Global health R&D strikes me as having very high expected value, but might be difficult for governments in low income countries to justify to voters when it could spent on urgent object level health interventions which produce benefits more quickly.
Does that mean donors should focus more on R&D (eg - give more funding to CEPI than to the Pandemic Fund)? Is this idea fleshed out in better detail somewhere in the global health world?
Normally I'd say yes, but my AGI timelines are now 50% in ~4 years, so there isn't much time for R&D to make a difference. I'd recommend interventions that pay off quickly, therefore. Bed nets, GiveDirectly, etc.
I think that the recent 80,000 Hours Podcast on high-impact climate philanthropy discusses this. See this section in the transcript, potentially.
And there's also this recent sequence (e.g. one post is about Estimating the cost-effectiveness of previous R&D projects).
Great post!
R&D about health in slums may be promising and neglected:
"Importance: Approximately 1 billion people currently live in slums, and it is estimated that a quarter of the world’s population will live in a slum by 2030. Slum conditions currently do not support a healthy or high-quality life. This is a very important issue. Tractability: Slum policy interventions appear relatively intractable. In contrast, In situ slum upgrading interventions largely deliver cost-effective results, but more research is needed.
[...]
As slums grow at an alarming rate, a better understanding of problems uniquely faced by slum populations is required. For this to happen, governments must consider slums as spatial entities and collect more extensive census data, while academia may contribute by focusing research directly on slum health."
Source: https://forum.effectivealtruism.org/posts/sgkvhJLFMDzBPgovp/shallow-investigation-slums
Normally I'd say yes, but my AGI timelines are now 50% in ~4 years, so there isn't much time for R&D to make a difference. I'd recommend interventions that pay off quickly, therefore. Bed nets, GiveDirectly, etc.