Back at the start of the Trump II administration, it was disorienting to watch powerful law firms preemptively surrender and genuflect before the King Turns out it was because the people in charge were close friends with Epstein and were worried about getting outed www.ft.com/content/7001...

This is not a new phenomenon for shipping companies, it turns out. www.hbs.edu/ris/Publicat...

Made a workout website

I must visit this museum

The Wright Museums Instagram page is HILARIOUS. Think The Office but it's the staff at a Museum for Black History. Better believe this is on my list of places to visit. Gentle Spades "I'm a veteran"

COONS: Joe, it's same rules NYPD follows. Do you see roving bands of NYPD w/ masks using bear spray on people? KERNEN: You lost the election! That doesn't give you right to shut down govt C: I have to fight for the core value of our democracy, including law enforcement that doesn't murder citizens

Heads-up, fellow New Yorkers! Big announcement here from @zachlowenba.bsky.social. Tix go on sale Monday at this link: www.theringer.com/events

Right, so last night I did this. Made a MSA-by-uyear database of 1) hpi 2) building permits 3) housing stock. Let it rip and it was done in about 21 minutes. See below if you want to see how it ran through it. paulgp.com/claude-code-... Repo is here: github.com/paulgp/housi...

I think the oracle view makes sense for app dev where the bottleneck is “does it fail loudly” vs in social science where the bottleneck is “can I explain this to a seminar audience if they ask how I did XYZ.” the moment I say “I have no idea but Claude said it made no mistake” my career is over.

George Stigler and Gary Becker, AER (1977): "De Gustibus Non Est Disputandum" www.jstor.org/stable/1807222

This is a great chart by the @economist.com summing up Kevin Warsh using LLMs and over 200 speeches/op-eds. I did a similar exercise showing same result (Warsh v FOMC) but with less data www.economist.com/finance-and-...

"Anecdotal evidence" is high IQ for "vibes"

New mission -- let's take our marginal paper ideas and make them blog posts instead thanks to AI!

Capable but error prone research assistant vs oracle sums up the bsky vs X views of Claude Code for social science divergence really well

LLMs/engineers go parser = argparse.ArgumentParser() parser.add_argument("--bread1", required = True) p.a_a("--spam") p.a_a("--onion") p.a_a("--cheese") p.a_a("--tomato") p.a_a("--bread2") burger.py --bread1 --spam --onion --cheese --tomato --bread2 vs make_a_sandwich() with sensible defaults

ski jump penis doping story only underscores the advantages of shape and drape over slim fit clothing www.theguardian.com/sport/2026/f...

If you use claude code, check out the /insights command Some cool stuff there.

Today, Anthropic released Claude Opus 4.6 with a #Finance plugin. I tested it on capitalizing VW's R&D expenses. In one vague prompt I got a full spreadsheet with amortization schedules and ROIC calculations.

Two development randomistas develop a secret love affair in the field Treated Rivalry

If you’re not using Claude Code, now’s the time to jump in!

This just seems like GasTown but officially

Anyone else tried Agent Teams? It is insane how fast it burns through tokens. I just did 400k in 3 min

Two game theorists cannot agree on the refinement concept for equilibrium selection but they will always keep playing. Repeated rivalry

The two best meteorologists on TV secretly pine for each other. Sleeted Rivalry

Wow. For the WFAs "2,213 papers were submitted, which is a record for an in-person conference, and we were only able to accommodate 156 papers on the program for an acceptance rate of 7%." That's a lot of papers!

This looks super cool

1000%. I had no idea how busy i would be

The Bulls are testing the limits of how many French people can fit onto a 15-man NBA roster

We have a new paper in JAMA Internal Medicine! Patient race is widely used in medical algorithms...but it's unclear how patients feel about this. We conduct the first nationally-representative YouGov survey to find out, producing four findings with practical clinical implications. 1/

My manuscript "The Economics of Noncompete Clauses" is now out in the Journal of Economic Perspectives. Few topics have garnered more debate and policy attention over the last few years. This article brings us up to speed on the debate and the current answers. pubs.aeaweb.org/doi/pdfplus/...

This took several years and I'm so happy it is finally out. We just released an open-source dataset of nearly 2M African speech records for speech recognition and vocalization (27 languages). Blog: blog.google/intl/en-afri... Paper: arxiv.org/abs/2602.02734 Dataset: huggingface.co/datasets/goo...

Enjoyed this track. Some real rock ballad vibes www.youtube.com/watch?v=oLzv...

Really nuts how many men were like "Me Too went too far, and I need to email Jeffrey Epstein about it."

The CIA just stopped publishing their World Factbook and took every page, including the archived copies of previous versions! This sucks. It was public domain, so I recovered the 2020 edition (the last one published as a zip file) and shared it to GitHub simonwillison.net/2026/Feb/5/t...

Website dashboard updated to incorporate finance journals and the Journal of Econometrics: paulgp.com/econlit-pipe...

dplyr 1.2.0 is out now and we are SO excited! - `filter_out()` for dropping rows - `recode_values()`, `replace_values()`, and `replace_when()` that join `case_when()` as a complete family of recoding/replacing tools These are huge quality of life wins for #rstats! tidyverse.org/blog/2026/02...

This part of our response essay is where I'm at. Boiling down all of politics to electoralism has been absolutely terrible for resisting rising authoritarianism. www.bostonreview.net/forum/how-no...

Looking for some more Lea testers, send me a DM!

this @kjhealy.co banger came up in Facebook memories and it deserves to live here too

I just came here to say that I'm 10 years out from my PhD and I'm "tenured" and the imposter syndrome is still strong. Maybe not for everything but for a number of aspects of my job.

Most unusual SI/methods write-up I've seen in a while: Digging in Utah scrapbooks for hair samples to document historic lead levels. Remarkable findings, note the log-scaled y-axis.

An unfortunate thing is that if you do a really bad job teaching, students will frequently assume that you’re clever and they didn’t understand it. Same thing for poorly written papers

People have been using birds to send messages for centuries. Given what, say, salmons and eels are able to do, I wonder why nobody has thought of using fish... Is this because they are unreliable as messengers? That seems like an easy problem to solve! Just use error-correcting cod

NSF GRFP applicants (and mentors): Was your application Returned Without Review and deemed ineligible despite fitting in the allowed topics? 1) Write NSF 2) Write your Congressperson 3) CC us at [email protected] so we can compile + follow up Details and template at grant-witness.us/grfp-letter

Lead (Pb) in archived hair in Salt Lake City. "Lead (Pb) in archived hair documents a decline in lead exposure to humans since the establishment of the US Environmental Protection Agency " www.pnas.org/doi/10.1073/...

Uhhh

Sounds pretty close to tearing down to me