Should donor lottery winners write reports?

byrk3mo22nd Dec 201826 comments


Summary: I was worried that donor lottery winners writing reports might have a net negative effect by discouraging entrants. After modelling the outcomes of reports, I provisionally think that writing reports is likely a good thing and should continue. This is mostly because the winner sharing their research work may improve the quality of others' donations.

EDIT/Update: Denise Melchin points out that people may overrate lottery winners' evaluation. This makes me less clear about the sign of reports' effects on others' donations, and so I'm now pretty uncertain whether reports are good or not.

NB: I discuss possible downsides of donor lottery winners writing reports. I definitely don't want to come across as criticising winners like Adam Gleave who wrote up a report. A lot of work went into the report, and reading it gave me useful object-level information and evaluation heuristics. I just wanted to look into a possible downside that occurred to me.

Initial worry

Section summary: if people think lottery winners have to publicly justify their choices, entries to lotteries may be reduced; if lotteries are the most efficient choice for small donors, this could be bad

I am choosing to use a donor lottery as a vehicle for my donations this year. When I was trying to decide whether or not to do so, I weighed the fact that my donations could be publicly scrutinised if I won. Would I have to follow in the footsteps of Adam Gleave's excellent 2017 report? I certainly didn't relish the thought of doing so.

Though I ultimately decided to use a lottery, if my reaction is common, some people could be dissuaded from entering the lottery. I think that economies of scale from donations may make the lottery the best choice for small donors. If that's the case, people being dissuaded from entering could be a big deal.

Possible upsides to reports

Section summary: reports could increase willingness to enter, induce more work from the winner and provide valuable information to other donors

Historical record

A report allows potential lottery entrants to see the sort of donations that have historically been made by winners. Is this a good reason to write a report?

On the one hand, in many situations, it shouldn't matter what other lottery entrants would donate to (search for "Does it matter what other donor lottery participants would do with their funds?").

Of course, the expected outcome if you don't win the lottery does in fact affect the overall expected outcome of the lottery. Yes, it's true that lotteries could make sense even if the expectation after losing is neutral, but they make even more sense if that expectation is positive.

In the end, I decided not to model the increased willingness to enter from a history of reports.

More work from the winner

If the winner intends to write a public report about their donations, they are probably going to do their due diligence and come up with good reasons for their donations. Of course, it's pretty likely they'd do that even without the report, but I think it likely increases the conscientiousness of the winner.

Improving others' donations

As I mentioned in the note at the beginning of this post, reading Adam's report gave me information about charities and some ideas about how to evaluate charities. Sharing the information from the evaluation is in fact the main reason Adam gives in the post for writing up the report.

I think this is likely to be fairly valuable. When I model this, I focus only on the information it provides to small donors. After all, larger donors already have quite a lot of evaluative power (Adam writes that he spent about 45 hours on the process. I don't think a marginal week of evaluation is very important for large donors)

A model

After thinking about this for a bit, I decided to make a Guesstimate model. The model is very rough, and feedback on parameter values or missing pathways from the model would be appreciated.

Despite the model's roughness, I find the benefit from writing reports to be large enough (~300x more benefit than cost) that I've changed my mind and think willing winners should probably write reports on their donations. In fact, even ignoring the knock-on effect of improving others' donations, I find writing the report to be a net benefit.

The summary from the model's page

I'm roughly estimating the value of donor lottery winners writing up a report of where they donated.

The causal pathways to improved outcomes that I'm modelling are (1) better donations by the lottery winner given that the winner knows they'll be writing up a report and (2) improvements to others' donations based on the lottery winner sharing their work.

The possible negative outcome is that it contributes to an expectation that the donor lottery winner should write up a report, leading people to not want to enter the donor lottery. If the donor lottery is in general more efficient than individual donations, this displacement of funds away from the lottery reduces the amount of good done.

A big problem with this model is that it goes through estimating the total size of the EA donor pool. This really isn't ideal, mostly because the estimate is very rough and changes would make a big difference to the overall improvement in donation.

A more conceptual problem is that I don't model changes in the donor lottery's size over time. I just consider the effect at current sizes. If an effect is slower growth of the donor lottery, that could be the largest contribution to the outcome of writing reports.

Evidence that would be useful

Here are the main things that would make me change my mind:

  • Arguments or evidence that reports may slow uptake of lotteries
  • Arguments or evidence against the claim that writing a report improves the quality of the winner's donations
  • Better information on the total amount of donations from small donors (on its own, this could only make me unsure whether reports are good or not; it couldn't make me think they might be bad)