Earlier this year, Bernadette Young and Peter Hurford recommended that I make a post like this once I had enough karma:


At an event hosted by the DC Effective Altruism meet-up group, we discussed if any organizations with large amounts of money, such as USAID, the Bill and Melinda Gates Foundation, or the World Bank, are actually more effective in choosing causes to benefit than GiveWell.


Many are skeptical of this, because it is easy to point to examples of failed projects, naivety, and bad funding priorities. However, these organizations seem to be pursuing some very promising projects, and there may be levels of effectiveness that are only feasible with such scale.


One argument was made that when you are a large organization, you have the ability to start a new initiative entirely and also guarantee that there is enough overall funding, which small individual donations can't. New organizations have a chance to be far more effective and aren't replaceable in many cases, so it might be the case that it is good for the rich to fund these instead of GiveWell charities.


Another argument was that if these wealthy groups did fill all the room for funding of GiveWell's top charities (they easily could), the top charities would naturally become less effective and not do as well at attracting other funding (I don't think it matters who funds a charity except if it changes the net amount donated to effective causes). The part of this argument that seems to actually matter is whether or not effectiveness will be reduced to the point that other causes would become better to fund, before full funding is reached.


An idea in common between these arguments seems to be that marginal contributions and large sum contributions will differ in cost effectiveness depending on the circumstances of a cause or charity.


If a cause requires $1,000 for its intervention, and you donate $1, you don’t have any effect unless $999 was or will be donated by others so your dollar bridges the gap to the next intervention. If others donate as much or more than is needed, your donation is replaceable. This is part of the problem of effectively coordinating small donors.


If the same cause already has $999 and you donate $1,000 the intervention will happen, but $999 might not be able to be spent on an effective cause. This is the inefficiency of donating too much, and possibly giving an organization room to be less cost effective with its interventions. The room for funding has been surpassed.


When there is high variance in the amount that may be donated, it pays to have a charity where each small contribution can make a difference. With each contribution making a difference, it becomes possible for many different levels of funding to be cost effective. The charities that GiveWell recommends seem to fit this pattern: they have very inexpensive interventions that are scaleable. Returns on investment are close to linear, even if there are diminishing returns.


On the other hand, with things like research, there may be larger brackets for what amount of money can make a difference. Once higher levels of funding are reached there is likely to be better cost effectiveness. If the next most meaningful way for a group to do better is to hire a new employee, they may need a lot of money. Meeting and not exceeding the need of such organizations requires coordination that small donors are unlikely to achieve. This may be where the rich have a comparative advantage. Returns on investment may be more like a step function, at each new threshold there is a sudden jump in effectiveness.

I don't have a strong opinion on these arguments, but I do think this is a good place to discuss them. Should what large aid groups focus on influence what we focus on? Should we work toward changing these organizations to be more effective if they aren't already? Should the amount of money you earn actually change where you donate?


(I will post some examples of what the Bill and Melinda Gates Foundation has looked into in the comments below, as well as an earlier comment from Peter Hurford responding to an earlier version of this post. If there is any mathematical naivety in the arguments presented above, please point it out. Some of the reasoning about marginal versus large sum donations falls into a similar category as the Sorites paradox, and I imagine it might be the sort of confusion you can defeat with math. For example, taking limits in math is a way to deal with Zeno’s paradoxes)






More posts like this

Sorted by Click to highlight new comments since: Today at 1:21 PM

Here are some examples of interesting things the Bill and Melinda Gates Foundation has looked into:

(1) Inexpensive lasers which only target the mosquitoes that can carry malaria

(2) Producing inexpensive meat substitutes that actually taste like meat

(3) Malaria vaccines and education about children’s health

I think there's a real question here and it is plausible that large donors do better.

I will point out that if it's correct that there are better returns from scale the conclusion probably isn't that small donors are best giving to charities like the GiveWell recommendations while large donors can do better. Instead small donors would be best getting to a position where their money acts like that of large donors.

Here are three possible mechanisms for that: (i) Give money to large donors, such as the Gates foundation; (ii) Coordinate with others, pooling money so that many small donors act as a single large donor; (iii) Gamble, exchanging certainty of a small donation for a small probability of a large donation.

Yeah if there are increasing returns to scale we can just head to the Blackjack table and have some excitement with our do-gooding!

How much of the large donor effect is due to leadership and catalysing teams to solve problems etc. by brand/authority/charisma/reasoning etc. rather than necessarily the funding behind it? I think we're already doing quite a lot of this, but there's probably room to stimulate more front line promising initiatives?

I'm especially interested in coordinating with others, something like a crowdfunding site with prescreened projects would be great.

Otherwise, gambling's cool too :)

One point to consider is that if there really are large gains from scale, small donors can make risky bets to buy a chance of getting large, e.g. if large donors can get triple cost-effectiveness, then it would make sense to go to Las Vegas, and then if you win buy out-of-the-money options, to convert a small amount of money into a chance at a large amount. I have blog post discussing this.

In the same vein, you can give to big donors, or coordinate with them to let them allocate your donations.


I think there is a danger in overlooking the small scale which sometimes cannot be readily duplicated or enlarged. A family friend, with lots of help from my mom, started an orphanage and school in Zambia. Projects like that are highly dependent on the individuals involved, their priorities, and their honesty. There is not another Kathi in the world and I would expect nearly everyone else to be less effective at trying to accomplish what she has done. Now, if you just want to count bodies, her project is not the most effective at saving lives per dollar. However, just as in nature, one must consider quality as well as quantity. Three of her orphaned formerly street children were in the top five on the national test for admission to high school the first year any of her kids took it. Her goal is to get as many through university as possible, but she also teaches farming and sewing and other skills. She could save more lives, but she wants to give Zambia the educated population it needs to thrive on its own.

A friend made the argument to me yesterday that large organizations have high costs to finding such investments, since people may try to scam them or just compete for their funding by exaggerating. As an individual, who already has knowledge of these small scale circumstances, you can spend time and money on such small projects without facing similar risks. This might be a comparative advantage for small donors who are good at evaluating persons working on such projects.

I agree that quality matters, but it does help accountability for progress to be measurable. Do you know if there are any proposed improvements to measurements like QALYs? If the EA movement proposed a new such measurement, backed by good data, it might be a very valuable contribution. Depression is now causing a very high DALY burden, so it seems plausible that the most effective interventions may be in reducing depression, rather than saving lives.

Yes. But its about how easy something is to treat also?

Small minnows definitely have some advantages (small is beautiful has a lot of arguments in this direction)

The issues is that we haven't found little charities that are making a difference as well / cheaply as SCI/AMF.

I think that actually, scale is very important unless you're doing knowledge based stuff. Economies of scale exist, and are often more important than the kinds of advantages small players have?

You raise an interesting point. If I have a quarter, what's the best I could buy with it? A chewing gum, perhaps. Certainly not a car. But if I successfully encouraged millions of others to do as I did (and buy chewing gum), I possibly could have convinced them to cooperate into buying a car instead.

If the EA movement had not hundreds or thousands of members, but millions or billions, would giving 10 percent of their income still be the best that they could do? Lets say that the EA movement grows enormously, and a decade from now comprises more than half the electorate of the US. Would the EA organisations recommend them to all vote for a specific candidate?

If that scale was achieved, I think we would be able to make a political party. When you have a large amount of people trying to be effective, the actions that we consider effective now may be the type that are replaceable in such an environment.

Earlier from Peter Hurford:

“I think I recall GiveWell agreeing that some of the Gates Foundation work is higher impact than GiveWell top charities, but it's already exceeded room for more funding (because of the Gates Foundation). Some of the vitamin fortification stuff seems like good examples, though GiveWell has recently recognized some vitamin fortification charities as standout charities.”

some of the Gates Foundation work is higher impact than GiveWell top charities

Hasn't GiveWell also said that large orgs tend to do so many different things that some end up being effective and others not? Does this criticism apply to the Gates Foundation?

I think GiveWell was initially much more critical of the Gates Foundation than they are today. Perhaps this is because during the OPP they found that what Gates does is (1) very difficult and that (2) Gates (or other foundations/govs) had already funded many of the most promising opportunities.

It's probably best to evaluate the GF like a venture capitalist rather than on a project by project basis.

I suspect that such a criticism does apply. I remember a friend criticizing the way the Bill and Melinda Gates Foundation funded charter schools and scholarships as ineffective. You can see some of the grants they have awarded here.