Spreading the Wealth Around

Conventional effective altruism wisdom is that however much money you are donating, you should give 100% to the best charity, because it is the best.  I think that is one perfectly good choice among several.  Until recently my explanation was “the estimated difference in effectiveness between these charities is many orders of magnitude smaller than the confidence interval of the estimates, so they are functionally the same, so I might as well do what makes me happy.”  Scope insensitivity makes donating $n to two charities twice as satisfying as $2n to one charity.  I would have given to several more charities this except my job matches donations by hand and the admin has other shit to do.  But recently I realized it is more complicated than that.


I spend a lot of time reminding people that estimates of genetic influence and heritability are only valid for the environment in which they are measured.  The same is true for charitable interventions.  The effects of any one intervention depend on the environment, which depends in part on other interventions.

Free condoms and instruction on their use doesn’t appear to make a big difference in teen pregnancy- but that study measured a single free condom program that existed in an environment with lots of existing programs.   Anyone who wanted condoms already had them.  That doesn’t mean such a program wouldn’t be useful in a population with no knowledge of condoms.

Interventions are synergistic.  Tostan’s educational programs won’t do much for anyone who died of malaria, but I’m also not excited about saving infants from death only to spend their entire lives in misery.  We could run around funding whichever need is most dire at any given moment, but organizations are costly to set up and a lot is lost when they disband.  Keeping the operational capital of the second and third best things live will let us react faster when we hit diminishing returns on the first.

And that’s when we know what to do.  Tostan and even GiveDirectly are very much works in progress, and because Tostan is so complex and culturally specific it’s slow to scale.  GiveDirectly can scale much faster, but too fast and corruption will become an enormous problem.  If we want those solutions ready to go when disease and nutrition are solved, we have to work on them now.  And that’s before taking into account the synergies.


100 small donors each dividing their donations among 5 charities is better for the charities than 20 small donors giving 100% to five different charities, because it’s more stable.  If a minuscule change in numbers causes half your donors to abandon your cause (and maybe come back two years later), your funding will swing wildly.  This is terrible for operational capital.

Risk of Neglect

And that’s assuming favorites are properly distributed.  If there’s an organization or cause that’s everyone’s second choice it should probably get some money, but under a favorites only system it never will.  My source at Effective Altruism Outreach says that’s exactly what the recent EA survey showed is happening to metacharities*; everyone has their favorite real cause, and then likes metacharities.  I’ve increased my estimate of metacharities’ value recently**, so I now think they’re underfunded, so this seems bad.

If you’re a very large donor none of this applies to you because you’re in a position to change charities’ behavior rather than just react to it.  If you’re a small donor who’s happiest donating to the charity with the single best numbers, keep going, I don’t think you’re doing any harm at the scale you currently operate on.  But if you’re like me or Brian  and will have more fun spreading your donations around, I think you’re doing a good thing and shouldn’t change.


*The publicly accessible survey summary doesn’t give numbers for individuals’ second choices.   This is still a good example if it’s not literally true so I haven’t bothered looking up the numbers, although I should do so before I actually donate to metacharities.

**I’ve also increased the number of friends I have working at metacharities.  This means I hear about the really cool stuff they do that can’t be publicized, but also that I’m more likely to be suffering from a halo effect or cognitive dissonance or simply a desire for my housemates to have more money because hiring a cleaner would make everyone’s lives easier.