In favour of finite meta

A while back I wrote in favour of fundraising for the best charities as an effective cause. Alyssa Vance, who also happens to be the President of MetaMed, which Robin critiqued recently, wrote a response titled ‘Against Infinite Meta’ strongly disagreeing. I think her points are, in order:

1. The fact that on average fundraising is effective today, doesn’t mean that additional fundraising ‘on the margin’, would also be effective;

2. Fundraisers for effective charities won’t do better than other fundraisers in general;

3. The reason high spending on fundraising is unappealing to donors is because funders want confirmation that a charity has a useful project on which to spend the money they raise, rather than because spending more money on fundraising would be bad for the charity sector as a whole;
   

4. There’s plenty of money around, so what matters is finding useful things to do with it, not advocacy about how it should be spent;

5. Starting a fundraising group is harder than I imagine.

I find her evidence for 2 and 5 reasonable enough and agree 3 is a better explanation, though for a different reason than the one she gives [1]. I just want to take issue with 1 and then 4.

While the high average effectiveness of fundraising doesn’t translate into a high ‘marginal’ effectiveness of fundraising for the charity sector as a whole, it probably does for any individual charity. The model I have in mind here is a fixed pool of donations D and a total fundraising spend F, which is some fraction of D. Different charities choose to contribute to F, and the share of D that each charity receives is proportional to their share of F. In this scenario, any extra money that one charity raises through fundraising comes wholly at the expense of other charities. But on the margin, their fundraising ratio is still approximately D/F, which is the average cost-effectiveness of fundraising. This is too optimistic insofar as each organisation will gradually run out of donors who can easily be convinced to fund their specific cause. But when you are as small as GiveWell’s recommended charities – which receive only millions of dollars each year – I don’t think that will be a significant effect. It is also too pessimistic, insofar as total donations to the charity sector could be increased through extra fundraising.

Regarding point four, there are a range of approaches one can take when trying to help others: the least meta would be to start or fund a project that used effective methods to directly help people (let’s just call this charity); another would be to try to identify organisations who are already using effective methods to directly help people (let’s call this meta-charity); another would be to draw greater attention to the work of meta-charities who have identified such organisations (let’s call this meta-meta-charity). In support of her skepticism about meta-meta-charity, Alyssa points out that the effectiveness of more meta approaches can’t go up forever:

This logic immediately falls apart, once one takes it a step further. If fundraising offers a 3x multiplier, why not fundraise to fundraise, and get 9x? 27x? 81x? Where does the tower of meta end? The answer, of course, is that doing things everyone is already doing (like generic “fundraising”) never has a 3x return.

While I agree that ‘meta’ approaches can’t get better forever [2], this objection can’t be right either, because it seems to demonstrate that fundraising must always be useless. It can’t be the case that you should never put effort into raising money for a useful and unfunded project you are aware of, even if you are trying to get it off the ground yourself! Instead, I suggest that operating effective charities, finding effective charities, and fundraising for those effective charities are complementary inputs which each have declining marginal returns. Which one will most benefit from additional resources depends on which component is comparatively neglected.

If there are a lot of organisations trying to find effective charities, and donors lining up to give to them, but no final-level charities doing a good job, we most need altruistic entrepreneurs to start promising projects. If there are many organisations that can make cost-effective use of money, and donors keen to give to them, but donors don’t know who they are, then we most need more meta-charity. On the other hand, if there are effective charities, and we know who they are, but they still face a ‘funding gap’, there is a place for meta-meta-charity.

Alyssa is right that there is a lot of money available on Earth, or as an economist would more naturally put it, there are a lot of potentially productive inputs like capital, land and labour. Insofar as these resources are already being well allocated, and there are few, or no, effective altruist projects that still need funding, my proposal to improve how those resources are allocated through fundraising cannot work. That would be the case if the charities GiveWell and others identified were fully or nearly-fully funded, or there was nobody out there who could be convinced to fund them. But last I checked, for some reason, their recommendations still had ‘room for more funding’, which is why GiveWell recommends giving to them. And the organisation I work for has found it fairly easy to convince folks to do so.

I actually think Alyssa and I agree more than it looks. We both think that finding the most useful things to do with money is hard work. We both think there are a lot of resources out there that would like to find something better to do. Hopefully that means that, once you identify a great project, a small amount of money dedicated to fundraising should convince a lot of donors and fill its funding gap. And once that work is done, we will indeed be back hunting for the best charities to give to.

[1] Once a distaste for charities spending a lot of money on fundraising is entrenched among donors, I don’t see how it can be undermined by any particular charity or donor ‘defecting’ from the norm.

[2] A simple reason such a tower couldn’t work would be that you would quickly stop finding willing donors. There aren’t many people easily convinced to fund fundraising for fundraising.

Update: I should have included something like the anecdote below – now helpfully provided by Nick Beckstead – to explain why I am much less skeptical than Alyssa about the fundraising multiplier working on the margin. This implies a multipler of at least 3 for a generic, simple and scalable fundraising operation:

“A data point: I have a friend who worked as a street fundraiser (aka a “chugger” or “charity mugger”). He told me he took home about 10% of what he raised, and about 20% of what he raised went to the company that did street fundraising. They would raise funds for a wide variety of different charities, about which they and the people they fundraised from knew very little. (Charities would contract with the street fundraising org, and pay about 30% of the amount the street fundraising org raised, on a commission basis, to that org.)

If what my friend has told me is true, it seems it would not be hard to make a charity that consisted simply of funding street fundraising for a charity like AMF, assuming the charity wanted to cooperate. You might have trouble getting your fundraising charity non-profit status, but that isn’t too relevant from the perspective of this argument. I think you’d have a hard time street fundraising for a charity that did street fundraising for a charity like AMF, so you could probably only get one layer of meta in this way. But it would require fairly limited marketing skills (just enough to get some people to donate to your meta-charity) and may even have very limited downside risk (my understanding is that the street fundraisers can easily raise money for many conventional causes, and the charity pays on a purely commission basis in any case).

Assuming cooperation from the object-level charity, it seems the main limiting factors for something like this would be: 1) dollars from people willing to donate to your pure fundraising charity, 2) your ability to raise those dollars, and 3) running out of the object-level charity’s room for more funding. There’s probably some unknown challenges I’m not thinking of, but I suspect (just barely) that they wouldn’t be decisive.”