Linkpost for April
Effective altruism, science, criminal justice, rationality, short stories and poetry, fun
Effective Altruism
Global Poverty
Actionable: research questions Open Philanthropy is interested in regarding global human health and wellbeing.
Animal Advocacy
Court cases about animal rights shift behavior by shifting perceived social norms: that is, what people think other people think about animal rights.
Saulius Šimčikas, a former wild animal welfare researcher, talks about why he no longer prioritizes wild animal welfare. In the short term, there are no interventions that outperform helping farmed animals; in the medium term, it seems unlikely we’ll convince governments to care about wild animal welfare; and in the long term, exotic moral patients like digital minds will probably far outnumber wild animals.
Existential Risk
There are roughly three kinds of AI risk. Misalignment risks occur during training of an advanced model: the model winds up wanting something we don’t want it to want. Misuse risks occur during deployment of a model: people use the model to do something dangerous, such as making bioweapons or locking in a totalitarian government. Systematic risks occur when an AI system has become a pervasive part of our society: systematic risks include mistreatment of AIs (if they’re moral patients), humans no longer having input into how society works, and the pervasive use of AIs eradicating important human values.
If incredibly powerful AIs can take over the world, surely less powerful AIs could take over 1% of the world. If AIs might want to take over a smaller part of the world, we’d expect AIs to be less likely to lie in wait and then attempt a full takeover, and for AIs to try to change human values (potentially in a positive-sum way).
We may soon create potentially sentient artificial intelligences. Researchers should prioritize understanding what kinds of beings have experiences and moral patienthood, how much capacity for welfare these beings have, what benefits and harms various beings, and how other concepts like rights or virtues apply to other beings. We can then use this knowledge to try to figure out which AIs are sentient and how we can advance their welfare.
Effective altruists no longer own a castle.
Actionable: questions you should ask about a job or internship in effective altruism.
Old but good post about why effective altruists should consider “fringe” ideas.
Science
It turns out that cystic fibrosis is no longer a death sentence! The article goes into more detail about what it’s like to expect to die young and suddenly discover you won’t.
Criminal Justice (etc.)
Planet of Cops: “The woke world is a world of snitches, informants, rats. Go to any space concerned with social justice and what will you find? Endless surveillance. Everybody is to be judged. Everyone is under suspicion. Everything you say is to be scoured, picked over, analyzed for any possible offense. Everyone’s a detective in the Division of Problematics, and they walk the beat 24/7. You search and search for someone Bad doing Bad Things, finding ways to indict writers and artists and ordinary people for something, anything. That movie that got popular? Give me a few hours and 800 words. I’ll get you your indictments. That’s what liberalism is, now — the search for baddies doing bad things, like little offense archaeologists, digging deeper and deeper to find out who’s Good and who’s Bad. I wonder why people run away from establishment progressivism in droves.”
Article from last September: Baton Rouge police officers were arrested for operating a “torture warehouse” in which they physically and sexually abused suspects. To some extent this is picking out an unusual case when the vast majority of police officers don’t operate torture warehouses. On the other hand, the correct number of torture warehouses for U.S. police to operate is zero.
Particularly Good: The people who visit inmates on death row.
The U.S. legal definition of “espionage” is extraordinarily broad, encompassing anyone who tries to get or communicate any information harmful to national security. For many decades, the U.S. government used its discretion to avoid prosecuting journalists and whistleblowers for espionage, but Obama began to dismantle this restraint.
Rationality
Adverse selection: “conditional on getting to trade, your trade wasn’t all that great.” If everyone has avoided a particular option and you can’t see why, they probably know something you don’t. You don’t interact with a random selection of people, but instead people who are incentivized to interact with you. If you don’t know what someone gets out of a trade, they might be taking advantage of you. In a competitive environment, you can’t beat the market.
Particularly Good: How to moderate a forum, from one of the moderators of the Questionable Content forum.
Short Stories and Poetry
The Afterlife: Life as if it were a movie.
Chicago: In praise of industry.
Why I Am Not Coming To Work Today: A checklist.
To be of use: On people who do useful work.
There’s Six Guys Who Live In This Flat: one of the most disturbing pieces of flash fiction I’ve ever read.
Only Some of True Love’s Miracles: An excellent fantasy Good Robot story inspired by Pygmalion. You might object that this robot is, in fact, murders an enormous number of people, but in my view robots can do anything they want.
Maybe The Stars: A short story about Lovecraft’s Deep Ones. Captures the awe and horror of the species perfectly. Content note for child abuse.
Fun
Chinese Doom Scroll: male webnovels are about acquiring status, female webnovels about acquiring a man. Fantastic cynicism about marrying for love instead of wealth. People seem weirdly likely to believe that other people will poison them out of jealousy??? Extremely petty sabotage as a result of business competition. Chinese people freaked out by involuntary commitment at American mental hospitals. The number of people speculating that the U.S. is separating out “Asian” into ethnicities on its forms as a prelude to a Holocaust is shocking to me. From my perspective, “ask about ethnicities and not just about whether someone is Asian” is a woke thing! It makes me wonder how seriously I’m misunderstanding other countries’ politics.
10-month-old’s letter to Santa.
The forum-moderation article is rather appalling. The section on how to handle suicidal members in particular is horrible, and a good reminder to use a VPN and be careful not to give personal information online.
The "email she wrote a decade ago" link probably goes to the wrong page.