Some thoughts on Effective Altruism

I’m not going to get into the more contested parts of Effective Altruism, like whether AI safety is an urgent problem, or whether we should accept stronger-than-conventional views about animals’ moral weight.

There’s a hard core of Effective Altruism that has an obvious broad appeal. Nearly all of us agree that preventing malaria deaths and addressing severe poverty are good, important goals. But lots of people (like me!) find themselves really put off by Effective Altruism, even though we may know and respect people in the movement.

There’s a stereotyped conversation that goes something like this:

EA proponent: How can you not see that encouraging people to fund bed nets and so prevent unnecessary malaria deaths is a good thing? EA has saved thousands of lives, how can that not be good?

Person like me: Well, holding all else constant, that is good! I certainly don’t object to people buying bed nets to prevent malaria. But it also strikes me that buying bed nets doesn’t really get at the problems that render so many people vulnerable to malaria in the first place. We need a kind of systemic change that simple philanthropy can’t overcome.

EA proponent: Of course we do! We totally agree, and EA-aligned entities are interested in research towards effective development solutions. Obviously people have different views at a political level, but many EA-aligned individuals fund organizations that advocate for reform and better governance in both the developing and developed world, in order to pursue systemic change. We adopt a “both/and” approach! And surely it is better to help at the margin than not to help!

Yet I walk away still unconvinced. I’m trying to interrogate for myself why. It’s easy to argue that, under a better system, much of what the EA community does would make no sense. Essential goods like public health and basic succor should be ensured by states. In general, in a reasonably arranged society, philanthropy would finance civil society in spheres very local to donors, where they have better information and enjoy the kind of fine-grained feedback that states are ill-placed to perceive and react to. But if Give Well knows how to solve development problems, state actors are perfectly capable of allocating resources according to its recommendations. Philanthropists have no comparative advantage.

But we’ll all agree we don’t live in a better arranged society right now. So shouldn’t we divert resources to the places Give Well directs if our states are underdoing it?

In the dialog above, “person like me” used the mealy-mouthed phrase “holding all else constant”. If we were holding all else constant, the answer would be sure, of course. But I think the problem with Effective Altruism is that in social affairs you can hold nothing constant.

An obvious example is what was once one of Effective Altruism’s core recommendations (recently I think they’ve deemphasized it), “earn to give”. If it really is important to give, but a donor has her own requirements for a reasonable life, then she must increase her earnings some margin above that. If the need for charity is truly urgent, then of course a donor should earn as much as she possibly can, in order to give as much as she possibly can.

But earning large sums of money may not be neutral with respect to larger systemic questions that Effective Altruists generally concede are very important. There are ways of earning high incomes that are directly harmful, like hustling opiates or establishing monopolies. More broadly, industries that pay high incomes work to reinforce and sustain the status quo in which they thrive, rather than to evolve towards more egalitarian systems in which high incomes might be harder to come by, but philanthropic transfers would less necessary.

I think that the current emphasis in EA circles, perhaps as a response to this critique, is to focus on impactful careers rather than earning to give. But does that really address the issue? If, for example, you devote yourself sincerely to a great charity that does development work, and you allow yourself to be underpaid relative to what you might earn elsewhere, isn’t that laudable? At an individual level, it may be. But the crucial flaw in much EA-reasoning is that social outcomes are not arithmetic aggregations of individual-level scores. If you work for a development charity, the good work you do still depends on the high earners you forewent to become. Your work must be tailored to avoid approaches or choices that would upset the system that elevated them. Again, you are deeply tethered to the systemic status quo. If your organization is ambitious to increase its scale and thrive, it will likely be called upon to promote and reinforce questionable arrangements. Welcome to the World Economic Forum!

Indeed, from the perspective of “person like me”, philanthropy and the plutocracy that, in dollar-weighted terms, must dominate it is so interwoven with all that is systemically wrong, trying to do net good in the world via donor-financed institutions is likely to be counterproductive. The trade-offs are hard! I suspect GAVI, chartered by the Bill and Melinda Gates Foundation, has saved many lives relative to a counterfactual where everything is the same as it is now except Gates never intervened. But Gates is famously supportive of an extremely strong and universal intellectual property regime that, from the perspective of “person like me”, helps condemn the developing world to dependence and poverty (and continued disease), that contributes to class stratification in the developed world, that could be replaced by better means of encouraging innovation. (During the COVID crisis, GAVI was notionally supportive of waiving intellectual property restrictions, but in practice the organization was understandably circumspect.)

If I am to restrict myself to “evidence-based” behavior, then surely it is a good idea to donate to, or even work for, GAVI! The evidence that it has done good things is real. But we can only gather evidence about circumstances very near to the status quo. Since we can’t run persuasive social science experiments about a world in which many variables would be different from the one we actually inhabit, “evidence-based” policy is a prescription for local optimization, fine-tuning. If we agree that the key source of the problems we wish to address are systemic, but we can also make clear marginal improvements to those problems at cost of supporting institutions that help reinforce the bad status quo, how does a good rationalist weigh the trade-off? You may be suckered in the way that a certain narrow economism tells you it’s irrational to vote. At an individual level, your vote almost certainly won’t make a difference, but it will certainly cost something in time and convenience. Similarly, your withholding contributions won’t lead to a new intellectual property regime or alternative forms of state action, while your contribution to GAVI really will do a bit of good at the margin. But that sort of individual-level rationality doesn’t compose to social outcomes. You should in fact vote. It might be wise to circumvent donor-financed philanthropy as a key element of an ultimately harmful constellation of institutions, rather than contribute to it, despite the real good it also does.

At an individual level, that’s a genuinely hard judgment call. But Effective Altruism is a social movement, not a collection of independent idiosyncratic choices. At a social level, to “people like me”, it is not so hard. A social movement should not reinforce reliance upon plutocratic finance. It should not encourage or help whitewash extractive wealth generation. It should not prod people into careers where they must adopt the perspectives of those who most benefit from systems that we seem to agree must fundamentally change. At an individual level, the tradeoff between the costs and benefits of voting may not seem clear, but it would be obviously dumb for an interest group to advise its own members not to vote. Effective Altruism is a movement of idealistic, unusually educated, often affluent and socially influential individuals. Is integration of this group into the philanthropic status quo a good choice?

Note that fundamentally, the problem here isn’t philanthropy. It’s discretionary donor finance. The same critiques apply to think tanks, activist collectives, publications, research organizations, and Congressmen. In theory they ought not apply to venture capital, as allocation should be based on objective valuation of prospects rather than investor discretion. In practice, though, valuation of startups is so dependent on contestable assumptions that the industry is best conceived as donor finance plus a lottery ticket. If you think plutocracy is an important component of our systemic problems, then working through institutional forms where relative advantage will be decided by plutocratic discretion seems less than ideal. Plutocratic money is fine. It’s as green (much greener) than anyone else’s. But it must be decoupled from the discretion.

I love Effective Altruists. They are people devoted to thinking carefully and clearly and creatively, and serving the greater good. But as a social movement, I think they’ve erred. They’ve succumbed to fallacies of composition on both the input and the output side. On the input side, they imagine total contribution can be measured as a sum of individual contributions, so it’s sufficient to ask people to maximize their own impact, holding the rest of the world constant. That’s partial equilibrium thinking, and mistaken in their domain. On the output side, well, the usual critiques of “aggregate utility” as a welfare measure apply. I’d love to see a larger menagerie of social welfare functions, as formalizations that try to approximate competing groups’ actual values (which we might compare, contrast, perhaps even average), or that capture our status quo revealed preference as a society. Summing putatively identical concave individual welfare functions is great for expressing egalitarianism as a social value, and it usefully captures individual risk aversion. But it doesn’t actually express our collective aspirations.

There is no “true” or “scientific” social welfare function. We have to make it up. There is no objective way to quantify how much good we are doing in the world. And even if we settle upon some measure that we acknowledge merely expresses our own particular values, we will find that our means are not separable from our ends. I think many Effective Altruists have not fully grappled with contradictions between the means that they adopt and the ends they hope to further.

Postscript 1: For the most part this essay has been critique. I try here at interfluidity to be constructive. If EA is getting it wrong, how should people, including the well-meaning very rich, actually work to do good? I’ll write more on this in a future post, soon, I hope! Also, I have presumed more than argued that plutocratically dominated, donor-financed philanthropy is important contributor to what is undesirable about the status quo. For now I’ll send you to Anand Giridharadas and Rob Reich, but maybe I’ll try to offer my own summary of the case. If I’m going to be constructive, it might help to first clarify the pitfalls to avoid. Kelsey Piper offers a measured take on billionaire philanthropy, but I perceive harms that she doesn’t discuss.

Postscript 2: I’ve been percolating this for a while, but it’s probably not coincidence that I’m getting around to it now, in the wake of the astonishing collapse of FTX, the crypto exchange led by Effective Altruist Sam Bankman-Fried. My original take was that FTX died like highly leveraged financial institutions that also take on risk for profit often die, so it would be unfair blame the shiny new thing for an old, common story. It turns out the FTX’s case is extraordinary even within its sordid genre, and a case can be made that SBF’s really immoderate utilitarianism contributed to it. Even if that is so, I don’t think SBF is representative of Effective Altruism, and it strikes me as cheap to blame the broader movement for SBF’s sins.

Disclosures: In the above (and here again!) I have nepotistically linked my brother-in-law’s book. I am an uncompensated board member and officer of a small, donor-financed nonprofit.

Update History:

  • 16-Nov-2022, 8:10 p.m. EST: “…philanthropy and the plutocracy that, in dollar-weighted terms, must dominate it is are so interwoven…”; “…at cost of supporting institutions that help reinforce a the bad status quo…”

13 Responses to “Some thoughts on Effective Altruism”

  1. Steve Roth writes:

    “we can only gather evidence about circumstances very near to the status quo.”

    This is a very crux IMO. Two examples:

    We have no idea what economic effects might emerge from a $20 or (the horror) $30 U.S. minimum wage. Because we’ve never tried it.

    Ditto “national debt” in the 200%++ of GDP range. (Except…Japan…which doesn’t seem to be crashing and burning…)

  2. Carol writes:

    EA is bad because it encourages inordinate wealth aggregation before the “philanthropy”. Wealth aggregation is inherently bad for the nation, because it concentrates power in the hands of the few, thus voiding the ability of the non-wealthy to direcf funds to actual immediate good causes. EA encourages the idea that somehow or other, the wealthy are superior in ability to choose what the rest of us need. Even so called research into “good philanthropy” isn’t very effective in the short term, and probably not in the long term, if only because it diverts resources to game playing vs immediate needs. How many children will starve, how many women will be denied health care, how many families will be denied reproductive choices, because the rich are spending money doing games to “”figure out how to help them”?

    EA, from all I can see, is an immense ego trip for rich people.

  3. Sorry if this is spammy, but I shared some thoughts back in August that I think intersect well with what you’re saying here.

  4. Andy B writes:

    This is the post I was hoping for when I criticized your earlier satire. Thank you.

  5. David writes:

    This is so elegant and correct: “ They’ve succumbed to fallacies of composition on both the input and the output side.”

    Another way to put it might be: if I told you that you could create a social movement of tens of thousands of the wealthiest, most influential people on earth, all committed to changing the world in a positive way, what would you want to be the ideology of that social movement?

    Clearly, you’d want it to be something more ambitious than EA.

    Of course if it were more ambitious you’d lose people, and maybe not just on the margin. And that raises another critique of EA which is that it’s a successful at gaining a certain type of (wealthy, influential) adherent because it sounds and feels radical but is in fact totally non threatening to the system that produced them.

    A contrast might be with YIMBYism, which probably has a lot of overlap with EA adherents, but which at least looks to systemic change of a certain kind and recognizes that in a democratic society, a group of people can exert more leverage than a bunch of individuals.

    Given EA’s myopic individualism, maybe we should see its cramped lack of imagination as a result of the past few decades of financialization and (I dislike the word but) neoliberalism. Is it radicalism for reactionaries? That’s too harsh: it’s radicalism for fans of the status quo.

  6. Detroit Ddan writes:

    I agree that with the thrust of the post that systemic change is needed. I wonder whether that is because our current Western status quo is exceptionally bad?

    My take is informed by The Revolt of The Public and the Crisis of Authority in the New Millennium, by former CIA analyst Martin Gurri. The Internet has exposed a lot of the lies promulgated by the ruling class. Probably most societies at all times around the world are corrupt and have elites promulgating good works which incidentally bolster the status quo. In our relatively open society in the Internet age, these are exposed to an unprecedented extent.

    Having said that, I do believe that our current Western status quo is exceptionally bad, and the FTX failure is one example. As the predominant global empire, the U.S. centered West has an extraordinary ability to propagate propaganda and related disinformation. Our powerful institutions are magnets for vested interests which have attended cancerous proportions, threatening global peace, health, and prosperity.

    So my opinion is that systemic change in the current West should take precedence over other worthy charitable goals.

  7. Brian Slesinsky writes:

    Why is voting good but deeper participation in the political process bad? This sort of rule seems too broad to be accurate. It depends what you do. Voting is only good if you vote for good things. Participating in politics could turn out well or badly depending on what you do (and probably a good deal of luck).

    Similarly, it seems like preferring “systemic change” could easily become an excuse for remaining powerless but (mostly) pure. An excuse to do nothing, or to do ineffective things that are likely to amount to nothing. Surely it matters what kind of systemic change you’re advocating for. These causes could be better or worse than bednets.

    Similarly, maybe a problem with the label “effective altruism” and associated discussion is that it’s too abstract and philosophical? Someone who wants to advocate for GiveWell as an excellent charity evaluator could do that directly. Similarly for other charities that you think are good and deserve more funding, whether they fall under “systemic change” or not.

  8. SivaDancer writes:

    I follow your thoughts weekly, but don’t often respond. IMHO, your best point here is that “… the crucial flaw in much EA-reasoning is that social outcomes are not arithmetic aggregations of individual-level scores.” I agree. But what are the implications of that statement?

    This flaw is not limited to EA-reasoning; it is deeply embedded in much reasoning used to explain and/or justify all sorts of systemic designs. In the rest of this essay (and in many of your other essays), you seem to still be trying to work within an intellectual system that falls prey to the same flaw, without realizing it.

    So, if you are right that social outcomes are not arithmetic aggregations of individual ‘scores’ (and I believe that you are right), then what are they? Would it be better to assume they are exponential aggregations? (I suspect not). Is it even reasonable to reduce social outcomes to mathematical formulae using existing analytical methods, given the complexity of inputs and outputs?

    What I hear you saying is that it is becoming increasingly clear that a model that assumes social outcomes are arithmetical aggregations is not really accurate or useful. What assumption would you use in its place? My untutored intuition suggests that, given their complexity, perhaps an assumption that social outcomes generate properties like emergence would be more helpful.

  9. Detroit Dan writes:

    Timely article in New York Magazine: Is Effective Altruism to Blame for Sam Bankman-Fried?

    This could have been me 40-50 years ago. I wanted to work in a non-profit to make the world a better place, and I probably could have convinced myself that becoming a billionaire would enable me to save the world. But the status quo is such that disaster results from exactly the kind of good intentions discussed here by Steve Waldman.

    The Ukraine war is exhibit A of how the cancerous system leads will intentionded people to commit monstrous crimes. The military-industrial-intelligence-media complex (MIC for short) has so much power and money that it is hard to be successful in life with working for it one way or another. Thus, SBF eventually started seeing support for this war machine as being an altruistic endeavor. He probably sincerely believed, due to propaganda, that helping Ukrainian Nazis was a good thing to do. After all, Big Tech and Big Finance dominate civilized discourse in the Empire, and it’s not easy to see how rotten the status quo is beneath the surface. But rotten it is, as being exposed by global powers outside of the West, as well as by failed technologies and corrupt economics.

    Thus, Effective Altruists such as SBF are losing the battle to make the world a better place because they cannot see beyond the cancerous MIC bubble. Ironically, these utilitarian visionaries are unable to see the Empirical truth (pun intended) in areas such as technology, finance, economics, politics, and military strength. As I said at the top, that could easily be me. It’s up to all of us to change the culture so that we recognize more clearly our capabilities, and channel our energy and resources into realistic endeavours.

  10. Detroit Dan writes:

    Brian Slesinsky makes a good point:

    it seems like preferring “systemic change” could easily become an excuse for remaining powerless but (mostly) pure. An excuse to do nothing, or to do ineffective things that are likely to amount to nothing. Surely it matters what kind of systemic change you’re advocating for. These causes could be better or worse than bednets.

    Absolutely right. Doing nothing is of course better than doing harmful things. Realistically, there’s not much that the average person can do. Remaining “(mostly) pure” is not a bad start. I believe in democracy — that people should have a say in how their community or country is run. For this to work, we need to talk to one another and try to learn the truth about societal issues.

    Donald Trump is popular because he TALKS a good game about challenging the status quo. See No, New York Times, You Don’t “Deserve Better” Than Donald Trump. It’s said that U.S. style democracy is superior to other systems because we give people a chance to throw the bums out. But we don’t, as evidenced by the aristocracy trying to delegitimize the Trump presidency and movement without recognizing its democratic genesis.

    Systemic change isn’t easy. Even when one becomes a billionaire, like some Effective Altruists, there are no easy solutions. The best we can do is to play our roles honestly and whole heartedly. Thus, we are here reading Steve’s blog where he comments constructively on politics and economics. We become smarter and better citizens and spread the word to family and friends.

    Slesinsky again:

    Similarly, maybe a problem with the label “effective altruism” and associated discussion is that it’s too abstract and philosophical? Someone who wants to advocate for GiveWell as an excellent charity evaluator could do that directly. Similarly for other charities that you think are good and deserve more funding, whether they fall under “systemic change” or not.

    Again, I agree. Effective altruists have done some high profile dumb things. They are in general, as I pointed out in my previous comment, deluded into thinking that money can solve social problems. It takes intelligence in addition to money. So Steve is right to point out that they are falling short, and you are right to point out that with more intelligent ideas they could be useful. We need specific ideas and monetary support to improve a failing system of governance.

  11. Detroit Dan writes:

    SivaDancer writes:

    IMHO, your best point here is that “… the crucial flaw in much EA-reasoning is that social outcomes are not arithmetic aggregations of individual-level scores.”… What assumption would you use in its place?

    Each of us has to start where we are in terms of what we can accomplish to make the world a better place, while also making a living and thriving as individuals and families. If we each do our best balancing these objectives, the world will be better place. The problem is not so much that good individual choices don’t aggregate to good social outcomes, but rather that our broken system encourages bad individual choices. The status quo is a cancer that benefits people who work to maintain the status quo. The empire of lies has grown out of control as a result, with overall social welfare disconnected from how people can make a living.

    SivaDancer may be on the right track in that the emergence of civilizational consciousness of some sort could provide a check on the unrealistic ideas of some powerful few individuals. In practice this might take the form of authoritative institutions that are above day-to-day politics yet dependent upon societal success for their continued existence over the long term. Frankly, Russia and China seem to be doing much better in this regard than the West. Western civilizational consciousness is embodied in institutions which are mired in corruption, and there is no authority above the fray which can put things right.

    What can we do to put things right? There’s space for effective altruism, for promoting systemic change, and for civil disobedience. The effects of such behavior won’t be additive, but they can promote emergence of more effective civilizational consciousness.

  12. davidly writes:

    Whenever I’m not suffering from an inferiority complex, I think I think pretty highly of myself. I’d hate to think of how highly I’d think of myself as a ponzi-scheming philanthropist.

  13. Detroit Dan writes:

    If you had billions to make the world a better place, where would you do with your money? I’d try to invent a cure for our cancerous national security state. Raise (inter)national consciousness through greater transparency. Maybe Wikileaks was on to something?