Anton Strezhnev

Gotta make sure it’s still got those SQL injections for true authenticity

a classic forum interface for bluesky

a classic forum interface for hacker news

a classic forum interface for the new york times

coding assistance lets you fast-track important projects that improve your life, such as reformatting every single site you read into the old vbulletin 3.x default template

Very curious to know what the spongebob SKILL.md does…

I don’t think the current bottleneck to good research is our ability to generate code and text.

Also, conditional on hiring an RA, that RA is probably doing coauthor level work - they should be a coauthor.

Perhaps we should change how we teach the undergrads, though I suspect there will be a lot of resistance - especially from the undergrads themselves!

The use of the two-way fixed effects regression in empirical social science was historically motivated by folk wisdom that it uncovers the Average Treatment effect on the Treated (ATT) as in the canonical two-period two-group case. This belief has come under scrutiny recently due to recent results in applied econometrics showing that it fails to uncover meaningful averages of heterogeneous treatment effects in the presence of effect heterogeneity over time and across adoption cohorts, and several heterogeneity-robust alternatives have been proposed. However, these estimators often have higher variance and are therefore under-powered for many applications, which poses a bias-variance tradeoff that is challenging for researchers to navigate. In this paper, we propose simple tests of linear restrictions that can be used to test for differences in dynamic treatment effects over cohorts, which allows us to test for when the two-way fixed effects regression is likely to yield biased estimates of the ATT. These tests are implemented as methods in the pyfixest python library.

arXiv📈🤖
When can we get away with using the two-way fixed effects regression?
By Lal

The deadly chemical in Michael Clayton is Culcitate and the evil metal in Andor is Kalkite, suggesting a connected Gilroyverse across which pernicious minerals share a linguistic herita—
*a taut cable snaps inside my skull—BANG!— foamy brain matter oozes out of my ears and nose*

My man, that’s not FDIC insured!!

Broadcast TV also very interesting

AI is going to let academics write so many more papers (derogatory)

Literally the Armstrong and Miller RAF pilots bit

Minnesota Nice The niceness of my adopted home state during oppression and beyond

sarahgardner.substack.com/p/minnesota-...

It’s weird to see the latent projection of someone you know into the space of “arXiV paper”

I’m not sure it’s actually consequential, but it’s annoying.

Oh sure, bigger issues - but like “you adjust for all the stuff they claim to and get nothing” is something that’s been bugging me about this debate.

Also - control for variables by putting them in the regression! bsky.app/profile/astr...

Both Y and D residuals get you FWL and you can also just residualize D and get the same coefficient. But Y residualized and not D doesn’t recover the same thing as the OLS coefficient w/ the Y, D and confounders.

Identification issues aside - I feel like something might just be messed up in the ST analysis from regressing WAR (a residual) on moderation (or any “causal” variable we’re interested in) without also residualizing the causal variable on those same things.

Causal Interpretation of Cross-Experiment Surrogate Regressions (PDF) We study the common practice in technology experimentation of regressing estimated long-term treatment-effect lifts on estimated short-term lifts across many experiments, and using the fitted coeffici

Discovered my favorite new euphemism for a hallucinated reference “classic viewpoint; citation omitted” - www.clawxiv.org/pdf/clawxiv....

Oh to hear the things biostatisticians say about doctors…

Jerzy Neyman: A Positive Role Model in the History of Frequentist Statistics Many of the facts in this blog post come from the biography ‘Neyman’ by Constance Reid . I highly recommend reading this book if you find ...

Always worth going back to this @lakens.bsky.social piece on an influential statistician who *wasn’t* personally a piece of shit. daniellakens.blogspot.com/2021/09/jerz...

@professormusgrave.bsky.social alert!

The Star Trek model is really helpful here in that “the Computer” often helps move the plot forward but it’s never the architect of the narrative and is not infrequently the foil

I do think a feed that’s \alpha, (1-\alpha) mix of main feed and quite posters could replicate the old model reasonably well

“Quiet Posters” feed is important IMO - academics just can’t match the output of other posters here and that’s the downside to a chronological feed. The magic of old Twitter was that the algorithm could mix in relevant posts from less active users.

My Minneapolis An ode to a city that knows who it is

Minneapolis is a place that means a lot to me, so I wrote a short Substack about the city, its people, and what we all owe them.

benansell.substack.com/p/my-minneap...

Population Estimates Thread:

1/ We got new Census Bureau population estimates today, and they confirm what we already suspected... population growth slowed dramatically in 2025.

Census estimates the population grew by 1.38M in 2025, vs. a projection of 1.90M a year ago.

Chocolate Giddy-Up from Black Dynamite

We already have a line city and it’s called Madison, WI