Arthur Katossky

Skills are among the most consequential new tools for AI, and Anthropic just released a very impressive nontechnical Cowork Skill that builds Skills, including doing interviews & providing benchmarks through parallel tests

I think you still need to add the human touch but this is a big leap forward

@coulmont.bsky.social I'm sure you wouldn't able to do this in R, would you?

I raise

You can use [[ instead of $ to avoid partial matching.

@monsieurphi.bsky.social

Crosstabs can be charted in this way with less information loss by making the horizontal bars thick according to the row sums.

Of course, there may be a trade-off between minimizing information loss in this way and maximizing audience engagement.

(Don't have the data to redo the whole thing).

Ever heard of Marimo?

Credit Where Credit is Due: Mary Eleanor Spear Exploring the life and work of an overlooked mid-century dataviz pioneer who may have discovered the Box Plot 17 years before John Tukey.

#TodayinHistory #dataviz #Onthisday #OTD 📊
🎂Mar 4, 1897 Mary Eleanor Spear born in Jonesboro, Indiana, USA 🇺🇸

A data vis specialist at the Bureau of Labor Statistics, she pioneered the boxplot and wrote books on effective graphic techniques (done by hand!)
👀 bit.ly/4baeFSY

Parfois, les recrutements à l'université se font sur plusieurs disciplines (par ex. histoire contemporaine et anthropologie, ou maths appli et informatique) : l'observation de ces croisements sur 12 ans donne cet espace de l'interdisciplinarité.

Lots of headlines use phrases like 'realistic possibility', but how do people interpret this?

New post, with insights from over 150,000 human judgements about probability-based phrases: kucharski.substack.com/p/insights-f...

Introducing CAPphrase (Comparative and Absolute Probability phrase dataset), an open access dataset containing over 150,000 probability-based language judgements: adamkucharski.github.io/CAPphrase/

This recent RCT of an "AI stethoscope" claims the technology "shows promise" for diagnosing cardiovascular conditions.

It does not.

It is a textbook example of the risks of conducting unprincipled 'per protocol analyses'. Once again, peer review at a major medical journal has failed.

🧵 1/

High impact political intervention right now:
Lobby states to build just the payment infrastructure to send 1 €/$/₱ to all residents. Use any type of program as an excuse.
The need for UBI will be obvious soon enough, but setting up verification and fraud-prevention will be a serious bottleneck.

"Errors in your code mean you're doing something cool!" 😊

(spotted in a tutorial)

Would anybody like to read one of my long threads about a 2.5bn year old rock, bacteria that could produce oxygen but not consume it, and tank production in wwii

I think that’s like 80% of the hate. 10% is the truncated printing. 5% is lack of row names.

Top panel: My plot of US adult female and male height distributions, which are much narrower, and overlap less than in the figure from Fuentes, which is in the bottom panel.

1. After I posted my critical review of @anthrofuentes.bsky.social Sex is a Spectrum, a colleague pointed out that his figure of adult heights by sex (bottom panel👇) can't be right: there aren't that many US adults shorter than 4' or taller than 7'

Turns out Fuentes' data are made up 🧪 #BioAnth 🧵

GitHub - paulgp/claude-container Contribute to paulgp/claude-container development by creating an account on GitHub.

In case folks are looking for a convenient way to containerize their claude code instances, here's a way to use Docker on your mac machine to isolate YOLO-ing claude instances.

github.com/paulgp/claud...

One element drawn = one line in the dataset. Easier to reason about when you want to see, say, the equivalence between a line graph and a scatterplot.

Tribune dans la Gazette du Palais : « OQTF : Mensonges et préjugés » Contenu de l’article trop long, accessible ici => https://bit.ly/4tICSIg

Tribune : « OQTF : Mensonges et préjugés » (in Gazette du Palais)

Face au déferlement d'inepties sur les OQTF, quelques rappels (basiques) s'imposent.

En particulier, parler d'« *un* OQTF » est erroné & déshumanisant, car réduit une personne à une simple mesure administrative.

=> bit.ly/4tICSIg

In the large: Mortality in France, 1816-2016.

Nice! Can you have plots in there? 😇

myRegressionModel_2.old

_La fin du bipartisme_ ou _Un diagramme en barre qui aurait pu être em un diagramme en ligne_ 😅

The motherhood penalty isn’t as large as we think And event-study designs aren’t the be-all end-all to empirical economics

On why the motherhood penalty may be overestimated. Simply put: The identifying variation comes from women having kids at different points in time, which might not be a valid counterfactual. open.substack.com/pub/plausibl...

Great graphic showing cities on a triangle graph showing if they are most oriented to public transit, active mobility or cars.

Have you seen this? This recent study of mode share (the % share of transportation trips that are by car, transit, walking, biking etc) relative to city size and income levels in almost 800 cities in 61 countries. Interesting results. HT @davidzipper.bsky.social
www.sciencedirect.com/science/arti...

🧪🏺 Update - authors have new paper showing how useless gen- #AI is for archaeological illustration.
All 400 images were multiply inaccurate (physically, socially, technologically, environmentally), even with improved prompts.

JUST USE HUMAN EXPERTS & ARTISTS

www.cambridge.org/core/journal...

LCP @LCP "En République il n'y a pas d'indésirables", martèle @NunezLaurent . "Qui allez-vous convaincre de ces contrôles au faciès qui n'existent pas ? Une amende forfaitaire délictuelle est toujours adossée à une infraction". #QAG #DirectAN [Extrait vidéo de Nunes à l'Assemblée nationale] 2:32 PM · 11 févr. 2026 · 39,4 k vues

Pour le ministre de l'intérieur, les « contrôles au faciès n'existent pas »... (🙃)

Pourquoi diable, alors :

- La @courdecassation.fr a condamné l'Etat pour contrôles aux faciès;

- En 2025, la CEDH a condamné la France pour ce motif;

- En 2025, le Défenseur des droits a souligné leur persistance.

Quand la police fait le tri à l’Assemblée nationale - Projet Arcadie Quand la police outrepasse ses droits à l’Assemblée nationale, en bloquant les invités d’un député. Récit d’un bras de fer surréaliste et démesuré.

Hier, le député Abdelkader Lahmar avait invité des militants du DAL à l'AN pour une conférence de presse.
Les CRS, sur ordre de la préfecture, ont refusé de les laisser passer, au point qu'il a fallu solliciter l'intervention de la première questeure.

Il y a plus de 28 000 policiers municipaux en France. C'est +45% en 10 ans.
Cette semaine, le Sénat étudiait un projet de loi pour renforcer les pouvoirs des polices municipales, y compris en matière judiciaire.
Explications pour @dallozactu.bsky.social www.dalloz-actualite.fr/flash/police...