When I was a visiting fellow at Oxford two years ago, I heard a story about effective altruism I keep going back to whenever somebody mentions Sam Bankman-Fried (one of the philosophy’s early champions) or colonizing Mars (one of its advocates’ principal obsessions).
According to the story, a classmate lent one of the most vocal advocates for effective altruism her toaster while they were both graduate students at Oxford. She reminded him to return it a week later, and a month later, and three months later, to no avail. Finally, invited to his apartment for a social gathering, she spotted the toaster on the kitchen counter, covered in mold.
“Why on earth didn’t you return the toaster to me?” she asked, in frustration.
“I ran the numbers,” he responded. “If I want to do good for the world, my time is better spent working on my thesis.”
“Couldn’t you at least have cleaned the damn thing?”
“From a moral point of view, I’m pretty sure the answer is no.”
I have no idea whether the story is true. Being just a little too perfect, I suspect that it is probably exaggerated, and possibly completely made up. But I share it here because the attitude it encapsulates goes to the core of how the movement of effective altruism went wrong—and why the original intuition that gave rise to it might just be worth rehabilitating.
Let me explain.
The Compelling Intuition Behind Effective Altruism
Effective altruism, at its core, is a simple idea. Many people are motivated to do good for the world: they volunteer and donate and engage in other activities that are meant to be genuinely altruistic.
But upon closer inspection, it turns out that many of these activities have little impact or are altogether pointless. People volunteer at organizations that fail to advance the causes to which they are supposedly devoted. They donate to their local cat shelter even though there are already enough organizations caring for stray pets in their affluent neighborhood. They buy their alma mater a fancy new gym even though the campus already has state-of-the-art facilities.
This is all the more galling because the same amount of money could make a vastly bigger difference if directed to more productive purposes. In America or Germany or Chile or South Korea, even a citizen with a perfectly ordinary job could, if they regularly donate a modest share of their income to a charity which provides people in malaria-infested regions with mosquito nets or distributes anti-parasite medications to people in worm-infested regions, save a human life. According to some calculations, the most effective charities take as little as $3500 to do so.1
The upshot, effective altruists argue, is simple: If each of us can save a human life with limited effort or generosity, it is grossly unethical for us to fail to do so. And if we do decide to engage in altruistic activities, we should do so in effective ways. Why spend a ton of money on some pet cause when that same sum could make a vastly bigger difference in improving human welfare?
This premise can hardly serve as a comprehensive guide for how to make the world a better place. There are all kinds of clever objections people might make—and indeed have made—to it. But in all of its original simplicity, it clearly has a lot going for it.
So why has effective altruism fallen into such disrepute over the course of the past years?
How Effective Altruism Fell from Grace
The most influential evangelist of effective altruism was none other than Sam Bankman-Fried, the cryptocurrency entrepreneur who made a name for himself as a generous philanthropist and political donor. But “SBF,” as he came to be known, turned out to be a fraud twice over: He misappropriated much of the cash with which users of FTX, his cryptocurrency exchange, had entrusted him—and, despite public protestations of great personal modesty, spent enormous sums on his lavish lifestyle.
The damage that SBF’s downfall wreaked on the brand of effective altruism was so big in part because many of the movement’s intellectual leaders were close friends of his, and most of its core institutions had come to rely on his philanthropy. The FTX Future Fund (on whose board Will MacAskill, one of the founders of the effective altruism movement, served until 2022) alone has given $30 million to a UK charity devoted to effective altruist causes (whose board MacAskill chaired).2
Since SBF was sentenced to 25 years in prison less than a year ago, this has given rise to two contrasting interpretations of the philosophy’s fall from grace. The opponents of effective altruism argue that the immoral actions of its most prominent advocate(s) reveal a deeper rot at the core of the philosophy. The defenders of effective altruism argue that we are in danger of throwing the baby out with the bathwater, condemning an insightful moral outlook for the unfortunate circumstance of being championed by somebody who turned out to be a crook.
The truth, I think, lies somewhere in between. To reject effective altruism because SBF is a fraud is far too simple. And yet, SBF’s fall from grace helps to illustrate some (though not all) of the fundamental shortcomings of effective altruism.
Why Effective Altruism Went Wrong
From the beginning, the leaders of effective altruism allowed their moral fervor to override any serious interest in the complexities of the real world. Consumed with the righteousness of figuring out the most clever way to alleviate suffering anytime and anywhere, they failed to ask basic questions about what makes human beings tick; how well we can predict the impact of our own actions; and whether we can influence events in the distant future in any meaningful way.
As a result, effective altruism fell into the classic trap suffered by all utopian movements: Well-intentioned though many of its leaders’ actions and prescriptions may have been, attempts to impose their neat ideals on the crooked timber of humanity rarely worked, and sometimes turned into disaster. More specifically, three fundamental defects explain why effective altruism went wrong: the problem of psychology, the problem of prediction, and the problem of providentialism.
The Problem of Psychology
When SBF was a student at MIT, he went through a crisis of meaning. A talented physicist, he was seriously considering an academic career. Then he came into contact with effective altruism, and began to reflect on ways in which he might make a bigger impact on the world. Ultimately, it was one of the key concepts of the early effective altruist movement which set him on his life’s trajectory: the idea of “earn to give.”
The idea of “earn to give” stands at the core of an influential organization founded by MacAskill. Named after the number of hours that an average professional might spend on their career, 80,000 hours sought to offer students and young professionals advice about what to do with their lives. But whereas you might expect that career advice from do-gooders would guide idealistic people towards working as teachers or aid workers, the organization’s trademark recommendation was to go into finance or other lucrative professions.
The average teacher or aid worker, the founders of 80,000 hours insisted, is only able to do a modest amount of good. A rich person donating to the right causes, by contrast, can have a transformative impact. And so their advice was to earn a ton of money in order to maximize how much you could later donate.3
The problem is that there’s a big psychological flaw lurking behind this seemingly clever prescription. “Earn to give” tacitly assumes that a 22-year old who devotes himself to making a ton of money for purely altruistic reasons will still be guided by a desire to do good thirty or sixty years later. But this doesn’t account for the ways in which the worldview of most people is shaped by their surroundings.
It is easy to see how the values of a young idealist who goes to Wall Street might weaken and drift and ultimately become unrecognizable. I don’t really care about wearing a fancy suit, our hero will tell himself at the beginning of his career, but if I want to keep advancing in the firm, I have to look the part. I don’t really care about owning a home in the Hamptons, he will tell himself when he has risen a few rungs, but if I don’t go to the right dinner parties, I’ll never make CFO. I don’t really care about the luxurious life we’ve built, he will tell himself as he nears retirement, but my wife and kids have become accustomed to it.
Before you know it, our hero resembles the senior banker portrayed in Margin Call, a movie loosely based on the downfall of Lehman Brothers. “Did you really make two and a half million bucks last year? What do you do with all that money?” a junior colleague asks him. “I don’t know really,” the senior banker responds. “It goes pretty quick. You learn to spend what’s in your pocket.”4
This kind of personal transformation doesn’t even have to take decades. In the case of those who are most successful at getting rich—and these are the people who most matter to the moral calculus behind 80,000 hours—a few years of phenomenal wealth may well suffice to skew their priorities. Take SBF. However tempting it may be to doubt his sincerity in retrospect, every biographical account of his early life seems to suggest that the aspiration to accumulate the money he would need to do a lot of good for the world was a big part of why he abandoned his academic aspirations to join Jane Street, a high-frequency trading firm.
SBF was 22 years old when he joined Jane Street. He was 27 when he founded his crypto exchange, started to get seriously rich, and became accustomed to his lavish lifestyle. He was 31 on the day he was apprehended for stealing from his customers at a massive scale.5
The Problem of Prediction
At the beginning, MacAskill and other effective altruists primarily focused on doing good in faraway places. They believed that inhabitants of affluent countries could save scores of lives in the poorest parts of the world today. Following a famous thought experiment first proposed by Peter Singer in 1972, they argued that our moral obligation to do so wasn’t diminished by their geographic distance.6 This was the “bednets” phase of the movement, when it focused on down-to-earth concerns like eradicating malaria.
In a later phase, MacAskill and other effective altruists instead started to focus on doing good in faraway times. Influenced by the thought of Derek Parfit, they came to embrace two provocative ideas. The first is that improving the life of a human being who lives 10,000 years from now is as morally significant as improving the life of a human being who is alive today. The second is that we shouldn’t just care about improving the lives of human beings who already exist; we should also value the creation of additional human beings who will go on to lead worthwhile lives. This was the impulse behind the “colonizing Mars” phase of effective altruism: in the minds of the movement’s leaders, the importance of saving a few millions people from malaria today started to pale in comparison to the importance of saving all of humanity from the risks of a nuclear winter or an AI apocalypse sometime in the distant future.
The bulk of the criticism of this pivot has been on ethical grounds. Some people have argued that there is good reason to value the lives of our contemporaries more highly than that of our distant descendants. Many more have questioned whether adding an additional human being who does not yet exist to the earth has inherent moral value. But in my mind, this misidentifies where effective altruism went so badly wrong when it took its long-termist turn.
The problem with long-termism isn’t that we shouldn’t care about avoiding a nuclear war that might wipe out humanity in 10,000 years. Nor is it that we should be indifferent between a world in which 10 billion people live happy lives and one in which one hundred people live happy lives. It’s that we know far too little about how the social world works—especially across such vast time spans—to do anything meaningful to pursue such long-termist objectives.
Saving people from serious but treatable diseases right now without doing damage is hard enough. Most well-meaning interventions fail. Even when they succeed, they can have unintended consequences. In some parts of sub-Saharan Africa, for example, noble efforts to reduce the prevalence of AIDS have devoted so many resources and the attention of so many local doctors on the fight against one deadly disease that mortality from other causes started to skyrocket. And then there’s the risk that an overreliance on foreign donations increases corruption, weakens demands for accountability from local elites, and thereby perpetuates the exploitative political and economic structures which are the root cause of abject poverty.
If the prevalence of such unintended consequences makes it difficult to save people in the here-and-now, just begin to imagine how hard it is to know what kind of action would make a meaningful contribution to the fate of humanity in the there-and-then. To be able to take effective action, you’d have to be sure about the direct impact of your work; to eliminate the possibility that it would have disastrous side effects; and to anticipate whether it might lead to some form of backlash which undoes all of your achievements. In other words, it would require an astonishing ability to imagine the future—and past attempts at prediction suggest that we are simply incapable of such foresight.7
A simple example illustrates that these concerns are by no means abstract. In the past decade, effective altruists have understandably become obsessed with how to avert the risk of superintelligent AI destroying humanity. Since the movement’s leaders and sympathizers did not believe that it would be realistic to halt the development of AI, they decided that the best course of action would be for the pioneers of the technology to share their moral commitments. And so they founded and lavishly funded a nonprofit whose mission was to guide the technology in a responsible manner. The name of that nonprofit is OpenAI, the entity that created ChatGPT and is now at the forefront of the AI revolution.
But things did not work out the way that OpenAI’s founders had imagined. Despite formally remaining a nonprofit, the organization now effectively operates like a classic corporation, pursuing profit and technological progress at any price. While the original purpose of the organization was to focus on AI safety, the need to stay ahead of peer competitors has made its leaders throw caution to the wind; according to whistle-blowers, the company is barely doing anything to contain the risks it was founded to combat.
The Problem of Providentialism
Let’s return to the famous philosopher and that toaster for a minute. The story may be apocryphal. But it gets to a persistent concern about philosophies that tell a grand story about how to transform the world—one that, perhaps surprisingly, applies both to effective altruism and to ideologies, like Marxism, that on their face are vastly different.
Both Marxism and effective altruism claim that they have figured out the true way to make the world a better place. And both Marxism and effective altruism flatter their adherents into thinking that this gives them a key role in a movement that will prove to have world-historical importance.
Taken together, these two beliefs provide a powerful justification for immoral action. The important role that each of us supposedly plays in bringing about world-historical change licenses an inflated sense of our own importance. Meanwhile, the overriding importance of the movement’s goals provides a good reason to dispense with the requirements of ordinary morality.8
This is the logic lurking behind the purported refusal by the graduate student who went on to become a famous philosopher to return the toaster. He had come to believe that he is vastly important to a cause that can make a vastly important contribution to the well-being of humanity. Surely, he thought, this must make it permissible to dispense with an activity as trivial as cleaning cheese out of an ordinary household appliance?
It is easy to see how this logic scales. The cause SBF took himself to be serving is also vastly important. In light of the huge financial contribution he was making to Important Charitable Causes, he had even better reason than the aspiring philosopher to believe that he is of great importance to that cause. So when he started to run into some financial headwinds, it was easy to tell himself: Surely, I’m morally justified—perhaps even required!—to dispense with such pedestrian constraints as SEC rules about how to handle customer deposits?
Effective Altruism Without the Hubris
The trouble with effective altruism has, from the beginning, been its hubris. The movement’s leaders treated doing good as a kind of intellectual game, one that is so deeply constituted by abstract rules about creating happiness and eradicating pain that it can dispense with such messy concerns as human psychology or the dynamics of the political world.
Any attempt to rescue effective altruism from these shortcomings has to begin with a big helping of modesty. It is very hard to predict how somebody will act three decades from now and virtually impossible to predict what challenges humanity as a whole will face in three centuries. The first step to doing good in an effective manner is to be honest about our cognitive limitations—and to take seriously the ever-present danger that even the most well-meaning and thoroughly “researched” interventions may prove to be counterproductive.
A reconstituted effective altruism must give up on long-termist ambitions that make for fascinating sci-fi but completely fail to guide action of which we can be reasonably confident that it will actually have a positive impact. It must understand that the human psyche is messy, giving us reason to be deeply skeptical about overly clever hacks to hard problems, like telling young idealists to turn themselves into efficient money-making machines. And it must dispense with the technocratic ethos of omniscient sages who are convinced that they are superior to their compatriots and, when push comes to shove, may even be justified in ignoring the ordinary rules of morality—an ethos that both leads good people astray and runs the danger of attracting some of the very worst people to the cause.
But none of this is a reason to throw the baby out with the bathwater. Effective altruists are right that people spend billions of dollars on charitable contributions every year. It is true that much of that money goes to building new gyms at fancy universities or upgrading local cat shelters. And it is hard to argue with the idea that it would, insofar as possible, be better to direct donors’ altruistic instincts to more impactful endeavors, potentially saving the lives of thousands of people.
Even in this more modest form, effective altruism will face some serious empirical obstacles. The history of seemingly obvious interventions that unexpectedly turned out to have adverse consequences is long. But when stripped to its core, the core intuition behind effective altruism does not depend on the more dubious assumptions about human psychology or our ability to predict the future that leading effective altruists have embraced. Rather, it merely claims that charitable donations can make a big impact in the world; that this gives people who are reasonably affluent by global standards good reason to donate a generous share of their income; and that they should think hard about what kinds of donations are most likely to make a real difference to people in genuine need.
Put in these simple terms—and stripped of the hubris to which the movement they inspired sadly succumbed—these premises are virtually impossible to contest.
There are, of course, serious questions about whether these calculations are right. For an excellent critique of most development aid, see my interview with William Easterly.
The effective altruist movement has also suffered from other serious scandals. The most consequential of them has been the discovery that Nick Bostrum, one of the key intellectual forces behind long-termism (on which more later) and the founder of the Future of Humanity Institute at Oxford University partially defended eugenics and used the n-word while he was a graduate student in the 1990s. Though a recent investigation by the university concluded that he is not a racist, the institute recently shuttered its doors.
To be fair, the organization’s advice was not always this black-and-white. When I myself was going through a crisis of meaning while completing my PhD, I wrote to the organization for some help in thinking through my options. Benjamin Todd, who co-founded 80,000 hours alongside MacAskill, kindly responded with some thoughtful advice that can’t be reduced to (though it also didn’t discourage me from) joining McKinsey. But perhaps this was in part because Todd sensed that my money-making potential was limited.
There’s also a more simple psychological problem with effective altruism. Members of the movement are motivated to do as much good in the world as possible. But they fail to account for the fact that most people who engage in altruistic actions don’t have the same motivations. They don’t donate money or volunteer because they want to maximize the balance of pleasure over pain in the world—but rather because they feel a connection with their local community or gratitude towards their alma mater. Telling them that they should direct their altruism to more effective causes may simply lead them to not donate at all.
My point here is that effective altruists are naive about how human motivation might evolve over time at the individual level. By extension, you could also say that they are naive about how institutional incentives might evolve over time at the collective level. They think that you can set up a movement like effective altruism and continue to attract people with genuinely other-regarding motives to it; but this underplays the likelihood that its popularity would eventually attract others who simply want to cloak themselves in its appearance of moral valor, or who are attracted to it by the excuse it offers for giving free rein to their technocratic arrogance. Just as individuals can stray from their mission over time, such organizations are likely to stray as well.
Imagine, Peter Singer exhorted his readers in “Famine, Affluence, and Morality,” that you are taking a stroll in a park. Suddenly, you notice a little girl struggling to stay afloat in a pond. Do you have a moral obligation to rescue her even though doing so might ruin the expensive suit you’re wearing? Most people won’t hesitate to answer this question in the affirmative, so this thought experiment doesn’t do much to change how we think about the world.
But now change the scenario a little. Imagine the girl has a different color of skin, or is a citizen of another country, or happens to live at a great geographic distance from you. She’s not literally drowning in a pond but she is at risk of starvation. Assuming that you are still in a position to save her life by sacrificing your suit—or some monetary equivalent of it—does the fact that she happens to be more distant render obsolete your moral obligation to save her?
Like Singer, effective altruists believe that the answer is no. While it might be easier to ignore our obligations to suffering people who are geographically distant, our reasons to come to their rescue are just as strong. And effective altruists conclude from this not only that we have strong moral reasons to make genuine personal sacrifices to help faraway people in distant lands—but also that, to fulfill this obligation, we should think hard about how we can actually effectively help people.
Science fiction has at this point been a popular genre for many decades. This makes it tempting to look back at some of the most prescient works of imaginative fiction, and wonder in awe about how much they got right. But that shouldn’t seduce us into thinking that we’ll succeed in predicting the future. For one, focussing on the books that, with the benefit of hindsight, have turned out to be most prescient skews the sample; after all, contemporaries couldn’t have known which of the many scenarios that science fiction writers imagined at the time would turn out to be most accurate. For another, it is striking how much even the most prescient sci-fi novels of the past got wrong; even futuristic works from the Victorian era that adequately capture some elements of contemporary civilization assume that our world would still run on steam and have travelers going to-and-fro in hot air balloons.
The most brilliant psychological portrayal of what these two assumptions can do to the moral life of a seemingly brave and idealistic man was painted by Arthur Koestler in Darkness at Noon. Rubashov, the novel’s tragic protagonist, has long been convinced that the goal of world revolution requires us to dispense with such bourgeois niceties as abstaining from imprisoning or executing the innocent. He has also believed that his own leading role in bringing about this revolution in Russia justifies him in sacrificing his own comrades when they stand in his way or fail to serve his cause. Now, the inevitable misfortune has finally befallen him: he himself has now been arrested on fabricated charges. The tension at the heart of the novel is less about what will happen to Rubashov, who eventually meets the fate that seems inevitable from the first page; it is whether, and how, he will be able to justify to himself the treatment to which his own comrades are subjecting him, and to which he had previously subjected so many others.
Beautifully done. I reread Darkness at Noon a month or so ago. It still effects me. Reading this inspired me to get Jonathan Aldred’s Licence to be Bad down from the bookshelf. It is a good companion to Easterly’s The Tyranny of Experts. Aldred might be a good person to interview on the podcast. One small quote “this urge to actuarial alchemy, dissolving the incalculable into the calculable, is strongest when everything else at stake is objective and quantitative, hence uncertainty is the only remaining obstacle to a seemingly perfectly rational mathematical decision-making process. And the urge to actuarial alchemy is even stronger when people are willing to pay a lot of money for it.” Effective altruism seems to be one more manifestation of actuarial alchemy.