Quick hits (part I)

1) John Cassidy on DeSantis:

This record suggests that Presidential races—primaries and general elections—are usually decided by major political forces that transcend the details of the campaign. In 2016, Trump successfully tapped into nativism, nationalism, economic disenchantment, and alienation from organized politics—all phenomena which were (and are) deeply rooted in American society. In 2020, Biden tapped into an even more powerful force at the time: revulsion at Trump’s Presidency. The question facing DeSantis isn’t whether he gives interviews to Newsmax or CNN. It’s what big political force does he have at his back that could sweep him to victory?

At the start of this year, the answer appeared to be electability. Following DeSantis’s runaway reëlection in Florida, the argument was that he could appeal to a broad spectrum of Republicans who believed it was time to move on from Trump. Today, that voting demographic looks a good deal narrower. Since April, when the Manhattan District Attorney, Alvin Bragg, indicted Trump on charges related to paying off the adult-film performer Stormy Daniels, many Republican voters have rallied around him. Meanwhile, DeSantis has undermined the electability argument by running to the right of Trump on cultural issues, evidently in the belief that “anti-wokeism” is the big force that will propel him to the nomination. It’s too early to say for sure that this culture-war strategy is doomed—he’s still in second place, and the Iowa caucus is nearly six months away—but one thing is certain: it has propelled him into the political netherworld of encouraging anti-vaxxers and inciting bigotry…

It’s sometimes overlooked that, in 2016, Trump twinned his race-baiting and attacks on cultural élites with criticisms of C.E.O.s, corporations, and bankers who had shipped abroad American jobs, and of the politicians who had enabled them. Although debates continue about the relative importance of economic and cultural factors in Trump’s victory, his championing of economic nationalism and protectionism surely played a role in persuading large numbers of working-class Americans to vote for him. Democrats have been forced to respond. The Biden Administration has retained some of Trump’s tariffs on Chinese goods and has adopted an aggressive industrial policy aimed at rebuilding U.S. manufacturing.

In running as the ultimate culture warrior, meanwhile, DeSantis has largely absented himself from broader debates about the country’s future. Back in March, when he delivered his annual State of the State address, he did talk about Florida’s economic record, boasting, “We rank No. 1 in the nation for new business formations. We are No. 1 in economic growth among large states.” Surprisingly, these themes haven’t figured prominently in his campaign to date. And, going forward, according to the NBC News story, he’s going to be making even fewer references to Florida.

Reboot or no reboot, DeSantis appears to be all in on the culture war. So far, this hasn’t worked for him, and appearing at more Iowa diners seems unlikely to change things. Like the rest of the Republican candidates not named Trump, he’s left hoping for a divine or judicial intervention.

2) I’ve done some work with “moral reframing” that didn’t seem to work. But other social scientists keep finding positive results:

Economically progressive candidates—candidates who champion redistributive policies designed to reduce inequality—rarely win elections in the United States. Here, we propose that progressive candidates achieve greater support by framing their policy platforms in terms of values that resonate beyond their progressive base. In two experiments (total N = 4,138), including one preregistered experiment conducted on a nationally representative probability sample, we found that a presidential candidate who framed his progressive economic platform using values consistent with the “binding” moral foundations—e.g. patriotism, family, and respect for tradition—as opposed to values consistent with the “individualizing” foundations, e.g. equality and social justice, received significantly stronger support. This effect was driven by increased support among conservatives and, unexpectedly, moderates as well. By comparison, a manipulation of how progressive the candidate’s platform was had small and inconsistent effects. Despite the potential gains associated with binding framing, analyses using presidential candidates’ debate speeches reveal that appeals to binding values are least common among progressive candidates. These findings show, however, that the alignment between values and candidate support is malleable, suggesting economically progressive candidates can build broader coalitions by reframing the values they associate with their platforms.

3) This chart is something else.  The American right is a global outlier on climate:

4) A quarter-century ago I read Song for the Blue Ocean and gave it to my mom to read, too.  What I most remember about it was just what an amazing and under-appreciated creature the bluefin tuna is.  And here’s the latest on the bluefin in the New Yorker.

By 2007, studies documented a collapsing bluefin population, and in 2008 the scientific arm of the iccat recommended that the allowable catch for bluefin be set at ten thousand metric tons. The iccat set it at thirty-two thousand metric tons. Almost an equal amount of illegal fishing could be expected, leading to a total that some estimated to be about a third of all the tuna to be found in the entire Atlantic.

Bluefin tuna are apex predators, at least if you leave humans out of the system. As we know from wolves and from sharks, apex predators are often keystone species. A tuna consumes eight to ten per cent of its weight in other species every day. “Without bluefin tuna, a phenomenon dubbed ‘trophic cascade’ would occur,” Pinchin writes. She cites a 2012 modelling simulation that saw the numbers of bluefin affecting the health of “swordfish, mackerel, Norway lobster, and bonito tuna”; damage to those populations affects the even smaller creatures such as “herring, squid . . . and phytoplankton,” and finally back to the true apex predator (humans), in increasing the likelihood of damage to coastal communities, which are already threatened owing to erosion, storm surges, and flooding.

In some sense, Pinchin’s book tells a story you might intuit: a species valuable both commercially and ecologically has been fished too much. The network of people and institutions meant to safeguard the fish, and the fishermen, and keep everything “in balance” are a jumble of well-meaning, corrupted by profit motive, corrupted by egotism, and plain unwieldy—all dwarfed, anyhow, by the broader forces of climate change.

5) Cathy Young with far and away the best and most nuanced take on Florida’s school curriculum on slavery.

Blogger Josh Marshall has also noted that the curriculum strongly emphasizes what he calls “other societies that practiced slavery and other places where slavery was arguably worse”—such as the existence of slavery and the slave trade in Africa before the European slave trade began, as well as slavery in the Caribbean islands (where labor was much harder and mortality rates were staggering) and in Central and South America. As Marshall notes, all of this is factual. Teaching these facts is valid; many young progressives raised on modern anti-racist activism seem to believe that slavery makes the United States uniquely evil, or that white people invented slavery. It’s important to put American slavery in global perspective (in which it is sometimes less and sometimes more inhumane than other forms of enslavement). But it’s also important to do so in a context that doesn’t lend itself to an “it wasn’t so bad” message. One could point out, for instance, that most other slave-owning cultures did not have the same hypocrisy problem, since they did not profess to believe that “all men are created equal” and that liberty is an unalienable right.

One may add that while the Florida curriculum wants to spread the blame for slavery, it doesn’t want to spread the credit for its abolition: the British abolitionist movement and its role in ending the slave trade rate no mention, and neither does the abolition of slavery in the British colonies thirty years before the American Civil War. (The only exception: two mentions of instances in which the British offered liberation to slaves in America as a strategy in military conflicts, in the Revolutionary War and the War of 1812.)

In other words, while this is not a curriculum that justifies or denies slavery or white supremacy, it is definitely one that tends to accentuate the positive in African American history, and more generally American goodness: abolition, the civil rights movement, the lives of free blacks, thriving and self-sufficient black communities, African-American heroes and cultural figures. Of course those are important things to learn; but such an approach can also result in a skewed rosy picture that reduces the oppression and atrocities to “bad stuff happened, but let’s look at how we got over it.” Yes, it’s pushback against a tendency on the left to treat racism and the legacy of slavery as America’s defining features; but it’s an overcorrection.

6) Okay, I didn’t actually want to read both of the essays this post summarizes, but it really does seem that some activists use the lack of a gender binary do claim there’s no sex binary, which seems kind of nuts. 

7) Good stuff from deBoer on standardized testing. I especially appreciated it as I had just read Levitz’s piece (whom I almost always agree with) that misapplied this, “Let Me Repeat Myself: The SAT’s Predictive Power for College Grades is Systematically Underestimated Because of Range Restriction”

Why do raw reported SAT-college GPA correlations look low? The fundamental problem is that correlations gathered at any given college can only include students who go to that college. Students who went to better or worse colleges aren’t in your data, and students who didn’t go to college at all aren’t in anyone’s data. This is absolutely basic range restriction stuff. Here’s a crude graphic representation of the possibilities for SAT scores and college GPA.

 

Think about it for a minute. Pearson correlations compare continuous variables to other continuous variables, but for the sake of conceptual ease, think in terms of these quadrants. Data points where a student has a high SAT score and gets good college grades make the correlation higher, as do data points where a student has a bad SAT score and gets bad college grades. Data points where a student has a high SAT score and gets bad college grades make the correlation lower, as do data points where a student has a bad SAT score and gets good college grades. So what’s the problem? The problem is that the kinds of students who have a bad SAT score and will get bad college grades didn’t get into your college in the first place! An entire quadrant of data that would help the correlation is thus excluded. And, at some schools, the obverse is happening – students who have high SAT scores and the ability and temperament to get good college grades went to a better school and thus aren’t in your sample. Since most kids are average, many college students will have a mixture of strengths, some better at tests like the SAT but worse at grades, some the opposite. Which means that most colleges are going to have more of the kids who are stronger in one or the other. But this tells us nothing about how strong the actual correlation is. (Berkson’s paradox, folks!) Meanwhile at the most elite schools where the average admitee has something like a 1550, there are ceiling effects.

8) Pretty interesting from National Geographic, “Why curly hair was an evolutionary advantage”

A research article by Lasisi and her Penn State colleagues, published this month in the Proceedings of the National Academy of Sciences, describes their measurements of how hair regulates scalp temperature in direct sunlight, using different wigs on a “thermal mannequin.”

The mannequin, heated to the average body temperature of 95 degrees Fahrenheit, was placed in a climate-controlled chamber within a wind tunnel that enabled scientists to study the amount of heat transferred between its skin and the surrounding environment.

Three wigs were made from black human hair sourced from China—one straight, one moderately curly, and one tightly curled—so that the researchers could observe how different hair textures affected heat gain and loss on the scalp. They also calculated heat loss at different windspeeds, after wetting the wigs to simulate sweating.

The researchers then made a model of heat loss under different conditions and studied it under the typical conditions in equatorial Africa where early hominins are thought to have evolved.

They learned that all types of hair gave some protection from the sun, but tightly curled hair gave the best protection and minimized the need to sweat—a significant finding, says Lasisi.

“Scalp hair is… a possible passive mechanism that saves us from the physiological cost of sweating,” she says. “Sweating isn’t free—you’re losing water and electrolytes. And for our hominin ancestors that could have been important.”

9) This is great (also, you really should subscribe to Brian Klaas— always so good), “The Funhouse Mirror of Trumpistan:
A new poll surveyed experts on a variety of claims about Trump’s criminality. Then, they asked Republican voters the same questions. The results give a glimpse of a warped bizarro world reality.”

There are several interesting findings here.

First, nearly 95% of experts believe that Trump committed a crime in the federal documents case. Political scientists are least convinced by the Stormy Daniels hush money case, but even in that instance, more than 70 percent of experts surveyed say they believe, based on the available evidence, that Trump committed a criminal offense.

The allegations related to trying to overturn the 2020 election and inspiring January 6th lie somewhere in the middle, with around 80 percent agreeing that Trump’s actions involved criminal behavior. Very few experts — around 10 percent — believe that Biden committed a crime in mishandling classified documents.

Second, Republican views are inverted from those of experts. In the first wave, less than 10 percent of GOP voters said that Trump had committed any crimes, and those numbers were largely similar in the second wave—with one exception. Now, around a quarter of Republicans acknowledge that Trump likely committed a crime in the federal documents case, a significant rise from the last round of the survey.

Unsurprisingly, the numbers are totally flipped for Biden: nearly 70 percent of Republicans think Biden committed a crime. (As a reminder, there is an enormous difference between these two documents cases).

The juxtaposition is even starker when it comes to questions about punishments for alleged criminality, as shown in the next graphic. In case you can’t see it with the fine print, the left bar in each color corresponds to support for “no crime or no punishment”; the middle is for “probation or a fine” and the right is support for Trump serving prison time for his alleged crimes.

 

A large majority of experts believe that Trump should serve prison time for three of the four major sets of allegations against him: the documents case, his attempts to subvert the 2020 election based on his lies, and for charges related to January 6th. Interestingly, there’s some substantial daylight between Democrats and experts on the hush money question, with a majority of Democrats favoring prison time for everything, while the majority of experts believe it would be justified in just three of the four cases, but not the hush money one.

Yet again, the Republican numbers are the outliers, with overwhelming majorities favoring no penalty whatsoever for any of the cases. None. Zip. Nada…

So, there’s a plausible objection here: what if the survey is just capturing partisanship rather than expert viewpoints? Maybe these are subjective opinions and it’s to be expected that the left-skewed field of experts align more closely with Democratic attitudes. That’s a fair question. But we can quickly deal with it by examining questions that relate to objective facts, which is what the clever boffins at Brightline Watch did next.

Lo and behold, for each question about objective statements, the experts got the facts right, while Republicans got them wrong. This isn’t a normative or partisan statement about whether Republican voters are good or bad, but rather a presentation of empirical evidence that Republican voters are badly misinformed about basic facts surrounding Trump’s alleged criminality.

For each of the four statements they surveyed, there are mountains of ironclad, publicly available evidence that these things happened. And yet, Republicans continue to exist in the funhouse mirror version of reality, in which allegations about Trump are either dismissed as lies…or they’re just never exposed to the damning facts of the case, cocooned as many are in a Fox News information bubble.

Astonishingly, only 52% of Republican voters surveyed indicated that they believed that Trump ever brought classified documents to Mar-a-Lago, even though Trump himself admits doing so. 

10) Damn I hate so much the idea that prosecuting someone who gave drugs to a friend who overdosed as a murderer! Maia Szalavitz:

But the recent history of mandatory drug sentencing — nationally and in New York — holds crucial lessons for those who want to end today’s crisis of illegally manufactured fentanyl.

Already, some 30 states and the federal government allow for the prosecution of street fentanyl suppliers as murderers. And at least five require a mandatory life sentence. During the 2023 legislative season alone, fentanyl-related crime bills were introduced in 46 states. In Tennessee, for example, when three teenage girls overdosed in a high school parking lot this May, the 17-year-old sole survivor was charged with her classmates’ murder.

Parental fury is understandable. But the history of tough drug laws shows that they have never significantly reduced the drug supply, or lowered addiction rates or overdose deaths. In fact, drug busts may increase overdose risk while saddling addicted people with criminal records that lower their odds of recovery by making them less employable.

Our history of enacting — and then reforming — harsh mandatory drug sentencing legislation illustrates its failure…

Law enforcement simply can’t stop addiction. Confirming earlier data, a 2018 report by the Pew Charitable Trusts found that states with more drug arrests and incarceration don’t have less drug use. Indeed, the United States, with the highest rate of incarceration in the world, still has the world’s highest rates of illegal drug use, along with the highest rates of illegal-drug-related death and disability.

The pain of parents who have lost children to overdose is crushing. But doubling down on counterproductive policies is not the answer. We know what happens when we sentence dealers on a par with murderers — mass incarceration of Black and brown people, and unaltered and ongoing widespread availability of drugs and death.

This year is the 50th anniversary of the Rockefeller drug laws. The worst way to recognize that grim marker would be by continuing to impose more and tougher mandatory sentencing laws.

11) Excellent from Jesse Wegman, “How to Know When a Prosecution Is Political”

First, what is the case about? Is there straightforward evidence of criminal behavior by a politician? Have people who are not powerful politicians been prosecuted in the past for similar behavior?

Second, what are top law-enforcement officials saying? Is the president respecting due process, or is he demanding investigations or prosecutions of specific people? Is he keeping his distance from the case, or is he publicly attacking prosecutors, judges and jurors? Is the attorney general staying quiet, or is he offering public opinions on the guilt of the accused?

Third, is the Justice Department following its internal procedures and guidelines for walling off political interference? Most of these guidelines arose in the aftermath of the Watergate scandal, during which President Richard Nixon ordered the department to go after his political enemies and later obstructed the investigation into his own behavior. Until recently, the guidelines were observed by presidents and attorneys general of both parties.

Finally, how have other institutions responded? Did judges and juries follow proper procedure in the case, and did they agree that the defendant was guilty? Did an agency’s inspector general find any wrongdoing by investigators or prosecutors?

None of these factors are decisive by themselves. An investigation might take a novel legal approach; an honest case may still lose in court. But considering them together makes it easier to identify when law enforcement has been weaponized for political ends.

12) Sorry, but really sounds like Biden’s dog has to go. I’m sure it’s tough living in the White House, but this is a completely unacceptable level of aggression, “Bidens’ Dog Has Bitten Several Secret Service Agents, Emails Show: One agent went to the hospital after being bitten by the Bidens’ German shepherd, Commander, in a series of at least 10 incidents of “aggressive behavior.””

13) Really interesting stuff from Yglesias, “New evidence that we’re solving more murders”

The big thing you see is that even though the clearance rate plunged between 1970 and 2000, the convict-to-victim ratio soared. Over the next 10 years it fell slightly, but remained much higher than it had been back in the good old days.

 

Their basic conclusion is that contrary to the impression given by the falling clearance rate, the odds of being held accountable for a murder seem to have gone up. The “response %” lines are showing what share of the national population and what share of homicides are in the survey. There’s a big leap in coverage between 1970 and 1991, so you might think the rising accountability ratio is some kind of statistical artifact. But the 19-state consistent sample covers about half the nation’s murders and shows the same trend. It’s a pretty miscellaneous collection of states — California, Colorado, Georgia, Hawaii, Illinois, Kentucky, Maryland, Minnesota, Mississippi, Missouri, Nevada, New York, Ohio, Oklahoma, South Carolina, Tennessee, Utah, Washington, and West Virginia — so the sample seems unbiased.

Again, ideally, someone would be tracking convictions in a more precise way. But given the data available, I think it’s convincing: the falling clearance rate almost certainly reflects a higher standard for arresting someone, not a falling likelihood of being held accountable.

14) I would be really surprised if any content producers successfully sue AI models, “The Generative AI Battle Has a Fundamental Flaw: Writers and artists want compensation from AI firms that they claim have trained their models on copyrighted works. But their legal fights miss the bigger issues.”

At the core of these cases, explains Sag, is the same general theory: that LLMs “copied” authors’ protected works. Yet, as Sag explained in testimony to a US Senate subcommittee hearing earlier this month, models like GPT-3.5 and GPT-4 do not “copy” work in the traditional sense. Digest would be a more appropriate verb—digesting training data to carry out their function: predicting the best next word in a sequence. “Rather than thinking of an LLM as copying the training data like a scribe in a monastery,” Sag said in his Senate testimony, “it makes more sense to think of it as learning from the training data like a student.”

This is pertinent to fair use, the part of US copyright law that generally protects the unlicensed use of copyrighted works for things like scholarship and research. Because if the analogy is correct, then what’s going on here is akin to how a search engine builds its index—and there’s a long history of Google using exactly this argument to defend its business model against claims of theft. In 2006 the company defeated a suit from Perfect 10, an adult entertainment site, for providing hyperlinks and thumbnails of subscriber-only porn in its search results. In 2013 it convinced a New York court that scanning millions of books, and making snippets of them available online, constituted fair use. “In my view, Google Books provides significant public benefits,” US circuit judge Denny Chin wrote in his ruling. In 2014, a judge found in favor of HathiTrust Digital Library, a spinoff of Google Books, in a similar case.

Sag reckons that defendants in similar generative AI lawsuits will use a similar augment: Yes, data goes in, but what comes out is something quite different. Therefore, while it might seem commonsensical that a human reading and a machine “reading” are inherently different activities, it’s not clear the courts will see it that way. And there’s another question mark lingering over whether a machine can make a derivative work at all, says Daniel Gervais, a professor of intellectual property and AI law at Vanderbilt University in Nashville, Tennessee: The US Copyright Office maintains that only humans can produce “works.”

15) Ugh! “Toxin-secreting hammerhead worms are invading the D.C. area. How to stop them.: The snake-like worms contain a potent neurotoxin and can grow more than 22 inches long.”

The hammerhead worm (Bipalium) is an invasive flatworm from Asia, probably spreading to many new locations through the exportation of exotic plants. A handful of sightings have been reported across the Washington area recently. According to the iNaturalist app, Arlington has 43 instances of the worms, D.C. has 16, Virginia has 248 and Maryland has 91 over the past 15 years.

Research shows the genus has spread across the world, especially along the East Coast of the United States. Computer models show the Eastern United States will continue to be a suitable environment for the worms, as the climate warms.

The good news is that the worm isn’t a threat to humans unless ingested or handled, biologist Amber Stokes said.

“Yes, they are poisonous, but they are so small that you would have to actually eat many of them to have any ill effects. I feel pretty doubtful that most people are interested in eating them,” Stokes, a professor at California State University at Bakersfield, said in an email.

Experts said handling them can cause issues like skin irritation. Additionally, like many flatworms, they can carry parasitic nematodes. They can also be dangerous to pets if ingested. But the worms are sensitive to light and don’t usually appear when the sun is out, leaving most of the day worm-free for people and many pets.

16) This is totally nuts and an extreme and absurd threat to academic freedom. An utter embarrassment for Texas A&M, “Texas A&M suspended professor accused of criticizing Lt. Gov. Dan Patrick in lecture: The professor, an expert on the opioids crisis, was placed on paid administrative leave and investigated, raising questions about the extent of political interference in higher education, particularly in health-related matters.”

17) Really enjoyed this from deBoer in Persuasion.  “Humans Suck at Seeing Into the Future: Breathless predictions about nuclear power put today’s AI “revolution” into perspective.” He’s not saying AI won’t change everything, but we should at least have some skepticism and humility in making these claims. And uses early optimism about nuclear power as a really interesting example:

I can anticipate an objection to my analogy here: nuclear power requires a tremendous amount of physical infrastructure to deploy, while AI requires far less; nuclear power (thankfully) involves a great deal of regulation and permitting, while AI development at present requires none. Therefore, an AI takeover is more likely than a nuclear-powered future was in the 20^th century.

I would respond by saying that, first, I think this line of reasoning underestimates just how vast the infrastructure powering AI really is, as these systems require massive amounts of computing power and thus large networks of server farms, which are expensive to build and run. But more to the point, I think this mindset misunderstands the point: the argument is not “nuclear power was likely and didn’t happen, so AI dominance is less likely.” The argument is “history is full of arbitrary turns and random occurrences, causality chokepoints, and chaos injections, and thus we should be far more humble about our predictions about the future.” AI looks inevitable right now. So have many things that haven’t happened.

Someday, people are going to look back at us and laugh, just as we look back and mock those 1950s futurists who predicted a nuclear power plant in every home by the end of the century. There is absolutely no prediction about AI that can be made with the same confidence as the prediction that people in the future—the near future—are going to look back in astonishment at how wrong our futurism was. The most reliable human prediction is that we will go on making bad predictions. The only reason we refuse to grapple with that certainty is because, frankly, our tendency to get lost in hubris is extraordinary. And that’s the greatest impediment we have to putting artificial intelligence and its consequences into their proper context, not bad media incentives or technological ignorance or our relentless drive to tell exciting stories, but the weight of our own egos. 

18) “The Retrievals” podcast is the best Serial season in a while. Really compelling. 

19) This was the talk of the internet much of the week.  You should just read it. Gift link, “Study of Elite College Admissions Data Suggests Being Very Rich Is Its Own Qualification”