Links For September 2023
...
[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: This past summer didn’t just break temperature records in the US and Europe, it was an unprecedented increase from previous years. Climate change explains why temperatures are going up in general, but not why the rate of change increased so much this summer in particular, or why it was centered on the North Atlantic. The explanation there might be a ban on sulfur emissions from container ships. Although sulfur has various bad environmental effects, it also blocks sunlight and cools the ocean; removing that effect caused a one-time large temperature increase. So should we start emitting sulfur again? Do more deliberate geoengineering?
2: Related: DIY Geoengineering - The White Paper
3: Related: The recent Hunga Tonga volcanic eruption probably didn’t affect the climate very much.
4: How is crypto going for sex workers? Sex workers have limited and erratic access to normal financial infrastructure due to a combination of government harassment and corporate reputation concerns. Crypto seemed like a solution. But the increasing centralization of crypto under eg exchanges has given it limited value; the same parties who strongarmed banks into dropping sex workers can strongarm crypto exchanges, or close offramps. I’m hopeful that in ten years crypto will have gotten its act together enough to be actually decentralized in a way that avoids this failure mode.
5: One of the largest planes ever seriously proposed was the Lockheed CL-1201 flying aircraft carrier, with a wingspan of 1120 feet:
6: I somehow forgot to mention this until now, but Rational Animations has made a video of my old poem “The Goddess Of Everything Else”:
7: “Adversarial examples” are a weird AI phenomenon where imperceptible changes to certain images can make AIs get them bizarrely wrong. For example, you can take a picture of a house, add a tiny amount of invisible noise representing “giraffe”, end up with a picture that still looks exactly like a house, but an AI will identify it as a giraffe. Now a new paper claims these sort of work on humans?! That is, given the following two pictures:
…and asked to choose which one looks “more cat-like”, people will choose the one with the adversarial example perturbation for “cat” slightly more often than the other. Or if you perturb one train picture to be “cat” and another to be “house”, people will be able to recognize better than chance that one of them seems more cat-like and the other more house-like. I have to admit I can’t see this, but some commenters say they can (maybe it’s my monitor)?
8: Related: AI art has gone from copying humans to inventing entirely new styles. Like images with hidden-yet-obvious spirals:
Or images with hidden-yet-obvious text:
I recommend moving your head closer vs. further from your monitor and squinting a bit for the full range of effects.
9: Related: AI can now translate videos, audio-to-audio, into other languages, but the results may lack nuance.
10: Claim: rodents have gotten smarter over the last century. I can’t access the full paper and I’ve previously found some of the authors lacking in rigor, so it’s probably nothing, but I’d love to hear more about this from someone who has full text.
11: Did you know: an honest-to-goodness Freemasonic conspiracy took over the French military in the early 1900s and may have damaged its ability to fight World War I. With a gravitas fitting its historical importance, it was dubbed “The Affair Of The Casseroles”.
12: You’ve probably heard this already, but: the Data Colada team of statistics bloggers discovered convincing evidence that Harvard professor Francesca Gino fabricated data for some of her studies, ironically on honesty; Harvard agreed and placed her on leave. Now she is suing Harvard and the bloggers for $25 million for libel and “Title IX gender discrimination”. Harvard can take care of itself but the bloggers are normal people without the hundreds of thousands of dollars it takes to defend a case of this magnitude. They have a GoFundMe which is already well-funded but every little bit helps.
13: Unfortunately related: Anti-Ukraine-war website Grayzone says that GoFundMe has frozen their account. They’ve been doing this for years for anti-woke sites, but anti-war sites feels like an escalation. I continue to think crypto is an important safety valve against this increasingly-used tool of control.
14: The Intercept writes about pharma companies’ efforts to get pre-Musk Twitter to censor campaigns to make COVID vaccines open-source. Read carefully, there isn’t much evidence that Twitter ever did censor them in a meaningful way. But the relationship was that the pharma companies made big donations to anti-Twitter-misinformation groups, then lobbied them to lobby Twitter to classify anything inconvenient to the pharma companies as “misinformation”. Again emphasizing that there’s no hard evidence that Twitter complied, this is not the sort of funding ecosystem that inspires confidence.
15: European fertility rates over time. What happened in Czechia in the past ten years? (no, it doesn’t seem to be immigration). H/T @RichardHanania and @benbawan
16: Freddie deBoer’s Derek Chauvin Defund Challenge asked defund-the-police advocates what their plan was for dealing with Derek Chauvin (the cop who killed George Floyd - presumably someone they think should face consequences, and presumably someone who wouldn’t voluntarily accept those consequences if there were no police to arrest him). The winning entry proposed a slightly modernized version of the medieval Icelandic system - a judge could assign a penalty like community service, and if he didn’t do it, the judge could declare him an “outlaw” and make it legal for any citizen to kill him. I agree this solves the problem, but it seemed more like the answer of a clever mechanism design appreciator and not a typical genuine defund-the-police advocate - I’m still curious what their answer is supposed to be.
17: A while ago I went back and forth on some informal studies showing that our intuitive estimate of animals’ moral value matches how many cortical neurons they have (eg a chimp has 100x more moral value than a chicken and 100x more cortical neurons). Rethink Priorities investigated in more depth and provides a long discussion of their findings here. My summary: it actually does match up well, but only after averaging out a lot of non-matching data and people who violently reject the premise, significance unclear.
18: How much does every country like every other’s country’s food? (source: YouGov)
19: A vegan reflects on the failure of the “Liberation Pledge”, an effort a few years ago where vegans would try to force change by refusing to eat at a table where meat was being served. Please stick to discussing this as social experiment instead of posting comment after comment about how much you hate vegans, I already know many of you have this opinion and you don’t need to express it every time the topic comes up.
20: Alex Kesin makes the case for etifoxine, a French anxiety drug which is like Xanax but safer, and discusses our prospects of ever getting it in the US.
21: Study: as far as anyone can tell, Republicans who assert the 2020 election was fraudulent/stolen really believe this, and aren’t just “emotionally responding” to the question. I think in general people should be skeptical that people who disagree with them are just faking their beliefs. This is a good study, but I’m irritated by their replacing “Trump’s claim of a stolen election” with “the big lie” (not even in quotes, they just call it the big lie throughout). While this is a lie and is big, it’s like insisting on calling Trump “Mr. Jerkface” every time you refer to him in a serious scientific paper. It’s not about whether he’s really a jerkface or not, it’s about dignity and avoiding a tedious forced signaling spiral about how willing you are to throw out any pretense of objectivity and fully optimize your language for propaganda.
22: From the subreddit: although it’s fun to invent clever ideas for efficient dating apps, the real challenge to any dating app is getting users and if you don’t have a solution here you have no business plan. I think this generalizes well beyond dating apps to proposals for new social media sites, scientific institutions, etc. Even if your version is better-designed than existing versions, that probably won’t be enough to lure people out of the existing versions where all the other people are. It’s not impossible to succeed at this, but you should consider it at least half of the challenge and focus at least half of your effort there.
23: India is discussing changing its name to “Bharat” (the Hindi word for India) on some level. Unconfirmed rumors about Pakistan being interested in claiming the name “India” for itself. No word yet on who would take “Pakistan”, but I hear Macedonia is looking for a new name.
24: Ethan Strauss on Twitter changes making it harder for posts to go viral. This is my experience too, but I think it started pre-Musk, or at least pre-recently - here’s Freddie de Boer complaining about the same earlier this year, and my impression is it’s been getting worse for years now.
25: Effective Altruist Forum: The charity Pure Earth, sponsored by GiveWell, claims to have reduced the prevalence of lead in Bangladeshi turmeric from 47% to ~0%. Previously, unsavory producers would add lead to turmeric spice to make it appear more brilliantly yellow, poisoning children who consumed it and lowering IQ. Pure Earth raised awareness among consumers, helped the government crack down, and is now declaring at least preliminary victory. “The preliminary findings are that this program can avert an equivalent DALY for just under $1.”
25: Related: EA charity Long Term Future Fund is fundraising. You can see some of their previous grants here, discussion on their grantmaking philosophy here, and an AMA with their grantmakers here.
26: A few years ago I reviewed Julian Jaynes’ Origin Of Consciousness In The Breakdown Of The Bicameral Mind, arguing that Jaynes was on the right track but disagreeing with some of his theories and terminological choices. I didn’t notice until now that orthodox Jaynesian Marcel Kuijsten has an argument here that Jaynes is right and my reinterpretation is wrong.
27: Jobs with the highest and lowest divorce rates (h/t rosey):
Gonna do some trauma re-enacting for a moment here, sorry: if you look at the lowest divorce rate, aside from clergy it’s all big nerd jobs. When I was young, all us nerds thought it was weird that nobody would date us, because we were nice and would probably be better at having good non-abusive relationships than all the successful serial-dater bartender types we knew. This got dubbed the “Nice Guy” argument and everyone agreed it was just a dog whistle for how we were actually rapists who believed we should be able to rape whoever we wanted with no consequences. And now someone gets data and finds that . . . actually nerds have unusually good and stable relationships. HMMMM SHOCKING WHO COULD HAVE GUESSED THIS?
28: Ascento robot security guard:
Thirty years after the Mega Man video games came out, we’ve finally invented the low-level robot enemies you have to kill a few dozen of before you get to the main boss! But (unlike Mega Man enemies) these don’t have weapons, so I wonder what the advantage is supposed to be over having lots of CCTV cameras. Maybe psychological?
29: Re…lated? Blogger/model Aella is offering aella.ai, an “AI girlfriend” based on her, as the flagship product of a company (?) that will help influencers create AI chatbot girlfriends based on themselves. I haven’t seen a lot of uptake yet - my trollish theory, which I might explain more later, is that the real killer app will be AI boyfriends (horny men want sex, horny women want attention / emotional validation; which of these can chatbots more effectively fake?)
30: DSL: have real US household incomes really fallen for three straight years? Does this mean there’s a productivity crisis? (AFAICT the answer is yes, but I welcome correction)
31: Leah Libresco interviews Freddie de Boer on his new book, partial transcript here.
32: A minister describes his experience with prayer: “In college, I started getting up at 5:30 every morning and praying until 9, because I wanted to hear God’s voice and it seemed crazy to spend less time on that than school work. Started hearing him about 5 months in. . . ” (h/t @eigenrobot)
33: Tomas Pueyo on “the loneliness epidemic”. He concludes that people are spending more time alone, but by choice, and they are happy with it.
34: Neuroscientist Erik Hoel discusses the new open letter condemning the integrated information theory of consciousness. I agree with Hoel: IIT is a weird theory, and I don’t personally believe it, but the few attempts to test it have been mildly supportive (including the most recent). Consciousness is inherently hard to study, but IIT proponents (including Tononi, a true great of neuroscience) are trying their best and have behaved entirely responsibly. The signatories’ attempts to (without any argument) go straight to the media and tar it as “pseudoscience” and “misinformation” don’t lower my opinion of IIT at all, but does lower my opinion of the letter signatories. (EDIT: the signatories defend their perspective)
35: Related: Adam Mastroianni: I’m So Sorry For Psychology’s Loss, Whatever It Is. A psychologist argues that the replication crisis doesn’t matter too much, but that this is bad. That is: social psychology as it currently exists is such a poorly-synthesized pile of scattered findings that debunking some of the findings has no effect on anything else (compare to eg physics, where if we learned that photons didn’t exist, we’d have to reassess everything else and plunge into total doubt). He suggests to continue to debunk bad ideas, but also come up with some kind of useful new idea. I give some of my thoughts here.
36: Newly popular Republican candidate Vivek Ramaswamy runs a pharmaceutical company; his track record there will probably become an issue if his campaign goes further. Biotech blogger Ruxandra Teslo has a good analysis.
37: In the US, big cities are further left than rural areas. Is the same true in other countries? See Urban–rural polarisation of social values and economic development around the world, key map (h/t @Whyvert) is:
38: @DrTimothyKelly’s theory of belief updating - I claim this is equivalent to my trapped priors and Friston/Carhart-Harris’ canalization, presented in a slightly graphier way.
39: Emil Kirkegaard has a good overview of claims of a “general factor of personality” and “general factor of psychopathology”.
40: Best of new Less Wrong: The Talk. Why does sex exist? Why do so many living things have two sexes, instead of some other number? Why do the sexes have differently shaped gametes? Why do species that have sex correlate so closely with species that have mitochondria? And other sexy questions.
41: AI company Anthropic announces partnership with Amazon (including $1.25 - 4 billion investment). This was predictable: the story of the AI industry so far has been that from 2015 - 2020, a few true believers founded early startups that ate up the talent and gained the institutional knowledge. Now that AI is the Next Big Thing, the big tech companies are trying to catch up, having a hard time, and choosing to partner with the prescient early startups instead. The early startups are finding they can’t keep scaling without more money and data, forcing them to accept the big tech companies’ offers. First it was DeepMind + Google, then Open AI + Microsoft, and Anthropic was the last holdout but has acknowledged economic reality. The safety movement is concerned that Amazon might have enough power to steamroll over Anthropic’s safety-conscious culture; this did happen with DeepMind and Google, didn’t with OpenAI and Microsoft, and my guess is Anthropic held out for a good enough deal (and had enough bargaining power) that it won’t happen there either.
42: Related: one joke I keep hearing is that Anthropic will single-handedly put FTX back in the black - FTX was one of Anthropic’s biggest early investors, and Anthropic’s valuation keeps jumping by billions of dollars. Could this be literally true? I think not yet: this article explains that FTX has $16.9B in liabilities and $9.5B in remaining assets, for a debt of ~$7.5B. We don’t know what stake they had in Anthropic, but they were lead investors in Series B, Series B is usually 25-40% of stock, I’m going to estimate about 25%. Amazon offered to pay $4 billion for some unknown stake in Anthropic; if it’s 49% (the same as Microsoft in OpenAI) that values the company at $8 billion. So FTX has $2 billion worth of stock, less if it’s been further diluted. That’s only enough to take care of about a quarter of their debt. Will Anthropic go up 4x in the next few years? OpenAI is already seeking (though hasn’t yet gotten) a valuation of $90 billion and it doesn’t seem unreasonable for Anthropic to be a third as valuable as OpenAI, so who knows?
43: Epstein . . . did kill himself?
44: Little known grammatical tenses: the prophetic perfect tense, “a literary technique used in the Bible that describes future events that are so certain to happen that they are referred to in the past tense as if they had already happened.”
45: Romana Didulo mixes sovereign citizenship, QAnon, and messianism in her claim to be Queen of Canada, by the grace of the US military and extraterrestrials; her followers (30 who literally follow her around, 30,000 on social media)considered dangerous. I think this is an especially interesting case for theorists of religion - the James Strang to the original Q’s Joseph Smith. The cult deficit must have something to do with the channeling of religious feeling into politics, and the difficulty of having political “unobservables” in the sense that God and angels are unobservable. But with sufficient paranoia, everything becomes unobservable. We will have schisms over the true nature of the Senate; crusades will be fought over which amendments are in the Constitution; martyrs will go willingly to their deaths over how many pages are in the Inflation Reduction Act. This will be bad, of course - but sociologically fascinating.
46: Related: @jeremychrysler discusses John Curry’s “Tragic Prelude”, a mural honoring John Brown in the Kansas capitol:
47: Lantern Bioworks plans to produce a genetically modified mouth bacterium that will outcompete your normal mouth bacteria and eliminate cavities (conflict of interest notice: my wife consulted on a version of this project).
48: Hyundai may have solved parallel parking forever (source):