Paul Hünermund

The program for this year's Causal Data Science Meeting #CDSM26 is out! 38 talks over two days, Nov 4–5: double machine learning, diff-in-diff, causal discovery, LLM agents, and applied causal work from academia and industry.

This isn't the first post I've written, but it's the first I'm making public. Short and sweet, I hope. It is about voting after not being able to vote for a while. descartos.substack.com/p/i-voted?r=...

As a proper European conference should be

Congrats to Harry Kane, well deserved! #BallondOr

Comments are very positive 😂

Antitrust law has, unfortunately, failed America. And they're trying to come for Europe next.

The Writers Guild on Musk blocking releasee of the documentary "Musk": "A small handful of billionaires now own virtually every major media company, news outlet, and social media platform and are demonstrating how this unchecked concentration over our news and media threatens democratic discourse."

We have to use this freedom responsibly and have to remain accountable. But pretty much, yes.

Charging for things is so so so so underrated

How your emails find me

Understanding Maximum Likelihood Estimation, now in 3D!: rpsychologist.com/likelihood #dataviz #stats #rstats

I'm doing a project in which I'm checking peer review reports for certain papers and this GIF is coming to mind a lot. Very anecdotally, it seems more famous researchers get a much easier ride than lesser-known names

Baby Made it until 11:15 today before her first nap 😣

Die lokale D-Jugend-Mannschaft sponsern und gleichzeitig Oma das Turbozertifikat verkaufen youtu.be/EJyb6g1kBlQ?...

if you decide not to do college, you have to bootstrap - a core professional knowledge base - proof thereof - a professional network - a sense of yourself as "good at" your field - habits for delivering work regularly college just hands you all that at the tail end of peak lifetime neuroplasticity

LLMs, asked to "peer" review 200 papers published this year in top science and medicine journals, either rejected or recommended major revisions 76% of the time. Telling the LLMs the paper was a revision paper made them much more generous. arxiv.org/abs/2610.31414

We acknowledge that no one has the bandwidth for this, but if it helps: -It’s ‘sleight of hand’ -It’s ‘bated breath’ -It’s ‘lo and behold’ -It's 'buck naked' -It’s ‘brass tacks’ -It’s ‘free rein’ -It’s ‘toe the line'

Excited by the "distinctively British" approach to AI, presumably we'll be nicking the bots from some other part of the world and allocating them by social class

The PHP Foundation, which coordinated fixes for two critical PHP flaws, operates on a shoestring budget with one full-time staff member and 14 contractors (Toby Sterling/Reuters) Main Link | Techmeme Permalink

Suppose in a couple centuries the genre of Fantasy isn't really around any more, and people try to describe it the way we now describe old genres that were popular for a long time, like picaresque. What would be the couple-sentence description?

Parenting tip: Don't do biceps curls if there's a chance that you have to rock your baby to sleep at night

😄

Labor Markets and Social Insurance @zew.de is looking for a postdoc in social insurance. Please share widely! jobs.zew.de/jobposting/7...

We have a new paper out. Knightian uncertainty ("unknown uknowns") may be an important boundary condition for ML in strategy, and we need more research. But progress requires conceptual clarity about what we mean by KU and how different ML tools might address it. doi.org/10.1177/2755...

here we go again

In my intercultural household we have fierce debates about the health effects of Durchzug

lmfao

A serious weakness in this paper by King and Worthington: The judges for the AI peer reviews are *not* experts in the field. They are former journal editors but no claim is made to substantive expertise in the reviewed papers. AI-generated reviews sound good *if you don't know the substance well.*

Remains poorly understood that humans hallucinate more than frontier AIs at this point bsky.app/profile/matt...

Interesting. Of course, humans judge sample was only N=3, and they only read the reviews, not the paper itself. But I have the suspicious that the pattern will hold even with larger tests. The problem is how AI would be able to judge a truly novel idea, though the same is true for humans.

Wäre kommenden Sonntag Bundestagswahl, würden CDU/CSU laut Umfragen rund ein Drittel ihrer Stimmen verlieren. Doch welche Wähler wenden sich ab und kann die Union sie zurückgewinnen? Ein paar Zahlen aus Paneldaten seit Februar 2025. Ein Thread 🧵 (und hier als Text) juliuskoelzer.de/union-verlus...

Gary King just dropped a study showing that AI peer reviews now dominate human peer reviews on every metric they tested. gking.harvard.edu/peerless/

Ich wusste gar nicht, dass die Biene Maja Zeichentrickserie in Japan produziert wurde

Tough luck if you're a one trick (DiD) pony

If a paper is written by multiple authors, and one is more famous than the rest, it is so insulting to act as if all the ideas in the paper come from the famous researcher. It is an important tell if you see a scientist do this.

Like all econ reporters, I've spent the past couple years trying to understand what A.I. will mean for the economy and workers. It's been hard to reach any firm conclusions, because so much is uncertain: how technology will evolve, how quickly it will be adopted, how policymakers will respond.

Lie-kurt scale

📢 We're very excited that our own Christopher Klamm (University of Mannheim; LMU Munich) will present: “Blind Spots: What AI Benchmarks Don't Tell Us About Social Science Tasks” at the next session of the TaDa Fall Speaker Series Sign up for our newsletter🗞️at tada.cool for the Zoom link!

One Tonie Box to rule them all

The review requests I'm getting from Springer are dubious

#EconSky: Going on the econ job market this year or know someone who is? Come to the free virtual job market workshop by Economists for Ukraine, featuring UC Berkeley's placement director Ben Handel and recent PhD graduate Yevhenii Usenko. Oct. 30, 11am EST econ4ua.org/events/job-m...

Musk’s wealth is a threat to democracy—not just in the U.S. but everywhere. He’s probably the greatest threat to the free world today.