Steffen

Posts Written Before Jumping out of an 8 Story Window Likes/Reposts ≠ endorsement

[Steffen @stesser1]

Gelesene Bücher 2025 / books read in 2025:

[Peter Hull @instrumenthull]

Causal inference is my passion

[Steffen @stesser1]

@Bumdeskanzler2 @keinName144 @fasanenkahn Danke! Fair, valider Punkt!

[Steffen @stesser1]

@Bumdeskanzler2 @keinName144 @fasanenkahn No front, aber was ist dein Punkt? Wäre cool, wenn du das ausführen könntest. Die DB AG (=UPE des DB-Konzerns) wird zu 100% vom Bund gehalten. Störst du dich an der Organisationsform (AG)?

[Dissent Magazine @DissentMag]

New @KnowYrEnemyPod: Matt and Sam talk to @Jasper_Craven about his new book, "God Forgives, Brothers Don’t: The Long March of Military Education and the Making of American Manhood."

https://t.co/3yZHX1aRy5

[Ihtesham Ali @ihtesham2005]

A Stanford economist paid 2,743 Facebook users to delete the app for a month and the people who took the money came back happier, less anxious, less angry at the other political party, and quietly valuing Facebook 14% less than they did before they left.

The lead names on the paper are Hunt Allcott and Matthew Gentzkow.

Gentzkow had already won the John Bates Clark Medal, which is given to the most important American economist under 40 and is considered a stronger predictor of future Nobel Prizes than almost any other award in the field.

He spent years studying media, persuasion, and political polarization. In 2018, he and his coauthors set out to do something nobody at this scale had ever done. Run a real randomized controlled trial on Facebook itself.

Most research on social media is correlational. It looks at people who use Facebook a lot and people who use it less, and tries to figure out which way the arrow points. The problem is obvious.

Unhappy people might use Facebook more. Lonely people might use Facebook more. You cannot pull cause and effect apart by watching from the outside.

So they ran the experiment the right way.

They recruited 2,743 active Facebook users through Facebook ads. They asked each person the smallest amount of money they would accept to deactivate their account for four weeks.

The four weeks were timed to end just after the November 2018 US midterm election, which mattered because it was the most politically charged month of the year. Then a computer randomly assigned the 1,661 people who valued Facebook at less than $102 to one of two groups. The Treatment group was paid $102 to deactivate. The Control group was paid nothing and kept using Facebook normally.

The team verified compliance by pinging each participant's public profile URL multiple times a day for the entire four weeks. Compliance in the Treatment group exceeded 90%. Less than two percent of the sample failed to complete the endline survey.

This is one of the cleanest natural experiments ever run on a tech platform.

Then they measured everything.

Time use first. The people who quit Facebook freed up 60 minutes a day on average. They did not spend that time on other social media. They spent it on television alone, and they spent it socializing with family and friends in person. Both went up. The narrative that Facebook substitutes for real human contact was not a vibe. It showed up in the time diaries.

News next. The deactivated group knew less about current events. Their score on a 15-question news quiz dropped by 0.19 standard deviations. They were less likely to follow politics closely and less able to identify what was actually in the headlines. Facebook is, for a meaningful slice of American adults, the primary place they encounter news. Take it away and the news disappears with it.

Then the part that started to get uncomfortable for Silicon Valley.

The deactivated group became less politically polarized. Their views on policy issues drifted closer to the middle. Their exposure to the kind of news content that makes people angrier at the other side dropped. The overall polarization index fell by 0.16 standard deviations in four weeks.

For context, a separate index of American political polarization rose by 0.38 standard deviations over the 22 years between 1996 and 2018. The experiment moved people in four weeks by a meaningful fraction of what an entire generation of political drift had done.

Then the part the researchers were most careful about.

Subjective well-being improved. The deactivated group reported higher happiness, higher life satisfaction, lower depression, and lower anxiety. The overall index moved by 0.09 standard deviations. That sounds small until you compare it to the benchmark.

It is roughly 25 to 40% of the effect size of professional interventions like self-help therapy, group training, and individual psychotherapy. Four weeks of not opening one app produced a fraction of the mental health benefit that a course of therapy produces. That is not a fraction to dismiss. That is a fraction to look at twice.

And then the finding that closes the case.

The experiment was supposed to end. The Treatment group was free to reactivate their accounts the moment the four weeks were over. The researchers kept watching. 9 weeks after the experiment ended, 5% of the people who had been paid to leave had still not come back.

The group as a whole was using the Facebook mobile app 22% less than the control group, an average of 11 minutes a day less. They had not just survived four weeks without Facebook. They had quietly decided they did not need as much of it as they had before.

The economists ran one more test. Before the experiment, they had asked each participant the smallest amount of money they would now need to deactivate again for four weeks.

After the experiment ended, they asked the Treatment group the same question. The valuation had dropped by 14%. People who had been paid to leave Facebook came back valuing Facebook less than they did before they left.

That is the sentence that ends the argument. If Facebook were worth what users said it was worth, the price would not move after a four-week break. The fact that it moves means a meaningful slice of what users were paying for, in time and attention, was the dependence itself. They were not buying connection. They were buying the habit of checking the connection. And once the habit broke, the price they were willing to pay for it dropped.

The standard calculation, the one tech companies love to quote, takes the average user's stated valuation and multiplies it by the number of users.

By that math, Facebook generates roughly $31 billion in consumer welfare every four weeks for American adults alone.

The Stanford team's data suggests that number is overstated, because part of what users say Facebook is worth is the cost of the dependence Facebook has already built in them.

The paper was published in March 2020 in the American Economic Review, the most cited economics journal in the world. It became one of the most discussed social media studies of the decade.

Inside the company, employees read it. Inside policy circles, regulators cite it. Inside academic departments, it is the reference point for every randomized trial on a tech platform that has followed.

80% of the people who were paid to leave Facebook said the four weeks were good for them.

The platform is still running. The trial is still on the public record. The single cleanest test ever run on what social media costs a human life found that the people who stopped paying that cost were happier, calmer, more connected to the people in front of them, and worth less to the platform afterward.

The price of Facebook was never the time. It was the version of you that started to need it.

And the study proved that version of you was never the one you would have chosen if anyone had asked.

[Dissent Magazine @DissentMag]

“Who Paid the Pipers of Western Marxism?” builds a faulty model of the world, interprets evidence poorly in support of that model, and does so in support of a delusional political program.

https://t.co/5GHAr1U9h5

[Jesús Fernández-Villaverde @JesusFerna7026]

Since I have posted so much on Marx vs. Weber, modernity, and development over the last few weeks, I have posted an updated slide deck of my lectures on Karl Marx and the Marxian Tradition (together with @ferarteaga) here:

https://t.co/TOGm7jXMKG

This is a long deck: 437 slides in the last compilation! (It also takes a few seconds to upload.) If I were to teach it carefully, with plenty of class discussion, I would require a whole semester. Even then, some topics (e.g., the Frankfurt School) receive only a cursory treatment because I focus more on economics and political economy, broadly construed. I hope to extend the discussion of those someday.

However, I cover topics rarely seen in these courses, such as Hans-Georg Backhaus and the Neue Marx-Lektüre, because most of the work is not translated into English and must be read in the original German.

I don’t have an equivalent slide deck on Max Weber, as I haven’t lectured on him. Hopefully, one day I will.

Comments and feedback are very welcome.

[Steffen @stesser1]

👇This!

[Jan Vogler @Jan_Vogler]

🚨 Coming soon: “The Political Economy of Public Bureaucracy” 🚨

My book is forthcoming in 2026 with @CambridgeUP! A preview of the cover art is below. Thanks to Dan Carpenter, @SeanGailmard, & @jrgingrich for their endorsements! 🙏

More details will follow in future posts. 😊

[Steffen @stesser1]

„Tariff increases are contractionary: imports fall sharply, exports decline with a lag, and output and manufacturing activity drop persistently.“

[NBER @nberpubs]

Using 180 years of US data to show that tariff increases reduce imports, output, and manufacturing, with effects operating through both supply and demand channels, from Tamar den Besten, Regis Barnichon, @drkaenzig, and Aayush Singh https://t.co/R1kdtfi3fO

[Benjamin R. Young @DubstepInDPRK]

Coming to (academic) bookstores near you this July. Published with @CornellPress

[Steffen @stesser1]

EU-maxxing

[The Rundown AI @TheRundownAI]

Sony AI just published the first autonomous robot to beat elite humans at a competitive physical sport.

Its name is Ace. The sport is table tennis. The paper dropped in Nature today.

9 cameras triangulate the ball in 3D. Three systems zoom in on the ball's logo mid-flight to read its spin axis.

End-to-end reaction time is around 20 milliseconds, roughly 10x faster than a human.

It learned by playing itself for 3,000 hours in simulation. No human demonstrations.

In April 2025, Ace beat 3 of 5 elite players and lost to both pros. By December, it had beaten a pro.

At one point, Ace returned an impossible backspin shot. Kinjiro Nakamura, a 1992 Barcelona Olympian, watched from the side:

"No one else would have been able to do that. I didn't think it was possible. But the fact that it was possible means there is a possibility that a human could do it too."

[Mark Koyama @MarkKoyama]

For the PPE senior seminar at GMU, I've been teaching a lot of 20th century economic history. I will upload all of the slides at the end of the semester. But here is a taster of my WW2 slides:

[Mattheus Berg @MattheusBerg]

Seid ihr Team SPRENGEN? Team LIEGEN LASSEN? Oder Team RETTEN?

Ich habe den https://t.co/ybq4yaxhu9 gebaut, damit ihr es herausfinden könnt.

[Steffen @stesser1]

Nature is healing

[Nayib Bukele @nayibbukele]

What must be done leaves no room for hesitation.

[Steffen @stesser1]

Demnächst - neues Buch von Freerk Huisken im Mai 2026:

https://t.co/vXcFdE8tKy

[nachmittag schwimmschule @kritik_d_maeuse]

cool dass sich lilian thuram mit sein popularität & standing für die erinnerung an addi bâ einsetzt

jetzt gibt es auch deutsche ��bersetzung eines französischen romans über ihn

[Michael Millerman @millerman]

Sorry, but I had to...

https://t.co/45MjP0HdmP

[Simpsons Quotes @Simpsons_tweets]

IMDB season by season episode ratings.

[Steffen @stesser1]

Ausgegooned

[The Wall Street Journal @WSJ]

Leo Radvinsky, the reclusive billionaire who reshaped the pornography industry by turning subscription service OnlyFans into an adult-content powerhouse, has died “after a long battle with cancer,” the company said. He was 43. https://t.co/asll3LVIou