Category: Uncategorized
Wednesday assorted links
2. Apartments that expand at the touch of a button.
3. Has this guy composed more songs than anybody else?
4. Psychoanalyzing Elizabeth Holmes.
5. AI wrote some of the Dallas Michelin reviews. And the reviewer responds.
6. New Yorker profile of James Daunt. The last event I was at I saw three Daunt bags, one of them mine (not in the UK).
The Causal Effect of Foreign Ideas on U.S. Innovation
Globalization is once again better than many people realize:
We study the role of foreign ideas in U.S. innovation by constructing a directed “genealogy” of technologies from the full text of 13 million patents filed between 1887 and 2024. By tracing which earlier technologies appear in the founding patents of new ones, we recover a large-scale input–output network of technological recombination that identifies the foreign technologies that serve as inputs into innovation in the United States. We report four main findings. First, foreign ideas are pervasive: 54% of U.S. technologies build on at least one foreign input, and the most fertile U.S. technologies rely on foreign inputs most heavily. Second, foreign ideas have large causal effects on U.S. innovation. Using the historical settlement of migrants as a source of exogenous variation in local ancestry, we show that U.S. counties more exposed to knowledge from a given foreign country absorb substantially more technologies from that country and, in turn, invent significantly more U.S. technologies building on them. Third, foreign ideas shape not only the level but also the direction of technical change: U.S. counties are far more likely to pioneer the specific technologies whose foreign inputs are already in their local idea stock, and the changing origins of foreign ideas accompany the shift of American innovation from its historical heartland toward the coasts. Fourth, these effects leave a lasting imprint on local labor markets: decades later, counties that absorbed more of an origin’s ideas host more jobs working with those ideas, and even more work with the U.S. technologies descended from them.
Here is the full paper by
Tuesday assorted links
1. Freddie as AI skeptic (and I have given my predictions in fact).
2. Claims about Shakespeare and Marlowe.
3. How did “artificial intelligence” get its name?
4. How is the econ job market this year?
5. AI can now ace econometric coding.
6. The quasi-Riemann proof is becoming intelligible pretty quickly. And new associated math results as a result, authored by humans.
Monday assorted links
1. More on Nigerian and African mega-cities, by the great David Pilling (FT).
2. Quantum gravity labor arbitrage.
3. New (non-AI) claims about consciousness. And more here.
4. “Stop Pink-Washing Patriarchy!”
5. How well do LLMs predict household behavior?
6. The Anthropic Public Benefit Fund. And a numerical estimate.
7. Yup.
*This Beautiful Country*
The author is Edward Cutler and the subtitle is How Italy’s Nature Shaped its Culture and Civilisation. This is an excellent book on how “the Italy you know” has been shaped by its natural environment, such as its mountains, rivers, geology, and so on. Ever wondered why Sardinia is different? The book is good at sticking to specifics. Excerpt:
This extraordinary overlap has made Italy the continent’s botanical crossroads and one of its greatest arboreal sanctuaries. Within Europe, few countries host such a variety of native trees: at least 140 species, more than almost anywhere west of the Balkans. Some, like the Sicilian fir, Lobel’s maple and the Etna birch, are totally unique. The structure of Italy’s forests is equally diverse — dense, high-altitude canopies, open oak pastures, cork scrub and coastal maquis, each shaped by centuries of disturbance and renewal. This is the legacy of a land that sreved first as Europe’s refuge during the Ice Ages and later as its great corridor of return, where northern, eastern and African floras met and mingled in perpetual exchange.
Recommended to all those with an interest in Italy.
Our fascination with screens, demand curves slope downward installment #1637
What we find here is that the quality-adjusted screen time price index is down 39 per cent since 2000, but that change is overwhelmingly a function of improvements in quality, not declines in prices paid for devices, subscriptions and so on. So the steep rise in screen time over the past 20 years has led to a 52 per cent rise in real-terms spending per person on digital technology.
What seemed at first like a story of screen time outcompeting socialising and reading thanks to a widening advantage on cost, now looks more like digital technology winning out through the sheer force of unparalleled convenience and attention capture, even if that comes at a mounting financial (let alone social) cost.
Here is more from John-Burn Murdoch at the FT, wwmas?
The archaeology tranche of Emergent Ventures
- Björn Öste is building Arkhub, an AI-powered platform that analyzes remote-sensing and other archaeological data to identify potential ancient sites in the Amazon and enable further investigation by citizen scientists.
- J. Cameron Monroe is leading the Allada Regional Archaeology Project, using lidar, remote sensing, and archaeological fieldwork to reconstruct the historical landscape of the Kingdom of Allada and strengthen cultural heritage preservation in Benin.
Previously reported grantees include:
- Benjamin Arbuckle is combining archaeology and ancient DNA analysis to reconstruct entire ecosystems of ancient cities, with the goal of understanding how human societies can thrive in balance with their environments.
- Jesse Casana, an archaeology professor at Dartmouth, is developing drone-based radar imaging methods to detect and map archaeological sites buried beneath desert sands. His work combines advanced remote-sensing technologies with efforts to preserve endangered cultural landscapes.
- Leila Character is developing drone-based imaging and AI tools to detect hidden archaeological sites, making discovery faster, cheaper, and more accessible to researchers worldwide.
- Bryce Hoenigman is developing an AI tool to help date ancient cuneiform tablets by analyzing how written symbols evolved over time, with the goal of making archaeological research faster and more accurate.
With thanks to Yonatan Ben Shimon for his generosity in support of this. We do still have funds remaining in this tranche…
Saturday assorted links
1. GLM-5.3 understands Bob Lucas (and then some). It amuses me how many basic principles of economics are being rediscovered in these new contexts.
2. Most famous book tied to every U.S. state. The NJ answer is crazy, how about Roth’s American Pastoral?
3. Adam Tooze on Perry Mehrling.
4. How employment is changing at Anthropic.
5. Susannah Clarke on fantasy writing.
6. Caracas is safe now? (NYT)
Friday assorted links
1. Can Dybantsa (and others) turn around the Wizards? (NYT)
2. Claims about interpretability. I have to say I do not quite “get it”…? Here is some exposition.
3. “90% of Americans’ working hours are spent on tasks not covered well by AI benchmarks. That’s our best guess from a new project mapping frontier benchmarks onto the US economy.” Study here.
4. Ross Douthat (The Free Press) on how to survive the robot apocalypse.
5. Jonathan Beauchamp podcast on genonomics.
6. More Bob Hall on monetary economics.
7. Arundhati Roy is upset at AI.
8. Flying car, at least for a bit. And here. As Roon points out, they offer you the characters too!
Thursday assorted links
China fact (?) of the day
A monograph by Yi Fuxian, a University of Wisconsin-Madison demographer and longtime critic of China’s population policies, argues that China’s actual population may be closer to 1.27 billion—not the 1.4 billion officially reported. The gap would be as much as 130 million people, roughly the population of Mexico. And Yi argues the distortion has warped strategic thinking in both China and the U.S., feeding overconfidence in Beijing and anxiety in Washington.
To make his case, Yi points to evidence in the country’s health and education records. Every newborn in China is required to receive a BCG tuberculosis vaccine within 24 hours of birth. Because multi-dose vials must be discarded 30 minutes after opening, each vial yields only about 1.2 to 1.6 usable doses in practice. Between 2018 and 2020, he noted, the government approved roughly 5.8 million vials a year. Yet official statistics claimed more than 14 million births annually. That would mean each vial somehow vaccinated 2.4 babies, which Yi said would be medically and practically impossible nationwide.
Yi also points to discrepancies in the reported size of individual birth cohorts. For instance, the National Bureau of Statistics said 17.7 million babies were born in 2000. But when that cohort reached age 10 in the 2010 census, only about 14.4 million were counted. By the time they graduated middle school in 2015, only about 14.2 million walked across the stage. About 3.5 million people from a single year were unaccounted for.
Yi said he cross-checks census records against school enrollments, hospital births, vaccine batches and household registrations. The picture that emerges, in his analysis, is a pattern of systematic inflation across multiple data sources.
Here is more from the WSJ, via Bob Davis.
Wednesday assorted links
1. Planets more habitable than Earth? (NYT)
2. Those new service sector jobs: “Cambiar de vida: cómo irte a contar pingüinos en la Antártida por hasta 2500 euros al mes.”
3. AI song about the China Shock debate.
5. “Fewer then 5 innovations could return 1920s era growth.”
6. Anthropic is hiring economists and political scientists, to teach their models those topics.
*Plunder: The Last of the War in Europe, 1945*
By Max Hastings, an excellent book, I benefited from every page. One recurring theme is simply how hard the Germans fought on the way down, and how unusual that behavior was. Here is one interesting passage of many:
In World War I, 150 German soldiers were sentenced to death for serious crimes, of which a mere forty-eight were executed. In World War II, thirty thousand death sentences were decreed by military courts, of which two-thirds were carried out, in addition to many more that took place without any judicial proceeding. By contrast the British executed forty men, the French 103, the Americans 146, almost all the condemned Allied soldiers having committed capital crimes as defined in civil law, rather than being sentenced for dereliction of military duty. Meanwhile the Japanese executed 22,253 of their own men, and the Russians at least 150,000, possibly as many as 300,000.
As I like to say, you can never read too many books about WWII.
Tuesday assorted links
1. Jon Hartley on Bob Hall. And Chad Jones on Bob Hall.
2. Nicholas Decker on who might have won the prize.
4. A parallel AI adjudication track?
5. Kevin Kelly on swarm intelligence.
6. What each country read on Wikipedia yesterday.
7. Is MIT economics good at talent evaluation?