genocide rider for average good consequentialism?

The horrible I have paraded out to discredit the average good version of consequentialism is a utility monster aided and abetted by homicidal consequentialists. Similar parades take place early in the consequentialist module of many classes on ethical theory. Isn’t it obvious, however, that we should call a halt to this dystopia-mongering by simply adopting a proviso prohibiting the taking of human life? 

 

This won’t work for two reasons: first, the utility monster problem would reappear without murderous prunings; second, and far more important, the proviso would be un-consequentialist. 

As we have already seen birth control population measures favoring utility monsters can lead to as consequentially unintuitive results as death camps. We will shortly see another, more pedestrian case, where the proviso would not cure the utility monster problem of average utilitarianism. 

From the side of theory, however, the decisive objection to the proviso is that it is alien to consequentialism.  “Don’t kill even if it maximizes the good” is obviously a non-consequentialist, i.e. a deontological, principle. At the end of the day, consequentialism may well need to be hemmed in by deontology for all sorts of reasons. Here, however, my concern is whether, in the face of such problems as utility monsters and barely happy multitudes, consequentialism can still claim to provide the fundamental moral principle. For that question, consequentialism can rely on nothing beyond its own resources.

5. An alternative technical definition of greatest good.

Perhaps there is a mathematical formula that will mix in a little total good consequentialism with the average good version so as to preserve the attractive features of each while suppressing their worst case applications. In looking for such a formula, I am not going to take as a starting point total happiness.  It is not happiness itself we are interested in, but happy people. That we can produce unlimited happiness by crowding in ever more barely net happy people shows a fundamental defect in total good consequentialism. By contrast, the utility monster sort of problem seems as if it might be philosophical sleight of hand, directing our attention away from a roughly sound idea. 

So is there formula that retains the virtues of average good consequentialism but vanquishes utility monsters and related devices?  A first thought is that the formula should somehow deflate the utiles count of the monsters. We certainly would like such a deflation when population policy is in question, most imperatively when offing is in the offing.

However, it is far from obvious that we want to deflate a utility monster’s utility in every case. Suppose the question is whether to subsidize the training of opera singers or the training of jockeys. (Of course, there should be a vote about this, but I am assuming that all our voters are average-good utilitarians. So each voter will be concerned with the question whether or not a deflator should be applied for averaging purposes to the happiness of the utility monster among them.)  

I submit that we don’t yet know. It will depend upon the monster. All the initial definition tells us it that a utility monster is capable of being, and generally is, preternaturally happy. That is what is important for population policy. For the present purpose, however, we need to know, at a minimum, about the sensitivity of the monster’s happiness to such external environmental changes as a dearth or plenitude of singers or jockeys.

Suppose that the utility monster is super-sensitive to changes in the opera scene. Although enormously happy in any case, the monster is very disappointed when exposed to second rate bel canto. The decrease in monster utility is so great as to swamp the gains and losses of happiness in the rest of the population relative to the singer versus jockey choice. Here, it may well be, a deflator would be in order. 

In an opposite scenario the celestial happiness of the monster might be affected by matters of the opera house or race track only by roughly the same number of utiles as any ordinary opera or race fan. The monster will remain ecstatic, with only a small percentage change in that ecstasy however the singer-jockey question is resolved. So the monster’s preferences will affect the average, but no more than do those of anyone else. No deflator would be appropriate. 

We could use a deflator only for people whose utility differential between singers and jockey’s was greater than some preset figure – a figure well in access of the median differential. Yet is this fair to the dedicated opera and track aficionados?  If you care too much about something, your happiness or unhappiness should get capped for utilitarian purposes? In addition  to its utter arbitrariness, such caps, again seem, again, un-consequentialist.

Rough as this is, it is enough to make me doubt that any  formula for adjusting the way the averaging of utility is to be done will work satisfactorily in all situations. The intuitively right way to average in a particular case is going to depend too much on details about the utility monster, or otherwise about the spread of preferences, as well as details about the choice alternatives.

6. Satisficing instead of maximizing consequentialism?  

Some theorists (Simon, Slote, Pettit) have argued for a consequentialism that seeks not the alternative that maximizes the good but an alternative that is good enough. The chief virtue of this version of consequentialism is supposed to be that it is not as demanding as the maximizing forms. To do at every decision point the very best thing for the global good really does ask quite a lot of us. 

It might be thought that satisficing consequentialism also solves the problems of utility monsters and barely happy multitudes. A policy that forswore shrinking population down to a small group of utility monsters might yet be good enough, even though it would not maximize average utility. Similarly a policy might well be good enough even if it did not maximize total utility with a trillion barely happy people. 

Still, if we want to satisfice with respect to average happiness, utility monsters will once again pose a threat to the theory. It is not that they force a dramatically pro-monster solution. Satisficing reduces drama with respect to what is required. Still, it has no such affect upon what is permitted. Policies that do not winnow the population down to the utility monsters are going to be good enough, but so are policies that do winnow. 

Of course genocide in the service of utility monsters is not really “good enough,” not being good at all.  Its obvious not-good-enough-ness, we may suspect, comes from deontological principles sounding in equal moral worth. This is not, then, an interpretation of “good enough” available to satisficing average good consequentialism, and so that theory remains stuck with the permissibility of genocide.

As you anticipate, the same thing is true of satisficing total utility consequentialism. It may be that a total quantity of utility less than that of the barely happy individuals of the multitude will be good enough for the satisficing total utility consequentialist, and that some of those alternatives will have many fewer people and a greater average utility. However, if total utility is what we are seeking to satisfice, then the trillion barely happy individuals will also be good enough if that is what maximizes total utility. The decision maker may choose one of the higher average utility options, but that would either be on a whim or because of some moral consideration other than satisficing total utility.

The move from maximizing to satisficing does not solve the problems.

7. But consequentialism is a practical, real world, moral methodology for which utility monsters and trillion person planets are as irrelevant as multiple spatial dimensions and talking giraffes!  

Purported paradoxes from fantastic thought examples will not keep anyone from being guided by consequentialist considerations in everyday life. Can you remember any policy discussion lasting more than two minutes without better or worse consequences being advanced in argument? Rarely, if ever, do those discussions turn on whether average good or total good is in question, and utility monsters have never been caught sight of outside philosophy classrooms. 

I am not wholly unsympathetic to this objection. Still, that we are so comfortable with consequential reasoning is no reason to avoid examining its deep structure and justification. The average utility and total utility problems cast doubt on the claim that, at bottom, morality equals consequentialism. Utility monsters and barely happy multitudes seem to show that morality must look to sources that are not consequentialist and sometimes conflict with and limit consequentialism.  

8. Conclusion

In this light, I do not see how we can give much credence to the claim that consequentialism provides the final answer for the fundamental question of moral theory. Even if we are agnostic as to exactly what is the deep structure of morality, we should bear the deficiencies of consequentialist theory in mind when its practical prescriptions clash with our intuitions. 

For, example, there is Bentham’s view that criminal sentencing should maximize the utility of the society. This seems initially plausible and would unquestionably license much less severe penalties than most of the current sentencing in the United States. Given the concerns about the fundamentals of consequentialism, we might canvass our intuitions about potential punishment reforms. 

Here is one. Society might well be happier if we sentenced all convicted shoplifters, including first offenders, to life without parole. Shoplifting would then be mostly deterred. It would not be entirely deterred. I once had a client who could probably have purchased the high end store from which they lifted a hundred dollar item. I doubt, however, that said client would have been deterred by the prospect of being conveyed immediately from the shop door to the gallows. (The client, I should add, was perfectly sane and competent under existing legal definitions in most US jurisdictions.)  

The near elimination of almost all shoplifting losses would be a very great economic gain, the benefits of which would be widely distributed.  The collection of all those who would have committed shoplifting under the old law would especially benefit. Of course, a few of them would still get caught shoplifting and their subsequent lives would be miserable. Those who were successfully deterred and spared a shoplifting conviction, however, would likely lead much happier lives, and they would vastly outnumber those who would shoplift in the face of the draconian penalty. 

 

Yet, even if it produced clear net positive utility, I have a strong intuition, which I suspect you share, that it would be wrong to sentence first time shoplifters to life terms. 

Sometimes, as Rawls argued, moral intuitions can be modified or overthrown if they conflict with an otherwise sound and well supported theory. Utility monsters and barely happy multitudes give us reason to doubt the soundness of consequentialism at its foundations, and so weaken its force against our intuitions in such cases as life terms for shoplifters.