Add new comment

I appreciate how honest, thoughtful, and transparent this post is. I’m very happy with OpenPhil’s approach and choices, despite disagreeing with several of their specific decisions, and this made me more confident that they’re doing the right thing. I’m not confident that any change in this process would make them better at doing good, but there is one I want to mention.

It seems like there’s one failure mode of this approach not mentioned in the article. Taking macroeconomic policy as an example, because it seems representative of this approach, let’s say someone has insider knowledge about lobbying, or works for the Fed, or has connections there. They are the sort of person who is tremendously valuable for OpenPhil’s goals there. Because of the constraints on the process that OpenPhil discusses in this blog post, the public information about OpenPhil’s grants in this area are unconvincing (by which I don’t mean that I am personally unconvinced, though I am, but that people with relevant expertise who I’ve encouraged to read it have also been unconvinced.)

OpenPhil says that’s a good thing: if an approach did not initially seem wacky and unwise to experts, there’d be other people doing it and it would be less neglected. They also say that if they expected their grant officers to be able to justify their grants fully, they worry researchers would be less likely to pursue out-there ideas. On the other hand, though, if a lobbyist with expertise in this area or a person who works at the Fed reads your grant explanations, it would be tremendously good if at least some of them come away convinced.

Because everyone knows that there’s lots of information not publicly available, people aren’t sure what is worth debating and arguing and when we should just trust that our concerns would be answered if we knew all the nuanced conversations that we know are happening at OpenPhil.

And while I’m pretty much willing to grant “they’re really good at what they do; I’d probably agree with them if I had the information they had, and they’ve probably considered my counterarguments even if nothing public answers them”, I’m willing to grant that because I trust and respect the people involved. A hypothetical insider who has expertise OpenPhil would benefit from doesn’t have the trust in OpenPhil to go “even though this decision is not justified by the information published in favor of it, the people involved are capable enough that I believe them anyway.”

Sounding weird to experts at first glance is a good thing to target: sounding weird to experts (or non-experts who have a highly related skill set which could be put to bear on the problem) who have read all of your public-facing information could be a failure of communication. I’m hopeful that with more time and expertise OpenPhil will figure out if there’s a way to gain in convincingness without costs to their internal process.
(crossposted from Facebook).