A donor lottery is a way to convert a donation of $X into a larger donation of $X/p by making a bet with probability p of winning. This is done by having each donor contribute to a pool (with a fixed size, set by a benefactor making up the difference), and a winner is randomly chosen proportionally to how much they donate. The Center for Effective Altruism runs a yearly donor lottery with pool sizes of $20,000, $100,000 and $500,000.
For a contributor, the expected number of dollars they will be able to donate is the same whether or not they participate. If they win the donor lottery, then they have more money to allocate, so if there are economies of scale, which there are likely to be (the typical example is that the value of information from extra time evaluating where to donate money is proportional to the size of the donation, so it's worth it for the winner to spend more time evaluating their grant and hence getting a greater expected impact per dollar), this means that they have a greater expected utility when participating in the donor lottery.
As such, I think that donor lotteries are hence obviously the most effective way for an individual small donor to turn their money into world improvement, and plan to give the remaining part of what I've pledged to give this year to the donor lottery.
In trying to figure out why not everyone is giving to donor lotteries, I quickly wrote down some reasons why someone might choose not to, then added some thoughts on each bullet point to make this post.
They don't trust the lottery administrator to pay out.
Right now, I could announce a donor lottery on my Twitter and run it. Probably not many people would trust it though, because they'd be hoping that I allocate a large sum of money on behalf of the winner out of the goodness of my heart. This is less of an issue for institutions like the Center for Effective Altruism who have established track records, are made of demonstrably trustworthy people, and can be sued.
They think they might be unhappy if they lose the lottery.
From the "some drawbacks" section of Why you should give to a donor lottery this Giving Season:
A donor lottery doesn’t have the same ‘warm fuzzies’ you might get if you were to know with 100% certainty that you had, for example, bought a certain number of bednets, or contributed directly to a cage-free campaign. In the case where you don’t win (which happens most of the time), you won’t have the knowledge of having directly contributed to a cause you care about. Hopefully this is offset by the knowledge that, in expectation, your donations are more effective than they would have been otherwise, but it’s understandable if this doesn’t feel as satisfying as donating directly.
I also want to add that this unrewardingness might lead to a greater chance of value drift compared to similar altruists who give to places with more certainty.
They think they might lose a lot of status if they admit to their friends they gave to a donor lottery.
The two aspects of this are firstly that donor lotteries are inherently unconventional so there's a cost in weirdness points paid for participating, and secondly there's an association with gambling, which is low-status (it's even called a lottery - quintessentially the domain of poor people who don't understand probability). Maybe the latter would be worth a change in naming, like calling it a "donor pool" where the a single member is "nominated as an allocator" "in proportion to their stake".
They don't trust the lottery administrator to send money where they recommend.
The Center for Effective Altruism has some limitations on where it is allowed to send grants:
The lottery is administered by EA Funds, which is a project of the Centre for Effective Altruism (CEA). CEA can only make grants that are within its charitable objectives, and retains sole discretion over where the final grants are made. This means that we won’t make grants that run counter to broad altruistic principles, or to projects that don’t satisfy our regular due diligence requirements. For practical purposes, we should be able to make grants to any registered charity in most parts of the world (provided we can easily verify their registration status). If you want to recommend the money be given to a project that is not currently a recommended charity, that’s probably OK, but the project will need to demonstrate that they’re providing a genuine public benefit, and pass our regular due diligence checks. We also can’t grant to organizations that are directly involved in politics.
A donor who thinks that one of these projects is the most effective way to improve the world would not want to participate in CEA's donor lottery.
Even for a nonstandard grant that's allowed by CEA, a winner of the donor lottery might find that there could be some overhead (on the side of the project in demonstrating their genuine public benefit, on the side of CEA in due diligence checks, possibly on the side of the winner of the donor lottery in back and forth emails) that makes it more effective to donate to the project directly rather than through the donor lottery and sometimes getting the value of information from extra evaluation of the project and a larger bankroll to donate with.
They think other donor lottery participants are evil in a way that increases superlinearly with bankroll.
This is one objection that remains after the conclusion in Doning with the Devil:
You should only worry about your fellow lottery participants insofar as you think they are significantly misaligned with you and have significantly greater economies of scale than you do.
I think that a proper donor lottery design avoids a lot of this problem; ideally, there's a benefactor who's infinitely wealthy and is willing to participate in the lottery because they're rich enough their utility is just straight-up linear in money who ensures that no matter who participates the winner gets the same amount of money.
The EA Funds donor lottery has such a benefactor; the winner always gets $20,000, $100,000 or $500,000, depending on which lottery they choose, so even in the case where you do enter a donor lottery with an adversary whose harm scales superlinearly with their bankroll your donation wouldn't affect their chances of winning or the amount of evil that they do if they can win.
They might be worried about fumbling their lottery win.
The winner of the donor lottery is allocating a large amount of money, which depending on where it's sent could be anywhere from amazingly impactful to harmful, and if it ends up being harmful or just suboptimal they might feel like the rest of the EA community is disappointed in them (especially if they anticipate large groups of internet people being cruel). This could put a lot of unwanted pressure on the winner, more than if they just quietly and privately donated their smaller amount of money without any reason for other people to pay attention.
For the EA Funds donor lottery, this is mitigated by donor lottery winners and payouts being potentially anonymous; some people have chosen to make their allocations public and wrote reports (some can be found under the donor lotteries tag), though this is not a requirement to participate.
Their payoffs are sublinear in the amount of money they can allocate.
It might be that someone knows of local hyper-effective places to donate that would benefit from $200 but not from $20,000. Luke Muehlhauser writes that his personal donation strategy includes:
Small one-time opportunities that seem (to me) to have higher expected value than Open Philanthropy’s last dollar, and for which I have lots of context that would be more time-consuming to communicate to another funder than the monetary cost can justify. An example would be to buy a new laptop for a likely high-impact person who I know well and who is low on savings because they are transitioning to a higher-impact career and had an unexpected financial setback.
If someone knows of opportunities like this, they might find their first $200 to be more impactful per dollar than their first $20,000.
The overhead of a donor lottery isn't worth the additional benefit.
There are costs associated with running a donor lottery. The administrator of the lottery needs to dedicate some funding and staff time to hosting the lottery itself, and the winner of the lottery has to dedicate some time to emails instead of just typing in a number and hitting the donate button.
This might especially be the case given the existence of good donor pools in the EA community already. For example, for a donor interested in animal welfare, the people running the Animal Welfare EA Fund and Animal Charity Evaluators have already put a great deal of thought into the best places to donate to help animals and will make grants according to these evaluations, so the benefit of putting additional thought into it is small (one might think of this as, rather than a single donor going from spending 1 hour evaluating a grant to 100, it's the EA community as a whole going from spending X+1 hours to X+100 hours, where X is large and done by experts).
They have donor matching opportunities elsewhere that beat the benefit from lotteries.
For example, during December 2020, around the time the donor lottery was being run, matching funds tripled donations to the Humane League, and doubled donations to the Good Food Institute, 80,000 Hours, Animal Charity Evaluators, Center for Effective Altruism, Clean Air Task Force, Founders Pledge, Givewell, the International Refugee Assistance Project, One for the World, StrongMinds, and the Center for Election Science, and for a few seconds every other charity, as well as probably some matches that I haven't listed here.
Some of these might not be counterfactually valid, and some of the remainder might have the matching fund fully spent anyway, but the possibility of a 2-3× boost in impact in the cases that remain might be greater than the value of information from the extra evaluation they would do as a donor lottery.
Donor lotteries are rare enough that they expect their values to change between now and then.
Joey Savoie's survey of '10% donors' and the EA Survey give a rate of people dropping out of the EA community of around 1/2 over five years. Someone might prefer donating each paycheck instead of waiting for the next donor lottery and potentially ceasing to be engaged with EA in the meantime, or just being tempted to dip into the pool of money that they've set aside for their yearly donor lottery donation.
Donor lotteries are rare enough that giving now might be better than giving later.
In 80,000 Hours' 2018 talent gaps survey, EA organisations reported high discount rates on future donations, with a median of 16% per year and a 90th percentile of 65% per year (though 80,000 Hours cautions against making major changes to ones' plans based on these due to some caveats they mention). These organizations would hence discount someone waiting until the next donor lottery is held (up to one year for the EA Funds donor lotteries) by this much.
The donor lottery might not be tax deductible in their country.
The EA donor lottery is tax deductible in the US, the UK, and the Netherlands; the lack of tax deductibility is another small but present discount a donor outside of those locations on the cost-effectiveness of their donations to the EA funds.
They just haven't considered donor lotteries.
Oh, that reminds me - have you considered donor lotteries?
They don't give anywhere and donor lotteries are a subset of this.
Also, some people just don't give to charity for other reasons beyond the scope of this post.