Effective Altruism: Not Effective and Not Altruistic

If you, like me, have stepped into Sci, Kohlberg, Singer, your dorm, Sharples, Trotter, or virtually any place on Swarthmore’s campus this semester, then you have probably seen a number of posters advertising Swarthmore Effective Altruism’s 8-week fellowship. “AAAAAAAHHHHHH,” scream the posters; I scream back. Effective Altruism is a global social and ethical movement which encourages people to not only do good, but to do the most good that they possibly can — be it through donating money to causes touted by the movement, choosing a career serving a high-impact cause, or selecting a high-paying career with the intention of donating the lion’s share of your income.

Beyond Swarthmore, charitable organizations aligned with the Effective Altruism movement such as GiveWell, GiveDirectly, and Giving What You Can, each report moving hundreds of million dollars, either in donations directly to the organization or donations inspired by the organization’s messaging. Billionaires such as Dustin Moskovitz, Bill and Melinda Gates, Warren Buffet, and Elon Musk have been linked to Effective Altruism with respect to their charitable donations. Though it is of course difficult to quantify the impact of a decentralized social and ethical movement, we may safely say that Effective Altruism is both responsible for somewhere between hundreds of millions and billions of dollars in charitable donations, and for directing those billions towards particular causes. Given the enormous reach of Effective Altruism, it is therefore critical to assess its aims, successes, and shortcomings.

As college students, we are neither as well-endowed nor as powerful as the aforementioned billionaires. The message Effective Altruism sends to college students is therefore slightly different. The nonprofit 80,000 Hours (oft-cited in Swarthmore Effective Altruism’s Fellowship curriculum) encourages young people to use their careers to solve the most pressing problems, and to do so in the highest-impact way. Fortunately for us future-do-gooders, just about every Effective Altruist organization keeps rankings of high-priority causes with detailed justifications — and some may surprise you. 

Any movement contains a huge diversity of views; Effective Altruism’s views on what its own priorities should be are no exception. For example, GiveWell’s Maximum Impact Fund splits donations between the causes of malaria prevention, vitamin A deficiency, childhood vaccination, deworming, and direct cash transfers. Giving What We Can gives priority to issues like factory farming, climate change, and promoting beneficial AI development. 80,000 Hours lists, as its top two priorities, more research into what Effective Altruism’s priorities should be, and promoting Effective Altruism itself — which may or may not reassure anyone wondering whether Effective Altruism’s organizational structure is pyramid-shaped.

In critiquing Effective Altruism, I do not intend to argue here that Effective Altruism’s preferred causes are bad ones, nor that any particular action which Effective Altruism recommends is harmful. I am also not saying that people should not try to do a lot of good, nor that they should not donate money to charitable causes, nor that they should not make career choices in virtue of concern for the common good. Indeed, I understand the motivation behind Effective Altruism all too well. In her book, “Trick Mirror,” essayist Jia Tolentino captures perfectly a feeling that I’ve had for as long as I can remember: “[T]he choice of this era is to be destroyed or to morally compromise ourselves in order to be functional — to be wrecked, or to be functional for reasons that contribute to the wreck.” What Effective Altruism promises is a third option: a lifestyle of levelheaded thinking and ethical action which shields both the doer and recipient of altruism from the vicissitudes of the modern world.

It is in this promise where Effective Altruism falls short. Effective Altruists can certainly do, and have done, a lot of good. But I argue that Effective Altruism is not true altruism, and that even if we recognize the positive impacts of Effective Altruism, we must also acknowledge that the grounds on which it defines its own effectiveness are incredibly shallow.

A litany of other criticisms of Effective Altruism are located only a short Google search away from the interested reader. The criticisms I’ll highlight here are the following and are to some extent, interconnected. First, philosopher Amia Srinivasan points out that Effective Altruism is a hyper-individualistic movement which places too much emphasis on private charity, and does so to the exclusion of addressing structural causes of suffering at the level of communities, classes, and states. Second, Effective Altruists are biased towards the sorts of suffering which lend themselves to easy quantification, maligning ineffable impacts like culture loss and self-determination. Third, like other humanitarian movements, Effective Altruism’s viewpoint towards the developing world is that of western saviorship, exacerbated by the fact that a significant majority of Effective Altruists are English-speaking white men. Although I do not focus on them in this article, I view these as valid criticisms which pose another sort of pressing problem for Effective Altruism. Naturally, there are reciprocal responses and dialogue surrounding these issues from within the movement.

Charity can cut both ways, so let’s take a charitable view of Effective Altruism. That is, let’s try for now to look past incredulous-stare-inducing arguments from the movement, such as Chapter Eight of the book, “Doing Good Better,” by William MacAskill (self-proclaimed co-founder of Effective Altruism), which quite literally defends sweatshop labor; let’s try to look past the movement’s apparent acquiescence to structures of domination and the overwhelming number-crunching involved in the perpetual triage of quality-adjusted-life-year (QALY) maximization. Perhaps it’s possible that Effective Altruists need not accept the conclusions at the logical extremities of the movement. Perhaps, some will argue, Effective Altruism is merely a useful practical framework for the modern person, driven by altruism and a desire to do good in the face of the world’s complexities.

The problem with this rhetorical move is that Effective Altruism’s deeply conditional way of viewing moral goodness is at odds with our general concept of altruism. Altruism is often defined as selfless action for the wellbeing of others. We should be careful to note that not all selfless actions are altruistic, and altruism need not be contrary to a person’s best interests (indeed, some philosophers argue that altruism is in everyone’s best interest, including that of the doer!). The reason why selflessness is still a part of the definition of altruism is because for the altruistic person, their concern for others’ well-being is great enough to overtake their self-interest, if the two should come into conflict. Well-being resists narrow definition: it could refer to a wide range of things from improved life expectancy to flourishing and self-actualization. Take as an example of altruism a person who, under conditions of food rationing, eats less so that others can have more. This person is acting altruistically by our simple definition because she knows that nutrition is crucial to any kind of well-being for people, and will sacrifice some of her own food towards this end. If you agree with me that this suffices as an example of altruism, then note what is not a part of the example: the person in the example does not impose on herself the condition of only giving food to someone for whom the food would cause the largest relative increase in well-being, yet we regardless still view her as altruistic. 

This reveals that Effective Altruism’s condition of optimizing well-being is not part of our organic concept of altruism. In particular, Effective Altruism is not actually altruistic because the purportedly altruistic choices the movement prescribes — donations, career choices, etc. — are all done conditionally. For example, an Effective Altruist making a donation to charity does so only on the conditions that the donation has some expected payoff in terms of QALYs, a comparative advantage over donations to different causes, and ultimately on the condition of the Effective Altruist maintaining both their moral high ground and immunity from risk. MacAskill’s own argument in “Doing Good Better” illustrates this perfectly: “Imagine saving a single person’s life: you pass a burning building, kick the door down, rush through the smoke and flames, and drag a young child to safety … You’d be a hero. But we [Effective Altruists] can do far more than that. According to the most rigorous estimates, the cost to save a life in the developing world is about $3,400 (or $100 for one QALY). This is a small enough amount that most of us in affluent countries could donate that amount every year while maintaining about the same quality of life.”

This example is designed to attract people to the cause of Effective Altruism by emphasizing its convenience, but therein lies the problem. We would call the person who runs into a burning building to save a child “heroic” and “altruistic” in virtue of precisely a set of conditions which would cause a doctrinaire Effective Altruist to advise people not to save children in burning buildings. Someone who rushes into a burning building does not stop to calculate the expected QALYs saved or consider the counterfactual; although an Effective Altruist would think this irrational and arbitrary, it does not matter to the person running into the building how old the child is or whether the child has a disability (QALYs decrease as people age and are lower in the case of disability — I address this later). Second, more importantly, the person who runs into the building does not even know if they will save the child at all, but are nonetheless willing to put themselves at risk. Third, the person knows that there are things which cause suffering on a scale far more vast than a house fire, as we all do, and does not let this distract them from acting altruistically on behalf of the child. MacAskill makes an improper comparison in saying that Effective Altruists are far more heroically saving children: maybe Effective Altruists affect more children than the person who runs into one building, but these two actions cannot be compared just with respect to scope because they are different kinds of actions altogether. Even worse for Effective Altruists, adding the condition of optimization onto altruistic action may undercut that which makes actions altruistic in the first place. A less-charitable reading of MacAskill’s argument, quoted earlier, reads MacAskill as bragging when he writes, “we can do far more than that.” If Effective Altruism’s focus on optimization plays into competitiveness, flattering oneself, or derision of others who ‘do less good,’ then Effective Altruism is not selfless but self-aggrandizing, and therefore goes against altruism.

Some will still argue that the movement’s focus on effectiveness is worthwhile. After all, we are limited beings in a world with finite time and resources. To this I say: Effective Altruism defines the effectiveness of its charitable effort with respect to a certain extremely narrow standard. If we call that standard into question — which I will argue we have good reason to do — then what we are really questioning is the effectiveness of Effective Altruism as a whole.

As I mentioned earlier, Effective Altruism uses quality-adjusted-life-years, or QALYs, as a metric for quantifying good which could be done by various charitable interventions. Conversion between dollars spent and QALYs pervades Effective Altruist reasoning; the pressing question when it comes to optimizing where to spend one’s money and time is which cause or charity saves the most QALYs per dollar. In calculating QALYs, the quality of life under a health state is determined by surveying populations on how desirable they would rate living in that health state for a year — so in short, QALYs are a descriptive tool which quantify the perceived desirability of a year of life as a measure of its quality. Effective Altruists, on the other hand, use the difference between amounts of QALYs as a normative or prescriptive metric to determine which causes to contribute to over others. For example, MacAskill makes a brief argument in “Doing Good Better” using QALYs to argue for contributing to someone’s blindness-preventing surgery over someone’s AIDS-treating antiretroviral therapy — surgery wins out, with 30 QALYs as opposed to therapy’s 6.5.

But, crucially, it is necessary that if a difference in the amount of QALYs provides moral reason for an Effective Altruist to value one cause over another, then they must agree that a year lived by a person at a lower number of QALYs is less morally valuable than a year lived by a person at a higher number of QALYs. This latter claim is different from and does not follow from the fact that people would prefer to live healthy lives — of course we all would like to be in perfect health. Effective Altruism conflates the desirability of a particular QALY with its worth and how morally important it is to give a person a chance to live out that year. It is a mistake for Effective Altruism to equate desirability with moral value, and this mistake is particularly evident in the case of chronic disability. To take a personal example, I am profoundly deaf in both ears, so the rest of my life by necessity contains strictly fewer QALYs than someone else who is like me in all other respects but not deaf. Given a scenario where only me or my hearing twin can survive, the Effective Altruist will say that it is a clear choice which one of us should live. I think this is not only wrongheaded but repugnant. Whether a condition is desirable is different from whether the lives of people living with that condition have equal moral value. And if the movement’s answer to the latter question is not a resounding yes, then Effective Altruism has gone terribly wrong.

Perhaps Effective Altruists have ways besides QALYs of calculating the value of different causes. For example, in the charity GiveWell’s spreadsheet analyzing the cost-effectiveness of its Maximum Impact fund, dollars spent by particular charities are evaluated in terms of the moral value produced. The moral value of various options come from a spreadsheet cited throughout by GiveWell containing “moral weights” of people in different age brackets. The spreadsheet is devoid of units for the values listed and contains no rationale for the numbers (which aren’t in a range where it would make sense for them to be QALYs), so evidently Effective Altruists have figured out some alternative. Setting aside the details of comparing different causes in a fine-grained way, even if Effective Altruism finds a satisfactory alternative to QALYs, I still think that the movement’s focus on maximization of good is, at best, a tough pill to swallow, and at worst, groundless.

I found the following thought experiment in a Forbes article by Davide Banis. Banis (I believe, since I did not see it in my reading of the book) wrongly attributes the thought experiment to MacAskill’s “Doing Good Better.” Still, it logically follows from Effective Altruism’s principle of optimizing the good one can do. Return to the previous example of the burning building, and now suppose that you know with certainty that there is a child in one room and a Picasso in another room. Effective Altruism argues that donating the countless millions you would gain from the sale of the Picasso would save countless more children than the one child in the other room, and that this choice would lead you to donate dozens of times more than you would otherwise be able to give in your lifetime. So run with the Picasso and make haste lest the fire damage it.

Most people, including me, will find this both counterintuitive and morally troubling. On one hand, an Effective Altruist might argue that you should still save the child. Perhaps the child could do good which equals or surpasses the good that the sale of the Picasso would do, factoring in various guesses as to the expected price of the painting, the probability and magnitude of the good the child might do, and so on. Of course, this argument would be highly speculative, which reveals a deeper problem: Effective Altruism’s claim to its own effectiveness depends on one’s trust in the assumption that reliable prediction of highly specific future circumstances lies within humans’ powers of reasoning.

On the other hand, an Effective Altruist might bite the bullet, arguing that their version of doing good simply entails that some sacrifices must be made along the way: to do the most good requires us to dispense of options promising less good. Though this seems to cater to what the average person wants — don’t we all just want to do good? — it is deeply misleading because Effective Altruists have equated the question of whether an action is a good one with the question of whether an action maximizes some component of what is good. Twentieth-century philosophers of virtue ethics such as Anscombe and Foot discuss the notion of “thick” and “thin” concepts: thin concepts, such as “right/wrong,” “good,” and “obligatory,” are merely evaluative, whereas thick concepts like “rude,” “pious,” and “open-minded,” do descriptive work on top of making an evaluation. When the average person is asked what they think moral goodness is, they will most likely gesture towards a diverse set of thick concepts without saying that goodness is reducible to any one of the concepts. An Effective Altruist, in contrast, is forced by their assumption that goodness is scarce and that tradeoffs must be made to zero in on only a few thick concepts (such as hedonic pleasure) and say that goodness is reducible to the maximization of that quality. Under the guise of satisfying a general and robust picture of goodness, Effective Altruism smuggles in a niche definition of goodness as maximization of a particular small set of properties. This takes advantage of the natural plurality, and therefore ambiguity, of the average person’s view of moral goodness.

As I mentioned earlier, I’m not arguing that people shouldn’t donate money to charity, nor that the actions which Effective Altruism recommends are bad ones in themselves, but rather I am arguing that the mindset or framework which Effective Altruists use to conceive of moral concepts like altruism and goodness is a flawed one. And if I’m correct, this means that people who just want to be altruistic and to do good should find cold comfort in Effective Altruism’s message. Effective Altruism is like a broken moral machine in that it runs over the wrong set of data and reads the data in a lopsided way. Unfortunately, even the best intentions of its user will not change its brokenness. So, as Swarthmore students, of course we should make conscientious choices and try to do well by others, both in our career trajectories and everyday lives, while still acknowledging that we are a part of a vast and complicated world. We don’t have to settle for a pared-down, frugal sense of ethics, because moral goodness is not a finite resource. I don’t mean to discourage doing good, so let me be clear that being an Effective Altruist is better than being amoral or evil, and the actions of Effective Altruists have done tremendous good. But if we want to do good in many ways — and I think we should want this — we should notice that goodness in all its forms becomes accessible to us only when we dispense with the moral lens proposed by Effective Altruism.