[Epistemic status: I have not independently verified each link. On average, about two of the links in each links post turn out to be wrong or misleading, as found by commenters. I correct these as I see them, but can’t guarantee I will have caught them all by the time you read this.]
A neural net trained on change.org tries to write its own petitions, eg “Help Bring Climate Change to the Philippines!” and “Donald Trump: Change the name of the National Anthem to be called the ‘Fiery Gator'”
New study fails to find any evidence that watching pornography as a teenager harms future sexual satisfaction.
The Pentagon vs. the Second Commandment: during the Gulf War, psy ops officers investigated various bizarre proposals, including “project[ing] a holographic image of Allah floating over Baghdad urging the Iraqi people and Army to rise up against Saddam”.
Although successful academics are less likely to have a psychotic disorder compared to the general population, their relatives are more likely, suggesting that psychosis genes are adaptive and maybe creativity-promoting up to some threshold. But looking in more detail, the study finds increased risk for academics’ children, siblings, and niblings, but not parents and grandparents; not sure what to make of this. Hot take: maybe exposure to an academic during your formative years makes you more likely to become psychotic.
The idea behind charter schools is that lots of people will start different schools, some will be better than others, and the good ones will take over. But for the good ones to take over, they would have to scale up from a single school to a whole chain of schools. A new study examines whether they can really do this – and finds that they can. “Estimates based on randomized admission lotteries show that replication charter schools generate large achievement gains on par with those produced by their parent campuses..charter schools reduce the returns to teacher experience and compress the distribution of teacher effectiveness, suggesting the highly standardized practices in place at charter schools may facilitate replicability.”
No opinion on the content, but I really like the name of Michael Huemer’s new blog, Fake Nous.
The regular market is a prediction market on asset values, so if asset values correlate with something we care about, we can use the market to predict how it will turn out. This is the principle behind Schlenker and Taylor on climate change, where they measure the prices of complex financial derivatives relating to air conditioning demand. They find that past investors did a good job predicting the extent of future (now present) climate change, mostly by trusting the IPCC predictions and ignoring doubters. Related: global warming skeptics could bet against future air conditioning demand and make a killing if they’re right – are they trying this?
Pulitzer-Prize-winning journalist Glenn Greenwald talks about animal rights. Also, Pulitzer-Prize-winning journalist Glenn Greenwald has between 24 and 26 dogs, and maintains “extremely personal relationships with each of them”.
If schools replace punishment-based models of discipline (especially suspensions) with “restorative justice” based on helping students build better relationships that prevent them from offending again, does this help students succeed? A big randomized controlled trial investigates the issue. Its own summary of its conclusions is highly positive, saying that it decreased suspensions and racial disparity in suspensions. But the skeptical perspective is that of course it did, just like never arresting criminals would decrease incarceration and racial disparity in incarceration. The real question was whether it made schools better or worse, and Twitter user Spotted Toad makes the case against, saying that it significantly decreased academic achievement at the relevant schools, hurt African-American students most of all, significantly increased the size of racial achievement gaps, and may have even resulted in the death of a student.
What Is Esther Wojicki’s Parenting Secret? Of Wojicki’s three daughters, one is CEO of YouTube, one is CEO of 23AndMe, and one is a UCSF professor of medicine. Wojicki says her secret is letting children be independent and make their own mistakes. Not mentioned as secrets: having kids with a Stanford physics professor; renting your garage out to Sergey Brin and Larry Page so they can make Google. I wonder if her genetics-tycoon daughter cringes every time Mom gives one of these “parenting secrets” interviews.
Scholar’s Stage: stories about sexual selection (like “women are naturally attracted to dominant-looking men, because throughout evolution they were better able to provide”) are meaningless, because for most of human history women did not choose their own mates, and so women are unlikely to have strong biologically-ingrained mate preferences.
Related to recent discussions here: mothers are less likely to breastfeed, seek prenatal care for later siblings. Regardless of how much these things matter, probably a pointer to generally caring less. Probably not an explanation for birth order effects given resetting after seven year gap, but interesting.
I’ve really been enjoying YouTube videos with visual representations of classical music. I know the score is supposed to be a visual representation of music, but these videos work better for me; I’m apparently so much of a visual learner that music makes a lot more sense to me when can I see it with my eyes. Start with Bach and Mozart, then follow the Up Next links if you want more.
McSweeneys: Obituaries For The Recently Cancelled. Suave upper-class jokes like this are doing a thousand times more for the anti-PC cause than Ben Shapiro fans openly protesting PC.
Related: the Internet analogized to Liu Cixin’s Dark Forest.
Supporters predicted that California bill SB50, the YIMBY measure to make it easier to build housing in urban areas, would have passed if it had made it to the state senate floor. Instead, it was killed in committee by state senator Anthony Portantino, “who represents a wealthy suburban bedroom community near Los Angeles” and “whose own solution to the housing problem is to authorize speciality ‘California Housing Crisis Awareness’ license plates” (h/t @webdevmason, follow her for more snarky housing-related updates). The bill may return in 2020.
New paper argues that that screening human embryos for polygenic traits won’t increase those traits much – with current tech and reasonable assumptions, it could only increase height by 2.5 cm or IQ by 2.5 points. But it’s kind of pointless discussing current tech levels when the tech is improving so fast. They also model a scenario where polygenic scores can predict 30% of IQ variability: selected embryos would gain 9 points on average.
The new president of Nintendo of America is named Doug Bowser.
A study demonstrating that homophobia cut 12 years off the life expectancy of gays has been retracted after years of criticism; the finding was the result of a variable coding error. Why did it take so long to discover? Because when other scientists first tried to point out flaws in the study, they faced media attacks like this ThinkProgress piece from 2016 titled Anti-Gay Researcher Now Tries To Claim Stigma Doesn’t Harm LGBT People.
A long time ago I wrote about the Wiseman/Schlitz experiment, where a psi believer and psi skeptic tried to do the same experiment, and found it detected psi whenever the believer did it, and found nothing whenever the skeptic did it. Alert reader Anna Mallett points me to a 2006 follow-up, where they tried to figure out why this happened. The result? This time nobody detected psi at all.
StreetRx tells researchers (and “researchers”) the going price for illegal street drugs in major cities throughout the US.
The military has raised eyebrows by admitting they’re concerned about some recent UFO sightings. There’s some good discussion on the subreddit, which mostly agrees that this article is the best introduction to / speculation on the issue. It theorizes that the military is deliberately trying to stoke public interest in UFOs for some reason, possibly because they’ve got some weird new plane designs they want people to avoid connecting to them if spotted. But assuming Russia/China/other rivals are smart enough to figure out what a random defense blogger figured out, why do they care whether Joe Average Citizen who looks up in the skies and sees a weird plane goes to the papers saying “UFO” or goes to the papers saying “new military prototype”? The Russians/Chinese will just read the article and assume it’s a new military prototype either way. I’d be interested in hearing our local defense experts’ thoughts.
Related: in 1997, the UK launched Project Condign, a top-secret investigation into British UFO sightings. They determined that some of them were inexplicable by normal standards, and posited “a supernormal meteorological phenomena not fully understood by modern science…referred to in the report as ‘Buoyant Plasma Formation’, akin to ball lightning”. Also, “the electromagnetic fields generated by plasma phenomena are also hypothesized to explain reports of close encounters due to inducing perceptual alterations or hallucinations in those affected”. This sounds like they’re optimizing for making it sound like a really pathetic cover-up; given the speculations above, maybe that’s exactly what they’re doing.
You probably knew that classical architecture originally had vibrant colors, but I’m still wowed by this reconstructed Etruscan temple roof.
RIP Norman Hardy, the first terminally ill person to cryopreserved directly after euthanasia. Would-be cryonauts have wanted something like this for a while, since it’s hard and takes time to go from whatever random place you die to cryopreservation; if your death is scheduled, you can be cryopreserved immediately with much less tissue damage. Hardy was an old friend of Robin Hanson’s, part of the original group of late-20th-century geek libertarian transhumanists, and has the most Web 1.0 homepage ever.
Megan McArdle: Europeans should stop mocking Americans’ air conditioning as environmentally unfriendly, because it shifts the US population to hotter places further south. This reduces the US need for heating, and heating is even more power-hungry than AC. High use of AC prompted the migration to Sun Belt cities where heating is unnecessary and so saves energy after all. Anyone disagree?
GeneSight has discontinued its pharmacogenetic tests for ADHD and analgesic tests, admitting they don’t work. I have no link to this because it was sent to me in an email, but their statement reads: “As the number of available pharmacogenomics tests increases dramatically, we believe it is important for clinicians to utilize PGx tests that have demonstrated efficacy and safety in clinical studies. The science for GeneSight ADHD and GeneSight Analgesic does not currently meet this standard. As a result, we will no longer offer these products. We will instead focus on helping clinicians improve the treatment of depression with GeneSight Psychotropic, which has been studied extensively in multiple clinical trials.” Keep in mind I don’t think that one works either, and the FDA agrees.
Related: how come some people take the sedative Benadryl and feel more excited instead of more sedated? Maybe because they’re an ultrarapid CYP2D6 metabolizer.
The rise in social-justice-related terms as a percent of words used in the media over time.
Penn study of school children finds those who nap midday are happier and do better in school. There’s a lot of medical advice against napping going around, but I think it depends a lot on the person.
To everyone’s surprise, there’s another Bitcoin boom, with values tripling over the past three months. Doesn’t seem to me like anyone really knows why, though speculation includes Facebook’s new cryptocurrency giving it an air of legitimacy, and monetary policy problems in India increasing demand for non-government alternatives.
Did you know: once Abercrombie and Fitch hired Slavoj Zizek to write ads for them, and somehow this just straightforwardly resulted in really Zizekian Abercrombie and Fitch ads. I am not making this up. This really happened.
@nostalgebraist has a very good introductory explanation to the transformer architecture, the key insight behind some recent machine learning advances including GPT-2.
Is Summer Learning Loss Real: How I Lost Faith In One Of Education Research’s Classic Results. Contra the title, the article doesn’t really seem to doubt that children forget things during the summer, only that this is a cause of student achievement gaps. Also, one of its points in favor of this is a study that shows achievement gaps are mostly present when kids enter school at age 5 and do not get any worse over the course of schooling. But taken seriously, this would suggest there is no advantage to going to good schools over bad schools at all. I don’t think this is impossible – it’s just the old “everything is genetic” argument – but it weirds me out when people say things that imply gigantic world-shaking conclusions but only use them to debate some minor point.
Speaking of which, Kelsey Piper at Vox has a good article on fact-checking errors in popular books. Her flagship case: a book saying women shouldn’t get married because married women were very unhappy. The author wrote about a survey where “married people are happier than other population subgroups, but only when their spouse is in the room when they’re asked how happy they are. When the spouse is not present: f***ing miserable”. It turned out the author was misreading a “spouse absent” condition on the survey, which actually meant people who were long-term separated from their spouses!
“Persistence research” tries to measure how modern outcomes may reflect long-gone historical events, like how the territories of two different medieval empires may remain different even today based on the long-term effects of their policies. A new paper challenges some of its most surprising results, suggesting that the field does a bad job adjusting for spacial autocorrelation. Potentially not great for “deep roots” work used to justify immigration restrictions.
Anti-Japaneseism is a fringe ideology in Japan which believes the Japanese people are uniquely evil and need to be destroyed, kind of like a single-country version of the Voluntary Human Extinction Movement. From there you can also go down the rabbit hole of weird Japanese leftist movements, including the “Armed Front of High School Students for Violent Revolution” and the United Red Army’s unusually heated “self-criticism sessions”, which “killed 14 of its 29 members in less than a year”.
I once attended a conference on AI risk where a skeptic said he wasn’t going to worry “until an AI could do Winograd schemas”. This referred to a test of common sense and linguistic ambiguity that AIs have long been famously bad at. Now Microsoft claims to have developed a new AI that is comparable to humans on this measure. Some more discussion here.
Vox: climate change will kill lots of people, but probably will not destroy the human race. Reassuring, but just a few days after the article, there was an announcement that Arctic permafrost is melting decades sooner than anyone thought, which probably increases risk of some kind of really disastrous methane/clathrate feedback loop thing. Any climate scientists in the comments want to comment on how worried we should be?
Sultan bin Salman Al Saud was the first (only?) person of royal blood to serve as an astronaut.
The Guardian: The Truth Behind America’s Most Famous Gay-Hate Murder. Argues that the brutal killing of Matthew Shepard was probably drug-related, not homophobia-related.
First controlled study of LSD microdosing. This writeup is mostly negative and describes it as “few benefits and some downsides”, but the actual study is more nuanced and mostly a grab bag of results on a bunch of weird tests that don’t necessarily seem to correspond to what we care about. I say jury is still very out.
If you ever wanted to stay on an Amish farm and experience not having technology, this is your chance: the Amish are now on AirBnB.