Is It Time to Take the "You" Out of Utilitarianism?

Sam Bankman-Fried was a charismatic fraud. I’m sorry for the investors who he ripped off and the causes which might have gotten cash if he’d better kept up a pretense of being a good person and competent manager of other people’s money. I don’t think the temptation to pedestal charismatic and powerful individuals is unique to Effective Altruism (EA) or philanthropy (see: politics, business, entertainment, sports, religion, war, science, the arts) nor are the risks attached necessarily inherently worse. I’d add that the two billionaires who are actually most closely involved with the EA movement really don’t seem to be much like Sam Bankman-Fried at all (at least to this outsider). But an EA approach that looks at global challenges and asks “what can *you*, individual X, do most to solve them?” may sometimes pedestal the individual in general in a way that I think might have unfortunate side effects.

I want to argue two things. First, the utilitarianism underpinning EA is a moral system—it is meant to guide the behavior of the individual—but also a political philosophy, designed to guide public policy. Second, an emphasis on maximizing the impact of the individual donor can lead to a focus on trying to match a relatively small amount of money with the very most effective causes in a way that asks too much of social science. Reorienting EA around political philosophy and broader public policy provides a way to fix that problem.

Bentham didn’t only see utilitarianism as a guide to private action but a guide to public action, a tool for the policymaker. The same was true of EA’s ‘founding philosopher,’ Peter Singer when he wrote about the 1971 East Bengal Famine. His essay Famine, Affluence and Morality made a strong call for individual donations to relieve suffering on that the grounds that money spent feeding the starving has a massively higher impact on wellbeing than the marginal dollar spent by a middle class Australian on personal consumption. But that is only the start of his complaint about inaction:

 “Generally speaking, people have not given large sums to relief funds; they have not written to their parliamentary representatives demanding increased government assistance; they have not demonstrated in the streets, held symbolic fasts, or done anything else directed toward providing the refugees with the means to satisfy their essential needs. At the government level, no government has given the sort of massive aid that would enable the refugees to survive for more than a few days… governments of affluent nations should be giving many times the amount of genuine, no-strings-attached aid that they are giving now…. giving privately is not enough.”

Singer imagined that Australia might be able to donate 40 percent of its GDP to relief and global poverty reduction, and of course he didn’t imagine Australia taking part alone. If all OECD countries donated 40 percent of their GDP to global causes today, it would add up to $24 trillion a year.

Sam Bankman-Fried’s supposed worth at its height was equal to a little more than two hours of OECD output (or 0.002 percent of OECD wealth). When you are dealing with Bankman-Fried-sized sums that amount to only about $2 per person on the planet, you really need to focus on maximum impact in order to make a dent in global problems. But figuring out maximum impact is hard. You are looking for causes that are big, tractable, neglected, and well-evaluated as such, when the last two rarely go together and the big and tractable can depend hugely on context. This is in part why GiveWell sometimes struggles to hand out cash. There’s whole bunches of big issues that we are nowhere near having a good understanding about how to shift significantly with relatively small amounts of money and may not for a very long time at least (see: economic growth). And a focus on precise attribution of impact fits particularly ill with collective endeavors where too many things are necessary but not sufficient to be able to slice up a result into causal percentages.

I’m not suggesting social science knows nothing, or that cost-effectiveness studies are useless. We know that spending on bed nets is a more cost-effective way to save lives than coronary bypasses. I use up a lot of time arguing against low-impact uses of development assistance. But enough of the world is unknowable enough that searching for ‘maximizing impact’ is likely all too often to turn into an exercise of looking for keys under the streetlight, when there might not be keys anywhere, and the streetlight is a guttering candle. (This may apply in particular to longtermism, which takes knowing there really is a small chance an existential event would be existential and might happen, and you know enough to spend on something that will actually, usefully, reduce the chance it happens rather than make it worse over the very long term. Singer’s partial embrace of population control to ward off global famine in his 1972 article is the kind of mistake even the smartest among us can make in predicting and responding to future threats.)

All of which brings us back to Singer’s core insight. Amongst all of this uncertainty, we can be very sure of the fact that an average dollar spent even not very efficiently improving the lives of people living extreme poverty will have a bigger impact on welfare than spending that same dollar in a rich country. That’s thanks to gross global inequity in income and the immense stock of cheap technologies that can improve quality of life. Combine that insight with a moral system that values all people’s utility at least somewhat equally, and the implications are pretty clear.

If you (first) are a good utilitarian and (second) accept the calculation that the utility impact of income follows a logarithmic curve, a calculation underlying some of Open Philanthropy’s work as well as Lord Stern’s analysis of climate change, you should be in favor of income transfers on a truly awesome scale. Singer suggested maybe 40 percent of rich country GDP, but my back-of-the-envelope calculation a few years ago was that a global utilitarian should be calling for the richest 10 percent of the world's population to face an average redistributive tax rate in the region of 82 percent.

Meanwhile, too much focus on trying to find the most high impact projects possible works against accepting a whole range of projects, many of which are difficult to evaluate with any great precision, are likely to have high returns in terms of global utility. The US PEPFAR program buying HIV drugs is an example: probably not the most effective possible way to spend money to save lives worldwide but saving a huge number of lives at low cost nonetheless. (Note also: it is likely the choice was ‘PEPFAR or nothing’ rather than ‘PEPFAR or deworming,’ or ‘PEPFAR or cash transfers,’ but it is hard to know with certainty). This is not a trap that Singer fell into in 1972. He didn’t go looking for the few projects that would have the most impact, he wanted to go (public policy) big. I think more Effective Altruists should embrace Singer’s ambition. Once you get off the ‘what can I do’ unicycle and join the ‘what can we do’ bandwagon, the need to figure out precisely which high-return intervention pays off the most is reduced: pressure to fund all of them.

What’s to love about the Effective Altruism movement (beyond the fact that it is a bunch of people who are committed to working to make things better for others) is that it thinks about everybody worldwide equally when looking for problems to solve. What’s to love a little less is that, when it comes to solutions, it often looks to the individual working through donations rather than through policy and institutional change (Alex Trembath makes a similar point here). To be very clear, there are lots of exceptions, more exceptions, and more exceptions still, and it is definitely worth noting this conflict of interest exception too: CGD gets funding from Open Philanthropy and Givewell.

The focus on the individual (to the extent it remains) is despite the fact a utilitarian calculus demands change on a level that is beyond delivery even by the whole Forbes rich list. I share the frustration that governments don't fund some incredibly obvious stuff. I wish rich governments were more cosmopolitan and I accept the facts that they aren't and that the world is hideously unequal creates the space for people who earn lots of money in rich places to do lots of good in poor places. But I don’t think EAs should work around the problem of anti-cosmopolitan government. Instead, they should try to fix it. The power of a movement that gets the collective action (and spending) of the world’s richest ten percent a bit closer to Singer’s ideal would be immense.

Of course, this only raises the question, how can you get rich people and governments to move toward your moral framework? The causal chains between effort and impact are certainly longer and more tangled than direct philanthropy (if still shorter and more certain than those involved in some longtermist endeavors). But I’ve suggested working in or around government yourself is one way. Singer suggested living global utilitarian values as an individual probably really helps too. In that regard, I don’t write from a position of moral strength. As a family we are not generous donors despite our income in the global two percent. I still have two kidneys. I hugely admire people who live closer to Singer’s ideal for the individual than I do. But whatever the exact mechanisms and approaches, I think it is time that effective altruism as a movement fully embraced the idea that we are more effective together and directed (even) more attention toward public policy solutions.