Utilitarianism, Effective Altruism, Population and Pigs

One ambiguity in utilitarianism, the philosophical claim that one should act in ways that maximize utility, is the question of whether what should be maximized is the average utility of the population of the world or its total utility. The average utility criterion follows, as Harsanyi pointed out,¹ from the assumption that the correct rule is the one that would be chosen in an initial situation where the chooser did not know which future life he was going to live, hence faced a lottery with an equal chance of being anyone — the same approach later followed by Rawls to a different conclusion.²

The problem with the average utility criterion was pointed out by James Meade seventy years ago:

Suppose two communities A and B to exist. Suppose that neither has any appreciable economic dependence on the other so that the disappearance of A would not appreciably affect the standard of living in B nor the disappearance of B the standard of living in A. Suppose, further, that the standard of living in B is somewhat lower than in A, though both communities are prosperous and enjoy high standards. The strict application of the objective of maximizing welfare per head would lead to the conclusion that the world would be a better place if community B ceased to exist, since output per head for all citizens of A and B would certainly be increased if that section of the community with the somewhat lower standard were to cease to exist (Meade, Trade and Welfare.).

The alternative of evaluating alternatives by total utility raises the problem termed by Parfit the repugnant conclusion, that a future with a very large population of very poor people may be preferable to one with a much smaller population of much better off people, a future where the world looks like India better than one where it looks like America.

The difference between the two criteria becomes important in the context of arguments about population policy, usually conducted on the implicit assumption that what is to be maximized is utility per-capita. I wrote a chapter on the question forty some years ago in which I proposed a criterion for comparing alternative futures with different numbers of people in them designed to avoid the problems with both total and average utilitarianism at the cost of making some pairs of alternative futures incomparable. The purpose of this post is not to revive that question but to point out an analogous problem in a different context: what altruists should eat.

Effective Altruism: The Problem of a Measure of Good

Effective altruism originated as a project to figure out how best to help people, how to get the most altruistic bang for the buck. Doing that requires a definition of bang, what you are trying to maximize. One candidate is lives:

GiveWell is an organization that does in-depth research to find the most evidence-backed and cost-effective health and development projects. It discovered that while many aid interventions don’t work, some, like providing insecticide-treated bednets, can save a child’s life for about $5,500 on average.

Collectively these efforts are estimated to have saved 159,000 lives. (What is effective altruism?)

Another is dollars:

Wave is a technology company founded by members of the effective altruism community, which allows people to transfer money to several African countries faster and several times more cheaply than existing services. It’s especially helpful for migrants sending money home to their families, and has been used by over 800,000 people in countries like Kenya, Uganda and Senegal. In Senegal alone, Wave has saved its users hundreds of millions of dollars in transfer fees – around 1% of the country’s GDP. (What is effective altruism?)

Saving lives and saving money are both good, but not all lives, or all dollars, are equally good. A long and happy life is a greater good than one that is short and miserable. A dollar that will be spent feeding a hungry child is a greater good than a dollar that will be gambled away at Las Vegas. If the objective is doing as much good as possible per dollar spent one would like a definition of “good” sufficiently general to distinguish among lives and among dollars, compare benefits in lives to benefits in dollars, compare both to benefits in other things an altruist might consider desirable. One would like a definition of amount of good done.

One candidate is utility.

Which brings us back to the total/average question.

Saving a life increases total utility, assuming that the life is worth living and does not have negative impacts on other lives. It lowers average utility if the life saved is less happy than average. If the same expenditure will save ten lives in Africa or one in Europe, and if Europeans have happier, Africans less happy, lives than average, saving African lives lowers average utility, saving a European life raises it. But unless European lives are at least ten times happier than African, saving the African lives raises total utility.

A different form of the same question is to ask whether an effective altruist should not only be against death but in favor of life, whether one more child being born counts in the same way one fewer dying does. Discussions of the effect of declining birth rates are usually put in terms of the effect on existing people, the average utility approach. Perhaps they should include the effect on people who don’t exist — but could.³

Digression on Quantitative Utility

What does “ten times happier” mean? If we use the Von Neumann/Morgenstern definition of utility, the same definition used by Harsanyi, and define zero utility as the suicide point, we can answer that question. Life A has ten times the utility of life B if an individual would be indifferent between a certainty of B and a gamble that gives him a probability of .1 of life A, .9 of death. Defined in that way, it seems unlikely that the average inhabitant of a rich society is ten times as happy as the average inhabitant of a poor one. Putting utilitarianism in formal mathematical terms feels odd but helps clarify the argument.

Having put the objective in terms of utility, a natural extension of the project is …

Applying Effective Altruism to Animals

People in effective altruism try to extend their circle of concern – not only to those living in distant countries or future generations, but also to non-human animals.

Nearly 10 billion animals live and die in factory farms in the US every year – often unable to physically turn around their entire lives, or castrated without anaesthetic.

…

Another strategy is to create alternative proteins, which if made cheaper and tastier than factory farmed meat, could make demand disappear, ending factory farming. (What is effective altruism?)

This again raises the average/total question. If, as the tone of the quote suggests, life as a factory farmed animal has negative utility, both the total and the average criteria imply that factory farming is bad. If, on the other hand, life as a factory farmed animal is better than no life at all, has positive even if low utility, the conclusion is less clear. Forbidding factory farming will raise the price of meat and reduce its consumption, so there is a tradeoff between making the lives of animals better and reducing the number of animal lives.

That point is made clearer if the issue is not factory farming but vegetarianism or alternative proteins. Should an effective altruist be a vegetarian in order to reduce the number of cows, pigs and chickens who are killed each year or oppose it in order to increase the number who live, even if not for very long? Is a not-very-good life better than no life at all? Should he, in a world without factory farming, prefer beef to pork on the theory that cows are larger and less intelligent than pigs, hence consuming beef kills fewer animals and animals less able to feel disutility, or should he prefer pork on the theory that the more pork is eaten the more pigs get to live?

Better not to have been born. But who could be so lucky? Not one in a million. (Leo Rosten. The Joys of Yiddish, from memory so possibly not verbatim)

---

My web page, with the full text of multiple books and articles and much else

Past posts, sorted by topic

A search bar for past posts and much of my other writing

1. John C. Harsanyi, "Cardinal Utility in Welfare Economics and in the Theory of Risk-taking."

2. Rawls, unwilling to accept Harsanyi’s utilitarian conclusion, instead assumed that an individual in that situation would choose as if certain of the worst outcome, hence prefer the future that maximized the welfare of the worst off person in the society. For my critique of the argument see Contra Rawls in this post.

3. Some effective altruists argue along these lines for policies designed not to improve present lives but to increase the probability of a future with a very large number of humans in it.