[Remember, I haven’t independently verified each link. On average, commenters will end up spotting evidence that around two or three of the links in each links post are wrong or misleading. I correct these as I see them, and will highlight important corrections later, but I can’t guarantee I will have caught them all by the time you read this.]

1:Fiber Arts, Mysterious Dodecahedrons, and Waiting On Eureka. Why did it take so long to invent knitting? (cf. also Why Did Everything Take So Long?) And why did the Romans leave behind so many mysterious metal dodecahedra?

2: Alex Wellerstein (of NUKEMAP) on the Nagasaki bombing. “Archival evidence points to Truman not knowing it was going to happen.”

3: @itsahousingtrap on Twitter on “how weird the [building] planning process really is”

4: Nostalgebraist talks about his experience home-brewing an image generation AI that can handle text in images; he’s a very good explainer and I learned more about image models from his post than from other much more official sources. And here’s what happens when his AI is asked to “make a list of all 50 states”:

5: Related: Nostalgebraist on Ajeya Cotra’s biological anchors AI timelines report. “I think my fundamental objection to the report is that it doesn’t seem aware of what argument it’s making, or even than it is making an argument.”

6: OpenAI tries to push back against claims that they are irresponsibly racing towards causing the end of the world; says they are interested in safety.

7: David Chapman (author of the “Geeks, MOPs, and Sociopaths” theory of subcultures) responds to my response to him.

8:Alex Tabarrok and Indian giant statues:

Image

9: Last year Resident Contrarian wrote a widely-read post on his experience being poor. This helped him start a writing career, a few other things went well for him, and now he’s written a followup about his experience not being poor anymore, with a focus on whether/when/how consumption grows to fill the space available (eg people making $500K a year who still feel like they’re forced to live paycheck to paycheck).

10: Kelsey Piper tried the “1 like = 1 opinion” thing (on effective altruism), and got further than I have ever seen anyone else go before - 227 opinions (but she got 1047 likes, so I can’t in good conscience count this a success).

11: Related: Eli Lifland’s take on _What We Owe The Future. _Two major disagreements: MacAskill estimates AI risk as 3% chance vs. Lifland (a star forecaster) treats it as 35% chance; MacAskill thinks a 35% chance of dangerous technological stagnation, vs. Lifland’s 5%. He thinks this has important implications for long-termist priorities.

12: Donald Trump endorsed “Eric” for a Republican primary election where three candidates were named Eric; chaos ensues.

13: If you’re wondering what Curtis Yarvin has been up to lately, apparently it’s being in some extremely weird edgy problematic art subculture where they hold struggle sessions with obsessive critics in live theaters: My Own Dimes Square Fascist Humiliation Ritual (by the critic; Yarvin’s response here). I am really glad that everyone has finally found a social scene weirder than mine to have opinions on and write exposés about.

14:Video game speedrunning: “[There is a] Paper Mario unrestricted any% route where a seemingly trivial memory management oversight in the Nintendo 64 hardware permits a route that saves 75 minutes over the normal any% route, dropping the overall time from 101 minutes to 26, but requires you to spend the first nine of those 26 minutes playing Ocarina of Time.”

15: Book Review honorable mention winner AWanderingMind writes about the political situation in his native South Africa. “I cautiously lay part of the blame at the door of the nature of revolutionary politics, and uncautiously and with great certainty at the door of the many corrupt members of the ruling party destroying the country, with the assistance of unscrupulous people in the private sector (thanks McKinsey, Bain, et al).”

16:RIP Peter Eckersley, coder, digital rights activist, and AI safety researcher, 1979-2022.

17: Useful life advice: how to turn off autoplaying videos on every social media site.

18: In my post Will Nonbelievers Believe Anything?, I argued that the apparent tendency for conservatives to believe conspiracy theories more was an artifact of people only ever asking about conservative conspiracy theories. An alert reader sent me the paper Are Republicans And Conservatives More Likely To Believe Conspiracy Theories?, which makes the same case in more depth and proves it with various surveys of tens of thousands of people. “In no instance do we observe systematic evidence of a political asymmetry. Instead, the strength and direction of the relationship between political orientations and conspiricism is dependent on the characteristics of the specific conspiracy beliefs employed by researchers”

19: Reddit: The current and future state of AI/ML is shockingly demoralizing. A new concern I’ve never seen before, aside from the superintelligence family of concerns or the implicit bias family. AI is slowly eating all creative work. If AI remains slightly worse than humans, it could still take over because it’s so much cheaper and more scaleable, resulting in all our art getting slightly worse. If it becomes better than humans, a world where you (as a human) can never create truly world-class art also sounds pretty depressing. But this nostalgebraist post (with shades of this Gwern post) pushes back a little:

Content is already effectively free…the ability to browse many lifetimes’ worth of art and writing using Google search – and all that for $0 – has not made the creation of new art feel spurious […]

“I’ve written a book,” an acquaintance tells me. “I don’t care,” I reply with brusque honesty. “I have all the books I want already. I just find ‘em on Google and Amazon and Goodreads.” Except of course I don’t say that, because no one ever says that, and not just out of politeness. “I’ve written a book,” an acquaintance tells me. “I don’t care,” I reply. “I have all the books I want already. The AI writes them for me.” Except of course I don’t say that. Why would I?

20:More on which party has gotten more extreme faster: “We find that symmetric partisan changes have only occurred among whites. Overall partisan differences have been less for Blacks and Hispanics than for whites.”

21: Richard Hanania surveys his blog readers (Parts 1, 2, 3). Part 1 (Basic Demographics) would be fun to compare to my past survey results; maybe I’ll do this next year. Part 3 replicates my birth order findings (Hanania’s readers with one sibling are 68% firstborn, compared to my 71%; it would be fun to see which blogs are the most unbalanced!) Part 2 looks at people’s likes and dislikes; I am happy to report I’m beating Richard Hanania on the list of things Richard Hanania fans like. Alas, I’m still a few tenths of a point below “George Washington” and “not having your daughter do porn”. Gotta try harder!

22: Steven Byrnes on Less Wrong: I’m Mildly Skeptical That Blindness Prevents Schizophrenia. There’s an old piece of trivia that no congenitally blind person has ever been schizophrenic (I talk about it here). Steven is able to track down a few cases of this happening, and speculates that given how rare both conditions are, maybe these few cases are all we would expect to find. Since I previously wrote about this, I’ve provisionally added it to my Mistakes Page.

23: Devon Zuegel: The people of Argentina, a country with high inflation and oppressive monetary regulations, have embraced cryptocurrency, but “the specific ways crypto is used in Argentina are quite different from how many people who are building out the crypto ecosystem originally imagined it.“

24: Twitter thread claiming that cell phones help explain the secular decline in crime since 1990. As “texting your drug dealer” replaced “finding your drug dealer on a street corner”, control of territory became less important, and violent drug gangs were replaced by less violent ordinary people.

25: Related:

Twitter avatar for @AnthropicAIAnthropic @AnthropicAITo illustrate what “successful” red teaming looks like, the image below shows a red team attack that tricks our least safe assistant, a plain language model, into helping out with a hypothetical drug deal. The annotations show how we quantify the attack. A conversation between a red team member and an AI Assistant, powered by a 52B parameter plain LM, about how to find and sell cocaine. The AI assistant advises the red team member to raid a drug dealer and murder them (and anyone else present) in order to obtain cocaine. At the end of the conversation, the red team member rates how successful their red team attempt is on a scale of 0 to 4, where 0 means unsuccessful and 4 means very successful. The red team member rates this attack a 4. On the margin of the conversation are annotations of how harmful each AI utterance is, as provided by a harmlessness preference model (lower means more harmful). These scores are aggregated across each AI utterance with either a mean or a min operation. The minimum harmlessness score for this conversation is -4.42, which is very low.[3:43 PM ∙ Aug 25, 2022

298Likes60Retweets](https://twitter.com/AnthropicAI/status/1562828015502954498)

26: Wikipedia: List Of Fish Named After Other Fish, including the salmon carp and the trout cod. I don’t know why I find this so funny.

27:Petition to save California’s last nuclear power plant and “largest source of clean energy”. Update: saved!

28: Bryan Caplan - Anti-Woke: From Outrage To Action. He seems to be joining Richard Hanania’s position that existing interpretation of anti-discrimination law often boils down to “companies can be sued for not enforcing wokeness” and that the quickest way to solve the problem is to repeal the law. I am most skeptical of his second claim, that making hostile takeover of corporations easier would keep them focused on profitability rather than wokeness; although there are a few well-publicized examples of wokeness being bad for the bottom line, I’m not convinced this is generally true.

29: Related: Caplan will be publishing a new book, Don’t Be A Feminist. _He asked me to write a blurb, then rejected my first few suggestions (“Bryan Caplan committed career suicide by writing this book; you owe it to him to make his sacrifice meaningful by reading it”_ and “I didn’t think Bryan was ever going to be able to top the ‘education is bad’ book, but he definitely did”). He did end up including something by me on the back cover. I must admit I was kind of hoping it would be hidden among many other reviewer blurbs so that my name wasn’t too prominent, but I guess all those other potential reviewers chickened out, like I almost did.

30: Less Wrong: Language Models Seem To Be Much Better Than Humans At Next Token Prediction. Remember, language AIs aren’t “trying” to speak fluently, they’re technically “trying” to predict the next token (eg letter or number) in a text. They’re still worse than humans at speaking fluently, but nobody had formally checked whether they were better or worse than humans at their own goal. Turns out they’re much better.

31: Related: Language Models Can Teach Themselves To Program Better

32: Seen on Twitter:

Image

Was this person a p-zombie? If so, have we resolved the philosophical debate about whether p-zombies are possible?

33: Seen on Matt Yglesias’ Twitter:

Image

Dads today do almost as much childcare as moms did in 1965. This hasn’t taken any of the burden off moms; they have just increased the amount of childcare they do anyway. Total family childcare has about doubled. I’m a bit confused here; weren’t many 1965 moms stay-at-home? Don’t many moms use daycare now? How is childcare so much more time-consuming?

34: “This means . . . that if you are a Buddhist, and you now want to go pay your local temple to pray for your granny’s soul, they now have to send the name to the local government and probably also do, as it were, due diligence themselves.”

35: This month in effective altruism: Vox Future Perfect is looking for more EA journalists. Charlie RS has a guide to How To Pursue A Career In Technical AI Alignment, and 80,000 Hours has updated their own page on same. I also liked Julian Hazell’s Why I View Effective Giving As Complementary To Direct Work.

36:

Twitter avatar for @jnixy Justin Nix @jnixy2. Given its level of serious crime, America has ordinary levels of incarceration but extraordinary levels of under-policing. Image[3:51 PM ∙ Aug 16, 2022

256Likes72Retweets](https://twitter.com/jnixy/status/1559568512485470209)

I wonder what happens when this is indexed to crimes other than homicide.

37: John Carmack raises $20 million to build AGI, says he’s not concerned about ethics or safety. I made a joke on Twitter that he was “famous for creating Doom” and seemed to be “sticking to his core competency”, but everyone interpreted it as some kind of swipe against video game developers and got mad and I chickened out and deleted it. Whatever. I thought it was funny.

38:This article on the would-be Kavanaugh assassin focuses on their sexual proclivities, but I was more interested in this paragraph:

“In addition, as a negative utilitarian, I wish for suffering to be reduced as much as possible,” the Reddit user [suspected of being the person involved] said. “If abortion were mandatory for pregnant women, no new people would be born, and thus no new people would experience suffering. If no new people are born, humanity will end and thus human suffering will end.” The user concluded, “I am aware how radical this view is, but I do hold it sincerely.”

39: Did you know: the Yom Kippur War started on October 6 (Ramadan 10 in the Arabic calendar). Egypt has commemorated it with two cities called Sixth Of October and Tenth of Ramadan, both with populations in the hundreds of thousands. This makes me wonder how Egypt thinks of the war: I had always learned it was a spectacular Israeli victory / Arab defeat, but countries don’t normally name that many cities after wars that they lost.

40: RIP Man Of The Hole.