Effective altruism is the philosophy of Sam Bankman-Fried, the crypto wunderkind now sentenced to 25 years in prison for fraud and money laundering. Elon Musk has said that EA is close to what he believes. Facebook mogul Dustin Moskovitz and Skype cofounder Jaan Tallinn have spent mega-millions on its causes, and EAs have made major moves to influence American politics. In 2021, EA boasted of $46 billion in funding—comparable to what it’s estimated the Saudis spent over decades to spread Islamic fundamentalism around the world.
Before the fall of SBF, the philosophers who founded EA glowed in his glory. Then SBF’s crypto empire crumbled, and his EA employees turned witness against him. The philosopher-founders of EA scrambled to frame Bankman-Fried as a sinner who strayed from their faith.
Yet Sam Bankman-Fried is the perfect prophet of EA, the epitome of its moral bankruptcy. The EA saga is not just a modern fable of corruption by money and fame, told in exaflops of computing power. This is a stranger story of how some small-time philosophers captured some big-bet billionaires, who in turn captured the philosophers—and how the two groups spun themselves into an opulent vortex that has sucked up thousands of bright minds worldwide.
The real difference between the philosophers and SBF is that SBF is now facing accountability, which is what EA’s founders have always struggled to escape.
***
If you've ever come across effective altruists, you’re likely fond of them too. They tend to be earnest young people who talk a lot about improving the world. You might have been such a young person once—I confess that I was. A decade before the founding of effective altruism, I too set out to save the world’s poorest people.I grew up like today’s typical EA. White, male, a childhood full of Vulcans and Tolkien, Fortran and Iron Man. I went into philosophy because it felt like a game, a game played with ideas. In 1998, with a freshly minted Harvard PhD, I was playing with the ideas of Peter Singer, today’s most influential living philosopher.
The idea of Singer that excited me was that each of us should give a lot of money to help poor people abroad. His “shallow pond” thought experiment shows why. If you saw a child drowning in a shallow pond, you’d feel obliged to rescue her even if that meant ruining your new shoes. But then, Singer said, you can save the life of a starving child overseas by donating to charity what new shoes would cost. And you can save the life of another child by donating instead of buying a new shirt, and another instead of dining out. The logic of your beliefs requires you to send nearly all your money overseas, where it will go farthest to save the most lives. After all, what could we do with our money that’s more important than saving people’s lives?
That’s the most famous argument in modern philosophy. It goes well beyond the ideas that lead most decent people to give to charity—that all human lives are valuable, that severe poverty is terrible, and that the better-off have a responsibility to help. The relentless logic of Singer’s “shallow pond” ratchets toward extreme sacrifice. It has inspired some to give almost all their money and even a kidney away. (...)
Aid organizations, I learned, have been through many cycles of enthusiasm since the 1960s. Every few years, an announcement—“We’ve finally found the thing that works to end poverty”—would be followed by disillusionment. (In the early 2000s, the “thing” was microfinance.) Experts who studied aid had long been at loggerheads, with Nobel laureates pitted against one another. Boosters wrote books like The End of Poverty: How We Can Make It Happen in Our Lifetime. Skeptics wrote books like The White Man's Burden: Why the West’s Efforts to Aid the Rest Have Done So Much Ill and So Little Good.
Hundreds of millions of people were living each day on less than what $2 can buy in America. Fifty thousand people were dying every day from things like malaria and malnutrition. Each of those lives was as important as mine. Why was it so hard to figure out what can help—to find out what works to reduce extreme poverty?
While I was learning about aid, real progress was made on the “What works?” question. People started testing aid projects like drugs. Give half of the villages bed nets, make the other villages the “control group,” and you can get a better idea of what benefits the bed nets are bringing. Experts still debate how much weight to give these results. But the drug trial innovation in aid was encouraging.
Still, even “control group” testing, I learned, gives only a close-up view of what’s happening. Extremely poor people live in complex environments—just as complex as our own, and usually more chaotic. Sending in extra resources can have all sorts of effects beyond what a close-up shows. And that’s the real problem in finding “what works.”
Say I give some money to a charity that promises to better the health of poor people in Africa or Asia. And let’s say that, in a close-up view, it works. What else might my money have done? Lots of things. Maybe bringing in the charity will boost the power of a local potentate. Maybe the charity’s donated medicines will just free up money in the budget of an oppressive regime. Maybe the project will weaken the social contract between the people and their government—after all, why would the state care for the health of its citizens, and why would citizens even demand health care from the state, if rich foreigners are paying for it?
Aid experts know all about these negative impacts, plus many others—they’re like the side effects of drugs. A blood-thinning drug may lower the risk of stroke, but it can also keep wounds from healing. Taking lots of painkillers might damage your kidneys. (...)
It took me, a philosopher, years to learn what might be obvious to you. The “close-up” effects of pills or bed nets are easy to advertise, but the side effects—political, economic, psychological—are just as important. Most important, of course, will be what the local people think about interventions into their lives. Yet their very poverty means they can’t hold anyone accountable for harms they suffer.
I drafted an article on what I’d learned about aid and called it “Poverty Is No Pond.” Making responsible choices, I came to realize, means accepting well-known risks of harm. Which absolutely does not mean that “aid doesn’t work.” There are many good people in aid working hard on the ground, often making tough calls as they weigh benefits and costs. Giving money to aid can be admirable too—doctors, after all, still prescribe drugs with known side effects. Yet what no one in aid should say, I came to think, is that all they’re doing is improving poor people’s lives.
***
Just as I was finishing my work on aid, a young philosopher from Oxford gave a lecture at my university, saying that all he was doing was improving poor people’s lives. This was Toby Ord, who was just then starting effective altruism.Like me a dozen years earlier, Ord was excited by Peter Singer’s “shallow pond” argument. What he added to it, he said, was a way of measuring how many people’s lives he could save. The simple version goes like this. Say there’s a pill that adds a year of life to anyone who takes it. If Ord gives $50 to an aid charity, it will give out 50 pills to poor foreigners. So with his donation, he has added a total of 50 years of life. And adding 50 years is like saving the life of one child drowning in a pond. So by giving $50, he has “saved the life” of one poor child.
Onstage with Ord that day was a former director of Christian Aid who’d written a massive book called Does Foreign Aid Really Work? This expert tried to persuade Ord that aid was much more complex than “pills improve lives.” Over dinner I pressed Ord on these points—in fact I harangued him, out of frustration and from the shame I felt at my younger self. Early on in the conversation, he developed what I’ve come to think of as “the EA glaze.”
It’s difficult to get a man to understand something, said Upton Sinclair, when his salary depends on his not understanding it. And all the more when his self-image does. Ord, it seemed, wanted to be the hero—the hero by being smart—just as I had. Behind his glazed eyes, the hero is thinking, “They’re trying to stop me.”
Not long before then, two hedge-fund analysts in their twenties quit their jobs to create an “effective giving” website, with an aim similar to the website I’d abandoned years earlier. They called it GiveWell. Like me in 1998, the two had no background in aid. But they’d found a charity that gave out bed nets in Madagascar. They checked how much it cost to give out bed nets and how likely bed nets are to prevent malaria. They used a method like Ord’s for measuring “lives saved” per dollar spent, with calculations that unfurled in a 17-row table of precise, decimal-pointed numbers. They put on their website that this charity could save a life for $820.
I added a bit about GiveWell to “Poverty Is No Pond,” asking about the possible side effects of its bed net charity. For instance, had its charity been taxed to support Madagascar’s corrupt president? Had their charity weakened the social contract by supplanting Madagascar’s health service, which had been providing bed nets for its own citizens?
I sent my draft to GiveWell. Its then codirector, Holden Karnofsky, replied he was confident that well-run charities, like the one that gave out bed nets, were beneficial overall—that the benefits to poor people minus the harms to poor people (maybe not the same poor people) was a positive number. I asked whether they’d be willing to mention possible harms on their website every time they asked for money. Karnofsky said it made sense to highlight harms, and they’d make a better effort to make the website clear about what went into their calculations.
That was more than a dozen years ago. Today, GiveWell highlights detailed calculations of the benefits of donations to recipients. In an estimate from 2020, for example, it calculates that a $4,500 donation to a bed nets charity in Guinea will pay for the delivery of 1,001 nets, that 79 percent of them will get used, that each net will cover 1.8 people, and so on. Factoring in a bevy of such statistical likelihoods, GiveWell now finds that $4,500 will save one person.
That looks great. Yet GiveWell still does not tell visitors about the well-known harms of aid beyond its recipients. Take the bed net charity that GiveWell has recommended for a decade. Insecticide-treated bed nets can prevent malaria, but they’re also great for catching fish. In 2016, The New York Times reported that overfishing with the nets was threatening fragile food supplies across Africa. A GiveWell blog post responded by calling the story’s evidence anecdotal and “limited,” saying its concerns “largely don’t apply” to the bed nets bought by its charity. Yet today even GiveWell’s own estimates show that almost a third of nets are not hanging over a bed when monitors first return to check on them, and GiveWell has said nothing even as more and more scientific studies have been published on the possible harms of bed nets used for fishing. These harms appear nowhere in GiveWell’s calculations on the impacts of the charity.
In fact, even when GiveWell reports harmful side effects, it downplays and elides them. One of its current top charities sends money into dangerous regions of Northern Nigeria, to pay mothers to have their children vaccinated. In a subsection of GiveWell’s analysis of the charity, you’ll find reports of armed men attacking locations where the vaccination money is kept—including one report of a bandit who killed two people and kidnapped two children while looking for the charity’s money. You might think that GiveWell would immediately insist on independent investigations into how often those kinds of incidents happen. Yet even the deaths it already knows about appear nowhere in its calculations on the effects of the charity.
And more broadly, GiveWell still doesn’t factor in many well-known negative effects of aid. Studies find that when charities hire health workers away from their government jobs, this can increase infant mortality; that aid coming into a poor country can increase deadly attacks by armed insurgents; and much more. GiveWell might try to plead that these negative effects are hard to calculate. Yet when it calculates benefits, it is willing to put numbers on all sorts of hard-to-know things. (...)
Onstage with Ord that day was a former director of Christian Aid who’d written a massive book called Does Foreign Aid Really Work? This expert tried to persuade Ord that aid was much more complex than “pills improve lives.” Over dinner I pressed Ord on these points—in fact I harangued him, out of frustration and from the shame I felt at my younger self. Early on in the conversation, he developed what I’ve come to think of as “the EA glaze.”
It’s difficult to get a man to understand something, said Upton Sinclair, when his salary depends on his not understanding it. And all the more when his self-image does. Ord, it seemed, wanted to be the hero—the hero by being smart—just as I had. Behind his glazed eyes, the hero is thinking, “They’re trying to stop me.”
Not long before then, two hedge-fund analysts in their twenties quit their jobs to create an “effective giving” website, with an aim similar to the website I’d abandoned years earlier. They called it GiveWell. Like me in 1998, the two had no background in aid. But they’d found a charity that gave out bed nets in Madagascar. They checked how much it cost to give out bed nets and how likely bed nets are to prevent malaria. They used a method like Ord’s for measuring “lives saved” per dollar spent, with calculations that unfurled in a 17-row table of precise, decimal-pointed numbers. They put on their website that this charity could save a life for $820.
I added a bit about GiveWell to “Poverty Is No Pond,” asking about the possible side effects of its bed net charity. For instance, had its charity been taxed to support Madagascar’s corrupt president? Had their charity weakened the social contract by supplanting Madagascar’s health service, which had been providing bed nets for its own citizens?
I sent my draft to GiveWell. Its then codirector, Holden Karnofsky, replied he was confident that well-run charities, like the one that gave out bed nets, were beneficial overall—that the benefits to poor people minus the harms to poor people (maybe not the same poor people) was a positive number. I asked whether they’d be willing to mention possible harms on their website every time they asked for money. Karnofsky said it made sense to highlight harms, and they’d make a better effort to make the website clear about what went into their calculations.
That was more than a dozen years ago. Today, GiveWell highlights detailed calculations of the benefits of donations to recipients. In an estimate from 2020, for example, it calculates that a $4,500 donation to a bed nets charity in Guinea will pay for the delivery of 1,001 nets, that 79 percent of them will get used, that each net will cover 1.8 people, and so on. Factoring in a bevy of such statistical likelihoods, GiveWell now finds that $4,500 will save one person.
That looks great. Yet GiveWell still does not tell visitors about the well-known harms of aid beyond its recipients. Take the bed net charity that GiveWell has recommended for a decade. Insecticide-treated bed nets can prevent malaria, but they’re also great for catching fish. In 2016, The New York Times reported that overfishing with the nets was threatening fragile food supplies across Africa. A GiveWell blog post responded by calling the story’s evidence anecdotal and “limited,” saying its concerns “largely don’t apply” to the bed nets bought by its charity. Yet today even GiveWell’s own estimates show that almost a third of nets are not hanging over a bed when monitors first return to check on them, and GiveWell has said nothing even as more and more scientific studies have been published on the possible harms of bed nets used for fishing. These harms appear nowhere in GiveWell’s calculations on the impacts of the charity.
In fact, even when GiveWell reports harmful side effects, it downplays and elides them. One of its current top charities sends money into dangerous regions of Northern Nigeria, to pay mothers to have their children vaccinated. In a subsection of GiveWell’s analysis of the charity, you’ll find reports of armed men attacking locations where the vaccination money is kept—including one report of a bandit who killed two people and kidnapped two children while looking for the charity’s money. You might think that GiveWell would immediately insist on independent investigations into how often those kinds of incidents happen. Yet even the deaths it already knows about appear nowhere in its calculations on the effects of the charity.
And more broadly, GiveWell still doesn’t factor in many well-known negative effects of aid. Studies find that when charities hire health workers away from their government jobs, this can increase infant mortality; that aid coming into a poor country can increase deadly attacks by armed insurgents; and much more. GiveWell might try to plead that these negative effects are hard to calculate. Yet when it calculates benefits, it is willing to put numbers on all sorts of hard-to-know things. (...)
Think of a drug company that’s unwilling to report data on harmful side effects, and when pressed merely expresses confidence that its products are “overall beneficial.” GiveWell is like that—except that the benefits it reports may go to some poor people, while the harms it omits may fall on others. Today GiveWell’s front page advertises only the number of lives it thinks it has saved. A more honest front page would also display the number of deaths it believes it has caused.
by Leif Wenar, Wired | Read more:
Image: Bill Mayer
