The mass replication studies published in Nature today are insane, exemplary and an enormous pile of work to improve science.

It's just so awesome how many people spent time on this in return to be 1/100 coauthors on a thing.

Just outstanding 🙌🙌🙌 1/ 150+ authors, the work of organizing and systematizing all this independent work is insane too.

They'll get too little credit, but understanding how social science functions as an institution is just insanely useful 2/

I was writing about land reform in West Bengal last night and was curious if it had persistent effects on the ownership distribution.

So I did what anyone would do, I* wrote an academic paper on it

Turns out — yes! 1/

The data are the complete landholding distribution of rural household from the 2012 SECC.

"Writing" this paper took approximately 2h of prompting.

Most of the time was spent vetting errors and outliers. The intro was helped by a DeepResearch query on the question.

2/

I'm enjoying this @alexolegimas suggestion so much.

In addition to being fascinating (who knew C. Elegans had so much in common with a Roomba), the author @maxsbennett has such a fine way with words.

Some examples in thread 1/ Fungal spores are all around us, patiently waiting for something to die.

I love this image, it's so dark, and inevitable. Take your time, we can wait.

2/

Here's an underrated theory paper on college admissions.

When the stakes get high, we end up selecting too many people who are good at gaming the system, and too few who are actually good. Link: journals.uchicago.edu/doi/epdf/10.10…

Ungated: navinkartik.com/Papers/FK-MudI…

👀👀

Very very interesting — data on disability accommodations in college, at last!

Many of us have suspected that high-income students are benefiting the most from disability accommodations. Some answers!

🧵 as I read 1/

https://twitter.com/_AlvinChristian/status/1989335685531369876

Data is from a single (unnamed I think) state school. Big enough to be representative.

Though I suspect behavior at elite schools might be different — I am afraid elite admissions selects (among other things!) for the people who are good at scheming these things.

2/

What happens when online job applicants start using LLMs? It ain't good.

1. Pre-LLM, cover letter quality predicts your work quality, and a good cover gets you a job
2. LLMs wipe out the signal, and employer demand falls
3. Model suggests high ability workers lose the most

1/n In April 2023, made it possible for workers to use AI in their cover letters.

Employers can't see if they used the tool.

Time spent to submit an application goes down, with a big increase in apps that took <30 seconds 2/ freelancer.com

In the 1980s, Nestlé was moving into infant formula markets in low-income countries.

With each new market entry, moms switched from breastfeeding to unclean water, and infant mortality increased substantially; 200,000 excess deaths per year. 1/n Amazing — Nestlé saleswomen dressed in nurses uniforms to pitch their product to women in hospitals right after delivery.

Marketing memos were very direct on this — "medical staff are more likely to influence mothers with regard to the [best food] for their babies" 2/n

David Roodman writes: consumers of economic research are more truth-seeking than the producers.

Here is Roodman's recap of his re-analysis of a set of papers on temperature and judge decision-making.

The comment process is not working that well!

Gory details in 🧵 1/ Here is the Gelman blog post:
statmodeling.stat.columbia.edu/2025/09/02/who…

Here is @davidroodman's reanalysis:
econstor.eu/handle/10419/3… 2/

Interesting: Konhee Chang studies what happens in suburban neighborhoods in Florida when large-scale property owners move in and convert suburban homes to rentals.
1/ Entry of these firms substantially changes the ownership composition of neighborhoods.

2/

The most essential stabilizing feature of a society is a status hierarchy that young people—especially young men—can reliably climb by following a clear and pro-social set of rules.

1/ It is thus much better to have a society where the young have low status and the old high, than a society where one group has high status at all ages, and another is always low.

Everyone is willing to do their time. But if you don't see a path up, you will tear things down. 2/

Deep meritocracy is fundamentally unstable, because elites can't tolerate their kids losing. They will engage in destructive status competition with each other, and also just directly corrupt the institutions. 1/ Anyway, non-elites don't stand a chance in a pure performance-based merit system, because elite parental investments have such high returns. In practice, we don't get meritocracy anyway. 2/

This @TheZvi post on AI cheating is worthwhile and asks many of the right questions.

But he's wrong to conclude that AI means what you learn in college doesn't matter.

This really misses the point. College builds brainpower like going to the gym builds muscle. 🧵
You don't learn algebra so that you can solve polynomials in your future job.

You learn it to develop the capability to formalize problems and solve them in abstraction.

Understanding how and when problems can be formalized is really really really useful.

2/n

Every econ referee report should include two scores out of 100: (1) Accuracy; (2) Importance.

These scores should be public and should follow the papers.

This would have all kinds of benefits:

1. It would reduce the incentive to submit to all 5 top five journals

🧵 1/n If your paper has 4 60's already from the other top 5s, maybe you'll get it into JPE, but everyone will know it's a marginal JPE.

This is better than our current system, where people whisper "oh that paper is really weak, it just barely made it to JPE." 2/

Intelligence matters way more for age 0–25 than for age 25–50. Since this is a formative period, people often overrate intelligence.

But the really successful people at 40 aren't the ones who were brilliant, they're the ones who kept at it, who kept working, who kept learning. For example, at age 20, there isn't so much difference between people who are into and aren't into reading books.

If you're smart, you can ace everything, be at the top, even if you're not investing that much.

This was surprising — respondents to the AEA social media survey are still 10x more likely to be reading Twitter than reading BlueSky.

Rumors of Twitter's demise are greatly exaggerated?

Some more notes from the report in the thread 1/ I'm doing fine, but everyone else is having a bad experience!

>70% are positive or neutral on their own social media experience, but consistently <50% think social media is beneficial. 2/

A spicy datacolada post on the challenge of evaluating papers where the data are proprietary and experiments are difficult to replicate.

My main thought: re-analysis like this is vastly under-supplied.

Link and some thoughts in thread. 1/ Link:

The general absence of public dialogue like this around most published papers is a major flaw of how econ works.

This kind of discussion does happen in (closed) seminars and in the (secret) referee process, but it is of public interest. 2/datacolada.org/122

One last try at this, I’m not giving up!

We revere Shakespeare because (1) he was a master of the language; (2) his plays strike to the heart of human experience.

When we teach Shakespeare to young people, they’re only experiencing the first. At best! 🧵 Shakespeare’s contemporaries spoke the language and got the best of all worlds: masterful dialogue and emotionally resonant stories.

When young’uns read Shakespeare today, they have to do a LOT of work! Read the footnotes, re-read scenes, get help, use a dictionary, etc. 2/N

For high schoolers, and probably college students too, all English lit from pre-1970 should be translated into modern English.

If you were reading a German novel, you'd want a recent translation. Why the punishing unequal treatment of English language writers?

https://twitter.com/alz_zyd_/status/1874863049037033784

Of course if the original work holds up, like Hemingway, then no translation is needed.

If the original work is tedious and overwrought (hello, Dickens), significant abridgements are in order as well.

Fascinating paper on where 6000 global elites went to college. Billionaires, CEOs, heads of state, central bankers, etc.

In a word: Harvard.

Fully 10% of global elites went to Harvard. Elite US schools are over-represented (23% IvyPlus), but nobody comes close to Harvard.

🧵 The paper is Ricardo Salas-Diaz and Kevin Young. They collected elite biographies across a few important international domains.

Corporate elites: CEO and board
IO: World Bank, IMF, Basel Committee, etc
National Elites: Head of state & central banker
3rd sector: think tanks 2/N

Next paper: Agarwal, Gaule, Jiang

How can we find people who are likely to excel in STEM, while they are still in their teenage years?

1/N

This conference is streaming btw: nber.org/conferences/ec… Einstein was a high-school dropout and failed the entrance exam at ETH in Zurich.

Finding Young Einsteins might be challenging!

2/N

Some tweets from the NBER Economics of Talent conference.

Conference is focused on when/how we create pathways for very talented people to get to places where make big contributions to society.


1/? nber.org/conferences/ec… Laura Giuliano is presenting a paper on tracking with David Card — studies Florida's advanced math program, which allows kids in the top 20% to take accelerated math (geometry in 8th grade).

2/?