Mo Putera's Quick takes

Abi Olvera's Golden rice delay dashboard, includes BOTEC calculations and sources, supplement to her Substack article A blocked GMO rice could have saved 100,000 children. The same tech makes pineapples pink:

Every year, vitamin A deficiency blinds 250,000 to 500,000 children. Half of them die within a year of losing their sight.

One third of children worldwide are deficient, roughly half of children in sub-Saharan Africa and Southeast Asia¹. Deficiency weakens the immune system, so children are more likely to die from common infections like diarrhea or measles.

Rich countries solved this kind of problem by adding nutrients to staple foods. The United States adds iodine to most salt, and folic acid to flour to prevent birth defects.

Rice is a critical staple among the world’s poor, so scientists improved rice. They added beta-carotene, the thing that makes carrots orange and that the body turns into vitamin A. The new rice cooks and tastes the same, but it’s yellow. They called it Golden Rice and licensed it for free to any farmer earning under $10,000 a year.

It has been ready since the mid-2000s. Today it grows nowhere.

The reason is that it is a GMO. Environmental groups, led by Greenpeace, fought it in country after country for two decades.

As far as I can tell, no one has calculated the cost of that delay. I’ve spent the last few weeks doing so. My rough estimate is that the delay has killed about 106,000 children and left another 210,000 to 425,000 blind.² Measured in healthy years of life lost, that is somewhere between 7 and 12 million.³ (My full calculations are here. I will update these figures as I receive feedback.)

That works out to roughly fourteen children dying every single day, for twenty years, from a nutrient we already know how to add to food. ...

Most GMO crops are changed in how they grow, so the part you eat is ordinary. In Golden Rice, the part you eat is the part that changed. That made it feel new.

Greenpeace framed it as dangerous, saying corporations were secretly behind it. They falsely claimed it was unproven and that it was unclear whether children could absorb the vitamin. Activists tore up test fields and filed lawsuits to block approval. Over a hundred Nobel laureates signed a letter asking Greenpeace to stop.

While Golden Rice sat blocked, other new GMO foods reached store shelves. One of them is a pink pineapple. It sells for about $10 in stores and up to $50 online. It uses the same chemistry as Golden Rice, run in the opposite direction. Golden Rice turns on the pathway that makes the vitamin children need. The pineapple turns it off, so the fruit stays a pretty pink.

The lifesaving technology is in the Western world, growing a nicer pineapple for parties.

The Frontier in 2025 (data), by Gavin Leech, Lauren Gilbert, and Ulkar Aghayeva, rated 202 of the biggest breakthroughs of last year. Some favorites, mainly public health- and society-related:

1. Diagnostics on a phone with no doctor needed (source)
2. Murder rates worldwide have fallen 25% since 2000 (source) ("On average! Potentially some confounding from improved trauma emergency care converting murders into attempted murders")
3. 5 factors explain most of the genetic variance in common mental illnesses (source)
4. Large effect for 5-MeO-DMT for treatment-resistant depression (source) ("Recall that major depression is maybe 2% of the total global burden of disease")
5. For the first time in recent history, China’s emissions might be falling (source)
6. The first evidence of a solar take-off in Africa (source)
7. A tiny number of people are functionally cured of HIV. The antibodies responsible may have been identified (source)
8. An E. coli vaccine is currently undergoing Phase III human testing (source): "E. coli is the second-most lethal bacterium in the world, with about a million deaths a year. There are currently no effective vaccines for it"
9. Extreme poverty drops from 27% of India to 5% in one decade (source)
10. Observational follow-up on the Covid vaccines shows a large decrease in all-cause mortality (source)
11. First-in-human 'prime editing' gene therapy. Cured an inherited immune disease (source)
12. Last year's biannual HIV shot available in low-income countries, $40/year (source)
13. AI designs antibodies that can turn on or off membrane signaling proteins implicated in many diseases (source)
14. AI for antibiotic design (source): "7 of 24 AI-designed and custom-synthesized compounds show selective antibacterial activity, including against N. gonorrhoeae and S. aureus"
15. AI generator for antibodies against specific protein targets (source)
16. Two promising drugs to prevent secondary and post-surgical stroke (source)
17. Challenge trial on a salmonella vaccine showed roughly 70% effectiveness (source)
18. Tiny demo of a 90% effective malaria vaccine which only takes one dose (source) ("However, the Leiden study was n=15. The followup PfSPZ-LARC2 study won't be finished until 2027 and is also n=22(!). Last year's R21 vaccine was 75% effective but takes 4 doses")
19. A candidate "gene drive" for eliminating malaria reduced parasite hosting from 80% to 30% (source)
20. A new class of treatment for malaria: 97% cured and it shouldn't suffer existing drug resistances (source): "the first new class of malaria drug approved in more than 25 years"
21. Three new countries certified malaria-free (source): Georgia, Suriname, and Timor-Leste
22. Four countries eliminated trachoma, a disease that causes blindness (source): Senegal, Egypt, Mauritania and Fiji
23. The cost to treat drug-resistant TB drops below $300 (source): "bedaquiline now available at $63 per treatment course, bringing the price of the complete BPaLM treatment to $284"
24. First successful transplant of a non-human lung into a human (source)
25. The oldest baby in history: an embryo frozen in 1994 was brought to term and resulted in a healthy baby boy (source): "The biological mother of the baby was 62 years old at the time of his birth. While incremental, this points towards eventually allowing for delayed IVF, which would be socially transformative"
26. First human infant cured of a lethal genetic disease with a personalized gene therapy (source)
27. Trevogrumab could potentially prevent muscle loss in the sedentary (source)
28. Approval of a strong non-opioid painkiller targeting a pathway specific to pain neurons (source): "It's strong, as strong as hydrocodone or low-dose morphine. It is claimed to be nonaddictive, which is the somewhat unlikely part. It does hit the brain much less, which might work"
29. Tooth regrowing procedure enters human trials (source)
30. Rising Internet access reduces prevalence of female genital mutilation (source)

Why they did this:

A couple of years ago, Gavin became frustrated with science journalism. No one was pulling together results across fields; the articles usually didn’t link to the original source; they didn't use probabilities (or even report the sample size); they were usually credulous about preliminary findings (“...which species was it tested on?”); and they essentially never gave any sense of the magnitude or the baselines (“how much better is this treatment than the previous best?”). Speculative results were covered with the same credence as solid proofs. And highly technical fields like mathematics were rarely covered at all, regardless of their practical or intellectual importance. So he had a go at doing it himself.

This year, in partnership with Renaissance Philanthropy, we took a more systematic approach. So, how did the world change this year? What happened in each science? Which results are speculative and which are solid? Which are the biggest, if true?

I just learned via Martin Sustrik about the late Sofia Corradi, 

the spiritual mother of Erasmus, the European student exchange programme, or, in the words of Umberto Eco, “that thing where a Catalan boy goes to study in Belgium, meets a Flemish girl, falls in love with her, marries her, and starts a European family.”

Sustrik points out that none of the glowing obituaries for her mention the sheer scale of Erasmus. The Fulbright in the US is the 2nd largest comparable program, but it's a very distant second:

So far, approximately sixteen million people have taken part in the exchanges. That amounts to roughly 3% or the European population. And with the ever growing participation rates the ratio is going to get even gradually even higher.

Is short, this thing is HUGE.

Sustrik argues that the Erasmus programme is gargantuan-scale social engineering done right:

Substantial portion of students actually does want to spend some time abroad. It’s no different from the Western European marriage pattern, where young people left their parental homes to work as servants, farmhands, or apprentices before they married and set up their own households.

The much-maligned idea of social engineering, in this case, doesn’t mean forcing people to do something they don’t want to do. It means removing the obstacles that prevent them from doing what they already want.

Before Erasmus, studying abroad was seen as having fun rather than as serious academic work, something to be punished rather than rewarded. Universities were reluctant to recognize studies completed elsewhere. Erasmus, with its credit transfer system, changed that and thus unleashed a massive wave of student exchanges.

The backstory to how Sofia came to focus on Erasmus is touching:

In 1957, in her fourth year of studies, she received the opportunity to study in the United States thanks to a Fulbright scholarship. She spent a year at Columbia University where she attended a master’s course in comparative university legislation.^[3][4] Upon her return to Rome in 1958, however, her degree was not recognised by the Italian educational system.^[4][5] She recalled how she felt humiliated in front of other students as her time in the US was dismissed as a "vacation", and how a functionary had told her "Columbia, you say? I've never heard of that before".^[5][6] She had to spend an extra year to obtain her Italian degree.^[5] The experience led her to the idea of creating a system of recognition of courses taken abroad and the promotion of university exchanges.^[5][7]

Such ideas had already been put forward in Italy, but without any concrete results.^[7] After graduating Corradi pursued research on the right to education at the United Nations and became a scientific consultant for the Association of Rectors of Italian Universities at the age of 30.^[5][6] It was a post she gained in part to her diploma from Columbia, and she used her position to lobby intensively for her idea of a university exchange programme and mutual recognition. ... (more on Wikipedia)

I've previously wondered what a shortlist of people who've beneficially impacted the world at the scale of ~100 milliBorlaugs might look like, and suggested Melinda & Bill Gates and Tom Frieden. (A "Borlaug" is a unit of impact I made up, it means a billion lives saved.) If you buy Corradi's argument that the Erasmus programme is at heart really a peace programme and that it deserves some credit for the long period of relative peace we've experienced globally post-WWII, then Sofia Corradi seems eminently deserving of inclusion.

Gemini 3 Pro's attempt to visualise Sofia Corradi's beneficial impact in Shapley value terms:

Let's define our terms:

• Y-Axis: "Level of European Youth-Driven Integration" (0-100%)
  • This is not economic integration (like the Euro) or political integration (like the Parliament).
  • It specifically measures the socio-cultural intermingling, mutual understanding, and reduction of nationalistic stereotypes among young Europeans. This is the "peace program" aspect.
  • It starts at a low baseline post-WWII, as even with the EEC, borders remained strong culturally.
• X-Axis: Time (1950 - 2025)
• Key Event 1: Treaty of Rome (1957) - Establishes the EEC. A step towards economic integration, but limited youth movement.
• Key Event 2: Erasmus Program Launch (1987) - The crucial inflection point.
• Key Event 3: Schengen Agreement (1995) - Eliminates internal border checks. Facilitates Erasmus, but Erasmus already laid the cultural groundwork.
• Key Event 4: Euro Adoption (1999/2002) - Further economic integration, making cross-border life easier.

Now, let's plot two scenarios:

1. "Actual Timeline: With Corradi & Erasmus" (Solid Blue Line): Represents the observed trajectory of youth integration.
2. "Counterfactual Timeline: Without Corradi & Delayed Erasmus" (Dashed Red Line): This is where we attribute Corradi's Shapley value.
   • Delayed Launch: As argued, without her tireless 30-year lobbying, a pan-European student exchange might have emerged, but likely much later (e.g., 2002).
   • Fragmented Design: Even if it launched, it would likely have been a collection of bilateral agreements, lacking the standardized credit transfer (her key design contribution). This means a slower, less efficient ramp-up of integration.

The "Area Between the Curves" will visually represent Sofia Corradi's Shapley Value, showing the accelerated and enhanced integration due to her efforts.

Alexander Berger's 2026 CoeffG annual letter describes their shift from marginalism to "inframarginal" funding, emphasis mine:

Our mission is to help others as much as we can with the resources available to us, and historically, that’s meant operating in a “marginalist” mode, thinking hard about how to equalize marginal returns within and sometimes across cause areas. (The idea is that each additional dollar we spend should produce roughly the same amount of good if allocated to any of our funds, because if it doesn’t we should be shifting resources toward the higher-return area until it does.)

The trade-offs that drive a focus on equalizing marginal returns remain. Funding still falls well short of the opportunity set across our focus areas, and we still face some zero-sum budget decisions across worldviews and funds.

But with nearly every fund working with a higher budget in 2026, and more growth expected in the future, we need to shift more of our attention away from marginal trade-offs, and toward more ambitious goals. Many of our best grants have been deeply inframarginal: we had to recruit a founder or — effectively — incubate an organization, but once we had and there was finally something fundable, the eventual grant was far above our bar. The kind of dedicated and proactive ownership from program staff required to enable these grants can pay off in a big way, especially over time, but it trades off with a focus on optimizing the marginal dollar in the near term.

I thought the importance of taking full responsibility for a problem was captured well by Nan Ransohoff ’s piece “There should be ‘general managers’ for more of the world’s important problems.” Lewis [Bollard], who Nan also points to in her post, exemplifies this in his work on farm animal welfare. Another colleague who really embodies this spirit is Andrew Snyder-Beattie, who runs our biosecurity and pandemic preparedness work. Andrew joined Coefficient in 2019 and has taken it upon himself to reduce risks from worst-case pandemics with unusual and impressive dedication and purpose. Recently, this has included work on reducing risks from mirror bacteria and a four-pillared plan to avoid engineered biological catastrophes.⁹

I’m inspired by these examples and want Coefficient to continue being a place where outstanding people’s responsibilities and ambitions can grow to match the scale of the world’s most important problems. However, I also want us to avoid the common “strategic funder” trap of thinking we have all the answers and just need to slot grantees into our vision, almost like subcontractors. An important virtue of the marginalist approach to funding is that it’s relatively strategy-agnostic: if an opportunity clears the cost-effectiveness bar, you should fund it. That tends to facilitate a healthy openness to grantees’ stronger local context and knowledge.

I want us to navigate this tension as thoughtfully as possible, bringing the ambition and ownership of general managers without losing the curiosity and humility of marginal funders. The right balance will vary across focus areas and individuals, but we should be intentional about trying to strike it.

(I do wish Berger gave a bit more detail than just "we should be intentional about trying to strike the right balance between GM and marginalist approaches", but I suppose the annual letter isn't the right place for this.)

Nan Ransohoff's piece on how there should be more GMs owning delivery of specific outcomes is a great read too (emphasis mine):

There’s a surprisingly big category of problems that are ‘orphaned.’ By ‘orphaned’ I mean: you can’t point to a specific person or organization who thinks it’s their responsibility to deliver the outcome in its entirety. Lots of people talk about the problem, and often many work on slices of it. But if you asked: ‘is there a hyper-competent person waking up every day feeling accountable for making sure this gets solved?’—the answer is very often, ‘no.’ ...

In my opinion, there should be ‘general managers’—GMs—for problems like these. These are founder-types who feel personally responsible for delivering a specific outcome (vs field-building generally); hyper-competent leaders who will pull whatever levers necessary to achieve the defined outcome. Most companies wouldn’t let an important initiative go unmanned or without a ‘directly responsible individual’ (DRI) — why are we OK not having GMs for even more wide-reaching problems? ...

[These GMs are] flexible on the details, constantly zooming in and out to readjust strategy and tactics as the problem evolves. They’re fast learners and quick to develop (or hire for) whatever skills are needed at the time. They feel deep personal responsibility to solve the problem, which means they’re likely to stick with it for years if not decades (the smallpox eradication effort took 11 years, marriage equality took 30). Great GMs have to carry the torch even when (especially when) political winds inevitably shift, public interest wanes, and funding environments worsen. They possess the conviction and stamina to lead through conditions both good and bad. 

As I've gotten more work experience (year 10 now, jeez) I've become increasingly a fan of the DRI approach, and by extension the GM ("super-senior-DRI") approach. You could think of incubators like AIM and SMA as "GM factories for orphaned problems". 

My favorite midsized grantmaker is Scott Alexander's ACX Grants, mainly because I've enjoyed his blog for over a decade and it's been really nice to see the community that sprang up around his writing grow and flourish, especially the EA stuff. His recent ACX Grants 1-3 Year Updates is a great read in this vein. Some quotes: 

The first cohort of ACX Grants was announced in late 2021, the second in early 2024. In 2022, I posted one-year updates for the first cohort. Now, as I start thinking about a third round, I’ve collected one-year updates on the second and three-year updates on the first. ...

The total cost of ACX Grants, both rounds, was about $3 million. Do these outcomes represent a successful use of that amount of money? ...

It’s harder to produce Inside View estimates, because so many of the projects either produce vague deliverables (eg a white paper that might guide future action) or intermediate results only (eg getting a government to pass AI safety regulations is good, but can’t be considered an end result unless those regulations prevent the AI apocalypse). Because we tend towards incubating charities and funding research (rather than last-mile causes like buying bednets), achieved measurable deliverables are thin on the ground. But here are things that ACX grantees have already accomplished:

• Improved the living/slaughter conditions of 30 million fish.
• Helped create Manifold Markets, a prediction market site with thousands of satisfied users, whose various spinoffs play a central role in the rationalist/EA community.
• Helped create thousands of jobs in Rwanda and other developing countries
• Passed an instant runoff vote proposition in Seattle.
• Saved between a few dozen and a few hundred lives in Nigeria through better obstetric care.

And here are some intermediate deliverables from grantees:

• Made Australian government take AI x-risk more seriously (estimated from 50th percentile to 60th percentile outcome)
• Gotten the End Kidney Deaths Act (could save >1000 lives and billions of dollars per year) in front of Congress, with decent odds of passing by 2026.
• Plausibly saved 2 billion chickens from painful death over next decade².
• Antiparasitic medication oxfendazole continues to advance through the clinical trial process.

And here are some things that have not been delivered yet but that I remain especially optimistic about:

• Creation of anti-mosquito drones that provide a second level of defense along with bednets.
• Revolutionize diagnosis of traumatic brain injury
• Improve dietary guidelines in developing countries
• Continue to support research and adoption of far UV light for pandemic prevention
• Reduce lead poisoning in Nigeria

I think these underestimate success since many projects have yet to pay off (or to convince me to be especially optimistic), and others have paid off in vague hard-to-measure ways.

This is a beautifully crosswise oriented slice of the entire collective endeavor of effective altruism, and quite a lot of good done (or poised to be done) helped by a not-that-large sum of $3M over 2 cohorts given that GW and OP move 2 OOMs more $ per year. 

It's also been quite intellectually enriching to just see the sheer diversity of proposals to make the world better in these cohorts; e.g. I was a bit let down to learn that the Far Out Initiative didn't pan out ($50k to fund a team to work on pharmacologic and genetic interventions to imitate the condition of Jo Cameron, a 77-year old Scottish woman who is both incapable of experiencing any physical or psychological suffering and has lived an astonishingly well-adjusted life despite that, by creating painkillers to splice into farm animals to promote cruelty-free meat and "end all suffering in the world forever").

Of Scott's lessons learned, this one stood out to me in light of the recent elitism in EA survey I just took, I think because I was leaning towards the same hope he had:

One disappointing result was that grants to legibly-credentialled people operating in high-status ways usually did better than betting on small scrappy startups (whether companies or nonprofits). For example, Innovate Animal Ag was in many ways overdetermined as a grantee - former Yale grad and Google engineer founder, profiled in NYT, already funded by Open Philanthropy - and they in fact did amazing work. On the other hand, there were a lot of promising ACX community members with interesting ideas who were going to turn them into startups any day now, but who ended up kind of floundering (although this also describes Manifold, one of our standout successes). One thing I still don't understand is that Innovate Animal Ag seemed to genuinely need more funding despite being legibly great and high status - does this screen off a theoretical objection that they don't provide ACX Grants with as much counterfactual impact? Am I really just mad that it would be boring to give too many grants to obviously-good things that even moron could spot as promising?

The other takeaway of his that gave me mixed feelings was this one, I think because I'd been secretly hoping for some form of work-life balance compatibility with really effective (emphasis) direct-work altruism:

Someone (I think it might be Paul Graham) once said that they were always surprised how quickly destined-to-be-successful startup founders responded to emails - sometimes within a single-digit number of minutes regardless of time of day. I used to think of this as mysterious - some sort of psychological trait? Working with these grants has made me think of it as just a straightforward fact of life: some people operate an order of magnitude faster than others. The Manifold team created something like five different novel institutions in the amount of time it's taken some other grantees to figure out a business plan; I particularly remember one time when I needed something, sent out a request to talk about it with two or three different teams, and the Manifold team had fully created the thing and were pestering me to launch a trial version before some of the other people had even gotten back to me. I take no pleasure in reporting this - I sometimes take a week or two to answer emails, and all of the predictions about my personality that this implies would be correct - but it's increasingly something that I look for and respect. A lot of the most successful grants succeeded quickly, or at least were quick to get on a promising track. Since everything takes ten times longer than people expect, only someone who moves ten times faster than people expect can get things done in a reasonable amount of time.

Edited to add: I appreciated this comment by Alex Toussaint, an ACX grantee: 

Tornyol (anti-mosquito drones) is based in France and we couldn't have got the support from ACX Grants from a local VC. ...

VCs, like potential employees or clients, have reading grids (i.e. rubrics, a transliteration of « une grille de lecture ») to evaluate pitches. The great thing I found about ACX Grants is that the grid is different, and encourages different kinds of projects. Founder obsession for a problem seems to be encouraged in ACX Grants, although it's clearly discouraged for very early VC funding. VCs like very well made slides, communication abilities, and beautiful people in general, while I've found no such bias for ACX Grants. Being based outside the US is a big minus for American VCs, but ACX Grants almost seems to be favoring it. VCs tend to think a lot by analogy (the Uber for X, the Cursor for Y ...) while I found ACX Grants to be much more thinking from first principles than the median VC I met.

I'm not criticizing the VC reading grid. It obviously comes from experience and it tends to work financially for them. But you have to remember that a large part of the decision comes down to the potential for a quite early (3-4 years) and billion-dollar exit option. Not all projects fit that and it's a good thing to support the other. The other advantage of it is that it selects founders that can go through the hoops of making their project fit the grid. That proves VCs the founders are capable of adapting their message to their interlocutors, which is highly necessary when raising further money, recruiting or discussing with any partner. That's something ACX Grants does not seem to value much.

All in all, ACX Grants is great in that it provides funding with a very unique reading grid, so it helps projects that could get no help anywhere else.

I liked GiveDirectly's recent update via GWWC's email newsletter:

Donations from Giving What We Can community members were delivered to Masauli, Chirtera, and Mtembo villages in Chiradzulu district in Malawi. Together, we funded transfers for 954 Malawians in poverty across all three villages. How did families spend their transfers? Here’s what follow-up surveys show:

Hear directly from Emily and David, who are just a few of the people in Masauli village who received transfers from you and other GWWC members:

Emily and her husband, Evance

“My husband and I relied on farming and casual labor to survive,” said Emily. “We grow maize, tomatoes, and cabbage, but without fertilizer, our harvests were always small. Our house also had no windows, and we used sacks instead of a proper door. I always felt unsafe because I worried people could steal from us.” “When I received my first cash payment, I used it to buy fertilizer. This changed everything. Our harvest increased from six bags to ten, and now we have enough food to last until the next harvest. This has brought a sense of peace since we know we have food. With the second payment, I focused on improving my home. I bought 11 bags of cement and used them to put in a proper floor, plaster the house, and add windows and doors. Now my house is safer, and I feel proud of it. I also started thinking about the future. I bought a goat as an investment and started a small business selling beans. I’m making sure I don’t just spend the money, but use it to build something for my family. My husband and I had been sleeping on a mat since we got married. Now, we sleep on a mattress. No more body pains when we wake up. We can sleep comfortably, something we only heard about from other people before. I also bought a bicycle, so I no longer have to borrow from my neighbors when I want to go to the hospital, market, or maize mill. When other women talk about how nice my house looks, it makes me feel proud. I can see the difference in my life now. I have more dignity, security, and hope for the future.”

David and his wife, Lucy, with their harvest

“Every year, we harvested between 15 to 20 bags of maize, but even with all that, we could not develop our home,” said David. “We had three children in secondary school, so all our money went to their education. As parents, we chose to sacrifice for them, but we always wanted to do more.” “We thought about starting a business, because farming is no longer reliable: the weather changes, and farm inputs like fertilizer are expensive. But we never had enough capital to start. Then GiveDirectly came. With our transfer, we opened a small shop using K200,000 (~$115). At first, we sold simple things, like eggs, drinks, and bread. These were the items that people bought most. The business started well, and we were making K40,000-50,000 (~$23-29) a day. We followed one important rule we were told by a friend: never keep money without restocking. Every day, we used the money to buy more goods: salt, biscuits, sweets, and more. The shop kept growing, and now we’re planning ahead. We want to build a proper shop structure. We’ve already bought 22 iron sheets and 3 bags of cement. We are waiting for the rains to pass so we can mold bricks. We’e also planning to build two rental houses for students from a nearby secondary school. We also want to buy livestock, goats, pigs, and cows, so that we can continue supporting our son who is still in school. For us, this money did not just start a business. It gave us a new path.”

All this for slightly over $800 per beneficiary. Hell of a benchmark, cash transfers.

Daniel Björkegren points out (h/t Deena Mousa's newsletter) that marginal returns to intelligence from advanced AI will be lower in LMICs due to scarcer AI complements, lower digital legibility, and smaller knowledge sectors, so AI that augments knowledge workers is likely to disproportionately benefit richer countries:

The economic implications of this transformation can be characterized by the marginal returns to intelligence (Amodei 2024): how much can we improve economic outcomes as we better generate ideas, process data, and apply knowledge? Intelligence allows us to solve scientific problems, design better products, better anticipate demand, and ensure the right quantities are stocked in the right places. Low-income countries will benefit from innovations developed in rich ones. But within many LMICs, the complements to advanced AI are scarcer, including data centers, reliable electricity, and digital records, as well as experienced knowledge workers. Data centers can be located in countries that already have good infrastructure (‘the cloud’) and accessed remotely. But LMICs are less digitally legible: AI will be less able to understand and act in markets, firms, homes, clinics, and schools that do not record data in structured forms. Overall, we would expect LMICs to be at a disadvantage in integrating advanced AI (Korinek and Stiglitz 2021).

A crucial distinction is that LMICs have much smaller knowledge sectors. LMICs employ fewer than 10% of workers in skilled knowledge work, like managers, technicians, and professionals, relative to 41% in high income countries (Silva 2026). Current AI tools require substantial human guidance. So, firms in rich economies are pursuing a grafting strategy: existing knowledge workers are being asked to integrate AI into their roles, starting from producing slides and emails, and scaling to more sophisticated tasks. In countries with smaller knowledge sectors, there are fewer workers and processes to graft AI onto. Thus a key question is whether advanced AI will mainly empower existing workers, or automate knowledge work completely. In wealthy countries, advocates concerned about jobs suggest that AI systems be designed to augment rather than automate (Acemoglu, Autor, Johnson 2026). But in low-income countries, the more urgent question may be how to provide knowledge services when few knowledge workers are available. Fully automating knowledge work could in fact augment less educated workers, who could ask AI to complete macro tasks like developing marketing strategies, rather than micro tasks like reformatting spreadsheets. However, even automated systems will likely require oversight from entrepreneurs and scientists with deep expertise, which may be sufficiently available only in wealthier countries like Brazil and India.

If AI allows LMICs to grow automated knowledge sectors, would the returns be high or low? One indicator is in wages paid to human workers. The wage returns to college education are slightly higher in lower income countries (Psacharopoulos and Patrinos 2018 and 2025), but educated people often earn higher wages abroad, and some domestic knowledge workers are working on rich countries’ knowledge problems in call centers and business process outsourcing. Lower income economies may not currently be structured to fully tap the decision making entailed in knowledge work (Engbom et al. 2025). If we tasked millions of data scientists with helping smallholder farms, the returns are unlikely to be large: agriculture is constrained elsewhere.

However, if the price of some forms of intelligence declines by orders of magnitude, it may become worth applying intelligence to problems that were never worth assigning a human to. Small manufacturers might generate nuanced designs that would have required a team of industrial engineers, and implement advertising campaigns that would have required large creative teams. Many regions have struggled to agglomerate sufficient human talent; since automated intelligence can be accessed anywhere, it could make businesses more mobile. These opportunities could more fundamentally change economic structure.

So what can LMICs do? Daniel suggests these:

The most capable AI systems currently require large-scale frontier models and large amounts of compute. Governments, firms, and NGOs will need to work with the frontier labs to ensure that the most advanced models speak local languages and understand local contexts. Ensuring that there are multiple suppliers for both models and data centers can reduce prices and risks of lock-in and geopolitical disruption (Athey and Scott Morton 2025).

Governments will also need to push to make economic activity digitally legible, from markets to clinics to schools.

It is also important to ensure that AI can be productively used. That may require training humans to be more productive users of AI, both in applying the tool and having the deeper world knowledge needed to direct it. Firms can also invest in developing AI tools that are complementary to the industrial structure of LMICs, including tools for small scale entrepreneurs who have less education, and for agriculture, like weather forecasting.

The diversity of institutional conditions in low- and middle-income countries may be a comparative advantage. Wealthy countries have evolved similar institutions around human knowledge work; tweaks may lead to local optima. In contrast, systems in low-income countries can differ greatly. Tailoring to different constraints can generate opportunities: for example, Kenyan entrepreneurs coping with unreliable network connections developed techniques to create on-device AI models that are seeing demand around the world (Fastagger). Or, also in Kenya, 90% of people resolve disputes outside the formal justice system (Kenya 2020), and just two doctors serve every 10,000 people, compared to 37 in the United States (WHO, 2022). Firms and NGOs may find creative new solutions, such as offering more efficient ways to settle disputes outside of court, or dynamic medical advice. Governments can take advantage of opportunities to design new regulation for AI, rather than retrofit regulation designed for humans. A lack of established institutions around human knowledge work could also allow harm: what happens when medical AI makes mistakes and there are limited mechanisms to address malpractice? It will take care to develop appropriate new institutions.

Relatedly, Daniel also has a great post on how the poorest use AI. A quote:

AI usage can provide a new window into the needs of the poor, analogous to Google Trends. This can help AI labs and a variety of organizations better serve these populations. We saw early examples of this among teachers in Sierra Leone, who submitted requests for not only facts and lesson plans, but also on handling reports for insurance claims and navigating interpersonal situations with students and supervisors. Another study found that one of the top uses of ChatGPT among gig workers in India and Brazil was for health queries.

What could be computed: An easy start would be to take the standard categorization of requests already reported by the labs (such as writing, technical help, or mapping to industries) and report them specifically for the subset of users in marginalized groups (defined by having cheaper devices, speaking local languages, or using from remote areas). However, these taxonomies are built around knowledge work, and may systematically undercount the ways poor people find the technology useful. Thus, it would be helpful to develop new taxonomies to understand poverty-specific needs, including particular uses within agriculture, health, navigating government bureaucracy, and business advice.

What we might find: The poor are likely to use AI differently from the wealthy: almost no software development, some use for navigating bureaucracy and social problems, more for help with homework, and less for writing assistance. Anthropic has already reported that Claude users in lower-income countries are more likely to request help with coursework. A further breakdown will help us understand if AI is being used only in the wealthiest schools or broadly, and help school systems ensure it is used in ways that support learning.

This topic became more salient to me after attending Deena's EAG talk on how LMICs should respond to AI, which feels like it should be a much bigger topic than it currently is.

Awhile back I came across this slide from the Money for Good project, which I thought was a sobering quantification of the rarity of donor decision-making based on nonprofit outperformance (cost-effectiveness etc). Hope Consulting got this data by surveying 4,000 US individuals with household incomes >$80k (top 30% income back in 2009, comprising 75% of overall individual donations), of which 2,000 were in the >$300k bracket.

Opportunity size for US retail donors in 2009 was ~$45B, so this works out to ballpark $1-1.5B which is still sizeable, e.g. it's more than total annual EA grantmaking has ever been:

How did Hope Consulting get the 3% figure? Top of funnel:

Middle of funnel steep drop-off:

and:

Bottom of funnel has even steeper drop-off, because confirmation bias is the default:

How to raise the 3% figure for donors who give based on nonprofit outperformance? Hope Consulting suggest this framing:

(I disagree with Hope Consulting on that last point, but the rest seems useful.)

What are midsized retail donors like? I used to work in marketing analytics so this piqued my interest. Max / diff to elicit donor value trade-offs -> cluster analysis (a few rounds) yielded these "donor personas":

The lack of demographic variation somewhat surprised me:

As a closing note, the Money for Good project was a major undertaking: 6 months, 4 major funders (including Rockefeller), 4 research orgs (!) partnering with Hope Consulting, etc. This makes me wonder what the 80/20 version of this could look like, with judicious use of Claude Code and such.

Ozy Brennan's Identifying healthy high-demand groups summarises takeaways from Abuses in the Religious Life and the Path to Healing, a book about spiritual abuse written by Dysmas de Lassus, the prior general (person in charge) of the Order of Carthusians. I've spent most of my life in high-demand groups of all kinds so this was interesting to read.

A high-demand group having a lot of people with good virtues isn't a sign it's healthy; toxic groups can have even more of these virtues:

There is no general pattern where people in healthy high-demand communities are, compared to people in toxic high-demand communities, more hardworking, generous, loving, self-controlled, courageous, honest, tolerant, clever, helpful, cheerful, or even compassionate (to outsiders).

Toxic high-demand communities often create a culture of competition to be the most ethical. ... So a toxic high-demand community can have more virtuous members and a greater positive impact on the world than a good high-demand community. 

At the same time, toxic high-demand communities generally pervert genuine virtues... For Catholics, humility becomes self-hatred; the desire to give of yourself to others becomes complete self-denial; forgiveness becomes forgetting the crimes of unrepentant abusers. For rationalists and effective altruists, consequentialism becomes tolerance of wrongdoing because of some far-off future benefit; agency and taking ideas seriously become hearing a bad argument and doing arbitrary bad stuff because of it... The members of a toxic Catholic community may well come off as humble, self-sacrificing, and forgiving; the members of a toxic effective altruist community, consequentialist, agentic, and dedicated to self-improvement.

Actual signs of a high-demand group being healthy:

A healthy high-demand community is patient with its members. It doesn’t expect perfection immediately. It doesn’t hold people to unreasonable standards. It accepts that mistakes and failures are part of the human condition. ...

One of the most important green flags in a high-demand community is the personality of the high-status people. High-status people should readily admit that they make mistakes, believe wrong things, and have personality flaws. ... High-status people should be aware of the suffering of those around them, particularly suffering that’s related to the beliefs and commandments of the high-demand group. If possible, high-status people should do something to alleviate the suffering of group members; if not possible, they should provide comfort and understanding. Most of all, high-status people should be genuinely kind. Not righteous, not self-sacrificing, not heroic, not good. Kind. ...

Any high-demand community is going to make a lot of rules about how you live your life—it’s inherent to the enterprise—but a healthy high-demand community limits its rules to those matters which are really important... As much as is practicable, a good high-demand community allows you to make your own judgments about how to put rules into practice in your own life. ...

In particular, the community and the high-status people should encourage you to take independent initiative... High-status people should praise you for coming up with your own ideas and projects. They should provide help, especially help that isn’t too costly for them (such as making introductions or publicly announcing your project at a meeting).

The simplest criterion de Lassus lays out is the most powerful: are you happy?

Young me used to be confused when people asked "are you happy?" in relation to the high-demand groups I was in. How was personal happiness at all relevant to the collective mission, from which purpose derives? Later on I would meet plenty of excited members of high-demand groups, which was quite the update; there were in fact people in the "ideal" quadrant:

No one can promise you a life without suffering. Being part of a high-demand group may well make you suffer more. ... But ultimately, most of the time, if you’re part of a healthy high-demand group, you should feel a sense of peace and joy. You should reflect on your life, or at least those parts influenced by the high-demand group, and think you know, I’m glad I’m doing this. When it comes right down to it, I like the way my life is going.

Toxic groups are aware that people prefer to stay in groups that make them happy, so they Goodhart it. Often, a group will teach that not being happy is a sin, or that crying yourself to sleep is real joy, a deep and pure kind of joy that the uninitiated would mistake for misery. But, even if the group is gaslighting you, you can still tell how you feel. Take a few hours by yourself or with a trusted friend and reflect: how do I feel about my life? Is my life okay? Do I feel simple pleasures, such as appreciation of a sunrise or companionship with friends or satisfaction at a job well done? If I look back on the past year or two, do I feel a sense of contentment about how it went?

If you are persistently unhappy, the high-demand group may be toxic or it might be all right. But it is clearly wrong for you.

(related)

I admire influential orgs that publicly change their mind due to external feedback, and GiveWell is as usual exemplary of this (see also their grant "lookbacks"). From their recently published Progress on Issues We Identified During Top Charities Red Teaming, here's how external feedback changed their bottomline grantmaking:

In 2023, we conducted “red teaming” to critically examine our four top charities. We found several issues: 4 mistakes and 10 areas requiring more work. We thought these could significantly affect our 2024 grants: $5m-$40m in grants we wouldn’t have made otherwise and $5m-$40m less in grants we would have made otherwise (out of ~$325m total).

This report looks back at how addressing these issues changed our actual grantmaking decisions in 2024. Our rough estimate is that red teaming led to ~$37m in grants we wouldn't have made otherwise and prevented ~$20m in grants we would have made otherwise, out of ~$340m total grants. The biggest driver was incorporating multiple sources for disease burden data rather than relying on single sources.1 There were also several cases where updates did not change grant decisions but led to meaningful changes in our research. 

Some self-assessed progress that caught my eye — incomplete list, full one here; these "led to important errors or... worsened the credibility of our research" (0 = no progress made, 10 = completely resolved):

• Failure to engage with outside experts (8/10): We spent 240 days at conferences/site visits in 2024 (vs. 60 in 2023). We think this type of external engagement helped us avoid ~$4m in grants and identify new grant opportunities like Uduma water utility ($480,000). We've established ongoing relationships with field experts. (more)
• Failure to check burden data against multiple sources (8/10): By using multiple data sources for disease burden, we made ~$34m in grants we likely wouldn't have otherwise and declined ~$14m in grants we probably would have made. We've implemented comprehensive guidelines for triangulating data sources. (more)
• Failure to account for individuals receiving interventions from other sources (7/10): We were underestimating how many people would get nets without our campaigns, reducing cost-effectiveness by 20-25%. We've updated our models but have made limited progress on exploring routine distribution systems (continuous distribution through existing health channels) as an alternative or complement to our mass campaigns. (more)
• Failure to estimate interactions between programs (7/10): We adjusted our vitamin A model to account for overlap with azithromycin distribution (reducing effectiveness by ~15%) and accounted for malaria vaccine coverage when estimating nets impact. We've developed a framework to systematically address this. (more)

(As an aside, I've noticed plenty of claims of GW top charity-beating cost-effectiveness figures both on the forum and elsewhere, and I basically never give them the credence I'd give to GW's own estimates, due to the kind of (usually downward) adjustments mentioned above like receiving interventions from other sources or between-program interventions, and GW's sheer reasoning thoroughness behind those adjustments, seriously, click on any of those "(more)"s)

Some other issues they'd "been aware of at the time of red teaming and had deprioritized but that we thought were worth looking into following red teaming" — again incomplete list, full one here:

• Insufficient attention to inconsistency across cost-effectiveness analyses (CEAs) (8/10): We made our estimates of long-term income effects of preventive health programs more consistent (now 20-30% of benefits across top charities vs. previously 10-40%) and fixed implausible assumptions on indirect deaths (deaths prevented, e.g., by malaria prevention that aren’t attributed to malaria on cause-of-death data). We've implemented regular consistency checks. (more)
• Insufficient attention to some fundamental drivers of intervention efficacy (7/10): We updated our assumptions about net durability and chemical decay on nets (each changing cost-effectiveness by -5% and 11% across geographies) and consulted experts about vaccine efficacy concerns, but we haven't systematically addressed monitoring intervention efficacy drivers across programs. (more)
• Insufficient sideways checks on coverage, costs, and program impact (7/10): We funded $900,000 for external surveys of Evidence Action's water programs, incorporated additional DHS data in our models, and added other verification methods. We've made this a standard part of our process but think there are other areas where we’d benefit from additional verification of program metrics. (more)
• Insufficient follow-up on potentially concerning monitoring and costing data (7/10): We’ve encouraged Helen Keller to improve its monitoring (now requiring independent checks of 10% of households), verified AMF's data systems have improved, and published our first program lookbacks. However, we still think there are important gaps. (more)

I always had the impression GW engaged outside experts a fair bit, so I was pleasantly surprised to learn they thought they weren't doing enough of it and then actually followed through so seriously, this is an A+ example of organisational commitment to and follow-through on self-improvement so I'd like to quote this section in full:

In 2024, we spent ~240 days at conferences or site visits, compared to ~60 in 2023. We spoke to experts more regularly as part of grant investigations, and tried a few new approaches to getting external feedback. While it’s tough to establish impact, we think this led to four smaller grants we might not have made otherwise (totalling ~$1 million) and led us to deprioritize a ~$10 million grant we might’ve made otherwise.

More detail on what we said we’d do to address this issue and what we found (text in italics is drawn from our original report):

• More regularly attend conferences with experts in areas in which we fund programs (malaria, vaccination, etc.).
  • In 2024, our research team attended 16 conferences, or ~140 days, compared to ~40 days at conferences in 2023.35
  • We think these conferences helped us build relationships with experts and identify new grant opportunities. Two examples:
    • A conversation with another funder at a conference led us to re-evaluate our assumptions on HPV coverage and ultimately deprioritize a roughly $10 million grant we may have made otherwise.36
    • We learned about Uduma, a for-profit rural water utility, at a conference and made a $480,000 grant to them in November 2024.37
  • We also made more site visits. In 2023, we spent approximately 20 days on site visits. In 2024, the number was approximately 100 days.38
• Reach out to experts more regularly as part of grant investigations and intervention research. We’ve always consulted with program implementers, researchers, and others through the course of our work, but we think we should allocate more relative time to conversations over desk research in most cases.
  • Our research team has allocated more time to expert conversations. A few examples:
    • Our 2024 grants for VAS to Helen Keller International relied significantly on conversations with program experts. Excluding conversations with the grantee, we had 15 external conversations.
    • We’ve set up longer-term contracts with individuals who provide us regular feedback. For example, our water and livelihoods team has engaged Daniele Lantagne and Paul Gunstensen for input on grant opportunities and external review of our research.
    • We spoke with other implementers about programs we’re considering. For example, we discussed our 2024 grant to support PATH’s technical assistance to support the rollout of malaria vaccines with external stakeholders in the space.39
  • This led to learning about some new grant opportunities. For example:
    • The $150,000 grant to International Rescue Committee (IRC) for desk-based scoping of programs to increase vaccination coverage stemmed from a conversation with an expert.
    • We connected with Zipline at a conference, and made a $55,000 grant to them for desk-based scoping of drones to increase vaccination coverage in December 2024.
    • We are currently considering a $4 million grant that we learned about through an expert conversation.40
• Experiment with new approaches for getting feedback on our work.
  • In addition to the above, we tried a few other approaches we hadn’t (or hadn’t extensively) used before. Three examples:
    • Following our red teaming of GiveWell’s top charities, we decided to review our iron grantmaking to understand what were the top research questions we should address as we consider making additional grants in the near future. We had three experts review our work in parallel to internal red teaming, so we could get input and ask questions along the way.41 We did not do this during our top charities red teaming, in the report of which we wrote “we had limited back-and-forth with external experts during the red teaming process, and we think more engagement with individuals outside of GiveWell could improve the process.”
    • We made a grant to Busara to collect qualitative information on our grants to Helen Keller International's vitamin A supplementation program in Nigeria.42
    • We funded the Center for Global Development to understand why highly cost-effective GiveWell programs aren’t funded by other groups focused on saving lives. This evaluation was designed to get external scrutiny from an organization with expertise in global health and development, and by other funders and decision-makers in low- and middle-income countries.

Some quick reactions:

• I like that GW thinks they should allocate more time to expert conversations vs desk research in most cases
• I like that GW are improving their own red-teaming process by having experts review their work in parallel
• I too am keen to see what CGD find out re: why GW top-recommended programs aren't funded by other groups you'd expect to do so
• the Zipline exploratory grant is very cool, I raved about it previously
• I wouldn't have expected that the biggest driver in terms of grants made/not made would be failure to sense check raw data in burden calculations; while they've done a lot to redress this there's still a lot more on the horizon, poised to affect grantmaking for areas like maternal mortality (prev. underrated, deserves a second look)
• funnily enough, they self-scored 5/10 on "insufficient focus on simplicity in cost-effectiveness models"; as someone who spent all my corporate career pained by working with big messy spreadsheets and who's also checked out GW's CEAs over the years I think they're being a bit harsh on themselves here...

Ben Kuhn has a great essay about how 

all my favorite people are great at a skill I’ve labeled in my head as “staring into the abyss.”¹

Staring into the abyss means thinking reasonably about things that are uncomfortable to contemplate, like arguments against your religious beliefs, or in favor of breaking up with your partner. It’s common to procrastinate on thinking hard about these things because it might require you to acknowledge that you were very wrong about something in the past, and perhaps wasted a bunch of time based on that (e.g. dating the wrong person or praying to the wrong god). However, in most cases you have to either admit this eventually or, if you never admit it, lock yourself into a sub-optimal future life trajectory, so it’s best to be impatient and stare directly into the uncomfortable topic until you’ve figured out what to do. ...

I noticed that it wasn’t just Drew (cofounder and CEO of Wave) who is great at this, but many the people whose work I respect the most, or who have had the most impact on how I think. Conversely, I also noticed that for many of the people I know who have struggled to make good high-level life decisions, they were at least partly blocked by having an abyss that they needed to stare into, but flinched away from.

So I’ve come to believe that becoming more willing to stare into the abyss is one of the most important things you can do to become a better thinker and make better decisions about how to spend your life.

I agree, and I think there's an organisational analogue as well, which GiveWell exemplifies above.

Ivan Gayton was formerly mission head at Doctors Without Borders. His interview (60 mins, transcript here) with Elizabeth van Nostrand is full of eye-opening anecdotes, no single one is representative of the whole interview so it's worth listening to / reading it all. Here's one, on the sheer level of poverty and how giving workers higher wages (even if just $1/day vs the local market rate of $0.25/day "for nine hours on the business end of a shovel") distorted the local economy to the point of completely messing up society:

[00:06:07] Ivan: I had a real moment when I had this construction crew that was rebuilding a wing of the hospital and there were 30 people on this construction crew. And at some point, my boss, the project coordinator says to me, "Ivan, why are you just so obsessed with the construction crew always working? Constantly working and, and you know, never lacking for something to do." And I'm like, "well, because you know, you have a whole crew of 30 people, it's terribly expensive if they're doing nothing, I mean, if they sit there and do nothing all day, that costs, oh wait, $30, huh? Maybe I'll just relax about that."

'Cause you know, my last gig I'd been a forestry project manager, crew of 75 people who cost $450 a day each. So if they, you know, lose an hour of productivity, that's like huge money. A day of productivity is unthinkable.

(Aside: I'm having trouble believing the $450/person-day cost for forestry crew back in the late 90s and early 00s, isn't that $90k/year or $155k today?)

[00:06:56] Ivan: So I bring that to this, you know, African construction crew and the construction crew themselves are kind of exhausted. Like, good lord, this guy's nuts. but that realization that... 30 people on the business end of sledgehammers and shovels and travels cost way less than one hour of my time for an entire day. Wow. That was shocking.

And we were paying more than the local market rate for unskilled labor. I mean, at that time, this is 2003, the, the local market rate for nine hours on the business end of a shovel was a shiny new quarter, 25 cents. We were paying a dollar. So we had this huge lineup of people to work. I kind of rotated through all the villagers, to give as many people as possible a chance for the real unskilled labor. I think the head construction crew guy was getting two bucks a day.

[00:07:52] Elizabeth: yeah, so maybe let's get into the economics of this. On one hand, it seems very generous to pay people four times their normal wage, and it's, you know, a trivial cost to MSF. On the other hand, that does distort the local economy. 

[00:08:07] Ivan: distort is putting it mildly. It just completely messes up the local society. I mentioned that I had done this back in the envelope calculation that we were 75% of the local economy. I mean, what that actually means is we destroyed and distorted the local economy completely; as development practice that would've been utterly and completely unethical.

The only justification for doing something like that is an acute emergency, which it was, it was nigh on a hundred thousand people with literally no access to healthcare whatsoever. The amount of avoidable suffering and death that was going on that we could actually alleviate was something that, you know, in sort of humanitarian practice, I guess we arrogate to ourselves the idea that we can, in a sufficiently emergency situation, justify doing things that would be unethical development practice.

[00:09:06] Elizabeth: Do you think the village was worse off for having the hospital located in their village?

[00:09:11] Ivan: oh yeah. Because we obviously brought this flood of money in, but where does the money go? The doctors and nurses, they're not even local. They're from the capital city. So you're bringing in people from the capitol who then lord it over the local people, price of food jumps up, price of accommodation goes insane. The trickle down opportunities are to be sex workers and cleaners and, you know, servants for these, for these newly created royalty. 

[00:09:46] Elizabeth: you might hope that if the price of food goes up, but their wages are also going up because they're working for the hospital or tangentially, then that would compensate?

[00:09:54] Ivan: Well yeah. For the people who are already, you know, have access to the labor market and are already able to sort of get in on that. Sure. I mentioned that I actually, I deliberately kind of rotated through the villagers to give lots of people a chance, but still, if you're not one of the people who gets a chance or even ever had a chance, or was somebody who's, you know, on the outs with the local powerful people, then we, as these foreigners providing these jobs, we never even see those people.

They don't even get to apply for a job with us. We never even know of their existence. So those people, now, the price of everything is jumped. There's a bunch of newly, much more wealthy people around them, and they're excluded from that. They don't see any of the benefit and all of the harm. So it's, it's terrible.

Counting people is hard. Here are some readings I've come across recently on this, collected in one place for my own edification: 

1. Oliver Kim's How Much Should We Trust Developing Country GDP? is full of sobering quotes. Here's one: "Hollowed out by years of state neglect, African statistical agencies are now often unable to conduct basic survey and sampling work... [e.g.] population figures [are] extrapolated from censuses that are decades-old". The GDP anecdotes are even more heartbreaking
2. Have we vastly underestimated the total number of people on Earth? Quote: "Josias Láng-Ritter and his colleagues at Aalto University, Finland, were working to understand the extent to which dam construction projects caused people to be resettled, but while estimating populations, they kept getting vastly different numbers to official statistics. To investigate, they used data on 307 dam projects in 35 countries, including China, Brazil, Australia and Poland, all completed between 1980 and 2010, taking the number of people reported as resettled in each case as the population in that area prior to displacement. They then cross-checked these numbers against five major population datasets that break down areas into a grid of squares and estimate the number of people living in each square to arrive at totals... According to their analysis, the most accurate estimates undercounted the real number of people by 53 per cent on average, while the worst was 84 per cent out."
3. David Nash's Nigeria's Missing 50 Million People argues that (quote) "Nigeria's official population (~220-230 million) may be significantly inflated and could be closer to 170-180 million (another article claims 120 million) likely driven by political and financial incentives for states". The comments are insightful too, e.g. David's comment that Uganda and Burkina Faso have the opposite problem ("in Burkina Faso the issue was that GDP per capita numbers were calculated from industrial output divided by population estimates so in order to look good, local government had an incentive to underestimate population so they seemed richer"), and Sjlver's comment comparing AMF's population data from distributing bednets to every household to UNFPA data; I've copied their table below: 

Reread Patrick McKenzie (patio11)'s inspirational oral history of VaccinateCA and thought to pull out a few quotes for my own edification. (Patrick posted about this on the forum awhile back, that's worth reading too.)

The following is what it looks like to bake in triage into org decision-making from the top down:

We had an internal culture of counting the passage of time from Day 0, the day (in California) we started working on the project. We made the first calls and published our first vaccine availability on Day 1. I instituted this little meme mostly to keep up the perception of urgency among everyone. 

We repeated a mantra: Every day matters. Every dose matters. 

Where other orgs would say, ‘Yeah I think we can have a meeting about that this coming Monday,’ I would say, ‘It is Day 4. On what day do you expect this to ship?’ and if told you would have your first meeting on Day 8, would ask, ‘Is there a reason that meeting could not be on Day 4 so that this could ship no later than Day 5?’

I started every meeting and status report to the team by reminding them what Day it was. Our internal stats dashboard had a counter of what Day it was. I had a whiteboard in my apartment showing what Day it was. I wrote that every morning as soon as I woke up, and updated the other two numbers right before I went to sleep. Those were: the number of locations we had published to Californians where they could currently get the vaccine, and the number we knew about elsewhere across the United States with the vaccine. 

The latter was zero at this point, of course. I brushed my teeth, wrote my emails, ate my meals, did media interviews, called my family, negotiated with funders, and said my prayers with the zero where I could see it.

A photo of the non-computer version of VaccinateCA’s dashboard, taken on Day 39. It shows VaccinateCA’s then-current understanding of where to find the vaccine: 1,025 sites in California and 0 sites outside of California. Image by Patrick McKenzie. 

I think in absolute terms plenty of orgs do this, Patrick just so happens to be a good writer. But in relative terms it's quite rare, and very meaningful to see, especially for folks like me with a bit of mission orientation. Also this:

After the workday was over and pharmacies stopped answering their phones, the workday began again immediately, as much of our engineering team turned off their day job computers and logged on to Discord to digest what we had learned. We worked into the night, and not infrequently through it, to be ready for 9:30am the next morning.

It was often a brutal crunch. I told the team to take care of themselves and keep an eye on one another. The mission wouldn’t be served by anyone ending up in the hospital. But subject to that constraint, we worked like men and women possessed, because people were dying.

On entrepreneurship:

Part of entrepreneurship is having a vision of something that is possible and figuring out what is necessary to bring it into the world. A cynic would say that the world has a secret: Building things is not actually possible, because different organizations have different timelines allowing access to different resources, and it is impossible to correctly sequence things to satisfy all the requirements in order to build anything. An entrepreneur would tell the cynic a secret in return: You can carefully titrate the amount of truth to various parties to dissolve these deadlocks.

Your donor-advised fund won’t let you donate unless we’re a 501(c)(3)? Well, you’d donate if we were a 501(c)(3), right? Great. We’re applying for approval as a 501(c)(3) from the IRS. Can I put you down for $25,000? Dear IRS examiners: I have a written commitment from a charitable allocator for a $25,000 donation contingent on 501(c)(3) status. As you are aware, IRS procedure says that this qualifies for expedited processing. Oh, yes, government actor whose cooperation we need, we’re a nonprofit. Look at this official paperwork from Delaware. It says that the State of Delaware is officially aware that I say we’re a nonprofit. Not good enough? Our 501(c)(3) status? The IRS is busy approving it, on an expedited basis.

This is somewhat reminiscent of what Scott Alexander wrote about a very different person, although what Patrick calls "carefully titrating the amount of truth to various parties" Scott outright labeled "blatant lies"; my takeaway is that it's possible to do a more ethical version of the description below: 

I started the book with the question: what exactly do real estate developers do? They don’t design buildings; they hire an architect for that part. They don’t construct the buildings; they hire a construction company for that part. They don’t manage the buildings; they hire a management company for that part. They’re not even the capitalist who funds the whole thing; they get a loan from a bank for that. So what do they do? Why don’t you or I take out a $100 million loan from a bank, hire a company to build a $100 million skyscraper, and then rent it out for somewhat more than $100 million and become rich?

As best I can tell, the developer’s job is coordination. This often means blatant lies. The usual process goes like this: the bank would be happy to lend you the money as long as you have guaranteed renters. The renters would be happy to sign up as long as you show them a design. The architect would be happy to design the building as long as you tell them what the government’s allowing. The government would be happy to give you your permit as long as you have a construction company lined up. And the construction company would be happy to sign on with you as long as you have the money from the bank in your pocket. Or some kind of complicated multi-step catch-22 like that. The solution – or at least Trump’s solution – is to tell everybody that all the other players have agreed and the deal is completely done except for their signature. The trick is to lie to the right people in the right order, so that by the time somebody checks to see whether they’ve been conned, you actually do have the signatures you told them that you had. The whole thing sounds very stressful.

But I digress. Relatedly:

That’s life in a start-up: trying to create enough impact very quickly to convince people to give you more resources, while understanding that the default case is running out of resources, and, by the way, everything is broken all the time.

On do-gooder precocity:

We saw peer projects sprout up in many states, with varying levels of effort and success. Many credited us as an inspiration. One peer project was ILVaccine.org, a project of Eli Coustan. We were working with him for a while before I learned he was in middle school. When I later blanked on his name and asked someone about the public health infrastructure coordinator who was a middle school student I was asked to be more specific.

On ownership and accountability, a case study:

Our scripts instructed callers to take down notes from pharmacists as to how to get an appointment for doses they had, when they had them. A Rite Aid pharmacy in San Bernardino asked our caller to sign up for an appointment at the county health department’s website. Our caller, who had been calling into San Bernardino frequently and had seen that website frequently, remarked that he had seen no Rite Aid listed as a possible vaccination location.

The pharmacist then swore into the telephone, hung up, and immediately called the county health department.

I want you to visualize the operations of county health offices during the middle of the pandemic. A stressed staff are busy coordinating a logistical challenge larger than any they’ve faced in their careers. Their phones are ringing off the hook. The consultancy that won the bid finally delivered the freaking appointment website, thank god. It is crappy and barely works but at least it is finally here. You just have to download all of the email attachments from the pharmacy chain corporate offices, maybe fix a few in Excel because those jokers can’t read clear instructions, then upload them into the administrative side of the portal, and finally people can register for appointments to get the doses sitting in pharmacy freezers.

Rite Aid’s data never made it into the system. Maybe Rite Aid forgot to send it and nobody followed up. Maybe it got eaten by the county’s spam filter. Maybe a public health worker with a million things to do did 999,999.

There were 13 Rite-Aids in San Bernardino county. None of them, despite being in possession of the most desirable object in the world, had received a single appointment. No pharmacist, with years of training in healthcare, noticed this before we told them.

Why would they? Every pharmacy has lots of tiny glass vials and bottles of pills and satchels of powder. Patients were coming in and getting healthcare. It was no one’s job to check that any particular vials got distributed quickly. Pharmacists are not pharmaceutical sales representatives; they do not generate demand. Pharmacists service appointments and prescriptions, deliver healthcare, and go on to the next patient. If you walked up to the counter or called in and asked about Covid appointments, they’d tell you to book one with the county and move on to the next customer. Just another day at the pharmacy.

You might object and say that it must have been someone’s job to actually get those doses injected. Someone who worked . . . at the White House? Okay, no, but at the CDC? Okay, no, but at the California Office of the Governor? Okay, no, but at the county health department? Okay, no, county health departments do not track individual SKU inventory levels at individual pharmacies, that’s actually not a thing. OK, then, Rite Aid – some logistics manager at Rite Aid should have opened a spreadsheet, seen an SKU like #DJFKJDF3285325 with 50 doses available out of 50 shipped at a location in San Bernardino, and immediately said, ‘Oh, #$*#(%. That drug being in supply is equivalent to a life-threatening medical emergency. I will now get out my emergency procedures binder.’ Nope, that is also not a reasonable expectation.

Each of these organizations wants someone else to be responsible for catching errors like this, and they want them to be effective at doing so. They want, and the nation wants, an organization to be accountable for delivering the vaccine.

VaccinateCA considered this bug, and anything else that kept vaccines in freezers while patients were still waiting, to be our problem.

This problem was fixed because a caller from VaccinateCA thought to say, ‘Wait, I notice that I am confused’. It was fixed within about half an hour of being noticed. We estimate more than 500 doses were quickly taken out of freezers, thawed, and injected into waiting arms. Those arms were often attached to people who had been refreshing the county website every few minutes hoping new appointments would finally open up.

This was early and dramatic evidence to me that California was benefiting from having an organization that felt itself accountable for delivering the vaccine.

On how much of Patrick's job in the early days as CEO was bringing in funding:

VaccinateCA had an extremely effective team, as good as any I have ever had the privilege of working with. They were instrumental in almost everything I did and probably could have done almost all of it without me. The main thing I uniquely brought to the table, and spent a lot of my cycles on, was finding the money.

Our earliest funding source was a prepaid debit card I spun up and posted in our public Discord, with $20,000 of my own savings on it. That would cover servers and software and similar for a while. That was not going to be sufficient to get a proper nonprofit with a paid staff off the ground, and I knew we would both need that and need to be read as that to get cooperation from some quarters.

I called in favors and plead our case up and down the tech industry, and scraped together about $1.2 million in funding.

This was below what I initially thought I could reasonably raise, and below what I thought we likely needed. For better or worse, it would have been a lot easier if I had pitched it as: ‘Just make a small angel investment in a promising technology company whose CEO thinks his job is burning through investor dollars as quickly as possible while driving the total addressable market to zero. You won’t make any money on it, but think of the story.’

But while that tech company would probably have been well funded, it would have smelled like a tech company to potential partners. To accelerate shots in arms we urgently needed the cooperation of people who, if confronted with the proposition ‘Big Tech is bringing about the end of constitutional democracy so that it can gather more of your data to sell’, would like that tweet from their iPhone. I have a different point of view, but debating would not have put shots in arms.

We were approximately the most privileged nascent nonprofit imaginable in terms of access to funding, and given that it directly unlocked our ability to help people find the vaccine, I don’t want to complain too much about the process of getting it. I will record, for the benefit of future charitable founders, that probably half of my time from Day 8 to Day 160 was spent chasing funding, dealing with funders and the nonprofit industrial complex, pitching (and pitching and pitching and pitching) large pots of tech money earmarked for pandemic response, filing required reports with funders and the government, and diligently accounting for every penny spent.

It was a bit of a culture shock coming from the technology industry. Tech isn’t exactly profligate, but it certainly empowers twentysomething engineers to spend thousands of dollars by typing a command into their terminals. An engineer who fumble-fingers a command and spends ten times what they expected to is told to type more carefully next time.

On the advantages private individuals and organizations have over official initiatives:

The government of the United States is an intrinsically political entity. We were formally nonpartisan (and even better, as a 501(c)(3) nonprofit, had to be). Informally, to quote a memo I wrote early on, we would do a deal with the devil himself if it got one more patient one more dose. We didn’t need to worry about compromising anyone’s reelection chances by being too maniacally focused on shots in arms to consider the big picture. We had no responsibilities to allies in our party, like not overshadowing their efforts, because we had no party and, for that matter, no shadow.

I knew that many political actors wanted to hoard the facts of their operations so that they could claim credit for them. I just didn’t particularly care. If an actor had a dose available to the public, it was going to be publicized anywhere we could cause it to be.

(It's hard to convey how much I like and appreciate that last paragraph.)

On funder-nonprofit misalignment:

Some potential funders were in after an emailed discussion that could fit in a tweet. Some were in after a single call.

Some potential funders had expectations that were misaligned with us. VaccinateCA was always designed as a rapid-response project that would spin up, cover for an urgent gap in US infrastructure, and then spin down once the work was done. We explained this at length while emphasizing that we wanted to work with funders who could reach a decision quickly. Every day mattered. Every dose matters.

We had some interactions where we were put through weeks or months of grant writing, which sounds like ‘turn in a paper and wait for it to be graded’ and is more ‘schedule sufficient meetings with sufficient backing documentation to buy a company for $20 million’. The funder eventually passed on us, saying they worried we would create no institution with enduring value after the pandemic.

I don’t begrudge anyone’s choices of how to spend their money, particularly charitably earmarked money. I will point out that the gap in expectations between a grant-reviewing team keen on institution building and a nonprofit with an urgent unmet need is a very, very common story in nonprofit fundraising.

Many pots of money have preferences with regard to how they allocate, and those preferences change with the seasons. Have I mentioned that health equity was all the rage in California in 2021? I put on my best face to funders, explained that the system of siloing vaccine information benefited only people who were professionally competent at navigating the American healthcare bureaucracy. I suggested that publishing vaccine locations to a website and Google and every other place we could think of was an improvement over that status quo. I didn’t engage with debates about how, and this was made absolutely explicit in some conversations – perhaps saving lives but failing to save lives in preferred demographic ratios would be considered worse than not engaging in the project at all.

(I'll be upfront that despite having spent a couple of my formative years in California, my bias leans so far in Patrick's direction that it'd probably be useful for me to hear out the strongest counterargument, especially in the context of triage.)

I just learned about Zipline, the world's largest autonomous drone delivery system, from YouTube tech reviewer Marques Brownlee's recent video, so I was surprised to see Zipline pop up in a GiveWell grant writeup of all places. I admittedly had the intuition that if you're optimising for cost-effectiveness as hard as GW do, and that your prior is as skeptical as theirs is, then the "coolness factor" would've been stripped clean off whatever interventions pass the bar, and Brownlee's demo both blew my mind with its coolness (he placed an order on mobile for a power bank and it arrived by air in thirty seconds flat, yeesh) and also seemed the complete opposite of cost-effective (caveating that I know nothing about drone delivery economics). Quoting their "in a nutshell" section:

In December 2024, GiveWell recommended a $54,620 grant to Zipline for a six-month scoping project. Zipline will use this time to review ways that they could use drones to increase vaccination uptake, especially in hard-to-reach areas with low vaccination coverage and high rates of vaccine-preventable diseases. ...

We recommended this grant because:

• Drones are an intuitive way to bring vaccines closer to communities with low coverage rates, especially when demand for vaccination exists, but conditions like difficult terrain, poor infrastructure, weak cold chain, or insecurity make it difficult for families to access immunizations.
• This grant aligns with our strategy of making several scoping grants to high-potential organizations to source promising ideas for solving bottlenecks in the routine immunization system, and then testing these concepts.

Okay, but what about cost-effectiveness? Their "main reservations" section says

Evidence on cost-effectiveness of drones for vaccine delivery is limited, and we have not modeled the cost-effectiveness of the types of programs that Zipline plans to consider, nor the value of information for this scoping grant. 

An internal review conducted in 2023 focused on drones for health was generally skeptical about there being many opportunities in this area that would meet GiveWell’s bar, although this scoping grant will focus on the most promising scenario (remote areas with high rates of vaccine-preventable diseases and low vaccination coverage rates).

Is there any evidence of cost-effectiveness at all then? According to Zipline, yes — e.g. quoting the abstract from their own 2025 modelling study:

Objectives: In mid-2020, the Ghana Health Service introduced Zipline’s aerial logistics (centralized storage and delivery by drones) in the Western North Region to enhance health supply chain resilience. This intervention led to improved vaccination coverage in high-utilization districts. This study assessed the cost-effectiveness of aerial logistics as an intervention to improve immunization coverage. 

Methods: An attack rate model, adjusted for vaccination coverage and vaccine efficacy, was used to estimate disease incidence among vaccinated and unvaccinated populations, focusing on 17 022 infants. Incremental cost-effectiveness ratios of US dollar per averted disability-adjusted life-year (DALY) were evaluated from societal and government perspectives, using real-world operations data. ...

Results: In 2021, aerial logistics averted 688 disease cases. Incremental cost-effectiveness ratios were $41 and $58 per averted DALY from the societal and government perspectives, respectively. The intervention was cost-saving when at least 20% of vaccines delivered by aerial logistics replaced those that would have been delivered by ground transportation, with potential government savings of up to $250 per averted DALY. Probabilistic sensitivity analysis confirmed the robustness of these findings. 

That's super cost-effective. For context, the standard willingness-to-pay to avert a DALY is 1x per capita GDP or $2,100 in Ghana, so 35-50x higher. Also:

... we calculated that aerial logistics facilitated the completion of an additional 14 979 full immunization courses... We estimated that 4 children’s lives (95% CI 2–7) were saved in these districts during 2021. ... the intervention averted a total of $20 324 in treatment costs and $2819 for caregivers between lost wages and transport. 

At a cost of $0.66 per incremental FIC (fully immunized child), this approach outperforms other delivery methods analyzed in the review, including the most cost-effective category of interventions identified, namely “Delivery Approach” interventions, such as monthly immunization by mobile teams in villages and the enhancement of satellite clinic immunization practices. 

(GW notes that they'd given Zipline's study a look and "were unable to quickly assess how key parameters like program costs and the impact of the program on vaccination uptake and disease were being estimated". Neither can I. Still pretty exciting)

What's the most cost-effective economic growth-boosting intervention? It's cat mascots. I just learned about Tama the calico cat (via @thatgoodnewsgirl on Instagram), who "gained fame for being a railway station master and operating officer at Kishi Station on the Kishigawa Line in Kinokawa, Wakayama Prefecture, Japan". 

The career section of her Wikipedia page astounded me:

Tama was born in Kinokawa, Wakayama, and was raised with a group of stray cats that used to live close to Kishi Station. They were regularly fed by passengers and by Toshiko Koyama, the informal station manager at the time.

The station was near closure in 2004 because of financial problems on the rail line. Around this time, Koyama adopted Tama. Eventually the decision to close the station was withdrawn after the citizens demanded that it stay open.^[3] In April 2006, the newly formed Wakayama Electric Railway destaffed all stations on the Kishigawa Line to cut costs, and at the same time evicted the stray cats from their shelter to make way for new roads leading to the stations. Koyama pleaded with Mitsunobu Kojima, president of Wakayama Electric Railway, to allow the cats to live inside Kishi Station; Kojima, seeing Tama as a maneki-neko (beckoning cat), agreed to the request.^[4] ...

In lieu of an annual salary, the railway provided Tama with a year's worth of cat food and a gold name tag for her collar stating her name and position. A station master's hat was specially designed and made to fit Tama, and took more than six months to complete.^[6] In July 2008, a summer hat was also issued to Tama for hotter weather.^[7] Tama's original gold name tag was stolen by a visitor on October 10, 2007, but a replica was quickly made to replace it.^[8]

The publicity from Tama's appointment led to an increase in passengers by 17% for that month as compared to January 2006; ridership statistics for March 2007 showed a 10% increase over the previous financial year. A study estimated that the publicity surrounding Tama has contributed 1.1 billion yen to the local economy.^[9]

That's actually lowballing Tama. If you click on footnote 9, it's ¥1.38 billion ($13.1 million at the exchange rate back then):

With 55,000 more people having used the Kishigawa Line than would normally be expected, Tama is being credited with a contribution to the local economy calculated to have reached as much as 1.1 billion yen (10.44 million dollars) in 2007 alone, according to a study announced last week.

Katsuhiro Miyamoto, a professor at Kansai University's School of Accountancy, said picture books and other merchandise featuring the feline stationmaster also produced significant economic effects. A television appearance and other publicity surrounding Tama -- who receives cat food in lieu of a salary -- was worth 280 million yen, according to Miyamoto.

In contrast, Tama was really cheap. Gemini 2.5 Pro helpfully estimated the following cost breakdown, which adds up to a median of ¥110,000:

• A year's supply of food probably cost ¥30,000 - ¥60,000
• Vet care could range from ¥30,000 - ¥70,000 annually to cover everything from routine check-ups to vaccinations, flea/tick prevention, and potential unexpected health issues
• Litter, bedding, toys, and grooming supplies probably cost ¥10,000 - ¥20,000
• Her promotion to "super station master" (superintendent equivalent) came with an office (a converted ticket booth containing a litter box) estimated at ¥50,000 - ¥110,000, amortised over Tama's nearly decade-long career
• Tama's two custom-made hats (summer and other seasons) weren't cheap, probably ¥10,000 - ¥30,000 each, but worth including in the compensation package to retain top talent
• In contrast her two gold-plated name tags (the first was stolen by a visitor) likely weren't as expensive, ¥2,000 - ¥5,000  

This works out to almost exactly 10,000:1 benefit-cost ratio just from increased ridership, and that's excluding merchandise, TV appearance and other publicity etc — well beyond even Open Phil's higher-than-ever 2,100x bar! 

In fact, a few hundred thousand Tamas would already double global economic growth, and the world's Tama-carrying capacity is probably 3-4 OOMs higher at least (the World Population Review's estimate of over a billion cats worldwide today should be interpreted as a lower bound). Talk about feline-powered explosive economic growth potential... 

Tama's office

Thought to share some infographics on animal advocacy org expenses from the Stray Dog Institute's 2024 State of the Movement report, which I learned about via Moritz's excellent post. 

Most org spending is in North America and Europe: 

North American and European orgs accounted for most of the spend in sub-Saharan Africa and LATAM & the Caribbean, despite spending (say) only ~1% of their total expenses in SSA: 

I don't have any good sense of how this Global North-dominated funding potentially skews priorities, but this drill down by animal category may be a start:

As well as this drill down by intended outcome. Naively it seems that SSA's allocation looks like North America's for instance, except that the latter has a greater proportion of org spending going to increasing availability of animal-free products, which makes sense given relative wealth:

For what it's worth, here's what the funding allocations look like for animal categories as a whole: mostly terrestrial animals, mostly farmed.

I'd be keen to get takes from folks in the know on what seems underfunded here. Farmed insects jump out: just $135k out of $260m overall (~0.05%) seems nuts.

I also wonder about the skewing of priorities due to outside funding. Moritz wrote

Why this could matter:

1. Strategy and local context
   Money shapes movements. It affects which projects get tried, which organisations survive, what gets measured, and what kinds of risks are acceptable. If most capital comes from a different region, that may (subtly, unintentionally) shape priorities and assumptions. Some strategic questions are deeply context-dependent (politics, culture, institutional incentives, reputational dynamics). Local donors may bring different assumptions and highlight different opportunities or overlooked risks.

which I agree with; another angle is Tom & Karthik's point that

• Most farmed animals live in low- and middle-income countries (LMICs), but traditional Western animal advocacy tactics often fail there due to fragmented supply chains, informal markets, and weak enforcement.
• Creating meaningful change for these animals requires exploring both building infrastructure for change and trying alternative pressure points like farmer cooperatives and local institutions.

although it also isn't clear to me from the infographics above whether meaningful change in their sense would be reflected in the drill downs.

It's mind-blowing to me that AMF's immediate funding gap is $462M for 2027-29. That's 56-154,000 lives (mostly under-5 children) at $3-8k per life saved, maybe fewer going forward due to evolving resistance to insecticides, but it wouldn't change the bottomline that this seems to be a gargantuan ball dropped. Last time AMF's immediate funding gap was over $300M for 2024-26, so it's grown 50%(!) this time round. Both times the main culprit was the same, the Global Fund's funding replenishment shortfall vs target, which affects programmatic planning in countries. I'd like to think we're collectively doing our part (e.g. last year GiveWell directed $150M to AMF, more than to any other charity, which by their reckoning is expected to save ~27k lives over the next 1-2 years), but it's still nuts to me that such a longstanding high-profile "shovel-ready" giving opportunity as AMF can still have such a big and growing gap!

I just learned that Trump signed an executive order last night withdrawing the US from the WHO; this is his second attempt to do so. 

WHO thankfully weren't caught totally unprepared. Politico reports that last year they "launched an investment round seeking some $7 billion “to mobilize predictable and flexible resources from a broader base of donors” for the WHO’s core work between 2025 and 2028. As of late last year, the WHO said it had received commitments for at least half that amount".

Full text of the executive order below: 

WITHDRAWING THE UNITED STATES FROM THE WORLD HEALTH ORGANIZATION 

By the authority vested in me as President by the Constitution and the laws of the United States of America, it is hereby ordered: 

Section 1.  Purpose.  The United States noticed its withdrawal from the World Health Organization (WHO) in 2020 due to the organization’s mishandling of the COVID-19 pandemic that arose out of Wuhan, China, and other global health crises, its failure to adopt urgently needed reforms, and its inability to demonstrate independence from the inappropriate political influence of WHO member states.  In addition, the WHO continues to demand unfairly onerous payments from the United States, far out of proportion with other countries’ assessed payments.  China, with a population of 1.4 billion, has 300 percent of the population of the United States, yet contributes nearly 90 percent less to the WHO.  

Sec. 2.  Actions.  (a)  The United States intends to withdraw from the WHO.  The Presidential Letter to the Secretary-General of the United Nations signed on January 20, 2021, that retracted the United States’ July 6, 2020, notification of withdrawal is revoked.

(b)  Executive Order 13987 of January 25, 2021 (Organizing and Mobilizing the United States Government to Provide a Unified and Effective Response to Combat COVID–19 and to Provide United States Leadership on Global Health and Security), is revoked.

(c)  The Assistant to the President for National Security Affairs shall establish directorates and coordinating mechanisms within the National Security Council apparatus as he deems necessary and appropriate to safeguard public health and fortify biosecurity.

(d)  The Secretary of State and the Director of the Office of Management and Budget shall take appropriate measures, with all practicable speed, to:

(i)    pause the future transfer of any United States Government funds, support, or resources to the WHO;

(ii)   recall and reassign United States Government personnel or contractors working in any capacity with the WHO; and  

(iii)  identify credible and transparent United States and international partners to assume necessary activities previously undertaken by the WHO.

(e)  The Director of the White House Office of Pandemic Preparedness and Response Policy shall review, rescind, and replace the 2024 U.S. Global Health Security Strategy as soon as practicable. 

Sec. 3.  Notification.  The Secretary of State shall immediately inform the Secretary-General of the United Nations, any other applicable depositary, and the leadership of the WHO of the withdrawal.

Sec. 4.  Global System Negotiations.  While withdrawal is in progress, the Secretary of State will cease negotiations on the WHO Pandemic Agreement and the amendments to the International Health Regulations, and actions taken to effectuate such agreement and amendments will have no binding force on the United States.  

Sec. 5.  General Provisions.  (a)  Nothing in this order shall be construed to impair or otherwise affect: 

(i)   the authority granted by law to an executive department or agency, or the head thereof; or 

(ii)  the functions of the Director of the Office of Management and Budget relating to budgetary, administrative, or legislative proposals. 

(b)  This order shall be implemented consistent with applicable law and subject to the availability of appropriations. 

(c)  This order is not intended to, and does not, create any right or benefit, substantive or procedural, enforceable at law or in equity by any party against the United States, its departments, agencies, or entities, its officers, employees, or agents, or any other person. 

THE WHITE HOUSE,

    January 20, 2025.

I really like Bob Fischer's point #4 from deep within the comment threads of his recent post and thought to share it more widely, seemed like wise advice to me:

FWIW, my general orientation to most of the debates about these kinds of theoretical issues is that they should nudge your thinking but not drive it. What should drive your thinking is just: "Suffering is bad. Do something about it." So, yes, the numbers count. Yes, update your strategy based on the odds of making a difference. Yes, care about the counterfactual and, all else equal, put your efforts in the places that others ignore. But for most people in most circumstances, they should look at their opportunity set, choose the best thing they think they can sweat and bleed over for years, and then get to work. Don't worry too much about whether you've chosen the optimal cause, whether you're vulnerable to complex cluelessness, or whether one of your several stated reasons for action might lead to paralysis, because the consensus on all these issues will change 300 times over the course of a few years.

Stuart Buck's new post over at The Good Science Project has one of the hardest-hitting openings I've read in a while: 

Many common medical practices do not have strong evidence behind them. In 2019, a group of prominent medical researchers—including Robert Califf, the former Food and Drug Administration (FDA) Commissioner—undertook the tedious task of looking into the level of evidence behind 2,930 recommendations in guidelines issued by the American Heart Association and the American College of Cardiology. They asked one simple question: how many recommendations were supported by multiple small randomized trials or at least one large trial? The answer: 8.5%. The rest were supported by only one small trial, by observational evidence, or just by “expert opinion only.”

For infectious diseases, a team of researchers looked at 1,042 recommendations in guidelines issued by the Infectious Diseases Society of America. They found that only 9.3% were supported by strong evidence. For 57% of the recommendations, the quality of evidence was “low” or “very low.” And to make matters worse, more than half of the recommendations considered low in quality of evidence were still issued as “strong” recommendations.

In oncology, a review of 1,023 recommendations from the National Comprehensive Cancer Network found that “…only 6% of the recommendations … are based on high-level evidence”, suggesting “a huge opportunity for research to fill the knowledge gap and further improve the scientific validity of the guidelines.”

Even worse, as shown in a great book co-authored by current FDA official Vinay Prasad, there are many cases where not only is a common medical treatment lacking the evidence to support it, but also one or more randomized trials have shown that the treatment is useless or even harmful!

(I'll refrain from quoting the rest and suggest instead you check out his post)

I spent most of my early career as a data analyst in industry, which engendered in me a deep wariness of quantitative data sources and plumbing, and a neverending discomfort at how often others tended to just take them as given for input into consequential decision-making, even if at an intellectual level I knew their constraints and other priorities justified it and they were doing the best they could. ...and then I moved to global health applied research and realised that the data trustworthiness situation was so much worse I had to recalibrate a lot of expectations / intuitions. 

In that regard I appreciate GiveWell's new guidance on burden note:  

Disease burden estimates, such as child mortality rates, are a key input in our cost-effectiveness analyses. Historically, for consistency and convenience, we've primarily relied on a single source for these estimates. 

Going forward, we plan to consider multiple sources for burden estimates, apply a higher level of scrutiny to these estimates, and adjust for potential biases or inaccuracies, like we do when estimating other parameters in our models. 

This change has already led to us making over $25m in additional grants we would not have otherwise. (Footnote: Our updated estimates of malaria burden in Chad have led us to allocate $3.3 million in grantmaking for seasonal malaria chemoprevention (more), and $25.9m for insecticide-treated nets (not yet published).) We expect to consider additional research to improve estimates of burden of disease in the future.

The rest of the note was cathartic to skim-read. For instance, when I looked into the idea of distributing low-cost glasses to correct presbyopia in low-income countries awhile back (a problem that afflicts over 1.8 billion people globally with >$50 billion in annual lost potential productivity annually in LMICs alone), the industry data analyst in me was dismayed to learn that the WHO didn't even collect data on how many people needed glasses prior to 2008, so governments and associated stakeholders understandably prioritised allocation of resources towards surgical and medical interventions instead. I think the existence of orgs like IHME and OWID greatly improve the GHD data situation nowadays, but there are many "pockets" where it remains a far cry from what it could be, so I appreciated that GiveWell said they're considering 

Fund data collection. This includes potentially funding additional nationally representative surveys (DHS/MIS/MICS) or additional modules to these surveys, or supporting more autopsy data collection to better understand cause-specific mortality, particularly for malaria in sub-Saharan Africa. Our guess is that part of the reason different models disagree is that the data underlying these models is limited. We may look for cases where we could fund additional data collection to improve burden of disease estimates.

Another example: a fair bit of my earlier analyst work involved either reconciling discrepant figures for ostensibly similar metrics (e.g. campaign revenue breakdowns etc) or root-cause analysing-via-data-plumbing whether a flagged metric needed to be acted on or was a false positive, which made me appreciate this section: 

Key uncertainties: ...

There are likely technical nuances we haven't captured. We've found that comparisons between sources are more complex than they first appear. For example, we recently learned that IGME and IHME define diarrheal diseases differently. Similar technical differences likely exist elsewhere.

Possible next steps:

Get a better understanding of what’s driving differences in models. This may come from bringing together modeling groups in regions with high disagreement to understand methodological differences.

Look for ways to improve model transparency. We’ve found it difficult to engage with burden of disease models, and think that finding ways to see inside the black box of how they produce estimates may make it easier to understand which estimates to rely on and how to improve them.

David Manheim's If AI is normal technology, history is not reassuring is a good read (emphasis mine): 

There’s a truism that technology is good - even if it creates winners and losers, it improves the world. Toby Ord argues that the conclusions about the benefits of technology is sensitive to the end of humanity - but this jumps over the transitions by starting from the assumption^[1] that “long-term progress in science, technology, and values have tended to make people’s lives longer, freer, and more prosperous.” That is, looking back historically, the net impact misses the immense immediate harms of large scale technological changes that can last for generations.

As I’ll explain, the largest technological revolutions in human history are arguably the agricultural revolution and the industrial revolution. In both cases, the vast majority of those immediately affected were harmed, not helped. Of course, the longer term impact was positive; those benefits are not in question^[2] - not that those alive during the transition should have cared.

The two obvious examples

The invention of agriculture led to increased food availability and around ten thousand years of greatly worsened health and lifespans^[3]. The wealthiest and most powerful people benefited immensely from the population explosion, and from the wars that larger populations enabled and required; the population suffered from both malnutrition, and that same increase in the scale of violence^[4].

The invention of industry was more beneficial to the consumer - but not to those directly involved. In 1840, over a third of the British population worked in a factory. This was bad, in part directly due to factory worker deaths, but also due to pollution and disease. Mortality shot up over the middle of the 1800s - the famed “urban penalty”, especially among children, albeit partially offset by reduced deaths because of sanitation later in the century^[5]. 

3 more examples via ChatGPT/Manheim (which provided 5 including the 2 above; I omitted them -- again emphasis mine):

1. Writing and external symbolic storage - Administration, law, history, mathematics, scripture, bureaucracy, long-distance coordination. Early writing mostly helped palaces, temples, tax systems, accounting, property claims, labor control, and bureaucracy before it helped ordinary people read novels or do science. So the near-term “users” benefited, but many affected subjects may have faced more legible extraction and administration. Evidence from early Mesopotamia links writing with larger government buildings and multi-level bureaucracies.
2. Metallurgy, especially iron - Tools, weapons, plows, empires, deforestation, intensified agriculture, military expansion. Bronze matters too, but iron’s scale and availability make it more transformative. Better tools helped agriculture and craft production, but weapons, fortifications, conquest, inequality, and elite control plausibly dominated early experience for many. The case is less clean because metal tools also had immediate productive benefits, but the war-and-hierarchy channel is very real.
3. Electricity + computation + telecommunications - I’d bundle these reluctantly as the “information-electrical stack”: telegraph, telephone, radio, electric grids, computers, internet, AI. This led to surveillance, labor displacement, attention capture, military command/control, financial acceleration, and dependence on fragile networks^[8].

Manheim's argument for what makes good technologies good:

There have been a couple of revolutionary changes in medicine and public health over the past couple centuries. The vaccine revolution, the advent of modern sanitation, and infection control each include a strong case that they were immediately beneficial, and stayed that way indefinitely^[9]. Refrigeration, washing machines, and bicycles^[10] are arguably more examples in this class. So some technologies really are just positive - but we need to ask which ones.

I think there’s a simple explanation; directly good things are good, but many other classes of transformative change end up disruptive in ways that hurt before they can help^[11]. Technologies that have first order impacts on coordination and production, or that empower groups in other ways, tend to differentially benefit the powerful in ways that are harmful to others, either directly or indirectly^[12].

I find myself instinctively resisting Manheim's explanation, since I'm generally keen on improvements to empowerment and coordination, but have to admit it parsimoniously explains the small-n historical track record above. The issue, as always, seems to be the gap between the beautiful ideal ("more empowerment! more coordination!") and the unavoidably-messy realities of implementation, people being people, etc.

Tangentially, this reminds me a bit of Holden Karnofsky's maximally-conservative utopia, which is just "status quo minus clearly-bad things". 

This isn't really a utopia in the traditional sense. It's trying to lay out one end of a spectrum.

Start here:

In this world, everything is exactly like the status quo, with one exception: cancer does not exist.

It may not be very exciting, but it's hard to argue with the claim that this would be better than the world as it is today.

This is basically the most conservative utopia I can come up with, because the only change it proposes is a change that I think we can all get on board with, without hesitation. Most proposed changes to the world would make at least some people uncomfortable (no inequality? No sadness?), but this one shouldn't. If we got rid of cancer, we'd still have death, we'd still have suffering, we'd still have struggle, etc. - we just wouldn't have cancer.

You can almost certainly improve this utopia further by taking more baby-steps along the same lines. Make a list of things that - like cancer - you think are just unambiguously bad, and would be happy to see no more of in the world. Then define utopia as "exactly like the status quo, except that all the things on my list don't exist." Examples could include:

• Other diseases
• Hunger
• Non-consensual violence (not including e.g. martial arts, in which two people agree to a set of rules that allows specific forms of violence for a set period of time).
• Racism, sexism, etc.

"Status quo, minus everything on my list" is a highly conservative utopia. Unlike literary utopias, it should be fairly clear that this world would be a major improvement on the world as it is.

I note that in my survey on fictional utopias, it was much easier to get widespread agreement (high average scores) for properties of utopia than for full utopian visions. For example, while no utopia description scored as high as 4 on a 5-point scale, the following properties all scored 4.5 or higher: "no one goes hungry", "there is no violent conflict," "there is no discrimination by race or gender."

Striking paper by Anant Sudarshan and Eyal Frank (via Dylan Matthews at Vox Future Perfect) on the importance of vultures as a keystone species. 

To quote the paper and newsletter — the basic story is that vultures are extraordinarily efficient scavengers, eating nearly all of a carcass less than an hour after finding it, and farmers in India historically relied on them to quickly remove livestock carcasses, so they functioned as a natural sanitation system in helping to control diseases that could otherwise be spread through the carcasses they consume. In 1994, farmers began using diclofenac to treat their livestock, due to the expiry of a patent long held by Novartis leading to the entry of cheap generic brands made by Indian companies. Diclofenac is a common painkiller, harmless to humans, but vultures develop kidney failure and die within weeks of digesting carrion with even small residues of it. Unfortunately this only came to light via research published a decade later in 2004, by which time the number of Indian vultures in the wild had tragically plummeted from tens of millions to just a few thousands today, the fastest for a bird species in recorded history and the largest in magnitude since the extinction of the passenger pigeon. 

When the vultures died out, far more dead animals lay around rotting, transmitting pathogens to other scavengers like dogs and rats and entering the water supply. Dogs and rats are less efficient than vultures at fully eliminating flesh from carcasses, leading to a higher incidence of human contact with infected remains, and they're also more likely to transmit diseases like anthrax and rabies to people. Sudarshan and Frank estimate that this led to ~100,000(!) additional deaths each year from 2000-05 due to a +4.2%(!) increase in all-cause mortality among the 430 million people living in districts that once had a lot of vultures, which is staggering; this is e.g. more than the death toll in 2001 from HIV/AIDS (92,000), malaria (53,000), and alcohol use disorders (14,000). 

(Cause X, anyone? Preventing a hundred thousand deaths a year for less than half a billion dollars annually clears the GiveWell top charity-level threshold, and half a billion is in the ballpark of Open Philanthropy's entire annual grantmaking...)

So what to do? For vultures in particular, Sudarshan and Frank say their results "inform current vulture recovery efforts in India, and conservation efforts elsewhere" e.g. parts of Africa and Spain, albeit without elaborating. More broadly, they hope their paper informs better policymaking by providing "a particularly stark example of the type of hard-to-reverse and unpredictable costs that must be accounted for when evaluating the introduction of new chemicals into fragile and diverse ecosystems", stating "it is plausible that a counterfactual policy regime in India that tested chemicals for their toxicity to at least keystone species might have avoided the collapse of vultures". They conclude:

In the absence of empirical estimates of the social benefits conferred by different species, conservation policy may be heavily influenced by existence values unrelated to utility. The vulture is not a particularly attractive bird and evokes rather different emotions at first sight than do more charismatic poster-animals of wildlife conservation such as tigers and panda bears. Nevertheless
our results suggest that subjective existence values alone may not be the best way to formulate conservation policy.

The remark that vultures are not particularly attractive reminds me of the overlooked plight of farmed chickens, shrimp, insects etc for not being charismatic fauna. (I am admittedly sort of emotionally conflating the welfare of vultures with their ecosystem importance as a keystone species here.)

Extended anecdote from My Willing Complicity In "Human Rights Abuse", by a former doctor (GP) working at a Qatari visa center in India to process "the enormous number of would-be Indian laborers who wished to take up jobs there":

Another man comes to mind (it is not a coincidence that the majority of applicants were men). He was a would-be returnee - he had completed a several year tour of duty in Qatar itself, for as long as his visa allowed, and then returned because he was forced to, immediately seeking reassessment so he could head right back. He had worked as a truck driver, and now wanted to become a personal chauffeur instead.

He had been away for several years and had not returned a moment before he was compelled to. He had family: a wife and a young son, as well as elderly parents. All of them relied on him as their primary breadwinner. I asked him if he missed them. Of course he did. But love would not put food on the table. Love would not put his son into a decent school and ensure that he picked up the educational qualifications that would break the cycle. Love would not ensure his elderly and increasingly frail parents would get beyond-basic medical care and not have to till marginal soil at the tiny plot of land they farmed.

But the labor he did out of love and duty would. He told me that he videocalled them every night, and showed me that he kept a picture of his family on his phone. He had a physical copy close at hand, tucked behind the transparent case. It was bleached by the sun to the point of illegibility and half-covered by what I think was a small-denomination Riyal note.

He said this all in an incredibly matter-of-fact way. I felt my eyes tear up, and I looked away so he wouldn't notice. ... 

I asked him how well the job paid. Well enough to be worth it, he told me. He quoted a figure that was not very far from my then monthly salary of INR 76,000 (about $820 today). Whatever he made there, I noted that I had made about the same while working as an actual doctor in India in earlier jobs. ... He expected a decent bump - personal drivers seemed to be paid slightly better than commercial operators. 

I asked him if he had ever worked similar roles in India. He said he had. He had made a tenth the money, in conditions far worse than what he would face in Qatar. He, like many other people I interviewed, viewed the life you have the luxury of considering inhumane and unpalatable, and deemed it a strict improvement to the status quo. He was eager to be back. He was saddened that his son would continue growing up in his absence, but he was optimistic that the boy would understand why his father did what he had to do. ...

By moving to the Middle East, he was engaged in arbitrage that allowed him to make a salary comparable to the doctor seeing him in India. I look at how much more I make after working in the NHS and see a similar bump. ...

... "exploitation" is a word with a definition, and that definition requires something more than "a transaction that takes place under conditions of inequality." If we define exploitation as taking unfair advantage of vulnerability, we need a story about how the worker is made worse off relative to the alternative - and the workers I spoke with, consistently and across months, told me the opposite story. ... They are adults making difficult tradeoffs under difficult constraints, the same tradeoffs that educated Westerners make constantly but with much less margin for error and no safety net.

I do not recall ever outright rejecting an applicant for a cause that couldn't be fixed, but even the occasional instances where I had to turn them away and ask them to come back after treatment hurt. Both of us - there was often bargaining and disappointment that cut me to the bone. I do not enjoy making people sad, even if my job occasionally demands that of me. I regret making them spend even more of their very limited money and time on followups and significant travel expenses, even if I was duty-bound to do so on occasion. 

Zooming out to the author's main argument:

The argument generally goes: This job involves intense heat, long hours, and low pay relative to Western minimum wages. Therefore, it is inherently exploitative, and anyone taking it must be a victim of coercion or deception.

This completely ignores the economic principle of revealed preferences: the idea that you can tell what a person actually values by observing what they choose to do under constraint. Western pundits sit in climate-controlled pods and declare that nobody should ever have to work in forty-degree heat for $300 a month. But for someone whose alternative is working in forty-degree heat in Bihar for $30 a month with no social safety net, banning Qatari labor practices just destroys their highest expected-value option. ...

The economic case for Gulf migration from South Asia is almost embarrassingly strong when you actually look at it. India received roughly $120 billion in remittances in 2023, making it the world's largest recipient, with Gulf states still accounting for a very large share, though the RBI's own survey data show that advanced economies now contribute more than half of India's remittances. For certain origin states - Kerala being the clearest case, alongside Maharashtra and Tamil Nadu - remittance income is not a rounding error in household economics; it is the household economy. The man sending money home from Doha is participating in a system that has done more for South Asian poverty alleviation than most bilateral aid programs combined.

It's a bummer that this situation is hard to improve affordability-, accessibility-, and safety-wise. CE-incubated Rahi Impact tried, but after deeper and frankly disheartening-to-read investigation concluded this was neither tractable nor cost-effective for them due to inability to source jobs at scale from Gulf employers due to their commitment to ethical recruitment disadvantaging them -- folks are so desperate to work in the Gulf that the resulting huge labor oversupply gives Gulf employers significant leverage which they use to monetise visas by demanding bribes from recruiting agents for access to job slots, which then get passed on to workers as exorbitant fees or garnished pay (worth 2.5 years of wages in their home country, a lot of this paid upfront). The recruiters also screw over the workers often enough to drive a ~30% migration failure rate, here's a real case study from Rahi's investigation:

Case Study: Fraudulent Practices in Migration  Asif (name changed), a 34-year-old from India, faced mounting debts that forced him to seek work abroad. In March 2023, he left behind his small garment shop, lured by the promise of higher wages and stable employment in the Gulf. However, his journey soon turned into a costly ordeal. Through a recommendation from his in-laws, Asif found an agent in a nearby city who charged $1,440 for a job as a helper in the GCC. Trusting both the recommendation and the apparent seniority of the agent, Asif paid the amount and traveled to Mumbai to complete the required migration steps. Once there, he was stranded—his agent disappeared, and phone calls went unanswered. Desperate to recover his losses, Asif sought advice from a family friend, who recommended a different agent. Acting on this recommendation, he paid $920 for a two-month visitor visa to Dubai, with the promise that it would be converted into a work visa. In Dubai, he was put to work in a laundry, packing clothes, but he faced severe exploitation. Although promised $290 per month, he received only $220 and worked 19-hour shifts without overtime pay. Since he was on a visitor visa, he had no legal recourse. When Asif could no longer endure the conditions, he paid for his own return ticket home. Unfortunately, at the airport he discovered he had overstayed his visa by 20 days and incurred a $290 fine.  In total, Asif believes he wasted two years and suffered a financial loss of over $2,300. Like many, Asif fell victim to false promises and desperation. Yet, with no opportunities at home and debts still unpaid, he remains willing to take the risk again—highlighting the dire reality for many low-skilled workers in South Asia.

Asif will still try it again, damn. 

As a silver lining, Rahi's cofounders list many other potential impactful solutions to be trialed for improving Gulf migration from South Asia for the >1.2 million workers who make the journey every year. But I think the ex-GP's point stands that taking away the option to work in the Gulf, exploitation and fraud and elevated death risk included, is a non-solution, it would make things even worse than not doing anything at all. 

GiveWell did their first "lookbacks" (reviews of past grants) to see if they've met initial expectations and what they could learn from them:

Lookbacks compare what we thought would happen before making a grant to what we think happened after at least some of the grant’s activities have been completed and we’ve conducted follow-up research. While we can’t know everything about a grant’s true impact, we can learn a lot by talking to grantees and external stakeholders, reviewing program data, and updating our research. We then create a new cost-effectiveness analysis with this updated information and compare it to our original estimates.

(While I'm very glad they did so with their usual high quality and rigor, I'm also confused why they hadn't started doing this earlier, given that "okay, but did we really help as much as we think we would've? Let's check?" feels like such a basic M&E / ops-y question. I'm obviously missing something trivial here, but also I find it hard to buy "limited org capacity"-type explanations for GW in particular given total funding moved, how long they've worked, their leading role in the grantmaking ecosystem etc)

Their lookbacks led to substantial changes vs original estimates, in New Incentives' case driven by large drops in cost per child enrolled ("we think this is due to economies of scale, efficiency efforts by New Incentives, and the devaluation of the Nigerian naira, but we haven’t prioritized a deep assessment of drivers of cost changes") and in HKI's case driven by vitamin A deficiency rates in Nigeria being lower and counterfactual coverage rates higher than originally estimated:

Buried deep in the PEPFAR Report's appendix - methodology section is a nice "introduction to global health programs" mini-article that also addresses some lay misconceptions about foreign aid and suggests a better way to think about it all in one go; it's a shame that most folks won't read it, so I'm reposting it here for ease of future reference.

Introduction to Global Health Programs

Many people are skeptical of foreign aid and other attempts to help the global poor—and they’re right to be! A lot of foreign aid is poorly targeted, counterproductive, or simply a waste of money. From PlayPumps to TOMS shoes to One Laptop Per Child, the news is full of well-intentioned programs that had nowhere near the effect their boosters advertised. Many prominent experts, such as William Easterly and Angus Deaton, question whether foreign aid works at all.

Development economists, charity evaluators, and other specialists perform “program evaluations,” which ask questions like:

1. Does the problem we’re trying to solve actually exist?
2. Why does the problem exist?
3. Is the program well-implemented?
4. Is the program having the effect that we expected?
5. Is the program too expensive? Can some other program get the same results for less money?

In general, program evaluations are interested in finding out what the effects of a program are. The effect of a program is the difference between the outcome (what actually happened) and the counterfactual (what would have happened without the program being implemented). It’s impossible to measure the counterfactual because the counterfactual is about the same people at the same time. The counterfactual can only ever be estimated. Program evaluators have come up with many different ways of estimating the counterfactual, which we’ll talk about on the main page.

Researchers have found that global health interventions are far more likely to work than programs like PlayPumps or One Laptop Per Child. It’s easy to be wrong about whether a school system needs laptops, especially in a country far away from your own; it’s much harder to be wrong about whether a country has sky-high rates of HIV/AIDS. We don’t know much about the causes of poverty or what makes countries develop economically; we know much more about the causes of HIV and what makes HIV progress to AIDS. PlayPumps were a brand-new invention that might not work; antiretroviral medications are well-tested, well-understood, and widely used in the developed world. For this reason, the charity evaluator GiveWell—which specializes in cost-effective ways of helping the global poor—mostly recommends charities that provide healthcare.

Foreign aid often has unintended consequences: for example, giving people shoes (like TOMS shoes did) can put local shoemakers out of work; foreign aid can lead to governments prioritizing the wishes of foreign donors over the wishes of their own people. Providing healthcare has many fewer negative unintended consequences than other forms of foreign aid: providing antiretrovirals is unlikely to put small local antiretroviral manufacturers out of work. It can have some other unintended consequences, like loss of democratic accountability.

But providing healthcare also could also have positive unintended consequences. Healthy children may be more likely to go to school. Then healthy adults may be able to work more and earn more for their families. These effects can be huge. For instance, the charity evaluator GiveWell estimates that a third of the benefit of insecticide-treated bednet distribution, a treatment to avert deaths from malaria, comes from increased income.

Poverty in America is horrible: no one should be unsure how to pay for rent, food, or healthcare. But Americans are extraordinarily rich compared to the rest of the world: a person at the poverty line in the United States is in the top 15% wealthiest people in the world, even if you adjust for how far money goes in each country. Many of the world’s poorest people live in the countries PEPFAR works in: about two-fifths of people in sub-Saharan Africa live on less than $2.15 a day, adjusted for how far money goes. Since these people are so poor, they don’t have many of the opportunities Americans take for granted. Of course, the first priority of the United States government should be to help American citizens. But if you’re used to charity at home, it can be shocking how cheap it is to help people abroad.

Most of all: just throwing money at foreign aid doesn’t fix anything. But that isn’t a reason to give up—not with millions of lives at stake. If we’re careful and thoughtful, and if we actually check whether what we’re doing does any good, then we don’t have to be PlayPumps or TOMS shoes. We can concretely, robustly make things better.

This passage from David Roodman's essay Appeal to Me: First Trial of a “Replication Opinion” resonated:

When we draw on research, we vet it in rare depth (as does GiveWell, from which we spun off). I have sometimes spent months replicating and reanalyzing a key study—checking for bugs in the computer code, thinking about how I would run the numbers differently and how I would interpret the results. This interface between research and practice might seem like a picture of harmony, since researchers want their work to guide decision-making for the public good and decision-makers like Open Philanthropy want to receive such guidance.

Yet I have come to see how cultural misunderstandings prevail at this interface. From my side, what the academy does and what I and most of the public think it does are not the same. There are two problems. First, about half the time I reanalyze a study, I find that there are important bugs in the code, or that adding more data makes the mathematical finding go away, or that there’s a compelling alternative explanation for the results. (Caveat: most of my experience is with non-randomized studies.) Second, when I send my critical findings to the journal that peer-reviewed and published the original research, the editors usually don’t seem interested (recent exception). Seeing the ivory tower as a bastion of truth-seeking, I used to be surprised. I understand now that, because of how the academy works, in particular, because of how the individuals within academia respond to incentives beyond their control, we consumers of research are sometimes more truth-seeking than the producers.

I had a similar realisation towards the end of my studies which was a key factor in persuading me to not pursue academia. Also I've mentioned this before, but it surprised me how much more these kinds of details mattered in my experience in industry.

Skipping over to his recap of the specific case he looked into:

To recap:

• Two economists performed a quantitative analysis of a clever, novel question.
• It underwent peer review.
• It was published in one of the top journals in economics. Its data and computer code were posted online, per the journal’s policy.
• Another researcher promptly responded that the analysis contains errors (such as computing average daytime temperature with respect to Greenwich time rather than local time), and that it could have been done on a much larger data set (for 1990 to ~2019 instead of 2000–04). These changes make the headline findings go away.
• After behind-the-scenes back and forth among the disputants and editors, the journal published the comment and rejoinder.
• These new articles confused even an expert.
• An outsider (me) delved into the debate and found that it’s actually a pretty easy call.

If you score the journal on whether it successfully illuminated its readership as to the truth, then I think it is kind of 0 for 2. ...

That said, AEJ Applied did support dialogue between economists that eventually brought the truth out. In particular, by requiring public posting of data and code (an area where this journal and its siblings have been pioneers), it facilitated rapid scrutiny.

Still, it bears emphasizing: For quality assurance, the data sharing was much more valuable than the peer review. And, whether for lack of time or reluctance to take sides, the journal’s handling of the dispute obscured the truth.

My purpose in examining this example is not to call down a thunderbolt on anyone, from the Olympian heights of a funding body. It is rather to use a concrete story to illustrate the larger patterns I mentioned earlier. Despite having undergone peer review, many published studies in the social sciences and epidemiology do not withstand close scrutiny. When they are challenged, journal editors have a hard time managing the debate in a way that produces more light than heat.

I have critiqued papers about the impact of foreign aid, microcredit, foreign aid, deworming, malaria eradication, foreign aid, geomagnetic storm risk, incarceration, schooling, more schooling, broadband, foreign aid, malnutrition, …. Many of those critiques I have submitted to journals, usually only to receive polite rejections. I obviously lack objectivity. But it has struck me as strange that, in these instances, we on the outside of academia seem more concerned about getting to the truth than those on the inside. 

Curious what people think of Gwern Branwen's take that our moral circle has historically narrowed as well, not just expanded (so contra Singer), so we should probably just call it a shifting circle. His summary:

The “expanding circle” historical thesis ignores all instances in which modern ethics narrowed the set of beings to be morally regarded, often backing its exclusion by asserting their non-existence, and thus assumes its conclusion: where the circle is expanded, it’s highlighted as moral ‘progress’, and where it is narrowed, what is outside is simply defined away. 

When one compares modern with ancient society, the religious differences are striking: almost every single supernatural entity (place, personage, or force) has been excluded from the circle of moral concern, where they used to be huge parts of the circle and one could almost say the entire circle. Further examples include estates, houses, fetuses, prisoners, and graves.

(I admittedly don't find his examples all that persuasive, probably because I'm already biased to only consider beings that can feel pleasure and suffering.)

What's the "so what"? Gwern:

One of the most difficult aspects of any theory of moral progress is explaining why moral progress happens when it does, in such apparently random non-linear jumps. (Historical economics has a similar problem with the Industrial Revolution & Great Divergence.) These jumps do not seem to correspond to simply how many philosophers are thinking about ethics. 

As we have already seen, the straightforward picture of ever more inclusive ethics relies on cherry-picking if it covers more than, say, the past 5 centuries; and if we are honest enough to say that moral progress isn’t clear before then, we face the new question of explaining why things changed then and not at any point previous in the 2500 years of Western philosophy, which included many great figures who worked hard on moral philosophy such as Plato or Aristotle. 

It is also troubling how much morality & religion seems to be correlated with biological factors. Even if we do not go as far as Julian Jaynes’s⁹ theories of gods as auditory hallucinations, there are still many curious correlations floating around.

From Rachel Glennerster's old J-PAL blog post, a classic worth resharing: "charge for bednets or distribute them for free?"

In 2000 there was an intense argument about whether malarial insecticide-treated bednets (ITNs) should be given out for free. Some argued that charging for bednets would massively reduce take-up by the poor. Others argued that if people don’t pay for something, they don’t value it and are less likely use it. It was an evidence-free argument at the time.

Then, a series of studies in many countries testing many different preventative health products showed that even a small increase in price led to a sharp decline in product take-up. Pricing did not help target the product to those who needed it most, and people were not more likely to use a product if they paid for it. This cleared the way for a massive increase in free bednet distribution (Dupas 2011 and Kremer and Glennerster 2011).

 

There was a dramatic increase in malaria bednet coverage between 2000 and 2015 in sub-Saharan Africa. At the same time, there was a massive fall in the number of malarial cases. In Nature, Bhatt and colleagues estimate that the vast majority of the decline in malarial cases is due to the increase in ITNs. They estimate there were 450 million fewer cases of malaria due to ITNs and four million fewer deaths due to ITNs. The lesson here is that testing an important policy-relevant idea can have as much impact on peoples’ lives as testing a specific program.

I appreciated reading these passages from Hannah Ritchie's Children in rich countries are much less likely to die than a few decades ago, but we rarely hear about this progress over at Our World in Data, for both the data and the shift in perspective:

Countries in the European Union, Japan, South Korea, the United Kingdom — the list goes on — have made childhood much safer in my own 30-year lifetime.¹ It’s just something we rarely hear about. I also don’t think that this is a “solved problem”; it is still too common for parents to see their children die, and there’s a lot more that we can do to save their lives. ...

It’s only when we look at the relative reduction in child mortality that we see that rich countries have also made impressive progress.

I think it’s important to highlight this point for two reasons.

First, the idea that progress on health has stalled (or even regressed) in rich countries is, I think, a common one. I’ve previously held that view myself. But it’s not true: improved treatments and vaccinations developed by scientists, dedicated care from doctors, midwives, and nurses, health policies developed by governments, and parents' choices have made things much safer for children even in the world’s richest countries. These efforts were not for nothing: they’ve given kids a future and spared many families the pain of losing a child.

Second, child mortality in rich countries is not a “solved problem”. 23,000 children still die in the United States every year. That’s around 50 times more than the number who die from natural disasters.³ And more than the total number of homicides.⁴ No one would say that murders in the US are a “solved problem”.

As someone predisposed to like modeling, the key takeaway I got from Justin Sandefur's Asterisk essay PEPFAR and the Costs of Cost-Benefit Analysis was this corrective reminder – emphasis mine, focusing on what changed my mind:

Second, economists were stuck in an austerity mindset, in which global health funding priorities were zero-sum: $300 for a course of HIV drugs means fewer bed nets to fight malaria. But these trade-offs rarely materialized. The total budget envelope for global public health in the 2000s was not fixed. PEPFAR raised new money. That money was probably not fungible across policy alternatives. Instead, the Bush White House was able to sell a dramatic increase in America’s foreign aid budget by demonstrating that several billion dollars could, realistically, halt an epidemic that was killing more people than any other disease in the world. 

...

A broader lesson here, perhaps, is about getting counterfactuals right. In comparative cost-effectiveness analysis, the counterfactual to AIDS treatment is the best possible alternative use of that money to save lives. In practice, the actual alternative might simply be the status quo, no PEPFAR, and a 0.1% reduction in the fiscal year 2004 federal budget. Economists are often pessimistic about the prospects of big additional spending, not out of any deep knowledge of the budgeting process, but because holding that variable fixed makes analyzing the problem more tractable. In reality, there are lots of free variables.

More detail:

Economists’ standard optimization framework is to start with a fixed budget and allocate money across competing alternatives. At a high-level, this is also how the global development community (specifically OECD donors) tends to operate: foreign aid commitments are made as a proportion of national income, entirely divorced from specific policy goals. PEPFAR started with the goal instead: Set it, persuade key players it can be done, and ask for the money to do it.

Bush didn’t think like an economist. He was apparently allergic to measuring foreign aid in terms of dollars spent. Instead, the White House would start with health targets and solve for a budget, not vice versa. ... Economists are trained to look for trade-offs. This is good intellectual discipline. Pursuing “Investment A” means forgoing “Investment B.” But in many real-world cases, it’s not at all obvious that the realistic alternative to big new spending proposals is similar levels of big new spending on some better program. The realistic counterfactual might be nothing at all.

In retrospect, it seems clear that economists were far too quick to accept the total foreign aid budget envelope as a fixed constraint. The size of that budget, as PEPFAR would demonstrate, was very much up for debate.

When Bush pitched $15 billion over five years in his State of the Union, he noted that $10 billion would be funded by money that had not yet been promised. And indeed, 2003 marked a clear breaking point in the history of American foreign aid. In real-dollar terms, aid spending had been essentially flat for half a century at around $20 billion a year. By the end of Bush’s presidency, between PEPFAR and massive contracts for Iraq reconstruction, that number hovered around $35 billion. And it has stayed there since. 

Compared to normal development spending, $15 billion may have sounded like a lot, but exactly one sentence after announcing that number in his State of the Union address, Bush pivoted to the case for invading Iraq, a war that would eventually cost America something in the region of $3 trillion — not to mention thousands of American and hundreds of thousands of Iraqi lives. Money was not a real constraint.

Tangentially, I suspect this sort of attitude (Iraq invasion notwithstanding) would naturally arise out of a definite optimism mindset (that essay by Dan Wang is incidentally a great read; his follow-up is more comprehensive and clearly argued, but I prefer the original for inspiration). It seems to me that Justin has this mindset as well, cf. his analogy to climate change in comparing economists' carbon taxes and cap-and-trade schemes vs progressive activists pushing for green tech investment to bend the cost curve. He concludes: 

You don’t have to give up on cost-effectiveness or utilitarianism altogether to recognize that these frameworks led economists astray on PEPFAR — and probably some other topics too. Economists got PEPFAR wrong analytically, not emotionally, and continue to make the same analytical mistakes in numerous domains. Contrary to the tenets of the simple, static, comparative cost-effectiveness analysis, cost curves can sometimes be bent, some interventions scale more easily than others, and real-world evidence of feasibility and efficacy can sometimes render budget constraints extremely malleable. Over 20 years later, with $100 billion dollars appropriated under both Democratic and Republican administrations, and millions of lives saved, it’s hard to argue a different foreign aid program would’ve garnered more support, scaled so effectively, and done more good. It’s not that trade-offs don’t exist. We just got the counterfactual wrong.

Aside from his climate change example above, I'd be curious to know what other domains economists are making analytical mistakes in w.r.t. cost-benefit modeling, since I'm probably predisposed to making the same kinds of mistakes. 

This is a top-level comment compiling scattered notes on quantifying benefits / costs and related things, for my own benefit. Future notes will be comments under this one.

The purpose of this compilation is to essentially fact-post / perpetual-draft my way towards a more nuts-and-bolts understanding of the "what are the most effective interventions?" question, which is a different starting point than the usual "what are the most cost-effective?" one, a perspective shift primarily spurred by Justin Sandefur's case study on PEPFAR and a desire to better internalise what "big EA" might look like (as opposed to taking a scarcity mindset as given), albeit in a lazy undirected way: these notes are from references I encounter in the course of doing other work.

Past relevant notes: 

• Global immunisation efforts have saved ~154 million lives and led to 10.2 billion full health years gained over the last 50 years, 95% of them children under-5, 2/3rds infants
• Tom Frieden likely helped saved tens of millions of lives by creating an international tobacco control initiative, and in his new role as CEO of Resolve to Save Lives is aiming to save 94 million more in 25 years
• Notes on Richard Parncutt's heuristic that someone in the next ~century is prematurely killed for every 1,000 tons of carbon humanity burns
• Notes on Sudarshan & Frank's paper on vultures as a keystone species: their accidental eradication in India led to more additional deaths each year (~100k) than HIV/AIDS, malaria, or alcohol use disorders
• Justin Sandefur's point that counterfactuals can be tricky to get right, with PEPFAR as case study, and how (sometimes) budget is not a real constraint even for multibillion-dollar initiatives
• Trying out RP's then-new cross-cause cost-effectiveness model for myself
• Notes on Max Dalton's 2014 writeup Estimating the cost-effectiveness of research into neglected diseases
• How GiveWell calculates AMF's cost-effectiveness, step by step 

Why did India's happiness ratings consistently drop so much over time even as its GDP per capita rose?

Epistemic status: confused. Haven't looked into this for more than a few minutes

My friend recently alerted me to an observation that puzzled him: this dynamic chart from Our World in Data's happiness and life satisfaction article showing how India's self-reported life satisfaction dropped an astounding -1.20 points (4.97 to 3.78) from 2011 to 2021, even as its GDP per capita rose +51% (I$4,374 to I$6,592 in 2017 prices): 

(I included China for comparison to illustrate the sort of trajectory I expected to see for India.)

The sliding year scale on OWID's chart shows how this drop has been consistent and worsening over the years. This picture hasn't changed much recently: the most recent 2024 World Happiness Report reports a 4.05 rating averaged over the 3-year window 2021-23, only slightly above the 2021 rating.

A -1.20 point drop is huge. For context, it's 10x(!) larger than the effect of doubling income at +0.12 LS points (Clarke et al 2018 p199, via HLI's report), and compares to major negative life events like widowhood and extended unemployment: 

Given India's ~1.4 billion population, such a large drop is alarming: roughly ~5 billion LS-years lost since 2011, very roughly ballparking. For context, and keeping in mind that LS-years and DALYs aren't the same thing, the entire world's DALY burden is ~2.5 billion DALYs p.a. 

But – again caveating with my lack of familiarity with the literature and extremely cursory look into this – I haven't seen any writeup look into this, which makes me wonder if it's not a 'real issue'? For instance, the 2021 WHR just says

Since 2006-08, world well-being has been static, but life expectancy increased by nearly four years up to 2017-19 (we shall come to 2020 later). The rate of progress differed a lot across regions. The biggest improvements in life expectancy were in the former Soviet Union, in Asia, and (the greatest) in Sub-Saharan Africa. And these were the regions that had the biggest increases in WELLBYs. In Asia, the exception is South Asia, where India has experienced a remarkable fall in Well-being which more than outweighs its improved life expectancy.

That's it: no elaboration, no footnotes, nothing.

So what am I missing? What's going on here? 

A quick search turned up this WEF article (based on Ipsos data and research, not the WHR's Gallup World Poll, so take it with a grain of salt) pointing to

• increased internet access -> pressure to portray airbrushed lives on social media & a feeling that 'their lives have become meaningless' 
• covid-19 mitigation-induced isolation curtailing activities that improve wellbeing (employment, socializing, going to school, exercising and accessing health services)
• urban migration to seek work -> traffic congestion, noise and pollution, demanding bosses -> less sleep and exercise -> higher anxiety and worsening health 

But I'm not sure these factors are differential (i.e. that they, for instance, happen much more in India than elsewhere s.t. it explains the wellbeing vs development trajectory difference over 2011-24)?