Cross-posted to my blog.
The common claim: Unlike more speculative interventions, GiveWell top charities have really strong evidence that they do good.
The problem: Thanks to flow-through effects, GiveWell top charities could be much better than they look or they could be actively harmful, and we have no idea how big their actual impact is or if it’s even net positive.
Flow-Through Effects
Take the Against Malaria Foundation. It has the direct effect of preventing people from getting malaria, but it might have much larger flow-through effects. Here are some effects AMF might have:
- Increasing human population size by preventing deaths
- Decreasing human population size by accelerating the demographic transition
- Increasing people’s economic welfare, which causes them to eat more animals
- Increasing people’s economic welfare, which causes them to reduce wild animal populations
Increasing population might be good simply because there are more people alive with lives worth living. Accelerating the demographic transition (i.e. reducing population) might be good because it might make a country more stable, increasing international cooperation. This could be a very good thing. On the other hand, making a country more stable means there are more major players on the global stage, which could make cooperation harder 1.
Some of these long-term effects will probably matter more than AMF’s immediate impact. We could say the same thing about GiveWell’s other top charities, although the long-term effects won’t be exactly the same.
Everything Is Uncertain
There’s pretty clear evidence that GiveWell top charities do a lot of direct good–but their flow-through effects are probably even bigger. If a charity like AMF has good direct effects but harmful flow-through effects, it’s probably harmful on balance. That means we can’t say with high confidence that AMF is net positive.
Among effects that are easy to document, yes, AMF is net positive (maybe). Maybe we could just ignore large long-term effects since we can’t really measure them, but I’m uncomfortable with that. If flow-through effects matter so much, is it really fair to assume that they cancel out in expectation?2 We don’t know whether AMF has very good or very bad long-term effects. I tend to think the arguments are a little stronger for AMF having good effects, but I’m wary of optimism bias, especially for such speculative questions where biases can easily overwhelm logical reasoning; and I think a lot of people are too quick to trust speculative arguments about long-term effects.
So where does this leave us? Well, a lot of people use GiveWell top charities as a “fallback” position: “I’m not convinced by the evidence in favor of any intervention with potentially bigger effects, so I’m going to support AMF.” But if AMF might have negative effects, it makes AMF look a lot weaker. Sure, you can argue that AMF has positive flow-through effects, but that’s a pretty speculative claim, so you’re not standing on any better ground than people who follow the fairly weak evidence that online ads can cost-effectively convince people to eat less meat, or people who support research on AI safety.
I don’t like speculative arguments. I much prefer dealing with questions where we have concrete evidence and understand the answer. In a lot of cases I prefer a well-established intervention over a speculative intervention with supposedly higher expected value. But it doesn’t look like we can escape speculative reasoning. For anything we do, there’s a good chance that unpredictable long-term effects have a bigger impact than any direct effects we can measure. Recently I contemplated the value of starting a happy rat farm as a way of doing good without having flow-through effects; but even a rat farm still requires buying a lot of food, which has a substantial effect on the environment that probably matters more than the rats’ direct happiness.
Nothing is certain. Everything is speculative. I have no idea what to do to make the world better. As always, more research is required.
Edited to clarify: I’m not trying to say that AMF is too speculative, and therefore we should give up and do nothing. I strongly encourage more people to donate to AMF. This is more meant as a response to the common claim that existential risk or factory farming interventions are too speculative, so we should support global poverty instead. In fact, everything is speculative, so trying to follow robust evidence only doesn’t get us that far. We have to make decisions in the face of high uncertainty.
Some discussion here.
Notes
-
I recently heard Brian Tomasik make this last argument, and I had never heard it before. When factors this important can go unnoticed for so long, it makes me wary of paying too much attention to speculation about the far-future effects of present-day actions. ↩
Why would we put more weight on current generations, though? I've never seen a good argument for that. Surely there's no meaningful moral difference between faraway, distant, unknown people alive today and faraway, distant, unknown people alive tomorrow. I can't think of any arguments for charitable distribution which would fall apart in the case of people living in a different generation, or any arguments for agent relative moral value which depend specifically on someone living at the same time as you, or anything of the sort. Even if you believe that moral uncertainty is a meaningful issue, you still need reasons to favor one possibility over countervailing possibilities that cut in opposite directions.
If we assign value to future people then it could very well be an exceptional way to make the long run worse. We don't even have to give future people equal value, we just have to let future people's value have equal potential to aggregate, and you have the same result.
Morality only provides judgements of one act or person over another. Morality doesn't provide any appeal to a third, independent "value scale", so it doesn't make sense to try to cross-optimize across multiple moral systems. I don't think there is any rhyme or reason to saying that it's okay to have 1 unit of special obligation moral value at the expense of 10 units of time-egalitarian moral value, or 20 units, or anything of the sort.
So you're saying that basically "this action is really good according to moral system A, and only a little bit bad according to moral system B, so in this case moral system A dominates." But these descriptors of something being very good or slightly bad only mean anything in reference to other moral outcomes within that moral system. It's like saying "this car is faster than that car is loud".
Carl's point, though not fully clarified above, is that you can just pick a different intervention that does well on moral system B and is only a little bit bad according to A, pair it off with AMF, and now you have a portfolio that is great according to both systems. For this not to work AMF would have to be particularly bad according to B (bad enough that we can't find something to cancel it out), rather than just a little bit bad. Which a priori is rather unlikely.