Let’s suppose there are 7bn people, and that we spend around $500bn each year helping total strangers. So each of us, on average, gets $140 per year from a stranger who wishes us the best.
Suppose also that you can save a child’s life for around $3500, meaning that child will live an extra 65ish years. That’s a cost of $53 per year of life.
This would give two possibilities:
- International aid and charitable donations are responsible for making a difference, to every single person in the world, 2.6 times as large as the difference between living a normal life and being dead
- Some of that money is not spent with maximum effectiveness.
There might be other explanations that are more complex.
One is that it’s not just humans that matter. There’s a lot of moral uncertainty around the exact status of nonhuman animals, but it’s not unreasonable to suppose that together they represent a pool of consciousness roughly comparable to humanity’s. (At least if you’re thinking of large animals. Insects somewhat complicate this picture: if there are 10^19 insects, then we’re down to asking some tough questions about how bad you would feel about squashing 1.4bn insects versus one human, and I don’t know the answer to that one but it doesn’t matter this time since basically zero charitable spending is going towards helping insects anyway).
And maybe we care about some things like biodiversity or culture or virtues or something, independent of any particular being’s experience of them. I don’t really know what to make of that, but I wouldn’t expect it to dominate all the other factors.
There are also things like space aliens and future digital consciousness and so on, which could be a huge deal but basically zero charitable spending goes on them either so let’s ignore them for now.
All this is getting at: what’s a reasonable estimate of the amount of good we’re actually doing right now?
A very high estimate would be given by the total population we seem to be trying to help, which I’ve given above as humans + largish animals + various intangibles, crudely estimated as equivalent to humans times 2.6ish (that’s a different 2.6 from before, I just chose that for simplicity). But realistically, if all charity were to suddenly stop we wouldn’t all die, nor would our lives get so much worse as to be equivalent to all dying. So maybe 10% of the value in the world is created by charity and aid? Even this sounds like a really high estimate, as it doesn’t include government spending within countries (the social safety net), nor helping your own family and friends.
My original figure suggested that if we were to stop giving in the way we currently are, and all switched to giving to the Against Malaria Foundation instead, we could do the same amount of good for about 10 times as cheap. Which sounds like a big win. But is it true?
Well, of course not. AMF can only absorb so much cash before it runs out of easy opportunities to help people. In terms of room for more funding, GiveWell is quoting numbers in the $20m/year-ish range. That’s one 25000th of our original budget, or one 2500th of the “we could do the same amount for 10x as cheap” budget.
I’m trying to decide if I’m surprised by that 2500 number. Somehow I was expecting it to come out larger, but on the other hand saying “if we had 2500 AMF’s we could solve aid” isn’t very interesting because that just isn’t going to happen.
Aside from insecticide-treated mosquito nets there are maybe 5-10 interventions that could plausibly be as cost effective, and my understanding – though I don’t know what to look up to get a firm source for this – is that after that, the effectiveness kind of tails off exponentially.
So giving 20m/yr to AMF would buy us the first 380,000 life years per year of goodness. Giving 200m/yr to other causes which turn out to be equally effective would get 3.8m of these human-sized units of goodness. The next 200m/yr would buy us substantially less than that because of the exponential dropoff problem, but we don’t know *how* quickly it drops of exponentially so I’m kind of stuck coming up with an estimate.
But suppose there’s a base of 1.06, so that the next 3.8m units of goodness cost us 212m/yr, the next costs us 224m/yr, the next costs us 238m/yr etc. By the time that we’re managing to provide the hundreds-of-millions of goodness units that we feel we might actually be providing, it will take 100ish of these steps and we’ll easily have hit our original $500bn budget.
This is a fairly shallow exponential – it assumes that once you’ve helped 45m people (and by “helped” I mean “do something as dramatic as save the actual life of”), it becomes twice as expensive to help the next 45m. Given many interventions will improve people’s quality of life by 10% or less, so you’re 10%-helping 450m with each doubling – that doesn’t sound unreasonably.
That means it isn’t necessarily going to be easy to do aid substantially better than we are now. Balls.
The reassuring part of this picture is that it suggests people aren’t idiots. If everyone had a mosquito net then that would be great, but that’s not all they need and maybe all the more complex development projects are still needed. It doesn’t prove that aid really is efficient (because I picked a lot of numbers out of my ass), and in reality there are a lot of complaints that aid is not efficient. But it’s at least *consistent* with a story where aid is efficient.
The thing that is bothering me, and has been ever since I first landed on GiveWell’s website, is this: why aren’t these top charities already fully funded? I’m looking at you, Gates Foundation, and to some extend you governments.
My first hypothesis was that GiveWell didn’t know what it was talking about. By now this is looking pretty unlikely. Their research looks extremely thorough and has been poured over by other independent researchers (who are not afraid to kick up a fuss, as with the Worm Wars or Giving What We Can not taking to GiveDirectly).
The second hypothesis, something of an oddball one, is that the big money players want to give effective altruists some toys to play with. We are cute after all, with our earning to give strategies and all of that, and if we aren’t given a few high bang-for-the-buck giving opportunities we might lose interest. In this scenario, the main value of the effective altruism movement is not in its direct donations – at least not those within the developing world health area – but rather as a sort of training ground for the right kind of minds.
A third hypothesis is similar to this. It supposes that everyone hopes that everyone else is going to donate instead of them, and since the EAs are taking care of AMF the other players don’t need to bother with that one and can focus on other things.
It’s going to be hard for me to come up with hypothesis here that aren’t, on some level, annoying.
A fourth hypothesis is that most individuals and organizations do not act like utilitarians. This might mean that the individual donors don’t have that kind of philosophy, or that even if they do at an organizational level they are somehow prevented from spending money in that way.
The pool of utilitarians could be much larger than the pool of self-identified effective altruists (who by and large seem to agree that AMF is where it’s at, at least if you exclude animal charities and things where impact is extremely difficult to accurately quantify). It’s possible that in EA were are locked into a set of epistemic assumptions, which are reinforced by belonging to a fairly close-knit community of people with similar assumptions. Other utilitarians might come with different assumptions, which in practice means they end up liking AMF less and other charities a lot more. (There are different versions of this kind of setup depending on whose view most closely corresponds to reality).
OK I’m going to stop because this is coming out super badly written, but I wanted to get the basic ideas down.