Retrospective funding, i.e. funding a project after it is completed, is not a new idea. Some organisations already do it, such as the Future of Life Institute. I used to think that this way of funding was a mistake. Since we want to change future behaviour, I thought it made sense (from a consequentialist perspective) to only fund future projects: We can’t change the past, so rewarding past work has no effect on behaviour.
But this line of reasoning is incorrect: Providing grants for past projects encourages people to create work without having any funding secured. It also allows for grant funds to evaluate finished products, rather than speculate on the quality of a project before it has even started, thereby allowing funds to better estimate the benefits of a research project.
In terms of implementing retrospective funding, I have a few tentative thoughts.
- New funding should account for the salary and level of funding a project has already received: If you would have paid $5000 for a project, but it has already received $4000, you’d only ever fund up to the 1000-dollar difference.
- People who are not resource constrained should not receive any funding. Let’s say Mary is the best researcher in the world, she’s working full-time on EA projects, and she has a personal assistant to handle any distractions (emails, grocery shopping, and so on). Providing extra retrospective funding to her can’t really increase her research output, and it’s unlikely to incentivize other people to work on EA causes, since they know Mary is more likely to receive new funding than they are.
- Funding should be done at the margin rather than be winner-takes-all. First, evaluate the value of a large number of projects, and retrospectively fund each project in proportion to project’s funding shortfall (i.e. the value of the project minus the amount it has already been funded). This reduces the variance of returns. And since rational people are risk-averse, you get a larger benefit (i.e. change in behaviour) for the same level of funding.
None of this is to say that upfront funding has no place. It does. Especially for projects that require such funding for upfront costs, such as equipment and so on. But aside from those cases, I think that retrospective funding is a better way to do things.
I’d like to hear everyone’s thoughts on the matter to see if I’m missing something. Should more grants be awarded to projects retrospectively, and why? Are there other circumstances for which upfront funding is superior?
I think retrospective funding is a really good norm to encourage, but the benefits only really start accruing once it's relatively widespread. Once it's common knowledge that one may get paid for self-initiated quality work, there's a very wide-but-weak incentive applied across the whole community. The norm-building itself, however, is a stag hunt with upfront costs.
Notice that this incentive affects the poorest of us most, and the people who don't already have jobs in EA. This means that it fulfills a funding niche that isn't already covered neatly by regular jobs and contract-based funding. (I'm especially thinking about those of us without jobs due to disability, but we'd still like to be paid for contributions during our uptime.)
I think the fact that so many EAs have independently arrived at this idea, is strong evidence in favour of there being a real potential for good here.
• I have previously independently arrived at the idea.
• Linda Linsefors wrote about it.
• Ben Kuhn, Paul Christiano and Katja Grace wrote, about, it in the form of impact certificates.
• Remmelt Ellen wrote about it and has a patreon page.
• Dony Christie has one too.
(If anyone know any other EAs with patreons (or similar), I'd like to know!)
I think we are the tip of the iceberg when it comes to who could potentially benefit from it being a universal expectation that if you do good work for the EA community, you can get paid for it.
Additionally, I think a norm for retrospective funding could open up occasional really high-impact giving opportunities for small-scale donors.