Linkpost for January
Effective Altruism
Global Poverty
I’m sure many of you, like me, have been following the chaos at USAID. Trump has ordered that all USAID employees return home to the United States by February 8 and has placed all nonessential USAID employees on administrative leave. Returning all USAID employees to the United States in less than a week is absurdly expensive, both directly and in terms of opportunity cost (for example, vaccines are expiring on warehouse shelves). I also worry about the reputational cost to America. How is anyone going to trust us to do as we promised ever again? A reputation as reliable and trustworthy takes decades to build but can be destroyed in a week.
Although PEPFAR is not technically part of USAID, its funding remains frozen (in spite of waivers intended to allow it to keep operating) and its future remains uncertain. PEPFAR is significantly more cost-effective than GiveWell Top Charities, preventing 10-20% of deaths from all causes in countries in which it works.
In 2023, GiveWell raised less than its 10th percentile forecast for how much money it was going to raise. This is dire news. If you have spare money I urge you to donate it to GiveWell’s cost-effective programs which help some of the poorest people in the world.
Scientific research grants that Open Philanthropy has made. I think it’s neat to get a sense of their giving in this area. I also feel a sense of effective-altruism-patriotic pride: malaria vaccines! We helped do that!
The malaria vaccine rollout is slow. We’re easily capable of manufacturing far more vaccines than Gavi (the primary organization that provides malaria vaccines) plans to purchase. But Gavi doesn’t have enough money to sustainably purchase that many vaccines. There are some easy ways to improve: Gavi continues to buy a more expensive and less effective vaccine, even though we have unused production capacity for the cheaper and more effective vaccine. But ultimately the problem is that Gavi doesn’t have enough money.
“Long-distance development policy” is advocating for policies that don’t seem related to global poverty but that have positive effects for the global poor: decreasing tariffs and “anti-dumping” penalties; increasing immigration; cutting foreign aid that props up horrific regimes; scientific research; freezing illicitly obtained funds from corrupt oligarchs and returning the funds to their home countries; making it easier to obtain travel visas if you’re from a developing country; improved payment processing for remittances.
Existential Risk
Historically, forecasting platforms have struggled with many of the questions users are most interested in, because they’re too vague: there are no unambiguous resolution criteria for, say, “is science stagnating?” The forecasting platform Metaculus has developed “indexes,” which combine answers from multiple very specific questions into a single number, which is intended to represent the answer to a vague question. The first index is an index of AGI readiness, but they plan to roll out more soon.
Some points on AI progress: We can in fact train models on more data (e.g. YouTube videos, synthetic data). “Inference scaling” (giving AIs more time to think) improves performance on problems with clear correct answers, but has a relatively small effect on problems without a right answer. Industry insiders constantly mispredict AI advances (in both directions). Whether AI transforms society depends less on its capabilities and more on whether it is packaged into a useful product.
I have been continuing to enjoy Joe Carlsmith’s series Otherness and Control in the Age of AI. Being nicer than Clippy articulates a problem with maximizing ideologies, ones that want to fill the world with only things they value; something like “liberalism” or “niceness” allows people with different viewpoints to live together. Loving a world you don’t trust is a beautiful meditation on atheist spirituality. I really recommend the series even to people who don’t care much about AI: Carlsmith does an excellent job thinking through ethics.
Animal Advocacy
Some animal advocates criticize “certifiers,” which label some animal products as coming from less cruel farms. However, farms that have been certified as less cruel actually do have better conditions than uncertified farms. The meat industry is fully capable of coming up with meaningless labels if animal advocates don’t come up with meaningful ones (in fact, they are already doing this). We need institutional solutions to animal farming, not solutions based on individual purchasing decisions. But in the meantime certifiers help.
Actionable: social science research projects related to animal advocacy that Open Philanthropy would like to see. If you’re in social science academia, consider taking up one of these projects as your next paper!
Meta EA
Everyone thinks that effective altruist organizations should seek more funding from donors that aren’t Open Philanthropy, but this is hard, because becoming obscenely wealthy is difficult and obscenely wealthy people are, like, way richer than everyone else. I was shocked to learn that ~90% of non-GiveWell EA funding comes from Open Philanthropy. I particularly recommend reading the top comment from EffectiveAdvocate, which adds important nuance about the social dynamics about what is considered “EA.”
Relatedly: Giving Pledge takers are (non-effective-altruist) billionaires who have agreed to give most of their money to charity. About two-fifths have been accused of serious misconduct, with 4% being imprisoned. Relying on billionaire philanthropy poses serious public-relations risks.
Some effective altruist cause areas benefit more than others from community building (e.g. they need direct workers more than donations, and community building tends to produce direct workers). Historically, funders in cause areas that benefit more from community building (i.e. AI safety) have tended to give to general community-building work and so subsidize cause areas that benefit less (i.e. global poverty). This is kind of unfair. But the other options also kind of suck. If people who prioritize AI safety have nice conferences and one-on-one career coaching, and people who prioritize global poverty don’t, that has weird knock-on social effects. And if the AI safety people stop having career coaching and nice conferences out of fairness, that leaves a lot of value on the table.
Where Open Philanthropy staff is donating in 2024. As usual, the plurality of donations are to GiveWell or GiveWell top charities. Animal welfare is making a good showing, with a mix of farmed-animal advocacy and “weird animal” causes like the Shrimp Welfare Project. Surprisingly little giving to existential risk causes. One of my big conclusions here is that the supposed “normal effective altruist”/“weird effective altruist” division doesn’t exist; weird effective altruists continue to donate to normal charities.
You shouldn’t say “effective altruism should do X.” Effective altruism is not a person and can’t do things. Instead, consider “[specific entity] should,” “someone should,” “you, the reader, should,” or “here’s why a specific claim is incorrect.”
Seventeen sentences to say to get other people to be more ambitious. People keep pulling these on me. >:|
Other Causes
Hayat Tahrir al-Sham, which has taken over Syria, seems weirdly… good? Or at least like they have a good PR department and are trying to appeal to the West? They collect taxes! They support free enterprise! They set up schools! They banned alcohol but didn’t create a morality police and tolerated private drinking! They started a war college and trained people following British military manuals that were posted online! (Apparently you can teach classes on military strategy from publicly posted manuals?)
Particularly Good: Clapping for Stalin. I cried.
Global emissions of local air pollutants have peaked and are now decreasing. (Global air pollutants, like carbon, haven’t peaked.)
Reality Has A Surprising Amount of Detail
Chinese Doom Scroll: Americans’ superstitious belief that they can call their congresspeople and accomplish something. Profoundly confused takes about what America is like. Don’t buy a pregnant woman soda or she’ll sue you for causing her miscarriage.
“Red tourism” in China presents a sanitized, patriotic history of the Chinese Communist Party. You can dress up in Red Army uniforms and swear your loyalty to Mao!
Today, the mafia is mostly elderly men, many of whom live in poverty. “They have lost control of the unions. Latin American gangs with more direct connections to producers dominate today’s drug trade. Many of the other vices on which they built their empires, such as pornography or gambling, are now legal and easily available on the internet. Electronic accounting has made it harder to fiddle the books.”
Particularly Good: Why Netflix produces an enormous number of movies that no one watches.
A sex therapist widely quoted in news articles appears to not really exist. She seems to be a fictional person created so that a sex toy company would be linked by major newspapers, and thus improve its SEO. To be honest, a lot of my feeling here is that major newspapers should ask me for quotes about sex. I would give much more interesting quotes!
Policymaking
Environmentalists stopped the building of a dam to protect an endangered species, the snail darter—which turns out to just be a population of a nonendangered species. The real problem here, I think, is the Endangered Species Act, which has trapped America in an outdated species-based conservation model. If we focused on preserving ecosystems—as our current knowledge of ecology would suggest—we could have a clear-minded discussion of the tradeoffs. Instead, we have environmentalists desperately pretending to care about the single endangered species they can find with the biggest range and rationalizing why whatever species-specific conservation methods couldn’t possibly work and the only thing we can do is preserve the whole ecosystem. And then sometimes the endangered species isn’t even real. Embarrassing.
Florida’s Stop WOKE Act is a blatant violation of freedom of speech: it goes so far as to forbid law school classes from hosting a guest speaker who argues in favor of affirmative action—without limiting professors who want to argue against affirmative action. If you had asked me in 2010 whether the Republicans or the Democrats would be the first to argue in front of a judge that racist speech isn’t protected by the First Amendment, I would absolutely have gotten the answer wrong. Fortunately, viewpoint discrimination remains severely unconstitutional, even if the viewpoint is racist. I recommend checking out the judge’s opinion linked in the article: it starts by quoting the first line of 1984 and only gets more outraged from there.
Chicago gave permission for a casino to open in Chicago, as long as 25% of the casino is owned by women or minorities. Since the casino is owned by Bally’s, a corporation, this requires some amount of… creative… structuring. And now Bally’s is scamming people in black churches into investing at an inflated valuation when they’re almost certainly going to lose their money. Not to mention that discrimination on the basis of race was illegal in the United States even before Trump won and wanted to crack down on DEI. Charming.
Nitrous oxide has long been legal for culinary purposes, but before 2020 people used it as a recreational drug relatively rarely. Cut off from their normal dealers by lockdown, people discovered that Amazon let you buy enormous multi-liter canisters of nitrous—which was much more convenient and allowed for a much more extended high than the tiny eight-gram canisters that were previously available to drug users. Soon, huge canisters of nitrous began to be sold in smoke shops. Unfortunately, nitrous abuse has serious health consequences, such as neurological impairment and spinal-cord degeneration.
Video Games
Particularly Good: Stimulation Clicker: Do you want to play a game that is horrible and unpleasant and no fun at all, but also a really interesting experience? Then you might enjoy Stimulation Clicker! ~40 minutes playtime, it’s worth it to play through to the end. The smart move is to get the true-crime podcast early and then avoid other sound-producing items so you can listen to the podcast, which is hilarious.
If you "liked" Stimulation Clicker, I would "recommend" eCheese Zone, especially if you run it on a projector at a party and work together to solve it.
HTaS in Syria opened by playing a massive COOPERATE once they'd taken over: promising not to attack other countries or allow their territory to be used to prepare attacks on other countries. Maybe we could return the favor, we can always put them back on the naughty list if that changes.