Utilitarian Suicide Pacts

I found Ran’s comment on my utility of life extension post sufficiently interesting to merit a post of its own. In response to my statement that someone maximizing the average utility of his life would commit suicide immediately if the expected utility of future years were lower than the average up to the present, Ran said:

I'm not so sure it makes sense for maximization of average utility to cover the past; it seems like the average-utilitarian's goal should be to maximize his average future utility.

That sounds sensible, but it also creates a problem of time inconsistency. Suppose you’re 30, and your life up to now has been “great.” You expect the next 20 years (up to age 50) to be “fantastic.” And you expect all years after that to be “pretty good.” Every year after age 50 will lower the average quality of future years. So from your current 30-year-old perspective, if you want to maximize the average quality of future years of life, you should plan to commit suicide at age 50. (Indeed, you should bind yourself to commit suicide at 50, if there’s any way to do so.) However, by the time you reach age 50, the intervening decades will have become the past and thus irrelevant to the goal of maximizing the quality of future years. You should therefore repudiate your plan to commit suicide.

This time inconsistency is, to me, further evidence that maximizing average utility of life is not the right goal. An analogous inconsistency occurs for average utilitarianism of the interpersonal variety, once we consider multiple generations. Right now, present and foreseeable future generations of humanity seem likely to have higher expected utility than the average of prior generations, since the mass of human beings throughout history have lived in conditions of squalor and deprivation. So average utilitarianism would recommend continued existence for humanity. But what if we reached a point in history at which humankind’s condition declined rapidly and seemed to likely to continue that way (perhaps as in Battlestar Galactica)? In that case, an average utilitarianism that considered all lives, past and future, would recommend immediate mass suicide. An average utilitarianism that only looked to the future, on the other hand, could allow continued existence for humanity for a period of time – but if humanity’s condition were expected to drop even further at some point in the more distant future, it would make sense to adopt a time-delayed suicide pact.

Read More...

Summary only...

The Utility of Life Extension

What if you could costlessly extend your life by one year? Suppose that the additional year would be a decent one – not a fantastic year, but definitely worth living. By “costlessly,” I mean that you could achieve this life extension with no sacrifice of utility in previous years, so you really would just be tacking one more decent year onto the end of your life. Would you do it? I sure would.

Now, suppose you’ve tacked this additional year onto your life. You now have the option of improving the quality of that added year by slightly reducing the quality of previous years – perhaps by saving more money or improving your nutrition. (This does not violate the costlessness assumption above. You can have the extra year at no sacrifice, but you also have the option of transferring some quality between years.) Moreover, the resulting increase in utility in the last year will outweigh the reduction in utility in the earlier years. Would you do it?

Your answer presumably depends on your rate of time preference. But set that problem aside by assuming no time discounting. (This assumption can be made realistic by assuming the early years that would suffer are still sufficiently far in the future. E.g., you’re 30 now, the sacrifice would occur when you’re 70, and the gain will occur when you’re 90. Given hyperbolic discounting, the rate of time of discounting between sufficiently distant years is essentially zero.) Now, would you do it? I think I would.

But if we repeated the logic above many times – add a year, redistribute utility, add another year, redistribute utility, etc. – we would eventually conclude that it makes sense to have a really really long life with each year just barely worth living. Regular readers know where I’m going here: This is an intrapersonal version of the mere addition paradox that afflicts utilitarianism (of the total, not average, variety) in the interpersonal context; see here, here, and here for prior discussion. (And now I should promise, once again, that Derek Parfit really is at the top of my to-read list.)

I’m not sure how the paradox can be resolved. But I’m wondering if the answer to the intrapersonal problem – if it can be found – might provide the seeds for resolving the interpersonal one. If you really had the option to repeatedly extend your life one year at a time, and then to redistribute quality from earlier to later years in a net-positive way, what would you do? Would you in fact opt for a very long life with years that are just barely worth living? If so, then maybe we should embrace the same conclusion in the social context (perhaps Parfit’s Repugnant Conclusion is not so repugnant). If, on the other hand, you would stop the process somewhere along the way, opting for a shorter number of higher quality years – crucially, with a lower level of total lifetime utility – then maybe your intrapersonal stopping rule could be applied to help strike a bargain between total and average utilitarianism.

At a minimum, I think the intrapersonal analogy helps to explain why average utilitarianism is not appealing: to maximize the average utility of your life, you would have to forgo any year whose utility was below average, no matter how high its utility in absolute terms. If you had a really fantastic life up to the age of 50, and you expected the rest of your life to be pretty good but not fantastic, maximization of average utility would require you to commit suicide!

Read More...

Summary only...

Utilitarian Implications of Gamete-Seeking

Via Scott Scheule, I find that Virginia may become the first state to ban anonymous donation of sperm and eggs. Interestingly, it turns out the ban’s leading advocate is Katrina Clark, a young woman who was herself the product of anonymous sperm donation:

”I just felt like something had been stripped away from me,” said Clark, a deaf studies and bioethics double-major at Gallaudet University in Washington, D.C.

What, exactly, does Clark think was stripped from her? Her mother, Janie Price, provides one possibility: the knowledge of one’s heritage:

"I talked myself into believing that if I loved her enough, it would be OK," she said. "What I didn't consider is that one's genetic component is very much a part of her identity. Why else would we spend so much money as adults researching our genealogy?"

Clark herself provides another: knowledge of genetic health risks:

Clark said she grew up not thinking she was any different from her friends. That changed when she was 15 and saw a show about a woman who died of a genetic heart disease that she had no idea she was at risk of developing because she had been adopted.

Now both of these are clearly things that Clark lacks – or at least lacked until she successfully tracked down her genetic father. But was she in any meaningful sense deprived of them? I think not, because with the proposed ban in place, she likely would not have existed at all. As the article reports, nations that have banned anonymity in sperm donations have seen large drops in the number of donations, as well as rising sperm prices. A representative of one U.S. sperm bank noted that only 29 of 265 donors chose to allow their identities to be revealed to their progeny. While it’s possible that some of the other 246 donors would have chosen non-anonymity, many others would not. In any case, the set of willing donors under anonymity surely differs from the set of willing donors under a ban. It follows that the set of children conceived via sperm donation under one regime differs greatly, if not entirely, from those who would have been conceived under the other.

What we have here is another practical application of the conflict between average and total utilitarianism, which I’ve discussed before. An anonymity ban most likely results in a smaller number of children, but those children might live somewhat better lives (through having knowledge of their heritage and possible genetic health risks). Setting aside the value of parentage to potential donation recipients, the average utilitarian should approve of the ban, since the utility of children who never exist doesn’t count. Caveat: this conclusion actually requires an additional assumption about the quality of life of children produced via donations relative to the average quality of life of all children. Assuming the lives of children from donated gametes are – in expected value – no better than those of typical children, then anything that reduces the number of such children while raising the quality of life for those that remain must increase the society’s average quality of life.

The total utilitarian, on the other hand, should oppose the anonymity ban unless (a) the quantity-reduction effect is substantially smaller than it appears to be, or (b) the gain from knowledge of one’s genetics is unexpectedly large.

Two factors might put the average and total utilitarians on the same page. First, even if some jurisdictions ban anonymous donations, others likely will not. If Virginia bans anonymity, Virginians can just go to Maryland or Washington, D.C. Sperm exportation is already common, largely because people seek out the most desirable donors, anonymous or not. If the anonymity ban extends to sperm imports as well, expect more sperm-seeking tourism to those jurisdictions without such bans. Local bans simply might not matter much in terms of the utility calculation, because anonymous donations will occur anyway, just in different places. The higher price of traveling for a donation might deter some gamete-seekers at the margin, but not as many as would be deterred by a blanket ban on anonymity that reduces overall donations.

Second, an effective ban on anonymity (say, one enforced by agreement among jurisdictions) could reduce the quality of life for the smaller number of children still conceived. As indicated above, gamete-seekers want the highest quality donors. A ban that reduces total donations has the potential to reduce the quality of the donation pool, especially if higher-quality donors are more likely to value their anonymity. (Entirely possible; imagine future politicians or celebrities who would like the cash now, but not if their progeny could come seeking handouts or publicity later.) The effect of lower-quality donations could outweigh the value of greater genetic self-knowledge. But if so, then the anonymity ban would result in a reduction in both the quantity and quality of lives of the children of donated gametes – a loss on both average and total utilitarian grounds.

(I will admit once again that I really should read Derek Parfit before doing any more posts like this.)

Read More...

Summary only...

How Is Teen Motherhood Bad for "the Child"?

In the course of an excellent post on the advantages and disadvantages of shame and stigma in inducing responsible choices, Julian makes the following common observation about the life prospects of the children of teen mothers:

On the other hand, having a child in one's teens is not only potentially disruptive for the teen, but generally not great for the child either. Some teens have strong support networks and end up being great parents, and their kids turn out as well as they would have if they'd been born later. But let's face it, statistically, they don't, even after you control for other factors.

Here’s the difficult question: Why does it matter how good the child’s life prospects are, so long as they are on net positive? The response would seem to be that the child’s life prospects are worse than if the child had been born under different circumstances, and so the child is in some sense harmed. But that logic doesn’t fly, because the particular person that results from one accidental teen pregnancy only exists under that circumstance. If the teen mother had abstained from sex, that child would simply never have come into being, so it makes no sense to say the child is “harmed” by the teen mother’s having gotten pregnant. Even if we assume that the teen mother would, if she hadn’t gotten pregnant in her teens, have had another child later in life under better circumstances, that child would be a different person, resulting from a different egg and different sperm, with a different genetic make-up and different characteristics and life experiences.

Of course, there are many good reasons to be concerned about the children of teen mothers, most of them relating to the effects such children may have on other people. If they are more likely to be criminals who predate on the rest of us, for instance, that’s obviously a relevant consideration. The teen mother herself is also likely to have a more difficult life; likewise for the father who is obliged to make child-support payments. But holding all those other-affecting factors constant, the question is whether the relative well-being of a child itself should make a difference to our judgments about the desirability of teens becoming parents.

If we take an average utilitarian position (that is, we should maximize the average rather than the total utility of society), then a concern with the relative well-being of the child of a teen mother makes sense. But adopting the average utilitarian position has some other rather unsettling implications – among them, the conclusion that you ought not bring any child into this world whose expected utility is lower than the society’s average utility, even if the child’s expected utility is strongly positive. This has implications for many, many children other than those born to teen mothers! (The total utilitarian position has problems as well, of course.)

My tentative take is that if we assume there is no “replacement child” later when a teen avoids having a child now, then the relative quality of the life of the child of a teen mother is not relevant; my reaction thus tracks that of the total utilitarian (thus possibly exposing me to some version of the Mere Addition Paradox). If, on the other hand, there is a “replacement child,” then both average and total utilitarianism give the same answer: it’s better to wait. However, the reason is not that the child of a teen mother is somehow “harmed” by her having given birth to him. The reason is that, if you imagine yourself having an equal chance of having either hypothetical child’s life, you would rather have that of the child born under better circumstances. But I wonder if that’s really the implicit reasoning of people who push the message that teen pregnancy is bad for the resulting child.

If I'm going to keep pondering issues like these, I really need to bump Derek Parfit to the top of my reading list.

Read More...

Summary only...

Buying Our Way Out of an Ethical Dilemma

Tyler and Alex are having an interesting debate at Marginal Revolution. Tyler poses one of those test-of-utilitarian-reasoning hypotheticals:

Let one person die a terrible and tortured death, but alleviate the headaches of billions of others by one second. ... If the billions are large enough in number, is this worth it? Or does the suffering of the lone individual hold special status?

And Alex replies:

The clearest reason to think that we should trade a terrible and tortured death of one in order to alleviate the headaches of billions is that we do this everyday. Coal miners, for example, risk their lives to heat our homes and to generate the electricity that drives this blog. We know that some of them will die horrible deaths but few of us think that we are morally required to give up electricity.

As several commenters (including me) point out, the analogy doesn’t really work. Unlike the (presumably unwilling) victim in Tyler’s hypothetical, coal miners voluntarily accept the risks and get compensated for doing so. Our willingness to allow miners to take such risks is not necessarily evidence that we’re all really utilitarians.

Alex’s coal miner story is still useful, however, because it highlights the way in which markets and voluntary exchange allow us to avoid unpleasant ethical dilemmas. Tyler’s hypothetical is a truly vexing problem; one person’s un-compensable suffering is required for great benefits. Fortunately, we don’t often encounter situations like that. Instead, we face situations like Alex’s, in which we can (at least ex ante) give the victims something in return for their sacrifice. Requiring their consent helps to assure that the benefits really are large enough to justify the costs; otherwise, those making the sacrifice would demand so much compensation that the rest of us would be unwilling to pay.

Even in those social situations where we can’t get true consent, as in the law of tort that deals with true accidents between strangers, we at least try to pay enough to the victims to make them “whole” (and simultaneously give the potential tortfeasors an incentive to weigh the benefits of their activities versus the costs).

None of this means that we do, or should, reject utilitarian ethics. It just means that we don’t always have to make tough utilitarian trade-offs, because it’s often possible to convert utilitarian gains into Pareto improvements.

Read More...

Summary only...

Voting, Altruism, and Total vs. Average Utilitarianism

Interesting debate over at the Volokh Conspiracy, where Ilya Somin has explained his theory of rational voting and is now defending it against Jim Lindgren. Here’s Somin’s original post and Lindgren’s first response.

Somin’s argument, in a nutshell, is that despite the incredibly small chance that your one vote will actually affect the outcome of an election, it’s nevertheless rational to vote if you’re even just a little bit altruistic. The reason is that you have to multiply the very tiny probability of your vote making a difference by the very large number of people who will be affected (beneficially in your opinion) by your favored candidate taking office.

Lindgren has challenged some of the mathematics behind Somin’s argument. Although Lindgren’s criticisms are cogent, I suspect Somin can probably rehabilitate his model. A somewhat more complex mathematical specification can probably produce qualitatively similar results.

My concern with Somin’s model, which I noted in the comments for one of Somin’s reply posts, is that it turns on the size of the population. The larger is the population of people who will benefit from your preferred candidate winning, the greater is your incentive to vote. Other things equal, that should mean a larger percentage of people should vote as the population increases – but that’s not what we’ve seen historically.

What I realized a short while later is that other things won’t be equal, since a larger population presumably means a smaller chance that your vote will make a difference. The countervailing effects of rising population could cancel out, so maybe my objection doesn’t matter empirically. However, it does matter philosophically. By treating the altruistic benefits of voting as a function of the sheer number of other people who will benefit rather than the percent, Somin has implicitly assumed that people are essentially total utilitarians rather than average utilitarians. Total utilitarianism has problems, the most important of which is the implication that we would rather have a very large population of people with lives just barely worth living than a smaller population of very happy people (as long as the total happiness of the latter was smaller). Average utilitarianism also has problems, such as the implication that we might want to kill people painlessly if doing so left a remaining population with higher average happiness.

But perhaps this philosophical issue doesn’t matter in the voting context, since total and average utilitarianism only differ when the population size is not fixed. Then again, at least some policies do affect the size and composition of the population; abortion springs to mind.

Read More...

Summary only...

IUC in the ICU

The purpose of my previous post on this subject was not to demonstrate that interpersonal utility comparisons (IUCs) are invalid. The purpose was to explain how commonplace observations that seem like IUCs (e.g., “My daughter will enjoy this baby doll more than my son”) might not really be IUCs after all, but instead a species of intrapersonal utility comparison. In other words, I was responding to a common argument in favor of IUCs, not positing an independent argument against them.

But I made the mistake of beginning that post by describing my attitude toward IUCs, and many (in the comments and elsewhere) took that description as a justification. In this post, I’ll take the question head on. Why do I think IUCs aren’t valid?

The most important point is utility, as the term is used by economists, is not synonymous with happiness or pleasure-net-of-pain. If it were, then I’d concede that we might be able to find some kind of scientific and interpersonally comparable measure of it, such as dopamine levels (or some other chemical aspect of the brain). But utility is used more broadly to mean satisfaction of preferences, where preferences are understood to include all sorts of values that might contradict or even transcend the dictates of mere pleasure and pain. And for this kind of utility, which is the kind I think matters, it’s far from clear that a single brain-chemical or brain-activity measure exists. People can value things in spite of, or even because of, the pain or self-denial involved. Will Wilkinson does an excellent job of making this point. Will gives the example of a fighter at the end of a brutal battle, in substantial pain and on the verge of death, who nonetheless thinks there is value in having fought well. I would add the example of a man who abstains from sex for moral reasons, even though he would derive substantial pleasure from having more sex.

In order to believe there is (or could be in principle) a scientific measure that encompasses both the pain from stab wounds and the value-satisfaction of having fought well – both the forgone pleasure of sex and the value-satisfaction of not being promiscuous – you have to believe that all values are “cashed out” in terms of the very same chemical(s) as simple pleasure and pain. Let’s say you think utility can be measured in dopamine levels. Then you have to think the value the fighter places on having fought well (perhaps for a good cause!) creates doses of dopamine that more than compensate for the physical pain of battle. And you have to think the abstainer’s valuation of non-promiscuity creates enough dopamine to make up for the forgone dopamine from sex. To paraphrase Will, you have to think that being good also feels good.

To be clear, I’m not saying value-motivated action lacks a chemical cause. It is brain activity, after all, so it must have some kind of neuro-chemical explanation. But I’d be very surprised to find it was the same chemicals and processes as those that correspond to the pain from stab wounds or pleasure from sex.

(Indeed, I wonder whether even raw pain and pleasure involve the same chemicals and processes. When someone experiences something that involves both pain and pleasure, they really experience both, not some net effect of pleasure minus pain. Doubt it? Ask someone who’s into sadomasochism. They don’t do it despite the pain, but in part because of it. Actually, I feel somewhat the same way about getting a back massage. I enjoy it most if my muscles are sore enough to hurt a little when they’re massaged. Does that make me a freak?)

In any case, different people have different lifeplans (Will’s terminology), which are essentially patterns that they wish to achieve. Such patterns may include the achievement or avoidance of hedonic pain and pleasure, but they consist of other things as well. “Utility” means congruence of one’s behavior and/or situation with the pattern. Now, what is the scientific unit for congruence, or lack thereof? I don’t think such a unit exists. Patterns cannot be compared in the same way that quantities can. Moreover, outcomes can be inconsistent with patterns in different ways, most likely resulting in different sorts of dissatisfaction in the brain. Thus, any attempt to measure the effect of incongruence will likely result in multiple measures. Imagine, for instance, the effect of our sexual abstainer giving in to temptation and having an orgasm; he may simultaneously experience ecstasy (perhaps measurable on one dimension in one kind of unit) and moral guilt (perhaps measurable on another dimension in another kind of unit).

Now, it might seem I’ve made the case against intrapersonal utility comparisons, since my argument seems to hinge on a kind of incommensurability of different experiences within a single brain. If so, that would certainly do the job: if intrapersonal comparisons don’t work, a fortiori interpersonal comparisons don’t either. But while jettisoning intrapersonal comparisons might be the correct conclusion, it isn’t strictly necessary for the argument. From introspection, it seems that somehow we negotiate our conflicting impulses, and we arrive at some notion of what kind of pattern we wish to achieve. That pattern might not correspond to any single measurable aspect of the brain, such as the amount of some chemical. It might, and I strongly suspect does, correspond to a kind of balance (or imbalance?) of different motivational forces, and the balance differs from person to person.

The result is that individuals arrive at various desired patterns, which we call preferences, whose satisfaction can only be understood in terms of their congruence with outcomes. We can say that certain changes will assist the achievement of one individual’s pattern while undermining the achievement of another’s, but we have no unit in terms of which we could say the increase in congruence for the former is “bigger” than the reduction in congruence for the latter.

Read More...

Summary only...

An IUD for IUCs

The boys at Catallarchy are having a little dust-up on the subject of interpersonal utility comparisons (IUCs). Are such comparisons possible? Brandon Berg, Jonathan Wilde, and Patri Friedman say yes; Brian Doss defends the conventional economic wisdom and says no.

Without offering too much in the way of justification, I’ll just say that philosophically I tend to agree with Brian. Making interpersonal utility comparisons is a kind of category error, akin to trying to add degrees and pounds. I don’t have a logical proof of this point; it just seems intuitively right to me. You can’t add my utility to yours, because my utility and your utility are just different sorts of thing. And appealing to objective measures, like dopamine levels or endorphins in the brain, seems to me to miss the point. We’ve got different brains, and an endorphin in my brain just ain’t the same as an endorphin in yours. I suspect there’s some connection here to Rawls’s “separateness of persons” objection to utilitarianism.

But on the other hand, as defenders of IUCs are fond of pointing out, we make such comparisons all the time. If you win a baby doll at the fair and give it to your daughter instead of your son, in the expectation that your daughter will enjoy it more, you’re making an IUC. Right? And if you’ll donate money to a homeless person rather than George Soros, on grounds that the homeless person will obtain greater benefit from it, you’re making an IUC. Right?

Since we make these comparisons all the time, they’re clearly not impossible. In that sense, the debate is suffering from ill-chosen language. The question is not whether we can make such comparisons, but whether they possess philosophical or scientific validity. People make IUCs, but they make lots of other mistakes, too. They say things like, “Mary is fatter than John is tall,” a statement that makes less and less sense the more your scrutinize it.

If we’re going to make sense of real-world IUCs, I think we have to reconstruct them more carefully. I think most of them can be understood as imagined “standing in the shoes of” comparisons. If you have a baby doll to give to one of your children, you imagine yourself as a girl with a baby doll, and then you imagine yourself as a boy with a baby doll, and you imagine your satisfaction with each situation. If you have a philosopher gene, you might even imagine that you have a 50% chance of occupying each child’s position. And then you think, “Which of these imaginings was the more satisfying?” Or if you have an economist gene, you might imagine that you also have a toy truck to give away, and realize that the daughter would prefer the doll to the truck and the boy would prefer the truck to the doll.

Similarly, if you have a dollar to give away, you imagine yourself as a homeless person gifted with one more dollar, and then as Soros gifted with one more dollar. In which of these imagined situations did you experience the greater “lift”? You might – again with the economist gene – imagine each of these persons as a possible future you, and then ask in which future you’d prefer to have that extra dollar. That is the sort of imagining that moves people to buy insurance.

Am I just slipping IUCs in the back door by this “imaginings” method? Maybe, but I think not. I am not claiming that the comparisons made by this method possess an objective reality. The comparisons are filtered through an internal, subjective assessment process, which is unavoidably connected to my particular background and character (though I might try to set aside my personal circumstances as much as possible). And that means people who perform such imaginings will naturally reach different conclusions in some cases. This is not because one evaluator is right and another is wrong, but because there is not an objective answer to the question.

Nevertheless, there will exist a range of cases where, despite the non-existence of an objective answer, most everyone will arrive at the same answer. That is, they reach intersubjective agreement. This is what makes the examples given by the IUC-defenders so intuitive; the examples are selected precisely because they fall into the regions where intersubjective agreement is easily reached. On the other hand, there exist other comparisons where different people will honestly reach different conclusions. The IUC-defenders wish to treat these as cases where the utility numbers are just too close to call, and thus people make mistakes. I would counter that in such cases, the differences are not merely honest but also accurate, being based on differing evaluative processes. (And yes, I realize that intersubjectivity is a post-modernist notion, so maybe Scott Scheule’s concerns are well-placed. But even if post-modernism is usually full of crap, even a stopped clock is correct twice a day.)

All this might seem odd coming from a confessed utilitarian like myself. I assure you that I have a fascinating and compelling account of how my positions are consistent, but it will have to wait.

Read More...

Summary only...

Deontology Meets the Mere Addition Paradox

One objection to utilitarianism is its difficulty in dealing with questions of population change. For example, is it desirable to add more people to a population if the new additions have lives that are (a) worth living but (b) not quite as good as the lives of those already living? The larger population would have higher total utility but lower average utility. (Set aside concerns about negative externalities from population growth. Holding the happiness of the already-living sharpens the philosophical question.)

Derek Parfit’s “Mere Addition Paradox” (MAP) shows how both total and average utilitarianism can lead to bizarre and (seemingly) unpalatable conclusions. This article by Kai Chan has the best explanation of the MAP I’ve found, all in the first three pages. Here’s my quickie explanation: Suppose you have an existing population with relatively high utility. Someone proposes an addition like that described above (new people with lives that are worth living but below the previous average). Should we favor the proposal? The obvious answer seems to be “yes,” or at least “there’s no reason not to.” Adopting average rather than total utilitarianism could allow you to resist this conclusion – but that’s precisely what’s bizarre about average utilitarianism. So the new group is added on. Once those people are in existence, someone proposes a slight redistribution of resources that will make the new people better off and the old people worse off, with an overall net positive result (the utility gains exceed the utility losses). This proposal, too, is difficult to resist. But now consider the addition of yet another group of new people, with lives worth living but less satisfying than the lives of those who already exist. Repeat until you have a very large population of people with lives barely worth living. This is what Parfit calls the “repugnant conclusion.”

Now, what I’m wondering is whether an analytically similar problem might afflict some deontologists (e.g., natural rights theorists) who think they’ve managed to dodge the problems of utilitarianism. Say you have a population of people whose rights are respected perfectly. And suppose the population could be increased through the addition of a new group of people whose rights are mostly respected, albeit with the occasional violation. On what grounds could the deontologist resist the addition of this group? These people would, presumably, prefer to live despite the occasional rights violation. And in any case, if the deontologist believes in a right of procreation (or opposes any intervention to stop procreation), then he would have to allow the entrance of the new group.

Next step: Once we have the larger population, suppose we have the opportunity, via some change in policy, to redistribute the rights violations so that the new group has fewer rights violation, the older group has more, and the total number and severity of rights violations has fallen. Now, here the deontologist might object: right violations are not interpersonally comparable. The lesser violation of the rights of some does not compensate for the greater violations of the rights of others. But I don’t think this position will withstand scrutiny. The deontologist would have to resist changes in enforcement policy that would, say, prevent 1000 forcible rapes while allowing one more act of shoplifting. Or, at a minimum, he would have to shrug his shoulders. Similarly, the deontologists would have to resist (or at least be indifferent to) any change in the form of government that would result in lesser protection of any person’s rights. In practice, I don’t believe deontologists actually think that way. (Notice, for example, the many deontologist anarcho-capitalists who favor abolishing the state in favor of private protection agencies, even though such a change would assuredly cause some people’s rights to be better protected while at least a few others’ rights suffered.) So I think that reasonable deontologists would have to accept the proposed redistribution in rights protection.

Now repeat the process. The endpoint will be a society with the maximum number of people that could be sustained at some minimal level of rights protection.

The problem, for both utilitarians and deontologists, results from an intransitivity driven by the distinction between actual and potential persons. For both groups, it seems strange to regard non-existent people as having morally relevant interests (for the utilitarians) or inherent moral rights (for the deontologists). Yet once such people do exist, their interests and/or rights suddenly matter. For utilitarians, that means weighing their interests against those of other real people. For deontologists, it means respecting and protecting their rights. That might seem to require no sacrifice of the rights of others – until we recognize that rights protection is costly, and therefore we can’t avoid trade-offs in rights protection. Within a given regime, the trade-off is felt in the allocation of law enforcement resources. Across regimes, the trade-off is felt in the choice of which regime to favor. To make the problem yet more vexing, the people who exist under each regime will differ (because, at a minimum, sensitive dependence on initial conditions means that different sperm will meet different ova).

For related thoughts about deontology versus consequentialism, see my post on lexicographic orderings.

Read More...

Summary only...

Utilitarianism vs. Intuitionism Dust-Up

I tried for a while to summarize the debate so far, but the summary threatened to fill pages on its own before I got to my own thoughts. So go read what Tyler said, then what Alex said, then what Will said, then what Carina said.

For a start, take Alex’s hypothetical, posed to him (and Robin Hanson) by a philosopher:

Suppose that you had a million children and you could give each of them a better life but only if one of them had a very, very terrible life. Would you do it?

Alex and Robin, both economists, said “yes” without hesitation. But to the philosopher who posed the question, the intuitively correct answer was clearly “no.” Who’s right? Alex suggests that we should simply overcome our gut intuitions and think logically here, and the logical answer is to favor the greatest good. Will and Carina both correctly reply that Alex, too, is relying on intuition – the utilitarian intuition that we can (in some rough-and-ready way) compare satisfaction across persons. So we can never fully escape the appeal to intuition.

But does that observation dispose of the matter? Can we just pick out our favorite intuition and run with it? Alex is still right that our intuitions are inconsistent. One intuition tells us that individuals have personal claims that should never be violated. Fine. But another intuition tells us it’s absurd to impose monstrous losses for miniscule gains. The lifeboat situation is a classic example of where this alternative intuition kicks in. See my post on lexicographic preferences for more on this point.

Moreover, it turns out the intuition that answers “no” to Alex’s hypothetical is highly dependent on the frame of reference. Consider a variant on the story:

Suppose that you could give one child a substantially better life, but only if a million other children had a worse life. Would you do it?

I’m guessing most people’s intuition says “no” on this question as well – thereby dictating the opposite result. The hypotheticals differ only because the language implies actively changing a status quo that – for some reason – we assume creates a rightful claim by one or more children. But in either hypothetical, you’re really being asked to choose between two situations:

A. One million children live relatively good lives, and one child lives a terrible life.
B. One million-and-one children live so-so lives.

Now what does your intuition tell you? Without a frame implying that A or B is the status quo, the answer doesn’t seem so obvious. Maybe I’m projecting, but I suspect most people would appeal to some form of utilitarian reasoning here (possibly asking questions about just how bad “terrible” is, what “so-so” is like, etc.).

So apparently the frame matters to us. But where does this frame come from? Why do we privilege the presumptive status quo? Take Alex’s original hypothetical, but add in some extra context:

The million children who stand to gain are the starving and oppressed people of an African nation. The one child who stands to lose is the son of the local tyrant.

Feel free to modify the context to explain why the tyrant’s son’s life will suck after his dad is dethroned. With the context filled in, I suspect most who said “no” before will say “yes” now. Why? Because we no longer think the status quo creates any rightful claims to its continuation.

Now, none of my reasoning thus far “proves” utilitarianism is the right moral theory. But the discovery of anti-utilitarian intuitions doesn’t disprove it, either. The question to ask now is, why do we care about the frame of reference? Why does it matter whether the society in question is a tyranny or something else? Once our intuitive commitment to particular frames is challenged, the hypothetical begins to look a lot like the A vs. B choice I posed above.

And just to be clear, we must challenge the frames, because that is the central question of ethics and political philosophy. What rights do people have, if any? What governmental and institutional structures are best? How should initial endowments be determined? Appealing to gut intuition is insufficient here, because the whole issue is which of our conflicting intuitions ought to be trusted, and under what circumstances. In this process, it’s useful to realize that not all intuitions are fundamental or axiomatic. Some have their origin in the interaction between more basic intuitions and facts about the world, such as the degree of scarcity or the typical costs and benefits of alternate courses of action. This is where utilitarian reasoning proves most useful. Now, that doesn’t mean that only utilitarian intuitions matter, but assuredly they ought to play a role – I’d recommend a relatively large one – in reaching our reflective equilibrium.

Read More...

Summary only...

A Matter of Coase

Will, an avowed non-utilitarian, sure does make a lot of compelling utilitarian arguments. In his latest flaying of Richard Layard (a hedonic utilitarian who assuredly deserves the flaying), Will deploys a fascinating corollary to the Coase Theorem that has implications far beyond his debate with Layard.

Layard has devised a new kind of negative externality argument. He observes that inequality of status makes many people – specifically, envious people – unhappy. As a result, any productive effort to improve one’s position creates a negative externality by reducing the relative status, and thus happiness, of others. Using standard Pigovian logic, Layard concludes that we ought to discourage (e.g., tax) productive effort in order to correct the externality.

Of course, Pigovian externality theory is almost 50 years out of date. Will brings to bear Ronald Coase’s “least-cost avoider principle,” which says that when parties A and B can both take actions to avoid a harm, efficiency dictates that whoever who can reduce the harm at the lowest cost should do so. For example, if an airport imposes noise pollution on nearby residents, the airport should relocate or restrict flights if and only if that’s cheaper than the residents moving away or installing sound-insulation. Will combines this insight with the observation that people can overcome their envy via sublimation or learning of new preferences. As Will puts it, “If you changing your preference is cheaper than taxing me, then you ought to change your preference.”

This reasoning opens a mighty interesting can of worms (and I mean that in a good way). In the traditional Coasean approach, preferences of agents are taken as given, and the only question is whose actions can reduce harm at lowest cost (keeping in mind that all harm is relative to the parties’ preferences). But if actions can change preferences, it follows that changing preferences may in some instances be the least-cost means of reducing harm. This is true even for less exotic forms of externality than Layard’s status-externality. Take the airport noise externality. Maybe the least costly solution is for people to start enjoying the sight and sound of airplanes. Or consider the visual externality created by a factory that obscures the view of a nearby mountain. Perhaps the least costly solution is for people to stop caring so much about natural scenery and learn, Rand-like, to appreciate the man-made beauty of mortar, block, and glass decorated with thousands of electric lights.

As Will’s post indicates, this approach also provides a ready-made response to anti-utilitarian arguments of the “what if lots of people have really monstrous preferences” variety. Consider a Southern town where lots of bigoted white people would be supremely satisfied if a black person were lynched. The Coasean answer: maybe they should just stop being bigoted! (This answer seems especially compelling if we take a multi-generational point of view.) Does an externality arise from people envying others? The Coasean answer: maybe they should stop being so envious!

Clever statists might seize on Will’s Coasean Corollary to justify some scary policy ideas. For example, a Marxist might claim the CC indicates people should stop being greedy and transform themselves – or be transformed by state power – into altruists. But remember, the CC doesn’t mean that preference change is always the preferred alternative, nor does it mean that only one party’s preferences can change. Preferences are malleable, but not infinitely so. Some preferences can change easily within a single person’s lifetime, others can change over generations, and yet others may be so built-in to human nature that changing them is essentially impossible. The Coasean argument is sensitive to the costliness of the actions involved.

Read More...

Summary only...

Utilitarian Retooling

Will argues that utilitarian ethical paradigms are not viable because they are inconsistent with the way humans actually think: “We are natural-born Aristotelians (or maybe Humean sentimentalists) unlikely to be moved by comprehensive schemes of utility maximization.” A few reactions, not necessarily in a logical order, as this post began as a too-long comment in Will’s comments box. Note: none of the following should be taken as an endorsement of the pure-Benthamite utilitarianism espoused by Layard, to whose position Will was responding.

1. Evolution, whether cultural or biological, operates on the phenotype, not the genotype. Suppose for a moment that selective pressures point in a utilitarian direction because they correspond to utilitarian factors (the need to produce food, clothing, and shelter, for example). I know this is a stretch, but go with me here; even if selective pressures don’t always point in that direction, they sometimes do. Even so, we should not expect evolution to produce a tendency to think in utilitarian terms – it should only produce either motivational states or specific actions that correlate with utilitarian results.

2. People don't usually think like utilitarians, and it's often difficult for them to do so. That, as Will says, puts a constraint on utilitarian theory. Fine. I don't think it follows, as Will says, that the “underlying principles” embodied in the biological constraints “do all the work.” In fact, that’s such a non sequitur that I’m not sure how to respond. It’s similar to saying that when a firm tries to maximize profits, the technology of production does all the work, without the need to ask about consumer demand or the supply of inputs. Moreover, the underlying principles themselves may (I would argue they often do, at least in the case of culturally as opposed to biologically embedded motivations) reflect utilitarian concerns. Yes, people think about the principles as absolutes, but so what? My argument above in (1) says we shouldn't expect cultural evolution to embed utilitarian reasoning so much as utilitarian results.

3. Example: Why do many societies have a moral prohibition on theft? Because in the absence of such rules, great effort will be expended on rent-seeking activities (stealing from others, defending one's own) instead of productive activities. Meanwhile, the marginal return of productive activities is reduced by theft, further reducing the incentive to be productive. So societies that somehow imbue their members with a moral aversion to theft will probably perform better. That's essentially a utilitarian argument for the prohibition of theft, but it doesn't require that individuals carry with them the utilitarian reasoning that supports the prohibition; a simple moral aversion to theft will do. (And before anyone jumps on my case for relying on group selection here, I’ll admit this is an incomplete argument. But I do think group selection can work under limited circumstances; see here and here. And does anyone have a better explanation for why norms against theft are so widespread?)

4. In his comments section, Will says that Joshua Greene’s experiments – which show that people are much more inclined to help people with whom they have personal contact than abstract strangers – show that people are motivated by anti-utilitarian principles. Not necessarily, I say. Any sensible ethical system, including one based on utilitarian concerns, needs to draw on local knowledge. Our knowledge of people in distant places is limited and often untrustworthy, and our knowledge of how best to help them is even more unreliable. So there's a sensible utilitarian justification for having people “take care of their own.”

5. Nonetheless, given our small-group evolutionary heritage, helping very distant strangers probably contributes little to reproductive fitness, so this is a case where evolution is highly unlikely to have equipped us with fully utilitarian norms. To the extent that “ought implies can,” we have to take these built-in anti-utilitarian motivations as given. But this is not an argument against utilitarianism. It's an argument for finding institutions that draw on our built-in motivations in a way that, nonetheless, contributes to the greater good. This is, I think, one reason capitalism is so fantastic: it harnesses self-interest (or small-group-interest) in the service of others, since opportunities to help others so often manifest as profit opportunities.

Read More...

Summary only...

Respecting the Dead for the Sake of the Living

Steve Landsburg’s economist’s take on the Schiavo case has been getting some attention. His basic argument is this: Terri Schiavo is effectively dead, and that means her preferences no longer matter because they no longer exist. What do matter, at least potentially, are the preferences of living people: her husband and parents. The rest of the article tries to sort out which preferences (of living people) should be counted and how competing preferences should be weighed against each other.

Tyler Cowen and Trent McBride both argue against Landsburg on essentially anti-utilitarian grounds, saying this is just the kind of case where weighing of preferences is not the right approach. Says Cowen:

Nor do I think that family decisions -- whatever your view in the Schiavo case -- should be decided by a real or hypothetical societal auction. If there is any "protected sphere" for human decision-making, surely it is here. The problem is that we don't agree on how to define the guardian of the sphere -- is it "Terry" or "husband as guardian of a no-longer-living Terry"?

Tyler’s point is sensible, but I think a counterargument to Landsburg can be made on utilitarian grounds as well. The reason is that a person’s happiness does not depend solely on current enjoyments (in philosopherese, instantaneous hedonic utility). It also depends on memories and anticipations. Anticipations are particularly important here. My current happiness is affected by what I think will happen in my future. Imagining a long and productive life enhances my current sense of well-being; imagining myself a quadriplegic diminishes it. Imagining my final years being spent hooked to machines, even without consciousness or pain, also diminishes my present happiness. Legal rules that eliminate or reduce that possibility of an undesirable death can therefore enhance your happiness during your life.

Lest we underestimate the importance of anticipatory utility, note that having our wishes about the fate of our bodies respected is one of the main reasons people get married. It’s among the leading arguments made by gay couples who want legal recognition of their unions: they want to put someone they love and trust in charge of their remains (whether those remains are dead, living but unconscious, or somewhere in between as in Schiavo’s case). The very same argument provides the justification for respecting a dead person’s expressed wish to be buried on hallowed ground, or to be cremated and have his ashes strewn over the Pacific: the likelihood that the wish will be respected gives some satisfaction or tranquility to the deceased before they die.

Once you’re dead (or approximately dead like Schiavo), of course, you don’t care. But fulfilling the wishes of the dead (or approximately dead) creates a precedent that affects the expectations – and thus the happiness – of the living.

Read More...

Summary only...

Lexicographic Lapses

Here is one of my favorite quotes, from a little-known 1979 essay¹ by Harold Demsetz:

Lexicographic orderings crumble in the face of scarcity.

Rather esoteric, I admit, and it took me a long time to appreciate it. But the more I think about it, the more profound I think it is. Let me try to explain.

“Lexicographic” means “ordered in the manner of a dictionary.” If you want to order two words, the first letters are the only thing that matters, unless the first letters are the same. If you’re comparing ‘azzzzz’ and ‘baaaaa,” those following z’s and a’s have no impact whatsoever.

Other things are sometimes ordered lexicographically. For instance, some authorities rank nations’ performances in the Olympics based on the number of gold medals received, looking only to silver medals to break ties in gold (and looking to bronze only to break ties in both gold and silver). The odd result is that a nation with 10 gold and 1 silver beats out a nation with 9 gold and 20 silver.

The latter example demonstrates the problem with lexicographic orderings: they’re great for organizing a list, but they not so great for ranking preferences or values. Why? Because they imply a willingness to make any sacrifice in category B, however large, in return for any gain in category A, however small. Demsetz’s point is that lexicographic value or preference orderings may seem sensible in the face of small trade-offs, but they become highly implausible in the face of large ones. (To broaden its application, I would reword Demsetz’s quote to read, “Lexicographic orderings crumble in the face of steep trade-offs.”)

The lesson has implications for people of many different ideologies – for instance, environmentalists who insist on preserving nature “whatever the economic cost.” But since I know many libertarians read this blog, I’ll give three applications that pose challenges to libertarians – in particular, those libertarians who base their beliefs on a dogged attachment to deontology or inviolable natural rights.

The first application is from philosophy. There is a famous thought experiment (I’m sure some commenter will tell me the original source) in which you and ten other people have been captured by a military junto in the South American jungle. The general gives you a choice: you can either execute one of the other prisoners yourself, or you can refuse, in which case he will execute all ten. This thought experiment is supposed to provide a kind of “utilitarian test case” (are you really willing to act on utilitarian precepts?). I’ve never found the test very demanding, because the answer is bloody obvious: taking the hypothetical’s conditions at face value, of course you should execute a prisoner. But say you disagree with me on this; how far are you willing to take your moral stance? If your refusal would lead to the execution of a hundred prisoners, would that change your mind? How about a thousand? A million? The insistence that you should never kill an innocent, regardless of the consequences, amounts to a lexicographic preference that places first priority on the number of people you kill, and only second priority on the number of people who get killed by others.

The second application comes from law. Suppose you fall out the window of your 20th story condo. On the way down, you catch hold of horizontal flagpole on the 19th floor. You could swing over and break through a window of someone else’s condo, or you could let go and plummet to your death. Which do you choose? And more importantly, how should the law treat your action? I think the answers are clear: you should break into the condo, and the law should require you to pay compensation for the broken window (and any other damage that results). But I’ve actually heard deontological libertarians argue that you should respect the 19th floor condo owner’s property rights come what may. Fortunately, the actual law is on my side: cases like these are called “necessity cases,” and they are indeed treated with a liability rule (which requires compensation for damage) instead of a property rule (which punishes with the intent of preventing violations in the first place). Those who insist on the property rule in this case are clinging to a lexicographic preference that places first priority on respecting material property rights, and second priority on… well, everything else.

The third application comes from political economy. Should libertarians favor a minimal state or anarchy? Some deontological anarchists (I exempt consequentialist anarchists like David Friedman from my criticism) oppose the minimal state because any taxation is theft, any exclusion of competing protection services is coercion, etc. This position commits them to supporting anarchy regardless of what consequences may occur, including consequences that impinge severely on individual rights the anarchists themselves hold dear. For instance, what if anarchy resulted not in the nicely balanced and competing protection agencies that anarchist libertarians imagine, but instead in a Hobbesian state of war? (Some anarchists dispute that prediction. That’s a perfectly good consequentialist argument, but not the issue here. My argument is directed at the deontological position.) Even if it is shown that anarchy leads to a much larger overall amount and severity of rights violations than a government, deontological anarchists must oppose the creation of a government because it legitimizes some of those violations. In essence, they adopt a lexicographic ordering that places first priority on minimizing legitimized rights violations, while placing second priority on all other rights violations.

I don’t wish to imply that treating rights as side constraints on human action is a silly idea. Most of the time, rights are an eminently practical and just way to characterize our obligations to each other. Performing a quick cost-benefit analysis in every moral, legal, or political situation would be a prescription for disaster (in cost-benefit terms!). Nonetheless, rights are not ideal for dealing with lifeboat situations. Our choices in tough situations should not be invariant to the trade-offs involved, especially when the trade-offs are steep.

1 “Ethics and Efficiency in Property Rights Systems,” in Mario Rizzo, ed., Time, Uncertainty, and Disequilibrium, Lexington Books (1979), 97-116.

Read More...

Summary only...