The case for donor lotteries is based on assumptions that don’t hold up to scrutiny. Based on how donor lotteries have performed in practice there is little reason to think that (as GWWC put it a few months ago when launching the 2024 lottery) “we believe donor lotteries are one of the most effective ways for smaller donors to give, provided you're comfortable with the approach.” In the past, even stronger claims have been made about the benefits of donor lotteries.

Questionable assumptions include:

• Donor Lotteries assume anyone can be a good grant evaluator. Even among a donor base of EAs, there’s no reason to think that everyone participating in a lottery is equally capable of being a good evaluator. If one assumes non-EA participants, this assumption becomes even less realistic. Different people have different strengths, but the lottery model assigns evaluation responsibility based on randomness rather than relevant skill.
• Donor Lotteries assume winners’ work will be leveraged. In practice, lottery winners rarely publish their work (the last published writeup is from the 2018-2019 lottery). It’s also rare for lottery winners to become professional grantmakers, so it’s questionable whether any additional evaluation knowledge they acquire is leveraged beyond the lottery.
• Donor lotteries assume there’s demand for the model. Donor lotteries have typically been in the ballpark of a few hundred thousand dollar, and often significantly smaller. One likely explanation is that the model is highly counterintuitive, and difficult to explain to casual donors (or anyone who isn’t already familiar with it).
• Donor lotteries assume the model improves impact/dollar significantly, and does so for a significant number of dollars. Lotteries are unlikely to improve impact/dollar significantly, as the model is mostly used by (or even understandable by) people who are already EAs. Casual donors whose impact could be considerably increased are unlikely to understand or participate in a lottery. And historically lotteries have only been modest size, so they are not impacting a large number of dollars.

Donor Lotteries were an interesting experiment because they address an important observation: there are economies of scale when giving. However, I’d argue EA Funds is a much better implementation model. EA Funds is easily understandable (to EAs and non-EAs alike), uses evaluators that are chosen for their ability in that area (rather than by random selection), has a better track record of publishing grant writeups than donor lotteries, and adds evaluation capacity more efficiently over time (since fund managers are relatively static as opposed to donor lotteries where a new evaluator is selected and needs to start training from scratch with each new lottery). EA Funds already does a much higher donation volume than donor lotteries, and it would be preferable to continue building these economies of scale than to divide resources across multiple projects with similar goals.