Thanks for replying! 

• I think you can just skip the submission step. There's no need to require submissions at all. My submitting to the FDA or equivalent, you're automatically deemed to have submitted.
• Yes, there are some context-specific elements, but they seem relatively small. Notably, once the FDA approves a drug, it is legal to use for any indication, and for any patient, in the US - even those the FDA did not label it for. I understand that there are biological differences in drug responses between different ethnic groups, but these are typically not that large, and note that the FDA already has a large population of African Americans, with ancestry from all across the continent, within its jurisdiction. Supply chain limitations in Africa, heat-resiliency, disease ecology etc. are real issues. But do they really justify multi-year delays for ordinary treatments? I think it is very unlikely that the cost-benefit analysis would come out this way. The EU runs EU-M4all, but this is focused on products that are focused on non-EMA markets - it doesn't mean than normal drugs approved for use in the EU are inappropriate for Africa.
• To the extent that post-approval surveillance is a problem, it is actually good to outsource approval to developed countries, because this would free up scarce resources for surveillance.
• The idea of an accountability seems more like a rhetorical / nationalist issue than a real one to me. The current degree of accountability national african regulators face seems pretty small in practice, and your preferred solution - an Africa-wide version - would also have accountability issues. I think ordinary Africans would be better served by a competent and (relatively!) swift international drug authority with no formal accountability to them over less-competent and slower national regulators with a veneer of accountability.
• It really doesn't seem that bad to me for African countries to indefinitely rely on others? Specialization, Division of Labour and Gains from Trade are some of the most important drivers of a prosperous modern world. But even if drug-approval-capabilities was something that autarky was desirable for, these countries should build the competence first, rather than imposing onerous restrictions on their citizens - and causing many premature deaths through denying safe and effective drugs. I would rather have them build capacity by focusing on surveillance, inspections, trial oversight, and procurement integrity.

Yes, sadly I think you're right, and the fact that this would be a good policy for Western countries also probably makes little difference in the rhetorical calculus.

Thanks for sharing this, it was a very interesting read!

I do want to question this claim though:

"But the alternative to a project like the AMA is that essential HIV treatments arrive half a decade late in places that needed them most."

It seems to me that an attractive alternative would be for African countries to simply give up on doing their own drug approvals, and outsource the decision making entirely to the FDA, MHRA, PMDA, EMA etc. Why not simply say that any drug approved by any of these agencies is automatically approved? This way drugs would be approved swiftly and with almost zero cost to both government and corporation.

If you weren't using a login, presumably you were using the lowest tier of models, which I don't think is a very good test.

You discuss institutions, but I don't think you discuss the right kind of institutions. If I am comparing Malawi to some of the other nearby landlocked African countries you mention, the first thing that jumps out to me is their dramatically worse economic freedom. Malawi has one of the worst scores in the entire world, ranked 147/165, a level more typical of central or saharan african countries.

https://www.efotw.org/?geozone=world&page=map&year=2023&countries=MWI,ZMB,RWA,BWA,LSO#country-info

(You didn't mention Eswatini or Burundi, but they also score very badly - unfortunately the map tool above will only let me display 5 countries so I focused on those named in the text plus Zambia as it is neighbouring).

This doesn't resolve the infinite-regress style question of what causes some countries to have more capitalist institutions than others, but when it comes to which institutions to investigate, I think it is their economic institutions we should focus on.

It seems very unlikely to me that there are AI safety projects that are both worth doing but also not worth doing if they spent $10k on compute. The human capital cost is typically far in excess of that.

If you don't gate access to your research resources (grant funding, GPUs etc.), and you have a non-trivial amount to give away, then I would expect almost all of them will end up going to projects you would not approve of. 

Note that UBI does gate access - it gates to a (small) finite quantity per person. 

I realise I am arguing with ChatGPT here, but you are equivocating between EA and the US government. We (EA) are preparing for having more money, and this doesn't in any way contradict the fact that we (USG) already have a lot and we (USG) have chosen absurd kidney market restrictions.

Thanks for this very interesting article.

One quick suggestion: assuming that 100% of distributed cups are used seems quite aggressive to me. Using a 7.5 year average life I think suggests you are implicitly selecting on very enthusiastic adopters; I would imagine a lot are distributed and then discarded.