Thanks for sharing this! I've also been working on this question of "what would better forecasting by AIs enable?" (or stated differently, "what advances could instantaneous superforecasting 'too cheap to meter' unlock?") I've come at this from a bit of a different angle of thinking about how forecasting systems could fit into a predictive process for science and government that imitates active inference in brains. Here're slides from a presentation I gave on this topic at Manifest, and here is a half-finished draft essay I'm working on in case you're interested. 

Descriptively I agree, but normatively it's not obvious to me which alternative it supports

I haven't had time to read all the discourse about Manifest (which I attended), but it does highlight a broader issue about EA that I think is poorly understood, which is that different EAs will necessarily have ideological convictions that are inconsistent with one another. 

That is, some people will feel their effective altruist convictions motivate them to work to build artificial intelligence at OpenAI or Anthropic; others will think those companies are destroying the world. Some will try to save lives by distributing medicines; others will think the people those medicines save eat enough tortured animals to generally make the world worse off. Some will think liberal communities should exclude people who champion the existence of racial differences in intelligence; others will think excluding people for their views is profoundly harmful and illiberal. 

I'd argue that the early history of effective altruism (i.e. the last 10-15 years) has generally been one of centralization around purist goals -- i.e. there're central institutions that effective altruism revolves around and specific causes and ideas that are the most correct form of effective altruism. I'm personally much more a proponent of liberal, member-first effective altruism than purist, cause-first EA. I'm not sure which of those options the Manifest example supports, but I do think it's indicative of the broader reality that for a number of issues, people on each side can believe the most effective altruist thing to do is to defeat the other. 

Different charities will have different effects but broadly speaking if you save someone's life, that person continues to live and generate economic value (they do work, other people benefit by associating with them, etc.) Some things (like animal welfare) may be more like consumption (though accomplishing animal welfare advocacy goals may change policy that continues on further in time).

But also there may be a category error because even if the nonprofit just pays people to do nothing. The money you gave doesn't cease to exist -- it just becomes income for the employees and is spent on some combination of consumption and savings by them and the government that taxes them. So i think it may lead to a question of: in the broader economy, is savings or consumption preferred? And I guess that would probably vary over time based on the macroeconomic sittuation.

Personally, I think specifically forecasting for drug development could be very impactful: Both in the general sense of aligning fields around the probability of success of different approaches (at a range of scales -- very relevant both for scientists and funders) and the more specific regulatory use case (public predictions of safety/efficacy of medications as part of approvals by FDA/EMA etc.) 

More broadly, predicting the future is hugely valuable. Insofar as effective altruism aims to achieve consequentialist goals, the greatest weakness of consequentialism is uncertainty about the effects of our actions. Forecasting targets that problem directly. The financial system creates a robust set of incentives to predict future financial outcomes -- trying to use forecasting to build a tool with broader purpose than finance seems like it could be extremely valuable. 

I don't really do forecasting myself so I can't speak to the field's practical ability to achieve its goals (though as an outsider I feel optimistic), so perhaps there are practical reasons it might not be a good investment. But overall to me it definitely feels like the right thing to be aiming at.

I think that’s the idea but I also don’t know that many details

I feel like as president of 1Day Sooner I should probably chime in -- first, I wanted to say this type of work -- critiquing advocacy campaigns and analyses from EA or EA-aligned groups -- is very valuable and should be encouraged. I'm appreciative of SoGIve for publishing this and think they should be commended for spending the time to conduct this analysis. I think creating a healthy ecosystem for disagreement and the right incentives to encourage criticism and full-throated debate is important. 

On the object-level question, I'm obviously biased but I think most of the difference in cost-effectiveness in the SoGive analysis goes away if you adjust for the fact that vaccines are only given to children under 5 but only ~15-20% of bednets cover children under 5. Because 75% of malaria mortality is in children under 5, bednets are cheaper per person protected but the vaccines are much more targeted to people whose protection is most valuable. (The development benefit effects of reducing morbidity in children are also age-skewed in vaccines' favor though that's less dramatic). 

Insecticide resistance (probably reduces bednets' effectiveness to about 80% of what they'd otherwise be) and durability (GW estimates each bednet purchased provides about 1.7 year of coverage) are probably also relevant. The AMF tab of the GiveWell spreadsheet is a useful resource in thinking through these questions. 

For more of my thinking, here's my side of the email correspondence with Sanjay at SoGive. (I didn't include text from other people on the thread because I haven't asked their permission to share). An interesting meta-question is what should be the norm about making these sort of red-teaming or adversarial post-review correspondences public. My guess is it's probably a good thing to default to because it incentivizes people to be on their best behavior (and the benefits to confidentiality of being able to speak frankly don't seem that strong in these cases). But I don't think it's obvious either way and would be curious what other people think.

Overall, I'm eager to see more analysis done digging into the Imperial/Oxford modeling of cost benefit of the R21 vaccine (which comes to about 630 lives saved per 100K vaccinated, see Table 2) and what's publicly available about the WHO estimate of 13% all-cause mortality reduction from RTS,S. (Here's an older preprint that finds a smaller benefit -- more recent data that was publicly reported is apparently higher). So I think generally the follow-on research plan Sanjay discusses makes sense from my perspective, and I'd be personally supportive of anyone who wants to contribute to that work. 

Thanks for your interest! I've copied a description of the pool work below to give a better sense, but basically it's mostly research tasks that are like "research how vaccine distribution (not purchasing doses) is normally funded for new vaccines and write a 3-5 paragraph summary)" or "take a 3-5 paragraph summary someone wrote and create 3-6 sentences of suggested text to include in the status report" or "cite-check a section of talking points to make sure all the facts mentioned have citations and that those citations actually support the facts." 

Overall, we're very much in a "more the merrier" stage and would love your help. 

Here are more details on the scheme:

Plan for a Pool System to Handle Research/Talking Points 

Our talking points are intended to be a live, continually updated document representing our best understanding of malaria vaccination and how to improve rollout. In a sense it is intended to be a “global workspace” for our campaign thinking, where new research on key questions is inputted and accurate and relevant information about vaccination is shared across the campaign. Stylistically the talking points are intended to emphasize brevity, simplicity, and ease of use by a general audience.

To create a manageable process to continually update and improve the document (i.e. by executing this rolling punch list of tasks), we propose a pool system where volunteers sign up for a five days per month where they are “on-call” and will be assigned a 1-2.5 hour task per day, with assignments going out the night before and due the following morning (e.g. a Monday pool task would go out Sunday night and be due Tuesday morning). The expectation would be pool members would ideally sign up for two 2-day blocks and one 1-day block in a month or 3-day and 2-day blocks. The blocks are so that larger tasks (3-5 hours) can be assigned over a two day period. 

We’d aim to have at least eight pool members and one pool manager.