Astral Codex Ten

Sean

13 min read

Sean

Article

Sean is a recurring person in the Astral Codex Ten archive, appearing 5 times across 5 issues between June 23, 2022 and April 01, 2026. The archive places it in contexts such as “Thanks to commenter Sean for hunting down this report and helping explain this”; “I’m not sure I agree with Sean on this one”; “Contact: Sean”. It most often appears alongside California, Chicago, Houston.

Metadata

  • Category: People
  • Mention count: 5
  • Issue count: 5
  • First seen: June 23, 2022
  • Last seen: April 01, 2026

Appears In

Source Context

Recovered passages from the original issue text. When the raw archive preserved outbound links inside the source passage, they are listed directly under the quote.

Book Review: San Fransicko
June 23, 2022 · Original source
The San Francisco districts with the highest (left) and lowest (right) homelessness rates. I correlated homelessness rate and population-adjusted density in the same cities I looked at above, but it didn’t add much predictive value to housing prices. Maybe this is restriction of range (all big cities are dense enough to have homelessness, compared to suburbs), or maybe the key feature is relative rather than absolute density (ie the homeless will go to the densest place nearby). Conclusion: No social phenomenon is ever caused by just one thing, but San Francisco’s homelessness rate is around where a housing-cost-based model would predict. San Fransicko briefly touches on this, but overall tries to de-emphasize it in favor of talking about drugs and mental illness. Critiques of patterns of emphasis are necessarily subjective, but the book’s pattern feels misleading to me. Claim 2: Standard Accounts Underemphasize The Role Of Drugs And Mental Illness In Homelessness Having argued homelessness isn’t just about poverty, the book goes on to say we’re neglecting the central role of mental illness and substance abuse: Over the last decades there were many visible signs that homelessness was about much more than poverty and housing. Between 2010 and 2020, the number of calls made to San Francisco’s 311 line complaining of used hypodermic needles on sidewalks, in parks, and elsewhere rose from 224 to 6,275. In 2018, footage of dozens of people slumped over in an entrance to a Bay Area Rapid Transit (BART) station, many with needles in their arm, went viral. “We call it the heroin freeze,” said one local. “They can stay that way for hours.” Said another, “It’s like the land of the living dead.” For decades researchers have documented much higher levels of mental illness and substance abuse among the homeless than in the rest of the population. It’s true that just 8 and 18 percent of homeless people point to mental illness and substance abuse, respectively, as the primary cause of their homelessness, but researchers have long understood that such self-reports are unreliable due to the socially undesirable nature of substance abuse, and the lack of insight that often accompanies mental illness. Using other methods, San Francisco’s Health Department in 2019 estimated that 4,000 of the city’s 8,035 homeless, sheltered and unsheltered, are both mentally ill and suffering from substance abuse. Of those 4,000, about 1,600 frequently used emergency psychiatric services. Shellenberger’s source for 4000 homeless people having these issues is this SF Chronicle article, which seems to based off of this report. The report does estimate 4000 homeless people with mental illness and substance abuse, but it uses a yearly rather than point estimate of homelessness, and finds 18,000 rather than 8,000 people. That means it only finds a 22% rate of these problems, not a 50% rate. Thanks to commenter Sean for hunting down this report and helping explain this. I looked for other statistics to provide context on this number. This 2013 San Francisco Homeless Count found that 29% admitted chronic depression, 15% PTSD, and 22% some other mental illness. About 30% admitted to a substance use disorder, although as far as I can tell this is just the number who admitted it was a disorder, so maybe more used drugs. This article by the Los Angeles Times describes an LA study finding that 25% of homeless people had mental health issues and 14% had drug issues. The Times re-analyzes it in a way that ups the numbers to 34% and 46%, respectively. But they don’t say exactly what choices they made differently, and the few they do give don’t really inspire confidence. Although in some cases they count questions clearly about mental illness which the official definition inexplicably refused to count, in others they decide to count anyone who has ever had mental illness, reversing a government decision to require the mental illness to be long-term (does this mean that if I lost my house tomorrow, the LA Times use me as an example of a “mentally ill homeless person” because I saw a psychiatrist for OCD when I was a kid?) Studies like these don’t show causation. Sure, mental illness can make people homeless. But homelessness can also cause mental illness. One SF study found psych diagnoses among the homeless to be evenly divided among depression, PTSD, and everything else. Homelessness is a depressing and traumatic environment. Just because someone who’s been on the streets for a year has depression or trauma, doesn’t mean that we should attribute their homelessness to mental illness. This study by the California Policy Lab does better. It asks what factors played a role in homeless people losing their homes, and finds that 50% of unsheltered and 17% of sheltered homeless point to mental illness (given SF’s balance, that suggests 37% of SF homeless would point to that problem). But I can’t help but notice that when you add up the percent of people who lost their homes due to physical illness, psych illness, and drug use, it totals 147%. Based on numbers from other studies, it looks like if you added in job loss, eviction, etc, the numbers would total well above 400%. This makes me think people are saying “yes” if the factor played even a minor role in their eventual homelessness, and this shouldn’t be treated as 37% of homeless having mental health issues being their main problem. The same study finds that about 66% of the homeless “have” some mental health problem, but this time they don’t tell us what question they asked or what criteria they use. What about psychosis in particular? This meta-analysis claims that in developed countries (a category to which San Francisco still nominally belongs) about 19% of homeless people qualify for diagnosis with a psychotic disorder, including 9% with schizophrenia in particular. Not all people with psychotic disorders are completely crazy all the time, and some very much are not, but this is at least a specific condition with real criteria. Conclusion: Overall, I’m disappointed in most of the published research on this question, which seems more interested in producing glossy brochures about funding disparities than in informing anybody what any of their numbers mean. But putting it all together and squinting really hard, I think we can tell a story where 10-20% of the homeless are seriously psychotic, and another 20-30% have contributing mental health conditions including depression, PTSD, and others. Somewhere between 25% and 50% of the homeless have substance abuse problems, and this probably mostly overlaps with the 25% - 50% who have psych diagnoses. I think San Fransicko gets this mostly right. Claim 3: “Housing First” Isn’t As Great As People Think, And Might Be Harmful The National Myth About Homelessness is that The Bad People are refusing to give people houses until they’ve “proven” they “deserve” them, thus perpetuating homelessness when they inevitably fail to qualify. The Good People have united under an exciting new banner called “Housing First” to push the revolutionary idea that people should get houses regardless of whether they conform to normal standards of respectability or not. Wherever this is adopted, homelessness rates fall, and the formerly homeless becoming healthier, safer, and more likely to re-integrate into society. Best of all, the program pays for itself in decreased health care and policing costs. The only impediment to solving homelessness everywhere is the Bad People who still insist on not housing the homeless until they’ve “earned” it. In real life, everyone important has been united under Housing First since the Bush administration made it national policy fifteen years ago, and most of the cities with spiraling homelessness crises have been pursuing Housing First policies for decades (eg San Francisco has been trying Housing First since the 1990s). The Obama and Trump administrations both set funding policies that penalized any non-Housing-First welfare programs. Still, everyone is sure that the reason there are still homeless people must be that some Housing First opponent still exists somewhere, ruining everything with their purity-testing ways. But actually these people have already been relegated to the conservative think tanks where moribund ideas go to die. I have looked through a lot of studies and articles to try to see how well Housing First works. I am most sympathetic to the conclusions of Tsai (2020), who basically says that: Homeless people who are given houses are more housed than homeless people who are not given houses. Way, way more housed. You would not believe how strong of an effect giving someone housing has on them being housed. The same is true for other outcome measures like “time spent experiencing homelessness”, “number of days spent in a temporary homeless shelter”, etc. You might think this is obvious, but this is used as the primary outcome in a lot of studies, and “success” on this metric is behind a lot of claims that “studies show Housing First works great!”
Inline links: this SF Chronicle article, this report, 2013 San Francisco Homeless Count, This article, This study, This meta-analysis, some very much are not, made it national policy, since the 1990s, funding policies, Tsai (2020)
Highlights From The Comments On San Fransicko
June 29, 2022 · Original source
Sean does some excellent detective work on both my and Shellenberger’s claims. I edited some of these into the post already, so you may not have seen the mistaken versions, but here are the corrections. First, on the Zillow study:
Inline links: Sean does some excellent detective work
I’m not sure I agree with Sean on this one - the site breaks down larceny into lots of different sub-crimes, and most of the rise is from car break-ins - which everyone already agrees have gone up a lot. I don’t think this leaves a lot of room for the growth in larcenies to come from shoplifting-not-officially-categorized-as-shoplifting.
Overall I’m very impressed with Sean’s work here. Sean, if you ever have anything this illuminating about some other topic, please send me an email and pitch it to me.
Spring Meetups Everywhere 2024
March 30, 2024 · Original source
TORONTO, ONTARIO, CANADA Contact: Sean A Contact Info: k9i9m9ufh[at]mozmail[dot]com Time: Saturday, April 13th, 2:00 PM Location: Mars Discovery District basement cafeteria. To get to the cafeteria, enter the Mars Atrium via University Avenue entrance. Enter from University Avenue and walk east until you see escalators. Take the escalators down. The food court is to the east of the escalators. If you are lost/confused, ask a security guard to direct you to the food court in the basement. I'll be wearing a bright neon yellow jacket. Coordinates: https://plus.codes/87M2MJ56+XP Group Link: https://www.lesswrong.com/groups/8ktnBi4AjxtCmGeXA
Inline links: https://plus.codes/87M2MJ56+XP, https://www.lesswrong.com/groups/8ktnBi4AjxtCmGeXA
SANTA BARBARA, CALIFORNIA, USA Contact: Sean Contact Info: acxsean[at]gmail[dot]com Time: Sunday, April 7th, 12:30 PM Location: Tables next to UCSB Lot 10 (near Engineering) Coordinates: https://plus.codes/8562C575+XW Group Link: https://discord.gg/vKuJ4NfHkF Notes: Please join the discord to help me coordinate/calibrate group size
Inline links: https://plus.codes/8562C575+XW, https://discord.gg/vKuJ4NfHkF
Meetups Everywhere 2025: Times and Places
August 29, 2025 · Original source
Contact: Sean Brocklebank Contact Info: astral[period]club[period]edinburgh[a t]gmail[period]com Time: Saturday, September 6th, 3:00 PM Location: 31 Buccleuch Place (Edinburgh University's School of Economics) Coordinates: https://plus.codes/9C7RWRV7+6J Group Link: https://chat.whatsapp.com/Bl5 [remove this bit] zIidSM2BA9VlBHbWxV3 Notes: The building is keycard access on weekends but Sean will come out to collect people and bring everyone in at 3pm (if you're late, try to message the WhatsApp group above and someone can come down and let you in). This is a monthly rationalist reading group. Please write to Sean at the address to be added to the list for readings, but for September, this is AI 2027 project material and Amodei's essay "Machines of Loving Grace". We will also meet on October 18 (in a different room on campus) to discuss a few recent articles from Works in Progress (again: message Sean for info).
Inline links: https://plus.codes/9C7RWRV7+6J
Contact: Abi O. and Sean Contact Info: k9i9m9ufh[a t]mozmail[period]com Time: Sunday, October 26th, 2:00 PM Location: Enter the Mars Atrium via University Avenue entrance. We'll meet at the food court in the basement. I'll be wearing a bright neon yellow jacket. Coordinates: https://plus.codes/87M2MJ56+XG Group Link: https://www.lesswrong.com/groups/8ktnBi4AjxtCmGeXA Notes: If for some unlikely reason the Mars Building is locked, which happens occasionally due to protests and other events, we will still meet outside of the University Avenue entrance for 30 minutes after the start time before relocating to somewhere more accommodating.
Inline links: https://plus.codes/87M2MJ56+XG, https://www.lesswrong.com/groups/8ktnBi4AjxtCmGeXA
Contact: Sean Carter Contact Info: sean[period]dan[period]carter[a t]gmail[period]com Time: Saturday, September 06th, 04:00 PM Location: We'll meet at Admiral Arleigh A. Burke Park, at the southeast corner at the gazebo. I'll be wearing a red shirt and there will be a sign with ACX MEETUP on it. Coordinates: https://plus.codes/85FPXQV7+XXF Group Link: https://groups.google.com/g/boulder-acx-ssc Notes: Everyone is welcome. Please note that the area will be outside with outdoor shelter. Snacks provided.
Inline links: https://plus.codes/85FPXQV7+XXF, https://groups.google.com/g/boulder-acx-ssc
Meetups Everywhere Spring 2026: Times & Places
April 01, 2026 · Original source
Contact: Sean Brocklebank Contact Info: astral[.]club[.]edinburgh[@]gmail[.]com Time: Saturday, April 25th, 3:00 PM Location: University of Edinburgh, Old College, Teaching Room 01 Coordinates: https://plus.codes/9C7RWRW7+X3 Group Link: https://chat.whatsapp.com/Bl5 [remove this bit] zIidSM2BA9VlBHbWxV3?mode=gi_t Notes: We meet about once a month to discuss ACX-related readings. Meetings are about two hours. At the April meetup, we will talk about: 1) https://www.astralcodexten.com/p/the-pentagon-threatens-anthropic 2) https://www.astralcodexten.com/p/all-lawful-use-much-more-than-you 3) www.dwarkesh.com/p/dow-anthropic The room is upstairs in Old College, but you can get inside going through the first big door on the right in the main courtyard (assuming you came in from South Bridge). We will have another meetup on Saturday 23 May. Please email me or join the whatsapp group for info about that one.
Inline links: https://plus.codes/9C7RWRW7+X3, https://www.astralcodexten.com/p/the-pentagon-threatens-anthropic, https://www.astralcodexten.com/p/all-lawful-use-much-more-than-you
Contact: Sean Contact Info: k9i9m9ufh[@]mozmail[.]com Time: Sunday, April 19th, 2:00 PM Location: Enter the Mars Atrium via University Avenue entrance. We’ll meet at the food court in the basement. I’ll be wearing a bright neon-yellow jacket. Coordinates: https://plus.codes/87M2MJ56+XG Group Link: https://torontorationality.beehiiv.com/ Notes: If for some reason the Mars Building is locked, which happens occasionally due to protests and other events, we will still meet outside of the University Avenue entrance for 30 minutes after the start time before relocating to somewhere more accommodating.
Inline links: https://plus.codes/87M2MJ56+XG