Stephen Darwall on effective altruism

Last week, Stephen Darwall, a moral philosopher at Yale University, published a critique of effective altruism in the journal Liberties (it’s behind a paywall, but if you sign up for an account with the journal, can read two articles for free each month). I’m always curious about such critiques, as I have a complicated relationship to effective altruism myself. I read about it in the mid-2000s in a newspaper article that highlighted the difference between what is achieved by giving £1000 to a national British charity like the National Trust, giving £1000 to an international health charity like the Against Malaria Foundation, and spending £1000 on something for yourself. Without much more thought, I was convinced of the broad commitments of what then was barely even a movement, more a handful of charity evaluator websites, and pledged, at least to myself, to give a proportion of my salary to their recommended effective charities. For about a decade, I didn’t think much more about it beyond checking the websites each year to see where I might send money for the next twelve months. I noticed when they started to draw attention to work that helps non-human animals, and particularly those suffering under factory farming, and I started sending some money that way. And I noticed again when they started to advocate for work that seeks to reduce the risk of human extinction, and I didn’t start to send any money that way—not because I am sure they’re wrong about that, but because I’m not sufficiently sure they’re right (see here).

Since my first encounter with it, effective altruism has decidedly become a movement, and it has grown enormously, both in terms of the number of people involved, either by pledging to donate or by researching the effectiveness of interventions, and in terms of the amount of money committed to its approach, largely as a result of persuading some billionaires of its merits. And, as is appropriate, with greater participation, funding, and power has come greater scrutiny—indeed, last summer I reviewed the first book-length critique of effective altruism, edited by Carol J. Adams, Alice Crary, and Lori Gruen. And, partly because of this greater scrutiny, and partly because of very public crimes and misdemeanours committed by very visible adherents, such as Sam Bankman-Fried, the movement has been rocked by a number of scandals recently.

Pink and Blue, Anni Roenkae

Darwall’s critique, however, doesn’t focus on these scandals. His concern is rather with the ethical principles that underpin the movement. At the beginning of his essay, Darwall talks as if all effective altruists are utilitarians, and proposes to object to effective altruism on the grounds that, as Mill himself saw, utilitarianism often seems to conflict with considerations of justice, which he thinks determine morally right action: “like utilitarianism more generally, effective altruism fails to appreciate that morality most fundamentally concerns relations of mutual accountability and justice”. But, as the essay proceeds, the scope of his critique broadens to encompass any moral action that is based on considerations of altruism, whether they arise from a utilitarian ethical theory or not. That’s a wise move, since I think many effective altruists (and certainly me, if I count as one) are not utilitarians in any sense. Many are consequentialists, for sure, but many aren’t. Indeed, part of what I want to do in this post is to argue that Darwall’s own favoured moral theory could underpin much—perhaps all—of what effective altruism currently does. It might be that Darwall agrees with that—he is much more concessive towards effective altruism than almost any other critic I’ve read, often pausing to acknowledge at some length positive aspects of the movement. It might be that Darwall sees himself as warning effective altruists where their ethical principles might lead them in the future, rather than objecting to where they’ve brought them so far. I’ll try to address that below as well, noting that Darwall’s favoured theory also runs risks, and that part of what is so difficult about the issues with which effective altruism grapples is how we navigate between the risks Darwall identifies and the risks I discuss.

It’s worth mentioning before we really get going that, while Darwall mentions longtermism, the recent development in effective altruism that asks us to devote a lot of resources to making small changes in the probabilities of different long-term futures for humanity, his critique is not focused on that specifically. Indeed, it receives very little attention. His critique covers it, of course, but only by covering the whole of effective altruism—that is his target. And, as I noted above about myself, it’s possible to be reasonably supportive of the broad commitments of effective altruism while remaining pretty sceptical about longtermism. When I argue below that Darwall’s ethical commitments can support what effective altruism currently proposes, I mean to exclude the longtermist side.

Darwall mentions three critiques of utilitarianism from the point of view of justice: it does not respect the separateness of persons; it places no value on autonomy; it considers the morally right to be the rational pursuit of the morally-relevant good, where the morally relevant good is restricted to the well-being of the morally-relevant individuals, whereas “morality is fundamentally about mutual accountability and respect”. In the remainder of this post, I want to ask whether the current recommendations of effective altruism really suffer from these problems.

Effective altruism and the separateness of persons

Those who appeal to the apparent separateness of persons to object to utilitarianism usually have in mind that it requires us to inflict great suffering on some small number of people if we have good reason to think doing so will create great happiness for a slightly larger number, or if we think it will very slightly increase the happiness of an absolutely vast number—see Ursula K. Le Guin’s ‘The Ones Who Walk Away from Omelas’ for the sort of case people often have in mind. But effective altruists don’t recommend anything like this. They don’t recommend interventions that actively cause the suffering of some to aid the well-being of others.

You might worry that, because they prioritize some interventions over others—donating to bed net charities over motor neuron disease charities, for instance—they trade-off the suffering of some against the well-being of others. But everyone must do this in a situation in which the resources to be distributed are limited and insufficient to alleviate all suffering. The objection to utilitarianism is that it requires you actively to cause suffering that wouldn’t otherwise happen if by doing so you can create in others happiness that wouldn’t otherwise happen; the objection is not that, when some suffering in unavoidable whatever you do, utilitarianism requires you to do whatever minimizes it.

Indeed, while the utilitarian is often condemned for caring only about the total well-being of the outcomes produced, and not how that well-being is distributed over the people in that outcome, in the world we inhabit, the recommendations of effective altruists will often agree with the recommendations of egalitarians. This is simply because of the diminishing marginal utility of money: the same amount of money will create much greater happiness when given to the very worst off in our global society than when retained by the wealthiest. And so considerations of maximizing well-being and considerations of equalizing well-being very often line up in our world. And this has usually been the case with effective altruism: the movement has typically championed interventions that involve the worst-off people in the world.

Effective altruism and autonomy

Those who worry that utilitarians place no value on autonomy make two points. First, they worry that utilitarianism can lead to an objectionable form of paternalism. Utilitarians will undermine someone’s autonomy, and make certain decisions on their behalf, if they believe the person would autonomously choose something that would bring them less well-being. But again, at least on the face of it, the recommendations of effective altruism aren’t like that. Providing insecticide-treated bed nets, anti-parasitic medicine, HIV treatments, and surgery for obstetric fistula do not reduce anyone’s autonomy. Indeed, if they have any effect of autonomy, they enhance it, because they make available to the people who now have access to those treatments a range of possible options that were not available to them before. And consider one of the most high-profile recent interventions that effective altruism has championed, GiveDirectly, which makes unconditional cash transfers to families living in extreme poverty in East Africa. This surely does not impair the autonomy of those receiving the cash.

Nonetheless, I do think Darwall identifies a genuine concern here. While it is true that the medical interventions I just mentioned don’t impair autonomy, the decision how to use the money donated is made not by the people who will benefit, but by the donors, who pick the charity they will support. This is different from an unconditional transfer of the funds that would allow the affected people to do with it what they would like. While donating to these charities will not undermine the autonomy of decisions that the affected individuals would be making in the absence of such donations, they might fail to adequately involve those individuals in decisions about how the money is spent. I think it is a difficult question what a respect for autonomy requires in that case: if someone donates to the Against Malaria Foundation, do they undermine the autonomy of the people who receive bed nets if those people would have chosen to spend the donated money on something other than bed nets?

Dark abstract, Anni Roenkae

The second worry about effective altruism and autonomy is that the movement rarely supports initiatives whose primary goal is to increase the autonomy of those with whom it works. Its interventions might increase autonomy as a happy byproduct of increasing well-being, as in the case of health interventions, or because of a belief that increasing autonomy will increase well-being, as in the case of unconditional cash transfers; but it doesn’t aim to rectify impairments to autonomy on the ground that doing so is a good in itself.

This is a version of the institutional critique of effective altruism, which says, in part, that the movement makes no attempt to change the political systems and institutional structures that give rise to the problems it seeks to alleviate—and, a stronger version of the objection alleges, it actively entrenches those harmful systems, so that they are less likely to change than if effective altruism hadn’t existed and the wealthy who now donate had simply kept their money. Effective altruism doesn’t attempt to create political systems in which the individuals with whom the movement currently works would have much greater autonomy.

To some extent, the movement has accepted this critique and now considers and sometimes recommends charities that seek to improve institutions and governance. But there is still resistance. One source of concern is that small-scale efforts to improve autonomy don’t really work, while large-scale ones are beyond the scope of private actors, charities, and NGOs. Helping to increase voter turnout in a country does little to improve autonomy if the candidates for which people can vote are not selected democratically; but changing the process by which candidates are selected is really beyond the purview of non-state actors.

Another concern is that some of the systemic institutional or political change effective altruists are encouraged to consider carries significant risks—a socialist or a democratic revolution might bring great happiness and autonomy if it succeeds, but it might bring immiseration and despotism as well if it is captured by tyrannical elements or corrupted or thwarted and punished. Effective altruists have, on the whole, been pretty risk averse in their recommendations. When it comes to health interventions, at least, there have been cases in which they have recommended something they’ve later come to think is in fact ineffective, but to my knowledge they’ve yet to recommend something that later turned out to be actively harmful.

I suspect such risk aversion is partly led by the desire to have very robust evidence in favour of an intervention’s effectiveness before recommending; but I suspect it goes the other way too, so that the desire for such evidence is partly grounded in the risk-aversion—they are natural bedfellows. It is, I think, an interesting question how much risk it is morally appropriate for one person to take on behalf of others (see Johanna Thoma and Lara Buchak on this question). The effective altruist will often say that the question doesn’t arise for them because they simply set the probabilities and then choose by maximizing expected utility. But, first, why think one must choose by maximizing expected utility and not using some more risk-sensitive decision theory? And, second, even if one is required to maximize expected utility, the evidence one has is rarely precise enough to pin down a single probability distribution as the unique rational response to it, and so risk aversion can guide effective altruists in their choice of probabilities.

Effective altruism and the nature of moral right and wrong

Darwall’s third critique of utilitarianism holds that it fundamentally misunderstands the nature of moral right and wrong. Whereas utilitarians say the morally right action is the one that constitutes the best means to the morally good ends, Darwall takes the more Kantian view that morally right action must flow from a particular sort of process or intention in the agent responsible for it. He writes: “questions of justice and moral right and wrong […] are irreducibly normative questions of what it would be morally right or wrong to do […], and these moral questions necessarily presuppose relations of mutual accountability and respect in the background.”

However, again, it’s not at all obvious that the effective altruist doesn’t make their moral decisions and recommendations with relations of mutual accountability and respect in the background. The belief that others whom you have never met and will never meet are just as deserving of happiness and well-being just as you are; the recognition that they can obtain more of it by getting resources that are currently in your possession; and the conviction that you must give over those resources rather than use them yourself. These are the basis of the effective altruist’s commitment to donating. And surely each of reflects a deep respect for those individuals affected by those donations. It is precisely the recognition that their distance from you—physically, temporally, socially—does not diminish the strength of respect you should have for them and the strength of obligation you have to bring them happiness that drives the effective altruist.

And it is clear how considerations of accountability to others with whom we share the planet might drive someone to act as the effective altruist recommends. Darwall mentions Peter Singer’s analogy between a child drowning in a pool that you walk past on the way to work, whose life might be saved at the cost of the clothes you’re wearing, and a person starving or suffering from severe but treatable illness in another country, whom you might save by donating the cost of those clothes. One way to read that example is as forcing us to see the legitimate demand made on us by someone whose intense suffering we might alleviate at little cost to ourselves, regardless of where they are. Its purpose is to hold us accountable on their behalf for the way we use the vast resources we have by chance been bequeathed by the current structure of the world.

But Darwall worries that the effective altruist must approach their relationship to those whose well-being will improve as a result of the money they donate in a way that renders that an inappropriate relationship for morally right action:

According to the doctrine of effective altruism, the fundamental ethical relation is benefactor to beneficiary; but from the standpoint of equal justice, the fundamental ethical relation is mutual respect. […] [T]he term “charity,” like the term “pity,” signals the potential in altruism to insinuate a disrespectful relation of superiority to a pitiable “charity case.” The perspective of charity, like the perspective of pity, is not the eye-to-eye relation of mutual respect; it comes, like God’s grace, from above.

But there’s no need to think of the effective altruist’s actions in this way. That the world has landed me within the top 5% of wealth and power globally, and someone else in the bottom 5%, reflects nothing about the relative worth of us as individuals. To think that one thing I should do from this position of enormous wealth and privilege is to give some of that wealth to someone not in this position should run for me no risk at all of viewing the other person in any way that conflicts with the eye-to-eye relation of mutual respect. I don’t deserve the wealth I have; they don’t deserve the penury they’re in—recognizing that is central to effective altruism and central to the recognition of equal respect. I can alleviate some of their suffering by giving money that will reduce my well-being by an amount significantly lower than the amount by which it will increase their well-being. Pity might motivate me to give that money, or playing God, or whatever; but it’s much more likely I’ll be motivated by straightforward considerations of respect for them as individuals, considerations of fairness, and a desire to prevent hardship.

Consider a member of a powerful group living in an unjust society that persecutes members of another group. And suppose that member of the powerful group advocates for legal changes that will prohibit, and so hopefully reduce, the persecution. They are moved by considerations of justice and the recognition that chance has put them in a position with the power to help effect this change. Would we say that they treat themselves as benefactor and the members of the persecuted group as beneficiary? I think not. And that is how I think many effective altruists think of the situation in which they find themselves: they recognize that chance has put them in control of money that they can either use for themselves or put in the hands of others who will help to alleviate suffering. They do the latter not because they pity those whose suffering will be alleviated, or disrespect them, but precisely because they recognise them as fully worthy of respect, and so possessing the same claim to happiness that they have.

Now, in line with his concessive tone throughout, Darwall doesn’t say that effective altruists do view those they work with in this way. Rather, he says they risk viewing them thus. And I agree that it’s important to be aware of those risks. But, for any moral theory, including Darwall’s favoured one, there is a risk for those who adhere to it that they’ll come to follow a corrupted version. I’ll turn to that now.

Effective altruism and interpersonal relations

Let’s turn to the final section of Darwall’s essay, where he discusses one of effective altruism’s more controversial recommendations: earning to give. Want to do the most good with your life? Perhaps you think you should work for a charity or NGO or become a healthcare worker or teacher. According to some effective altruists, that’s wrong. If possible—if you have the talents, connections, social position, or whatever is needed—you should become a hedge fund manager or stock trader instead. And then you should donate the vast majority of your enormous paycheck to fund others to do good. That’s the principle behind earning to give.

Darwall, who is himself a philosophy professor, finds it depressing that he should be in the business of preparing his students for this sort of work. Part of his objection is that people who plan to earn in order to give often become corrupted by the culture of these high-paying jobs and renounce their altruism. But another part of his objection is this: “[e]ven those who maintain their allegiance to EA while working in offices on Wall Street never get to see how the people they seek to benefit actually live or get to live or work with them.” Instead, he hopes that his students might go in greater numbers to “badly needed but talent-starved fields such as public K-12 education.” There, of course, these students will get to see the effects they have.

But, as I argued in my review of The Good it Promises, The Harm it Does, the recent critique of effective altruism’s work in animal welfare that I mentioned above, there is a risk to those ethical theories that hang much on the importance of the interpersonal relationships between the person who does a moral action and the people affected by it. The ethics of care that plays a central role in The Good it Promises, and Darwall’s own proposed theory, seem to share this in common. At one point, Darwall says: “Even the term ‘the global poor,’ risks condescension, since it signals not a respectful relation of mutually accountable collaboration to establish more just relations as equals, but regarding someone as an object of charity that, by definition, no one has the standing to demand.” Moral action, then, ideally takes place in the presence of a respectful relation of mutually accountable collaboration to establish more just relations as equals. The problem with this is the world we inhabit. In it, those who are extremely wealthy by global standards—which includes a large proportion of people living in the US, Canada, the UK, and so on—have such interpersonal relations only with other people who are extremely wealthy by global standards; and that is unlikely to change, because ties of family and friends and other duties will prevent them from leaving their home country to work elsewhere. And so such a theory risks driving people towards jobs working directly with those most in need in their home country, instead of doing good indirectly for those in much much more dire need in other countries, because they lack any interpersonal relationship with those. There is, of course, nothing wrong with each case in which someone chooses to do this—and indeed Darwall is surely right that it would be wonderful to see more people keen to teach in America’s schools, and Britain’s, though my own limited experience is that there is no lack of talented people teaching in schools, but bad systems and poor working conditions. But, as a pattern, this would perpetuate the inequality in the world. The wealthy countries would see increases in quality of life, due to the improved education system, and less money would flow to the less wealthy countries.

A further risk of such a theory is that it leads us to ignore those with whom we simply cannot have a relationship of mutual accountability and respect. For instance, it might lead us to ignore those who do not have the capacity to hold us accountable or show us respect, and those we are unable to hold accountable: think, for instance, of factory-farmed chickens, young children, people who are severely mentally disabled, and future generations.

Just as Darwall doesn’t think that current effective altruists do in fact show anything but equal respect for those with whom they work, and worries only that their underpinning moral theory risks this, so I don’t think that Darwall himself commits the sort of parochialism about moral action that might arise if you took too seriously the requirement that we must stand in relations of mutual respect and accountability to those to whom we have moral obligations. But it seems to me that this is a risk of that moral theory, just as it is a risk of viewing actions through the lens of altruism that you come to pity or disrespect those whose suffering would be alleviated by that altruism.