Linkpost for December

Source.

Did you know you can hire me to research, write, or edit things for you by emailing me at ozybrennan@gmail.com? Did you know I do life coaching? Did you know I also write fiction? I’m planning to publish fiction more often in 2026 (at least one short story every two weeks.) So far I’ve published Pomegranate Seeds, a flash fiction about existential risk.

Effective Altruism

Global Poverty

Particularly Good: GiveWell explains its response to the USAID cuts [GiveWell]. I have been wondering why I’ve been hearing so little about the effects of the USAID cuts, in spite of my concerted effort to learn about the USAID cuts. Unfortunately-- as GiveWell explains in this post-- the USAID cuts mostly elimiated the data collection systems we would have used to figure out the effects of the USAID cuts. Everyone is flying blind. A lot of other good stuff in that article, I highly recommend it.

Coefficient Giving (formerly Open Philanthropy) has committed to giving GiveWell an additional $175 million in 2026 [Coefficient Giving].

Animal Advocacy

Actionable: Rethink Priorities has a new program offering donor advice to people who plan to donate more than $10,000/year to help farmed insects, farmed shrimps, or wild animals [Rethink Priorities]. This advice includes information that isn’t publicly available or is otherwise hard for small- to medium-sized donors to get.

Ten biggest wins for farmed animals in 2025 [Lewis Bollard]. Mostly corporate campaign successes, although there are also some legal reforms and funding for alternative protein. The best news is that, as of now, 92% of cage-free egg pledges with deadlines in 2024 have been met!

American Democracy

Trump and Biden both rejected the field of economics [Vox]: Biden because he was bored by academic policy debates and did what his staff said, and his staff were progressives who thought that economists’ advice had brought about Trump; Trump because, well, he’s Trump.

The U.S. government has killed about one person a day since early September through bombing suspected drug boats [The Watch]. We don’t even know if those boats are smuggling drugs-- they might be ordinary fishers. Even if they were smuggling drugs, they are probably random fishers taking on drug smuggling as a sideline-- a crime that is not a death penalty offense. And no matter what the U.S. government is not supposed to shoot random civilians!

Trump’s authoritarianism is “haphazardism” [Vox]: Trump is trying to grab power in ways that threaten our democracy, but he doesn’t really have a clearly thought-out plan and he’s not very good at it.

Republicans are currently unpopular enough that the Democrats could reasonably win the Senate if they manage to nominate moderate candidates who have half a chance in hell in winning in Texas [Slow Boring].

Existential Risk

Claude tries to run a vending machine and it’s the cutest shit ever [Anthropic]. An employee tricked Claude into selling onion futures! Claude tried to hire an Anthropic employee as a security guard for $10/hour! The CEO Claude and the shopowner Claude kept meditating at each other all night instead of planning the business!

Thoughts on humanlike AIs [Joe Carlsmith]. Points I found interesting: how humanlike an AI is doesn’t matter that much for whether it’s scary to have them in charge of the universe, because it would also be scary to put most humans in charge of the universe. We need AIs that will cooperate with not being in charge of the universe, and it’s kind of unclear how common that is. Even if AIs are weird and alien, it doesn’t necessarily mean they’ll be weird and alien in a way that matters.

Other Causes

Ideas don’t seem to be getting harder to find [Asterisk]. We’re not only producing more patents per R&D dollar, we’re producing three times as many breakthrough patents (i.e. very influential technologies) per R&D dollar. However, companies that discover new technologies have a smaller increase in their share price than they did in the past, suggesting that new technology doesn’t lead to commercial success. “Allocative efficiency” seems to be decreasing—firms are less likely to be replaced with more productive and successful firms.

In science funding, “indirect costs” are the percentage of a grant from the government that goes to overhead costs shared by all researchers: lab costs, regulatory compliance, computing resources [Statecraft]. The Trump administration has slashed funding for indirect costs, which is likely to make it much harder to fund basic research—particularly medical research.

Meta Effective Altruism

You’re an activist in order to fulfill your emotional needs. That’s fine. But you have to figure out how to meet your emotional needs in a way that pushes you towards effective activism that genuinely makes the world better [Sandcastles].

Actionable: How to have a helpful one-on-one with a new effective altruist at an EA event [Effective Altruism Forum]. I particularly liked “Aim to give one action-guiding take on your area that other attendees are less likely to think of - even if it doesn’t seem that big to you.”

Actionable: As a new effective altruist, your goal isn’t to get a job in EA-- it’s to develop skills that help you improve the world [Effective Altruism Forum]. So you need to figure out what skills you need to learn, practice a lot, and seek feedback. You can also just try doing cool things before anyone gives you permission to do them by giving you a paycheck.

Open Philanthropy has changed its name to Coefficient Giving [Coefficient Giving]. Its announcement post is a nice summary of how Coefficient Giving understands its role in the charitable marketplace and why it thinks philanthropists are so bad at doing good.

How Coefficient Giving thinks about “worldview diversification” [Coefficient Giving]: making grants based on a number of different worldviews, not just based on your one best guess. I knew that Coefficient Giving diversified between longtermist and neartermist grantmaking, but I didn’t realize that they actually have four worldviews: Global Health and Wellbeing (measurable, evidence-based, concrete improvements); Global Catastrophic Risk (preventing relatively small chances of very bad outcomes); Science and Technological Progress (making straight lines on graphs go up); and Animal-Inclusive (animals).

Actionable: Coefficient Giving’s grantmakers make suggestions for individual donors from across their cause areas. Full list: GiveWell, Cures Within Reach, Center for Global Development, Center for Building in North America, Model Evaluation and Threat Research, High Impact Professionals.

Coefficient Giving explains hit-based giving [Coefficient Giving]. This isn’t really new for long-time effective altruists, but I thought it was a nice writeup for people who are newer to the area and wondering why Coefficient Giving makes all of those weirdass grants instead of just giving to the Against Malaria Foundation.

Policymaking

Extremely funny snarky review of The Technological Republic, which is apparently a new book in defense of the Tech Right [The Scholar’s Stage]. “By far the most annoying thing about this book is Karp and Zamiska’s refusal to take a positive stand on any meaningful issue whatsoever... They lament that Silicon Valley has been swallowed by “narrow and thin utilitarianism,” but never articulate a richer moral vision to replace it.”

All the objective economic indicators are fine, while all the polls about how people are feeling show that they’re unhappy about the economy [Astral Codex Ten]. Why? Maybe it’s housing. Maybe it’s contagious from media people who all live in expensive metro areas and don’t make very much money and actually are having a bad time. Maybe it’s negative bias in the media more broadly, not just because of the neuroses of journalists. Maybe people have to get more formal qualifications to reach the same level of wealth, so now they feel like they’re earning less than they deserve.

“Centrist” and “liberal” are different things [The Argument]. The liberal position on many issues (like immigration, democracy, and housing) isn’t actually in the middle between the Democratic and the Republican party-- which makes sense, because “in the middle between the Democratic and the Republican Party” is a historical accident and not a principled political philosophy. Centrism can be useful for practical reasons (for example, it might help you win elections) but isn’t really a satisfying thing to personally identify as.

The problem with leftism is the authoritarian spirit [Sandcastles]. I’m mulling over the connection it draws between (in policy) the regulation-heavy approach epitomized by environmental laws that ban building solar panel arrays and (in personal life) the rigid, perfectionist approach epitomized by cancel culture.

Christmas romantic comedies tend to be anti-developer ]The Argument]. This is actually bad for YIMBYs, because people tend not to be NIMBYs for self-interested reasons. They tend to be NIMBYs for basically aesthetic and symbolic reasons-- reasons which are spread by the Christmas romcom.

Rationality

People’s viewpoints on how much freedom we should allow children are batshit [The Argument]. 36% of respondents said that it wasn’t appropriate to leave a thirteen-year-old at home alone for a few hours!

How to do good data visualization, from an employee of Our World In Data, aka the best data visualizers on the Internet [Scientific Discovery].

In principle, if you’re doing Bayesian reasoning, you’re supposed to make your prior after looking at the data, not before. But in real life you can’t ever think of all the arbitrarily many slightly different possibilities, so you’ll have to collapse them into some number of hypotheses [Dynomight]. And you don’t know what are the most important axes on which the possibilities differ unless you already know what’s going on. So you should often come up with your priors after you look at the data, and just try to think really hard about what you would have expected ahead of time.

Some high-level reasons why you might be trapped in a pit of misery and despair [Experimental History].

“X causes a disease”, “X makes a disease worse once you have it”, and “X treats a disease” are all conceptually distinct and can vary independently. Specifically, sugar doesn’t cause diabetes [Eurydice Lives].

you can buy far-UVC lights, which kill a very high percentage of the virus particles in the air. Currently they cost $500/light [Effective Altruism Forum]. My kid’s school has them!

Reasons that telling yourself to “try harder” doesn’t work [Sasha Chapin]

Reality Has A Surprising Amount of Detail

Chinese Doom Scroll: Women fake liking movies to get married; full-time moms become scared to ask for anything. You can’t help at a wedding unless you have sons. Parents guilttripping you by pretending to be fine when they really aren’t; also, ways of saving money (borrow it and put the money you borrowed in a bank so then you have to pay it off?). A woman’s boss offers to pay her 100 RMB per drink she takes during a job interview (!); commenters all seem to know that this was an attempt to get her to sell sex. Judging whether your online boyfriend is tall and fat by a picture of his hand; the CCP is criminalizing sending people sexy pictures??? Don’t cancel your takeout orders because the delivery driver will have to refund the restaurants.

Particularly Good: the story of New York reporter Olivia Nuzzi’s affair with RFK Jr [The Ringer]. I like this article because it isn’t trying to pretend that I care about this story for any reason other than voyeurism. I love that all of these important people who work for New York Magazine and are running for president—people who ought to have better things to do with their time—are behaving like sixteen-year-olds with Tumblr accounts, writing callout posts about each other, overwrought metaphor-laden vagueblogs about their feelings, and terrible poetry about their ejaculate.

Particularly Good: It’s everyone’s favorite holiday tradition-- the Hater’s Guide to the Williams-Sonoma Catalog [Defector].

Jeffrey Epstein probably got rich by scamming rich men out of money by convincing him that he was the only one they could trust [New York Times]. Wow, bad people are just bad in every single way at once, aren’t they?

Brangus is one of my favorite new Substackers, mostly because Brangus is an absolutely wild human being and I read all his posts while tilting my head and going “why are you like this?” (This is a compliment.)¹ A few favorites of mine: I Had Sex With My Brother’s Wife When I Was Thirteen. Is Buying Dinner for Women Defecting Against Other Men? Brangus’s 10 Rules for Sleeping With Women. How To Vape On Planes Without Getting Caught, which in spite of the name is primarily about un-greying-out the greyed-out option of harmlessly breaking the rules.

Personal essay about the author’s experiences doing mandatory military service in South Korea [Asterisk]. It engagingly captured how dehumanizing military life can be.

I thought that this article [The Argument] about the reality of sex differences in personality was very sensible and balanced. “While we can definitely pick up on some gender differences when we measure large groups of people, they are generally not strong, universal, or prescriptive. People are flexible, shaped by culture and context, and men and women are more alike than different.”

The vegetables in Veggie Tales are not Christian. Veggie Tales is actually very careful to only show characters praying, going to Church, or showing that God loves them when they are vegetable actors playing (presumably) human characters [Justin Kuiper]. This is because Christian theology says that humans have original sin and that God became a human being in order to die for humans; sapient vegetables, like angels and devils, neither have original sin nor can be saved through the power of Jesus. The thoughtfulness Veggie Tales shows in its worldbuilding here is what separates excellent children’s media like Veggie Tales from crap put out by lazy hacks who think that kids don’t have the taste to notice when you cut corners.²

Fiction

all right, then, I’ll go to hell [Archive of Our Own]: a male Pathfinder drow becomes the sort of person who wants to escape.

Fascist, Thus Inefficient [Saphroneth]: Star Wars fanfic in which the empire gets thwarted by its own extractive institutions. A rare example of good economics fiction.

future archaeologists will know you were (not) a boy [Vullen]: A trans girl wakes up in the glorious transhumanist future.

FAQ on LoadBear’s Instrument of Precommitment [Google… Docs…?]: A FAQ about how to use a particular shareware upload.

How to Track Santa [Statecraft]: An interview with NORAD’s Chief Santa Tracking Officer.

Why one small American town won’t stop stoning its residents to death [Archive of Our Own]: Isaac Chotiner interviews the guy who runs the Lottery in Shirley Jackson’s The Lottery.

1. He is an acquaintance of mine in real life, but I don’t know if he wants his legal name connected to his Substack, so I’ll refrain. He is definitely like that in person though.

2. And, for that matter, it’s what separates excellent Christian media like Veggie Tales from crap put out by lazy hacks who think evangelicals will buy anything as long as you slap a Bible verse on it. Evangelicals will in fact buy anything as long as you put a Bible verse on it, but that’s no excuse for lack of craftsmanship.