Saturday assorted links
1. Jonathan Ross podcast. While Jonathan is doing very, very well, he remains underrated.
2. Colombia vs. “Brayan” (NYT).
5. Estate sale with beautiful Philly town home and 100,000 books.
6. Georg Friedrich Haas, innovator (NYT).
7. Learning to live with Chinese surpluses (FT).
8. This Renoir is a bargain. Overall some strong items in the sale.
AI Scientists in the Lab
Today, we introduce Periodic Labs. Our goal is to create an AI scientist.
Science works by conjecturing how the world might be, running experiments, and learning from the results.
Intelligence is necessary, but not sufficient. New knowledge is created when ideas are found to be consistent with reality. And so, at Periodic, we are building AI scientists and the autonomous laboratories for them to operate.
…Autonomous labs are central to our strategy. They provide huge amounts of high-quality data (each experiment can produce GBs of data!) that exists nowhere else. They generate valuable negative results which are seldom published. But most importantly, they give our AI scientists the tools to act.
…One of our goals is to discover superconductors that work at higher temperatures than today’s materials. Significant advances could help us create next-generation transportation and build power grids with minimal losses. But this is just one example — if we can automate materials design, we have the potential to accelerate Moore’s Law, space travel, and nuclear fusion.
Our founding team co-created ChatGPT, DeepMind’s GNoME, OpenAI’s Operator (now Agent), the neural attention mechanism, MatterGen; have scaled autonomous physics labs; and have contributed to important materials discoveries of the last decade. We’ve come together to scale up and reimagine how science is done.
The AI’s can work 24 hours a day, 365 days a year and with labs under their control the feedback will be quick. In nine hours, AlphaZero taught itself chess and then trounced the then world champion Stockfish 8, (ELO around 3378 compared to Magnus Carlsen’s high of 2882). That was in 2017. In general, experiments are more open-ended than chess but not necessarily in every domain. Moreover context windows and capabilities have grown tremendously since 2017.
In other AI news, AI can be used to generate dangerous proteins like ricin and current safeguards are not very effective:
Microsoft bioengineer Bruce Wittmann normally uses artificial intelligence (AI) to design proteins that could help fight disease or grow food. But last year, he used AI tools like a would-be bioterrorist: creating digital blueprints for proteins that could mimic deadly poisons and toxins such as ricin, botulinum, and Shiga.
Wittmann and his Microsoft colleagues wanted to know what would happen if they ordered the DNA sequences that code for these proteins from companies that synthesize nucleic acids. Borrowing a military term, the researchers called it a “red team” exercise, looking for weaknesses in biosecurity practices in the protein engineering pipeline.
The effort grew into a collaboration with many biosecurity experts, and according to their new paper, published today in Science, one key guardrail failed. DNA vendors typically use screening software to flag sequences that might be used to cause harm. But the researchers report that this software failed to catch many of their AI-designed genes—one tool missed more than 75% of the potential toxins.
Solve for the equilibrium?
New data on social media
It has gone largely unnoticed that time spent on social media peaked in 2022 and has since gone into steady decline, according to an analysis of the online habits of 250,000 adults in more than 50 countries carried out for the FT by the digital audience insights company GWI. And this is not just the unwinding of a bump in screen time during pandemic lockdowns — usage has traced a smooth curve up and down over the past decade-plus.
Across the developed world, adults aged 16 and older spent an average of two hours and 20 minutes per day on social platforms at the end of 2024, down by almost 10 per cent since 2022. Notably, the decline is most pronounced among the erstwhile heaviest users — teens and 20-somethings…
Additional data from GWI trace the shift. The shares of people who report using social platforms to stay in touch with their friends, express themselves or meet new people have fallen by more than a quarter since 2014. Meanwhile, reflexively opening the apps to fill up spare time has risen, reflecting a broader pernicious shift from mindful to mindless browsing.
Here is more from John Burn-Murdoch in the FT. I was just doing as Aspen podcast two nights ago, where I spoke of social media as a problem that, in time, largely would solve itself. You also may recall my recent post about declining rates of depression for young adults. That said, you might wonder what exactly is the correct definition of social media (MR comments section?), and whether this study is tracking the proper conception of it.
For the pointer I thank Adrian Kelly.
Türkiye’s Homemade Crises
Türkiye’s response to post-pandemic inflation is a cautionary tale of how political pressure for low interest rates can create macroeconomic instabilities. While central banks worldwide raised interest rates to combat inflation in 2021-2023, Turkish authorities pursued the opposite strategy: cutting real rates to deeply negative levels while implementing financial engineering tools, FX interventions, and financial repression to stabilize markets. The centerpiece was a novel FX-protected deposit scheme (KKM) that guaranteed depositors against currency depreciation, shifting exchange rate risk to the government balance sheet. We provide a detailed account of this policy experiment and develop a theoretical model focusing on how KKM functions and creates vulnerabilities. Our model reveals that pressure to keep interest rates below inflation-targeting levels can lead to an interconnected destabilizing sequence. Low rates generate inflation, current account deficits, and exchange rate depreciation. KKM provides partial stabilization by effectively raising rates for savers while maintaining low rates for borrowers. However, this creates growing contingent fiscal burdens and vulnerability to self-fulfilling currency and sovereign debt crises. This explains additional policies adopted including capital flow management, financial repression, and return to orthodox monetary policy. As central banks worldwide face renewed pressure to set lower policy rates, Türkiye’s experience illustrates the consequences.
That is from a new NBER working paper by A. Hakan Kara and Alp Simsek.
Do LLMs favor outputs created by themselves?
Here is part of the abstract, I will not ask who or what wrote this:
We focus on the hiring context, where job applicants often rely on LLMs to refine resumes, while employers deploy them to screen those same resumes. Using a large-scale controlled resume correspondence experiment, we find that LLMs consistently prefer resumes generated by themselves over those written by humans or produced by alternative models, even when content quality is controlled. The bias against human-written resumes is particularly substantial, with self-preference bias ranging from 68% to 88% across major commercial and open-source models. To assess labor market impact, we simulate realistic hiring pipelines across 24 occupations. These simulations show that candidates using the same LLM as the evaluator are 23% to 60% more likely to be shortlisted than equally qualified applicants submitting human-written resumes, with the largest disadvantages observed in business-related fields such as sales and accounting. We further demonstrate that this bias can be reduced by more than 50% through simple interventions targeting LLMs’ self-recognition capabilities.
Here is the full paper by Jiannan Xu, Gujie Li, and Jane Yi Jiant, via the excellent Kevin Lewis.
Friday assorted links
1. How many legal moves can a reachable chess position have?
2. French guy with opinions visits the UC salaries website.
3. Sam H. on LDS.
4. The job market for economists is weakening.
5. Climate advocacy should focus more on the hard problems.
7. The AI productivity index, with Sunstein, Topol, Summers, and others.
Valuing free goods
There is a new AEJ Macro paper by Brynjolfsson, et.al. on how to value free goods. Here is one of the concrete measures:
Using this approach, we estimate the reservation price [for giving up Facebook] to be $2,152 in 2003 US dollars.
That is for the 2017 version of Facebook. Note this does not measure “whether Facebook is really good for you on net,” but it does indicate some fairly strong demand. And:
…the estimate contribution to welfare due to Facebook in the US over the period 2003-2017 is $231 billion (in 2017 dollars), which translates to $16 billion on average per year.
What I’ve been reading
Marcus Willaschek, Kant: A Revolution in Thinking. A very good book, perhaps the best introduction to Kant? Though for me it is mostly interior to my current knowledge set.
Matthew Bell, Goethe: A Life in Ideas. A beautiful book, now in English we have Nicholas Boyle’s work and also this. Bell is wise enough to understand and value Iphigenia auf Tauris, a good test for Goethe appeciation. Although I had a library copy out to read, I went ahead and bought a copy of this one to own.
Benjamin Wilson, Strange Stability: How Cold War Scientists Set Out to Control the Arms Race and Ended up Serving the Military-Industrial Complex is both interesting and has plenty of information on early Thomas Schelling and his precursors.
Very well researched is The Highest Exam: How the Gaokao Shapes China, by Ruixue Jia and Hongbin Li, with Claire Cousineau.
Peter Baxter, Rhodesia: A Complete History 1890-1980. The most complete history of the country I have been able to find. Many of the other books contain a few dominant, non-false narratives, but one gets tired of that? I say LLMs come especially in handy for learning this history.
Luka Ivan Jukic, Central Europe: The Death of a Civilization and the Life of an Idea. I took this sentence to encapsulate the main lesson of the book, namely that this does not usually work: “Central Europeans were, as ever, masterfully adept at rearranging polities into new configurations.”
I enjoyed Maxim Samson, Earth Shapers: How We Mapped and Mastered the World, From the Panama Canal to the Baltic Way.
On politics and gender
I also had other opportunities to meet with conservatives in DC. With a foot in both worlds, I noticed certain social differences that stood out to me. They center mainly around the ways in which individuals perform gender and are worth reflecting on.
When I talk about differences between conservatives and liberals here, I’m talking about people in politics who hang out in Washington. They may work as campaign managers or speech writers, or have jobs in think tanks, journalism, government, or sometimes academia. The following analysis doesn’t apply to San Francisco rationalists, or Brooklyn Hipsters, or rural church folk in Kentucky. And this doesn’t even apply to all conservatives and liberals in politics, but the ones I happened to spend some time with. So the scope of this analysis is limited, but readers will recognize some of what I’m talking about in other contexts.
The women at Abundance dress business casual. I don’t have the eye for these things to be Vanessa Friedman, so I can’t give a sophisticated analysis of what people wear, but the main difference is the degree to which dress accentuates secondary sexual characteristics. Among the MAGA crowd, cleavage lines are lower and skirts higher, with pants all but unthinkable. There is more makeup and the hair is longer. None of the women wear glasses; among liberals they all have very fancy frames. You don’t have to meet many conservatives or liberals to know this. Roger Ailes famously banned female talent on Fox from wearing pants until 2017. He also of course ended up being brought down for using the workplace as a harem, which he probably would not have been able to do running MSNBC.
In terms of behavior, left-wing women discuss their personal lives or ideas. If they flirt, it’s very subtle. Eye contact that lasts too long, a conversation that continues past the point at least one participant would have ended it under normal conditions, standing unusually close to the other party or looking for an innocent seeming pretext to see one another again. Conservative women, in contrast, flirt as their default style in loud, high-pitched voices. “Oh, you didn’t TEXT ME BACK, I’m so sad!!!” “Would you say I’m Low Human Capital? he he he he.” Of course, any particular signal shouldn’t be taken too seriously as an indicator of interest since they are like this with a lot of men.
You shake hands as a default when meeting liberal women, while with conservatives it would be strange to shake their hand instead of giving them a hug, which they will usually initiate. Liberals bring the norms of HR into social life. Anything too forward or that can be interpreted as showing sexual interest is potentially perilous. Meanwhile, with conservative women, men have the option of coming on to them, and then brushing off the rejection if they are shut down.
While not engaging in ostentatious displays of femininity, liberal women will sometimes drop these hints that subtly remind you they are still women. She might have a pixie haircut and thick glasses on, but will find a way to mention that she likes baking or the color pink. I’ve noticed that liberal women like to discuss how their sons are more aggressive than their daughters, which is the opposite of what must go on in the imaginations of many conservatives who probably picture them all bragging about their children being trans. I think that this stuff is a way to create a little room for gender expression in an environment in which feminist norms and HR culture push towards androgyny.
That is all from Richard Hanania. I too have noticed the hugging point.
Thursday assorted links
1. Lisa Cook stays in office at least through January.
2. “There are 19,000 private equity funds in the US. There are 14,000 McDonald’s in the US. How are there more private equity funds than McDonald’s? That’s actually crazy, right?” (Bloomberg link)
4. Is the stablecoin duopoly breaking down?
5. Worries about bank lending to shadow banks (FT).
7. “Prospect of life on Saturn’s moons rises after discovery of organic substances.“
Uri Bram on throwing a good party
Interesting throughout, here is one excerpt:
14) Couples often flake together. This changes the probability distribution of attendees considerably, and so your chance of losing a quorum in a small-group setting. Small-group couple-events (e.g. 3-4 couple dinner parties) are very hard to manage in a high-flake society, as a result.
15) Create as much circulation at your party as you can. People circulate more when standing than when sitting, so try to encourage standing for those who can e.g. by having high-top tables, or taking away chairs from around tables, or leaving shelves and counter-tops open for people to rest their plates and drinks.
16) Put the food in one part of the room and the drinks in another, or spread the food and drinks out around the space, so that people have lots of excuses to move around the room.
Via (duh) The Browser.
My excellent Conversation with John Amaechi
Here is the audio, video, and transcript. As I said on Twitter, John has the best “podcast voice” of any CWT guest to date. Here is the episode summary:
John Amaechi is a former NBA forward/center who became a chartered scientist, professor of leadership at Exeter Business School, and New York Times bestselling author. His newest book, It’s Not Magic: The Ordinary Skills of Exceptional Leaders, argues that leadership isn’t bestowed or innate, it’s earned through deliberate skill development.
Tyler and John discuss whether business culture is defined by the worst behavior tolerated, what rituals leadership requires, the quality of leadership in universities and consulting, why Doc Rivers started some practices at midnight, his childhood identification with the Hunchback of Notre Dame and retreat into science fiction, whether Yoda was actually a terrible leader, why he turned down $17 million from the Lakers, how mental blocks destroyed his shooting and how he overcame them, what he learned from Jerry Sloan’s cruelty versus Karl Malone’s commitment, what percentage of NBA players truly love the game, the experience of being gay in the NBA and why so few male athletes come out, when London peaked, why he loved Scottsdale but had to leave, the physical toll of professional play, the career prospects for 2nd tier players, what distinguishes him from other psychologists, why personality testing is “absolute bollocks,” what he plans to do next, and more.
Excerpt:
COWEN: Of NBA players as a whole, what percentage do you think truly love the game?
AMAECHI: It’s a hard question to answer. Well, let me give a number first, otherwise, it’s just frustrating. 40%. And a further 30% like the game, and 20% of them are really good at the game and they have other things they want to do with the opportunities that playing well in the NBA grants them.
But make no mistake, even that 30% that likes the game and the 40% that love the game, they also know that they like what the game can give them and the opportunities that can grow for them, their families and generation, they can make a generational change in their family’s life and opportunities. It’s not just about love. Love doesn’t make you good at something. And this is a mistake that people make all the time. Loving something doesn’t make you better, it just makes the hard stuff easier.
COWEN: Are there any of the true greats who did not love playing?
AMAECHI: Yeah. So I know all former players are called legends, whether you are crap like me or brilliant like Hakeem Olajuwon, right? And so I’m part of this group of legends and I’m an NBA Ambassador as well. So I go around all the time with real proper legends. And a number of them I know, and so I’m not going to throw them under the bus, but it’s the way we talk candidly in the van going between events. It’s like, “Yeah, this is a job now and it was a job then, and it was a job that wrecked our knees, destroyed our backs, made it so it’s hard for us to pick up our children.”
And so it’s a job. And we were commodities for teams who often, at least back in those days, treated you like commodities. So yeah, there’s a lot of superstars, really, really excellent players. But that’s the problem, don’t conflate not loving the game. And also, don’t be fooled. In Britain there’s this habit of athletes kissing the badge. In football, they’ve got the badge on their shirt and they go, “Mwah, yeah.” If that fools you into thinking that this person loves the game, if them jumping into the stands and hugging you fools you into thinking that they love the game, more fool you.
COWEN: Michael Cage, he loved the game. Right?
But do note that most of the Conversation is not about the NBA.
Some simple economics of Sora 2?
I do not have access or any kind of inside information on what it can do, or not. Still, from my distance it seems quite possible that the “slop” side of the equation is a simple way to fund AI “world-modeling” (and other) skills in a manner that is cross-subsidized by the consumers of the slop.
That is good, not bad. Let us hope it is true, and so shall all the glass bridges break yet again.
Wednesday assorted links
Higher education is not that easy
More than two years into a conservative takeover of New College of Florida, spending has soared and rankings have plummeted, raising questions about the efficacy of the overhaul.
While state officials, including Republican governor Ron DeSantis, have celebrated the death of what they have described as “woke indoctrination” at the small liberal arts college, student outcomes are trending downward across the board: Both graduation and retention rates have fallen since the takeover in 2023.
Those metrics are down even as New College spends more than 10 times per student what the other 11 members of the State University System spend, on average. While one estimate last year put the annual cost per student at about $10,000 per member institution, New College is an outlier, with a head count under 900 and a $118.5 million budget, which adds up to roughly $134,000 per student.
Here is the full story. Maybe you think this is exaggerated, but I never hear from anyone that the venture is going well. There is a reason for that. A tiny bit you can blame on the FAA.