Antoine Vernet
LightNews LightNews
antoinevernet.com
Antoine Vernet
@antoinevernet.com
850 followers 950 following 3.6K posts
Associate Professor in Management at UCL.
Networks and Organizational Design.
Mostly found reading.
Very occasionally on Youtube: www.youtube.com/@antoinevernet
Posts Replies Media Videos
Reposted by Antoine Vernet
alexselbyb.bsky.social
I failed to finish my French cheek-kissing map in time for Valentine's Day so I'm having a sale on my equally romantic cholera deaths map instead. Now 15% off!

www.etsy.com/uk/listing/4...
3D French cheek-kissing map—coming soon! 3D map of cholera deaths in Soho in 1854
February 14, 2026 at 2:03 PM
antoinevernet.com
You'll never guess where this is going!
andreamatranga.bsky.social Andrea Matranga @andreamatranga.bsky.social · 19h
When the first dinosaur bones were found, artists didn't really know how to render them realistically, so they do the artistic equivalent of stretching skin around them as tightly as possible, without considering where muscle and fat must have been.
February 14, 2026 at 2:46 PM
Reposted by Antoine Vernet
andreamatranga.bsky.social
When the first dinosaur bones were found, artists didn't really know how to render them realistically, so they do the artistic equivalent of stretching skin around them as tightly as possible, without considering where muscle and fat must have been.
February 14, 2026 at 1:56 PM
antoinevernet.com
The 4 biggest parties by vote intentions are 2 nazi-adjacents parties, one "we can be racist too" party, and a party led by someone who used to defraud women by promising to grow their boobs with hypnosis. No wonder British politics feels depressing
jack-bailey.co.uk Jack Bailey @jack-bailey.co.uk · 22h
UK Poll of Polls, 14 February 2026

Reform: 29% (27-31%)
Labour: 20% (18-22%)
Conservative: 19% (17-20%)
Greens: 14% (13-16%)
Lib Dem: 12% (11-14%)
Other: 3% (3-4%)
SNP: 3% (2-3%)
February 14, 2026 at 2:41 PM
Reposted by Antoine Vernet
stephenturner.us
Introduction to building (better) R packages www.youtube.com/watch?v=Y8Kr... (R/Pharma 2025 #Rstats)
Introduction to building (better) R packages
Introduction to building (better) R packages - Nicola Rennie Abstract: There are many benefits to turning your R scripts or functions into a package, like making your code easier to re-use, easier…
www.youtube.com
February 13, 2026 at 9:01 PM
Reposted by Antoine Vernet
alaburda.bsky.social
I unashamedly took a swing with Claude:
gist.github.com/Alaburda/8e3...
gist.github.com/Alaburda/115...

I think you might have meant something like this though?
gist.github.com/Alaburda/6a5...
annotating-gt.R
GitHub Gist: instantly share code, notes, and snippets.
gist.github.com
February 13, 2026 at 6:53 PM
Reposted by Antoine Vernet
simonwillison.net
I had some fun pulling OpenAI's mission statement out of their IRS tax filings from 2016 to 2024, loading them into a git repo with fake commit dates and then taking a look at the diffs simonwillison.net/2026/Feb/13/...
The evolution of OpenAI’s mission statement
As a USA 501(c)(3) the OpenAI non-profit has to file a tax return each year with the IRS. One of the required fields on that tax return is to “Briefly …
simonwillison.net
February 13, 2026 at 11:40 PM
antoinevernet.com
Labour always finds new ways to be worse 👇

www.theguardian.com/politics/202...
Dual nationals to be denied entry to UK from 25 February unless they have British passport
New border controls require ‘certificate of entitlement’ to attach to second nationality passport that costs £589
www.theguardian.com
February 13, 2026 at 4:56 PM
Reposted by Antoine Vernet
vincentab.bsky.social
You can now use LLMs to learn about any chapter of my 𝑀𝑜𝑑𝑒𝑙 𝑡𝑜 𝑀𝑒𝑎𝑛𝑖𝑛𝑔 book, or any function in the `marginaleffects` package for R or Python.

Check out the cool demo video below (sound on).

Install here: github.com/vincentarelb...

#marginaleffects #RStats #pydata
February 13, 2026 at 3:52 PM
antoinevernet.com
The fascists are despicable, but they are also giant toddlers, like their boss
paleofuture.bsky.social Matt Novak @paleofuture.bsky.social · 2d
The details of the Kristi Noem blanket incident are just fucking perfect
In the blanket incident, Noem had to switch planes after a maintenance issue was discovered, but her blanket wasn't moved to the second plane, according to the people familiar with the incident. The Coast Guard pilot was initially fired and told to take a commercial flight home when they reached their destination. They eventually reinstated the pilot because no one else was available to fly them home.
February 13, 2026 at 4:43 PM
Reposted by Antoine Vernet
rmkubinec.bsky.social
One way academic research practice should change with the advent of claude code:

Every academic with stats code (R/Python/Stata) should make Claude Code run "unit tests" for code files.

These tests can catch accidental data coding errors. Not perfect, but much better than the status quo.

#rstats
February 13, 2026 at 2:27 PM
Reposted by Antoine Vernet
markhisted.org
In terms of what cures are being lost:

- Epstein-Barr virus is perhaps the major trigger for multiple sclerosis
- herpes simplex virus causes cold sores, genital herpes, infections in babies, deadly meningitis
- shingles virus causes an intensely painful disease
"Moderna's CEO announced the company will no longer invest in new Phase 3 vaccine trials for infectious diseases: 'You cannot make a return on investment if you don't have access to the U.S. market. Vaccines for Epstein-Barr virus, herpes, and shingles have been shelved.” 9:12 AM Feb 12, 2026 604.8K Views
February 13, 2026 at 5:44 AM
Reposted by Antoine Vernet
beenwrekt.bsky.social
Unbiased with respect to what? The primordial data-generating distribution of pdfs?
February 11, 2026 at 11:19 PM
Reposted by Antoine Vernet
volts.wtf
Pretty much the most important thing happening in the world right now.
www.carbonbrief.org/analysis-chi...
February 12, 2026 at 8:07 PM
Reposted by Antoine Vernet
stephenjwild.bsky.social
Instead, the academic publication system serves as a way for academics (primarily, though also other researchers to a lesser extent) to produce currency (articles) with which to buy things that matter to them. Think tenure, grants, speaking engagements, influence, etc.

4/N
February 12, 2026 at 2:45 PM
Reposted by Antoine Vernet
youngvulgarian.marieleconte.com
right, so, some quite insane professional news from me: last month I went to Paris to interview Gisèle Pelicot about what happened to her, and what her life looks like now - out in print on Sunday
February 12, 2026 at 2:46 PM
antoinevernet.com
In other news, crypto is a very normal financial product 🤪
tyleraking.com Tyler King @tyleraking.com · 3d
A Korean bitcoin market intended to give away prizes worth ₩620,000 or about $425. Instead, they gave away 620,000 bitcoins worth roughly $40 BILLION at the time—despite not actually holding that many tokens. It’s a massive failure of crypto safety protocols. Now they’re asking for the coins back.
A Bitcoin Blunder for the Ages: $40 Billion Accidentally Given Away
A botched prize giveaway has landed a South Korean cryptocurrency exchange in hot water.
www.wsj.com
February 12, 2026 at 9:00 AM
Reposted by Antoine Vernet
emollick.bsky.social
I get yelled at a lot on BS for saying so, but this is 100% true. There are people overhyping AI, but the alternative is not that AI is useless, or even the average of the two positions.

A lot is going to change dramatically even with today's AI. Ignoring that means no chance to shape what's next
chrislhayes.bsky.social Chris Hayes @chrislhayes.bsky.social · 3d
I understand why people are exhausted by AI hype, and why those of us squarely in the corner of "human dignity uber alles" see AI doomerism as self-serving hype, but I *really* think people on the left broadly need to start thinking seriously about the possibiltiy of the hype being...true.
February 11, 2026 at 10:56 PM
Reposted by Antoine Vernet
andrew.heiss.phd
(this is also part of my secret plan to get people to be bayesian)
This all feels vaguely Bayesian? Yep. All this simulation thinking is actually part of my clandestine plot to get more people curious about Bayesian statistics. This way of thinking prepares you for Bayesian inference, without actually being Bayesian. This simulation-based approach mirrors a lot of the computational thinking behind Bayesian statistics. In both approaches, we (1) start with an explicit data-generating process and (2) think about uncertainty with simulation instead of formulas. But there’s an important difference in what we’re simulating! With traditional null hypothesis testing, we simulate a null world and ask: If this were the world we lived in, how surprising would our observed δ be? Bayesian statistics flips this. Instead of asking how extreme our data (or δ) is in a single null world, we model uncertainty about the parameter itself. After combining prior beliefs with observed data, we get a posterior distribution that we can simulate from to ask: What range of values of δ are most plausible, given the data we’ve observed? If that sounds fun—and if you’d rather say things like “there’s an X% probability that δ is positive” or “δ is most likely between X and Y” instead of the convoluted “in a world where δ is 0ish, there’s an X% probability of seeing a δ at least as extreme as what we observed”—check out Bayes Rules! for the best introduction I’ve found to Bayesian modeling.
February 11, 2026 at 9:26 PM
Reposted by Antoine Vernet
andrew.heiss.phd
I just whipped up this little #QuartoPub site last week that demonstrates how I teach p-values/hyp-testing through simulation both with live OJS and with #rstats, and I think it's super neat! It has examples for diff-in-means, diff-in-props, and regression slopes nullworlds.andrewheiss.com #statsky
Simulated null distribution for data with a sample size of 100, difference in group means of 5, and a p-value of 0.142 Simulated null distribution of a slope of 0.8 and p-value of 0.002 Finally, we have to decide if the p-value meets an evidentiary standard or threshold that would provide us with enough evidence that we aren’t in the null world (or, in more statsy terms, enough evidence to reject the null hypothesis). There are lots of possible thresholds. By convention, most people use a threshold (often shortened to α) of 0.05, or 5%. But that’s not required! You could have a lower standard with an α of 0.1 (10%), or a higher standard with an α of 0.01 (1%). Statistically significant The p-value is < 0.001 and our threshold for α is 0.05 In a world where there is no relationship between x and y, the probability of seeing a slope of at least 0.901 is < 0.1% Since < 0.001 is less than 0.05, we have enough evidence to say that the slope is statistically significant. Evidentiary standards When thinking about p-values and thresholds, I like to imagine myself as a judge or a member of a jury. Many legal systems around the world have formal evidentiary thresholds or standards of proof. If prosecutors provide evidence that meets a threshold (i.e. goes beyond a reasonable doubt, or shows evidence on a balance of probabilities), the judge or jury can rule guilty. If there’s not enough evidence to clear the standard or threshold, the judge or jury has to rule not guilty. With p-values: If the probability of seeing an effect or difference (or δ) in a null world is less than 5% (or whatever the threshold is), we rule it statistically significant and say that the difference does not fit in that world. We’re pretty confident that it’s not zero. If the p-value is larger than the threshold, we do not have enough evidence to claim that δ doesn’t come from a world of where there’s no difference. We don’t know if it’s not zero. Importantly, if the difference is not significant, that does not mean that there is no difference. It just means that we can’t detect one if there is. If a prosecutor doesn’t provide sufficient evidence to clear a standard or threshold, it does not mean that the defendant didn’t do whatever they’re charged with†—it means that the judge or jury can’t detect guilt.
February 11, 2026 at 9:14 PM
Reposted by Antoine Vernet
rohanalexander.bsky.social
Monica's latest! read.dukeupress.edu/demography/a...
The Limits of Predicting Individual-Level Longevity: Insights From the U.S. Health and Retirement Study | Demography | Duke University Press
read.dukeupress.edu
February 11, 2026 at 2:11 PM
Reposted by Antoine Vernet
paulgp.com
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...
GitHub - paulgp/claude-container
Contribute to paulgp/claude-container development by creating an account on GitHub.
github.com
February 11, 2026 at 1:52 PM
Reposted by Antoine Vernet
theukka.bsky.social
cuplyr version 0.1.0 is now out!

A GPU-accelerated dplyr backend for R, powered by RAPIDS cuDF.

Write familiar tidyverse code, execute on GPU. Lazy eval with AST optimization.

In my benchmarks 60x faster than dplyr on 50M rows.

github.com/bbtheo/cuplyr
#rstats #cuda #DataScience
GitHub - bbtheo/cuplyr: GPU powered dataframes in R
GPU powered dataframes in R. Contribute to bbtheo/cuplyr development by creating an account on GitHub.
github.com
February 10, 2026 at 9:47 PM
Reposted by Antoine Vernet
sarahpyke.bsky.social
anyway the consultation shuts tomorrow, one minute to midnight UK time: www.gov.uk/government/c... Amnesty have some good guidance here. Go on: media.amnesty.org.uk/documents/Am...
media.amnesty.org.uk
February 11, 2026 at 8:49 AM
Reposted by Antoine Vernet
sarahpyke.bsky.social
the whole thing is of course completely revolting. 20 YEARS to settlement for refugees? no settlement if you’ve ever had a criminal conviction or a tax debt? 10 YEARS added if you’ve ever received benefits? partners and children on their own routes to settlement? What are you even TALKING about
February 11, 2026 at 8:43 AM