Excerpts:
In 2024, the Against Malaria Foundation funded the distribution of 55 million insecticide-treated bed nets across sub-Saharan Africa. Those nets protected roughly 99 million people and, by the foundation's estimates, prevented about 25,000 deaths, mostly among children under five. The cost per life saved, according to GiveWell's models, was approximately $5,500.
In April 2023, Kenneth Griffin donated $300 million to Harvard, an institution sitting on a $56.9 billion endowment. Both gifts were called "charity." Both donors received praise, tax deductions, and warm feelings. Only one of them was making any serious attempt to help people.
And then Sam Bankman-Fried's conviction in November 2023 and sentencing to 25 years in prison in March 2024 was supposed to be the death blow to this idea. One high-profile fraudster wrapped himself in EA branding, stole $8 billion from customers, and got caught. Therefore, the logic apparently went, it's perfectly fine to keep donating your money to symphony orchestras and university endowments while 610,000 people died of malaria in 2024 alone, three-quarters of them children under five. That inference might be the stupidest non sequitur in the history of moral reasoning.
And I am going to make a claim that sounds hyperbolic but is, if anything, conservative: the difference in cost-effectiveness between the best and worst charities working on the same problem is not 10% or 50% or even double. It is routinely a factor of 100 or more.
Concretely, if you have $5,500, you face a choice. You can donate it to the Against Malaria Foundation and, in expectation, save a child's life. Or you can donate it to a well-intentioned but unmeasured charity and have essentially no idea what happened. Or you can donate it to Harvard, where your $5,500 will constitute approximately 0.00001% of their endowment and change precisely nothing about the institution's operations.
In a study that should haunt anyone who cares about rationality, William Desvousges and colleagues asked people how much they'd be willing to pay to save migratory birds from drowning in oil ponds. One group was told 2,000 birds would be saved. Another group: 20,000 birds. A third: 200,000 birds. A hundredfold difference in scope.
The willingness to pay was $80, $78, and $88 respectively.
So what, exactly, is the alternative? If you reject the idea that effectiveness matters in charity, you must believe one of the following: (a) all charities are equally effective (demonstrably false), (b) it's morally acceptable to choose the less effective option when actual lives are at stake (very hard to defend), or (c) we can't measure effectiveness at all (false in many domains, though true in some). None of these positions is remotely as defensible as they'd need to be to justify the status quo of charitable giving (i.e. doing whatever the hell you want).
I am not saying these things have no value. (Some do; some don't.) I am saying that when we use the same word, "charity," for a $300 million gift to a $56.9 billion endowment and for a $5,500 donation that saves a child's life, we are committing a kind of conceptual fraud. We are telling ourselves that these are the same kind of act. They are not. One is reputational laundering. The other is saving lives. And the tax code treats them identically.
Basically, the question isn't whether effective altruism survived Sam Bankman-Fried. The question is the same as it always was: can you read those numbers and go back to pretending that all charity is created equal?
Posted by lakmidaise12