Links For October
[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: The history of the exocentric compound noun: although English usually combines verbs and nouns in as $NOUN-$VERBER (eg “firefighter”, “giftgiver”), some lower-class medieval people used an alternative form, $VERB_$NOUN. Their dialect survives in a few words most relevant to seedy medieval life, like “pickpocket”, “turncoat”, and “cutthroat”. (EDIT: see here for corrections and for a more detailed discussion)
2: File under “inevitable”: YouTuber builds a computer in Minecraft that you can play Minecraft on.
3: Harsimony on Less Wrong: Georgism . . . In Space! “Extending the Georgist paradigm into space neatly solves problems with sharing resources and ensures that colonization proceeds at an appropriate pace.”
5: Among new Italian PM Meloni’s political allies is Benito Mussolini’s great-grandson, Gaius Julius Caesar Mussolini.
7: New flying car project, Jetson One, advertises 20 minutes of flight time (at up to 63 mph) per charge, ~$92,000, no pilot license or skills necessary, technically available now-ish but sold out until 2024.
8: What explains this? (h/t @WaltHickey)
9: How Jon Stewart Made Tucker Carlson. Good but hard to summarize. The news used to be staid, neutral, and formulaic, Jon Stewart discovered that a news show could get more viewers by pitching itself as the antidote to the news rather than the news itself, and others (like Tucker Carlson) took that insight in unexpected directions. Also offers an unexpected possible explanation for polarization: there were some regulations and business incentives pushing the news in the direction of being boring until about 1990, but not so much afterwards.
10: Vadim Albinsky on EA Forum: Is it really true that economic growth doesn’t increase happiness? Concludes that this is a pretty weak finding which goes away if you choose to do the analysis a little differently.
11: Wedding Of The Waters: To celebrate a new highway opening in 1937, California arranged a ceremony in which water from North America’s highest (then known) lake was transferred from an Indian to the Pony Express to a mule to a covered wagon to ( . . . ) to an airplane to North America’s lowest lake.
13: Related: Hanania on polls vs. prediction markets for the midterms. The markets give the Republicans better odds, and seem to be counting on pollsters missing paranoid anti-pollster Republicans. “[The Salem/CSPI forecasting tournament] thinks there's only a 22% chance that the 538 deluxe model will favor Republicans on election day, [but that] there's a 47% chance Republicans will actually win the Senate.”
14: Kind of feels like something from Unsong:
15: Related: Lula’s opponent, incumbent Jair Bolsonaro, is under fire after a video surfaced of him saying that he would eat human flesh, and that in fact he tried to do this once but his friends talked him out of it. Is Brazil okay?
16: Yitz on EA Forum: Public Facing [AI] Censorship Is Safety Theater. Big AI companies have whole teams that spend months making sure their AI will refuse to draw boobs or swastikas. Then two weeks later some scrappy open source team releases a copy that can draw as many swastika-covered boobs as you want. Given that all the “Trust And Safety” stuff seems more about protecting AI companies’ reputations than really preventing boobs or swastikas from being drawn, what are we actually doing here? Is it damaging public trust in AI safety? Producing false confidence? Muddying the waters? I guess I should be in favor of this if it wastes AI talent that would otherwise be going to capabilities research - but as a social phenomenon it’s pretty strange.
17: Re…lated? Someone creates an AI filter that makes the black mermaid in the new Little Mermaid reboot white (for a few seconds - I think it’s more of a test / proof of concept than a full alternative movie). I am less interested in this particular trolling attempt than in the idea of customizable media - the idea that if you want a movie character to have a different race or gender or just be played by a different actor, you can make it happen. Or even if you as an individual can’t, maybe a studio can - eg trivially generating a new version of their film with all Chinese actors for the Chinese market, or a Special Feminist Director’s Cut where all the characters are women, or et cetera. How long will it be until we have this? Or will legalities and artistic/ethical concerns prevent it?
18: Travelogue of a science journalist trying to kayak down an environmentally devastated river in California’s Central Valley. Full of interesting stories and facts - this is the piece I got the “Central Valley poorer than Mississippi stat from. Also, there’s some federal regulation protecting “navigable” rivers, and part of the definition of a “navigable river” is “a river someone has successfully navigated”, which produces an incentive for environmentalists to try to navigate “rivers” that no sane person would otherwise try to go down, and for polluters to come up with wacky plans to stop them. I think more laws should also operate as quest hooks.
19: Maxim Lott tries to do an actually good analysis of worldwide COVID death rates, adjusting for reporting biases.
20: Related: plausibly the entire dictatorship “advantage” in GDP growth is just dictatorships lying about how much growth they had.
21: The Supreme Court will probably strike down affirmative action next term, but academics are already planning to replace it with a system of ranking applicants on diversity statements, how many racial-equity-related classes they took as an high schooler, and how many racial-equity-related extracurriculars they did. I can’t tell whether the goal here is to increase admission of minorities (under the assumption that they will either naturally do better at these things, or that the assessment can be weighted in their favor), to increase admissions of racial activists regardless of their own race, or both. [EDIT: A commenter suggests real decision-makers are not taking these kinds of proposals seriously]
22: Stuart Armstrong argues that the “Sadistic Conclusion” - one of the potential alternatives to the Repugnant Conclusion usually considered even worse and not worth thinking about - is actually underrated.
23: Did you know: the steps of the Ethereum development roadmap are called “the Merge”, “the Surge”, “the Verge”, “the Purge”, and “the Splurge”.
24: The British have a proud tradition of discovering (or “discovering”) new lands and naming them after their kings and queens - Victoria Island, the Queen Charlotte Islands, etc. Who was the most recent British monarch to have a new land named after them? It’s actually the current king, Charles III - Prince Charles Island, a 3,600 square mile island in the Canadian Arctic, was first mapped in the year of his birth (1948) and named in his honor. I can’t imagine there are many undiscovered lands left, so William is out of luck at least until we get off Earth.
28: More sentimental cartography of the greater rationalist community (author anonymous by request)
29: An interesting Twitter debate on permission structuring, a communication / persuasion / propaganda technique where you try to ease someone into accepting your position by validating their previous belief but saying this is a special case where you are right. EG “It was reasonable to doubt vaccines back when the original studies were published, but now there’s more research proving they work”, or “It was reasonable not to get vaccines before, but now that there are more dangerous variants we need to change our minds”. To me this just seems like good hygiene - don’t cold open with “YOU AND EVERYONE YOU TRUSTED ARE IDIOTS” - but other people seem to find it more sinister.
30: Stephen McAleese on EA Forum: Estimating the current and future number of AI safety researchers. There are probably about 300 AI safety researchers in the world [insert appropriate caveats], and their number seems to be increasing by about 28% per year (or 10x per decade); at this rate there will be 1500 in 2030. Does that mean that If we can make it another 60 years, every living human will be an AI safety researcher, there will be nobody left to work on capabilities, and the problem will be solved by default?
32: A National Geographic article claims that a newly-discovered rare plant may be the long-thought-extinct ancient miracle herb silphium, which the Greeks and Romans used for cooking, medicine, etc. Only problem: the plant is in Turkey, and all the ancients agree silphium grew in North Africa. The article suggests that maybe Greek traders transplanted some to Turkey successfully (even though the ancients all said silphium couldn’t be transplanted). But according to Wikipedia, genetic studies show the Turkish plant is related to other Turkish plants and not to North African plants, which I think is near-fatal for this theory. Still, according to the National Geographic article, they tried the plant as a spice, and it tastes amazing (as the ancients say silphium did). Confirmation bias, or extreme good luck?
33: Claim spotted on Twitter: Wikimedia Foundation gets $100 million + dollars per year and only needs a single-digit million to keep Wikipedia up. All those ads begging you to donate to “protect their independence” actually give them a huge surplus, some of which gets redirected to leftie culture warrior causes. For example, they gave $250,000 to a group promoting an “intersectional scientific method” that argues that objectivity is “colonialist”, and another $250,000 to a group promoting police abolition. Is this claim true? The very small amount of research I’ve done suggests that it’s true that Wikimedia spends a lot of its budget on things other than hosting (estimates of how much maintaining their websites costs range from 8% to 43%, I haven’t looked deep enough to know who’s right), that some of the remainder goes to grants (this isn’t a specific line on their budget, but seems to be some part of the 32% going to “direct support to [Wikipedia-related] communities”), and that some of these grants do go to “racial justice” type charities, including the two above. Wikimedia says this is about increasing minority representation in Wikipedia/academia/knowledge/whatever, but the charities do also fund controversial work like opposing scientific objectivity or trying to defund the police. I don’t know if any Wikimedia money ends up at those causes. How would people be thinking about this if it went to right-wing culture war causes instead? Or is it cancel culture to worry about this?
34: Why does a line of waterfalls in Pennsylvania track the boundary between right- and left- leaning electoral districts? (answer: because navigable-by-river vs. not determined when an area was settled, who settled it, and how much trade it got)
36: Here’s a variant of the Hollow Mask Illusion I hadn’t seen before:
37: Scott Sumner on the neoliberal wave of the 1970s. I would still like to see a good analysis of whether the neoliberal wave was inevitable (because the mid-century statist policies which seemed to work so well for so long were unsustainable) or an overreaction to a contingent recession and if we hadn’t done it we could have returned to mid-century-style statist policies and they would have gone back to working well. I suspect inevitable but I haven’t seen any really good treatments of this question.
38: Claim: mitochondria look nothing like the textbook illustration of a little capsule; they’re shaped a lot like the endoplasmic reticulum, kind of a web or net. [EDIT: Potentially more complicated?]
39: Very large Pew poll of black Americans. Findings include: 60% say racism is an extremely big problem, 15% say they are regularly (64% from-time-to-time, 21% neither of the above) discriminated against, 65% say increased focus on racial inequality has not improved black people's lives, 35%/39%/23% want local police funding increased/the same/decreased. Many more results at the link.
40: When Trump passed his tax cut, I complained that you could do a lot better things with the $100 billion/year price tag. So I guess it’s only fair for me to link that same post again now that the cost of Biden’s college loan forgiveness has been estimated at ~$500 billion (equivalent to something like ~$25 billion/year, I guess?) But looking over that post from 2017 again I find I am now much more skeptical of all the people who say that homelessness could be solved for $23 billion/year or whatever.
41: Reading that Kardashian-related Substack article the other week paid off, in that now I understand jokes like this: