Linkpost for April

Source.

Regular announcements: did you know you can hire me for life coaching and general consulting? You can also buy my novella Her Voice Is A Backwards Record wherever fine ebooks are sold (except Google Books).

Effective Altruism: Moral Circle Expansionism has been translated into Portugeuse.

Effective Altruism

Global Poverty

While malaria bednets are sometimes used for fishing, there is little evidence that they cause harm to people or fish, and such misuse of bednets is accounted for in studies of their effectiveness.

Economic growth in developing countries has enormous benefits, but we’re not sure how to speed it up and a lot of people are already working in the area.

The Trump administration laid off all experts on preventing mother-to-child transmission of HIV.

You know that your USAID cuts are a disaster when William Easterly—who has spent his entire academic career calling for cuts to USAID—calls them illegal, undemocratic, and horrific.

Rethink Priorities’s framework for making donations given the USAID cuts. (Primarily useful for large donors and grantmakers.)

USAID is unpopular among conservatives because it spread pro-LGBT messaging. I think pro-aid-reform people should have reached out proactively to conservatives who have these concerns. We could validate the concerns (DEI musicals in Ireland are indeed a dumb thing to spend money on) and provide a framework for criticizing them that doesn’t mean defunding important global health aid.

Animal Advocacy

Actionable: Take a survey on vegetarianism and your personal philosophy.

Lessons from farm animal economics: If you block a factory farm in your area, animal products might be imported from elsewhere in places with worse animal-welfare regulations. Meat-company mergers can reduce the quantity of animals farmed by decreasing competition. Taxes and subsidies can improve animal welfare. Reducing fishing may just increase aquaculture and cause more fish to be fished in the long run as populations increase.

Ways animal advocacy heuristics can be misleading: even if there are more farmed animals than lab animals, there might be more of some species of lab animal (e.g. rats) than of some species of farmed animal (e.g. cows). Interventions that target a large group of animals (e.g. animals in China, shrimp) might not be very effective if the intervention can only reach a small percentage of those animals. A large population of animals still might suffer less than a smaller population, which would mean that interventions targeting the smaller population are more important even though they affect fewer animals.

A simple explanation of cultivated meat for people who aren’t scientists.

Researchers who work with animals experience intense pressure to be okay with hurting them, which can cause trauma [cw: explicit descriptions of animal cruelty].

Existential Risk

AI startup founder says that AI performance hasn’t improved, even as AIs continue to score highly on benchmarks. “Benchmarks get saturated faster than you think, applications are slower than you think” seems to be a principle of AI development; I’m not sure why.

Anthropic’s proposals for AI regulation reflect concern about national security, particularly around bioweapons. Conversely, OpenAI emphasizes the importance of voluntary cooperation and corporate freedom. Google has a more conservative viewpoint than OpenAI or Anthropic about the potential future effects of AI.

What might go wrong if, instead of a single artificial general intelligences, we have a whole bunch? The paper goes into coordination failures (AIs with similar goals can’t cooperate to achieve them); conflict (AIs with different goals compete with each other in a negative-sum way); and collusion (AIs work together to achieve goals they couldn’t achieve on their own and that we’d rather they not achieve). It includes a lot of interesting real-world case studies that helped me understand misalignment in much greater detail.

Actionable: Journalism about AI is potentially very valuable, because journalists can make readers more aware of risks and hold powerful people accountable for safe AI development.

I recommend the essay Slopworld 2035 with caveats. I think it has a lot of really unnecessary dunking on AI safety advocates and too much of the author’s politics. But I think it’s interesting to see a model of what can go wrong if AIs continue to be rather incompetent.

Meta Effective Altruism

Particularly Good: a guide to doing a cost-benefit analysis.

Actionable: Probably Good is reopening its one-on-one advising program.

Actionable: An effective altruist is offering assistance in becoming a nuclear power plant operator as an earning-to-give career. Many places only require a high-school degree.

“I remember in the early days of 80,000 Hours, we spent a whole day hosting an [ultra-high-net-worth individual. He ultimately gave £5000. The week afterwards, a one-hour call with Julia Wise - a social worker at the time - resulted in a larger donation.”

A personal story about experiencing rejection from effective altruist jobs and incorrectly deciding that you can’t help.

Open Philanthropy’s global health and wellbeing¹ movement building program has donated $28 million, ~70% to charities which promote effective giving. They plan to continue funding effective giving going forward, but are considering expanding more into programs which recruit talented people.

Relatedly: projects that Open Philanthropy’s global health and wellbeing movement building program is interested in.

I’ve been really excited by the new direction the Centre for Effective Altruism is taking, and their recent release of their strategic plan has only made me more excited. They intend to prioritize growing the effective altruism community, improving effective altruism’s brand, increasing effort towards fundraising, and further improving CEA itself. They also cut spending by $5.9 million while adding twenty staff, which is insane.

80,000 Hours has announced a pivot to prioritizing AI safety, with a particular focus on the possibility of AGI by 2030. Fortunately, Probably Good exists as a more general-purpose effective altruist career organization.

Reasons not to defer to funders: You might hear negative information about projects and people that funders don’t have access to. Funders might have different values than you do. Grantmakers often don’t have experience directly doing work in a field.

History of diversity efforts in effective altruism.

American Democracy

The Trump administration is deliberately going after unpopular or polarizing targets to establish a precedent that they can use to further eradicate our civil liberties.

This post on DOGE came out in early March, which is about a decade ago in Trump years. But I still think it has some sharp analysis about how difficult it is to know what’s going on in the Trump administration, about USAID/PEPFAR, and about how DOGE’s actions reflect common Republican beliefs.

A ketamine addict’s perspective on Elon Musk’s ketamine use and how it might be affecting his thinking.

Other Causes

Japanese cherry trees blossom more than a week earlier than they did before the Industrial Revolution—a result of climate change.

In China, the government sets a policy directive, various states compete to fulfill it, and the best pilot programs are scaled up. Unfortunately, this works much better for GDP development than for environmental protection or food safety, because it tends to lead to lack of coordination. The Chinese government in general tends to be very uncoordinated, because they crack down hard when things go wrong. That means that no one has any incentive to report bad news, and the system can’t engage in probabilistic risk management.

Rationality

Minimal-trust investigations—learning about a topic without putting any trust in anyone—are a useful (but extremely time-consuming) way to learn what kinds of arguments, people, and institutions you should trust.

On the definition of math: “Math is a human mental activity based on playing the “Game of Truth”: when we do math, we make as if notions had precise definitions that were perfectly stable over time, as if statements had binary “truth” values, as if one could play Lego and deduce new true statements from existing ones.”

The story of the whistleblower who discovered that Francesca Gino committed data fraud. “Business academia needs to reckon with this inconvenient truth: Committing fraud is, right now, a viable career strategy that can propel you at the top of the academic world.”

Scientists spend a collective 15,000 years peer-reviewing papers every year, and yet scientific fraud constantly slips through and is almost never discovered in the peer-review stage. Peer review also seems to have had no effect on the number, size, or importance of discoveries—certainly not an effect in line with the amount peer review costs. Is it time to declare the experiment over?

A natural experiment shows that people who take the shingles vaccine are 20% less likely to get dementia.

If you act the same way over and over again, try experimenting to see if this is the action that gets you the best results.

There is no one set of social norms around dating; there are many different subcultures and social groups, each with their own norms. I particularly like the analysis of why so many rationalists are poly but only have sex with one person.

Social Justice

Programs that allow high-achieving people to skip college are feminist because they make it more likely women can take time off in their thirties to have kids.

Forty years passed between first-wave feminism and second-wave feminism. In this period, the National Women’s Party preserved feminist ideas, particularly focusing on the Equal Rights Amendment. The National Women’s Party was made up of a small, homogeneous group of dedicated women, mostly wealthy and elite, who were bound together by friendship and community and centralized under its leader, Alice Paul.

I stumbled across this obituary of my grandfather, Bart McCash, a history professor who taught one of the earliest Black Studies courses in the United States (and certainly one of the earliest in the South). He was a great man who sadly died young, only ten months before I was born. According to family legend, he did in fact face opposition to his Black Studies course, which he countered by playing dumb until the administration gave in. “But I’m a Civil War historian! Are you saying black people aren’t relevant to the history of the Civil War?”

Reality Has A Surprising Amount of Detail

Chinese Doom Scroll: scuba companies threatening to kill people unless they buy photos, plus how can it be rape if you’re engaged? Many Chinese cops don’t accept cases in the last half of December because they don’t have much time to solve them and unsolved cases look bad on their yearly statistics. Using phones as standard weights so you aren’t cheated by shopkeepers. Insider secrets from professionals: cheap beef is fake, it’s easy to transfer house deeds, and something about different kinds of red sugar.

Particularly Good: Patrick McKenzie read an article where a woman withdrew $50,000 cash and thought it sounded suspicious. So he investigated. It turns out that absolutely nothing was wrong, but I’m not going to spoil the ending. (Featuring a cameo from occasional Thing of Things reader Samuel Cottrell. Hi, Sammy!)

American prisoners compete in state-sponsored gladiator games????

Japan built a 3D-printed train station in six hours.

A voting error in Belgium may have been caused by a cosmic ray.

The Bureau of Fiscal Services processes at least 90% of the United States government’s payments. Bureau of Fiscal Services is important to national security: if they mess up, people could steal the government’s money or Americans’ private data, and if they lose continuity of operations then the government would fall apart. The article also has a lot of interesting information about their anti-fraud efforts.

Fiction

A few favorites from April Fools at the Effective Altruism Forum: the Wasting AI Researchers’ Time Initiative. The Centre for Effective Altruism Is No Longer “Effective Altruism”-Related. Pledge to spend at least 11% of your income on productivity improvement which is way better than spending it on all that ‘donations’ stuff. We’re running out of low-hanging fruit; is this solvable by genetically engineering dwarf trees, or perhaps getting the fruit to grow on the ground like strawberries?

Where Babies Come From: A (fictional) history of noncongerism—the idea that humans don’t come from sex. A sample: “Pregnancy occurs when a man and a woman share an intense emotional experience at the same time, and any sufficiently intense emotion will burst the bounds of the individual subject, forming an entirely new person… Al-Saghir concludes by commending the “lofty path to securing an heir,” in which husband and wife go to opposite sides of their house, and simultaneously bend their foreheads to the ground in sincere and fervent prayer.”

The Red Mother: A Viking witch battles a dragon, featuring riddle-games.

1. i.e. everything that isn’t global catastrophic risks.