iiinova — where your digital self keeps evolving

Reposted by Martin Modrák

Grass Toucher Georg @avengingfem.me · 07/10/2026

i have a new theory of Social Media Bad. a lot of ink has been spilled about the addictive qualities of twitterlike social media, including many allegations of developer use of "dark patterns" and self conscious "engagement maximizing", but i'd like to propose something a little more mundane.

613828

Reposted by Martin Modrák

Grass Toucher Georg @avengingfem.me · 07/10/2026

because, as long as you actually care about having the conversation, which i do, *the rational response* is to start multitasking. do part of a chore, until you get a notification, or you check the app. write another reply, and then back to the chore. all day long. because that's how long it takes.

1754

Martin Modrák @modrakm.bsky.social · 20h

Broadly agree, though metrics evaluating pairs of studies (esp. "replication estimate within original prediction interval" - which they use) can be made reliable (i.e. control error rates).

100

Martin Modrák @modrakm.bsky.social · 04/10/2026

Loved the pod as usual! Wouldn't a big barrier for siRNA be that some (many? most?) rare diseases are caused not by bad protein product but by absence of the correct protein product, which siRNA cannot fix? I was curious that you didn't discuss this at all - am I missing something?

110

Reposted by Martin Modrák

Saloni @scientificdiscovery.dev · 02/10/2026

Can we switch off diseases? siRNA drugs are, in my view and @jacobtref.bsky.social’s too, one of the most exciting developments in medicine. So we decided to do an episode!

open.spotify.com

Can we switch off diseases?

Hard Drugs · Episode

13914

Reposted by Martin Modrák

Ian Coldwater 🧊🚫 @lookitup.baby · 30/09/2026

BREAKING NEWS: “Do geese see God” is a palindrome

14780781990

Reposted by Martin Modrák

Sanho Tree @sanho.bsky.social · 26/09/2026

Happy Stanislav Petrov day! He saved the world from nuclear annihilation on this day in 1983. My freshman year in college would have been so much worse without his courage.🧵

5599185

Reposted by Martin Modrák

Os Keyes @okeyes.bsky.social · 15/09/2026

Since we’re doing another round of “AI will destroy humanity” doom-hype, it’s a good time to announce a new paper myself and my postdoc advisor just had accepted by BioSocieties on just this. TL;DR: almost all the “AI can make bioweapons” literature comes ultimately from the Effective Altruists.

ironholds.org

17752271

Reposted by Martin Modrák

Robert Komaniecki @komaniecki.bsky.social · 27/09/2026

“Why shouldn’t I custom-generate all of my child’s media?” Because the point of children’s literature, music, art, and movies is not to optimize leisure time, it is to provide your children with cultural, emotional, societal, and epistemic touch points via which they can relate to the wider world.

617643

Martin Modrák @modrakm.bsky.social · 25/09/2026

It's the one where they then start using humans as missile guidance computers because it's cheaper or is that another one? Oh, it's been long since my Asimov phase...

030

Martin Modrák @modrakm.bsky.social · 24/09/2026

I want this to be true, and it is an important question overall, but I admit the discontinuity plots are IMHO quite underwhelming and to me appear quite consistent with no discontinuity...

021

Reposted by Martin Modrák

Stand Up for Science! @standupforscience.net · 18/09/2026

Your life saving clinical trail could be TERMINATED by the Trump administration. We scraped through the data and found that 50% of all clinical trials are at risk of being cancelled. It's time to talk about it and what you can do to stop this. 🧵 🧪

217697

Reposted by Martin Modrák

Catalina I. Fernández, PhD (she/her) @catferna.bsky.social · 22/09/2026

New in the saga of our causal model of human growth 🚼🧬📈at AJBA: can we reconstruct growth trajectories when kids’ ages are uncertain & all we have is cross-sectional data? Turns out, YES! 🎉 Accurate estimates, with exciting (?!) implications for bioarchaeology and beyond. doi.org/10.1002/ajpa.70358

doi.org

2104

Reposted by Martin Modrák

Alex Holcombe @alexh.bsky.social · 21/09/2026

Consequences of the flood of AI manuscripts: "An early-career researcher at an unbranded institution loses the one mechanism that ever let them be read on merit — anonymous review of the work itself, however badly it functioned in practice.

410823

Reposted by Martin Modrák

Per Engzell @pengzell.bsky.social · 21/09/2026

Need new examples for teaching confounding and reverse causality to my undergrads. Send any published statistical crimes my way please.

7238

Martin Modrák @modrakm.bsky.social · 20/09/2026

I don't think that's true. Non-randomized causal inference should still work. My answer was to the narrow question: it is AFAIK possible to see regression to the mean in some/all groups. This does not mean it is impossible to run an analysis that is not misled by it.

020

Reposted by Martin Modrák

Ben Recht @beenwrekt.bsky.social · 18/09/2026

@dsquareddigest.bsky.social "13 Theses on AI and regulation" are exceptionally good. My favorite set of thoughts I've read on this topic so far.

backofmind.substack.com

13 theses on agentic AI and regulation

thinking out loud and showing traces of reasoning

1234

Martin Modrák @modrakm.bsky.social · 18/09/2026

I'd guess that if you have non-trivial selection at baseline (or other time point) you'd see regression to the mean.

130

Reposted by Martin Modrák

Mike Murphy @mjamurphy.bsky.social · 18/09/2026

I would take the alignment stuff more seriously if anyone in the space appeared remotely interested in the cases of AI being used to kill people *which have already happened*

042

Reposted by Martin Modrák

Richard McElreath 🐈‍⬛ @rmcelreath.bsky.social · 18/09/2026

Happy to see others emphasizing traceability (Nachvollziehbarkeit) over simple reproducibility. Let’s make fetch happen!

15511

Reposted by Martin Modrák

Gautam Kamath @gautamkamath.com · 16/09/2026

Nihar Shah did a heroic experiment for TMLR: he spent 20-25 hours over two weeks interviewing authors of seemingly low-quality submissions about their own papers. He confirmed what we all suspected: people submitting these papers have *no idea* what is going on in them.

5293117

Reposted by Martin Modrák

Peter Ellis @freerangestats.info · 16/09/2026

Inspired by a @f2harrell.bsky.social comment, I did some #rstats simulations about how big a sample needs to be for the central limit theorem to kick in and the sample mean be normally distributed for inference purposes. Yes, with skewed data the answer is "lots". freerangestats.info/blog/2026/09...

Scatter plot with 8 facets, each showing a different distribution, and the actual coverage of a 95% confidence interval for the mean at different sample sizes. Most values are well below 95%.

33315

Reposted by Martin Modrák

Greg Wester @gwestr.me · 16/09/2026

LOL

3490146

Reposted by Martin Modrák

Joe Bak-Coleman @jbakcoleman.bsky.social · 16/09/2026

I replied to @zachlowenba.bsky.social (whose work I love) but feel like it probably deserves a longer thread a QT. Setting aside the AI bit, I don't think the question(s) are coherent, but they sure are harmful. 🧵

1147

Reposted by Martin Modrák

Danielle Navarro @djnavarro.net · 15/09/2026

In her ongoing quest to convince #rstats users to stop invoking the cursed "rm(list=ls())" incantation at the top of their analysis scripts, she has released sessioncheck 0.2 on CRAN. Mostly unchanged, some bug fixes, some new reporting features that are most useful in the regulatory context

blog.djnavarro.net

Release: sessioncheck 0.2 – Notes from a data witch

In which the author realises she doesn’t know how to write package release posts

4439

Martin Modrák @modrakm.bsky.social · 14/09/2026

I will be looking for a PhD student in Bayesian statistics (in collaboration with Paul Bürkner), located in Prague. Forward to students if you have somebody who could be interested. www.martinmodrak.cz/grammo-call/

martinmodrak.cz

Grammo - Let's Research Bayes!

03324

Reposted by Martin Modrák

Julia M. Rohrer @dingdingpeng.the100.ci · 14/09/2026

Okay before I do write my screed about cluster analysis, please send me your favorite *positive* examples that illustrate the utility of the method!

304011

Martin Modrák @modrakm.bsky.social · 13/09/2026

The road to IKEA is paved with good intentions.

021

Reposted by Martin Modrák

Richard McElreath 🐈‍⬛ @rmcelreath.bsky.social · 12/09/2026

I do wonder what David Hilbert would think about all this. He was accused of trying to automate mathematics, in the 1920s! The "Hilbert program" failed in an exact sense, because Gödel, but it succeeded by making formal proof systems like Lean possible. And Lean makes the AI-sausage-machine work.

Two panels from the graphic novel Logicomix.
Left: He wants a machine to feed it axioms and make theorems, like one where a pig enters the one side...
Right: ...and the sausages come out from the other!! (laughter)

3749

Reposted by Martin Modrák

Felipe Fontana Vieira @felipefv.bsky.social · 11/09/2026

Next week I will give a free webinar about how I have been using Nix (by itself) but mainly through {rix} for my current articles. If anyone is intested, my plan for the structure of the webinar (in the comments):

11913

Reposted by Martin Modrák

tj mahr 🤘 @tjmahr.com · 10/09/2026

i'm using my solarizeddocx which i made way back when because the default RMarkdown MS Word output looked bad github.com/tjmahr/solar...

github.com

GitHub - tjmahr/solarizeddocx: A refence docx file for pandoc with solarized syntax highlighting

A refence docx file for pandoc with solarized syntax highlighting - tjmahr/solarizeddocx

021

Reposted by Martin Modrák

Mel Andrews @bayesianboy.bsky.social · 11/09/2026

You can now read this invaluable book on methods in philosophy of science open-access. Find me in Chapter 16!

0364

Reposted by Martin Modrák

Luna Fazio @lunafazio.bsky.social · 09/09/2026

Had a Zoom meeting earlier today where a lot of great exchanges were happening in the chat, so we wanted to save the log for later. Zoom exports a plain text file, so I made a tool to convert that into something that's more convenient to interact with. Sharing in case anyone else finds it useful!

Example of what a zoom chat log looks like (fictional conversation)

The chat log after being converted into an interactable html

153

Reposted by Martin Modrák

ponder @ponder.ooo · 08/09/2026

lmao what the fuck do you mean it's unlikely but you can't rule it out. do you not know what you're training your models on? are you guys unaware of what's going on in your own operation? does openai research not have any knowledge of what data they train their models on?

we cannot rule out that de-identified data derived from their usage of our products helped improve our models⁠

1528132

Reposted by Martin Modrák

Noah Haber @whaleactually.com · 08/09/2026

... and that's where Registered Revisions comes in The Registered Revisions Meta Trial Pilot Report is OUT NOW!! osf.io/preprints/me... But new(ish) policy involves developing a new(ish) multi-headed trial design, workflows, infrastructure, piloting, mistakes, and a super team. And it's all here.

22117

Reposted by Martin Modrák

Martin Modrák @modrakm.bsky.social · 17/07/2026

Multiverse analysis, abdication of responsibility and manufacturing of doubt: I have written on some downsides I see with multiverse analysis (which I like in principle): arxiv.org/abs/2607.14623 I was inspired/provoked to write it by @dingdingpeng.the100.ci (thanks!)

Abstract: I argue that multiverse analysis is highly suited to two undesirable uses: abdication of researcher's responsibility for their conclusion and manufacturing of doubt. A review of multiverse analyses published in 2025 provides tentative empirical support that abdication of responsibility is present in the literature and I mention anecdotal evidence that multiverse has been used for manufacturing of doubt about Covid-19 precautions. To mitigate negative effects if multiverse analysis becomes widely used I suggest the community adopts two conventions for evaluating multiverse analyzes: evaluating multiverses by the single worst universe they contain and considering large size of a multiverse as a sign of weakness rather than a praiseworthy achievement.

45818

Martin Modrák @modrakm.bsky.social · 07/09/2026

It is a reflection of the remarkable success of the Roman commitment to diversity and inclusion* that almost all of the people it conquered ended up identifying with the empire, even as it's western part was falling apart and even beyond. (*by pre-19th century standards)

020

Reposted by Martin Modrák

Richard McElreath 🐈‍⬛ @rmcelreath.bsky.social · 04/09/2026

It's a systemic problem in science reporting that these summaries exaggerate the evidence. In the actual candy study, the participants were all students, status was manipulated in the lab by telling some students they were better than others, and "nearly twice" means 1 piece vs 2 pieces.

Piff and his colleagues found that upper-class drivers were more likely to cut other cars off at junctions and ignore pedestrians at crossings. Wealthier people also appeared more willing to lie at work. In one experiment, volunteers were left alone with a jar of sweets ostensibly meant for children. When the researchers returned, upper-class people had snaffled nearly twice as many candies as lower-class people.

1329166

Reposted by Martin Modrák

Kate Starbird @katestarbird.bsky.social · 03/09/2026

Seattle Times comes out against I-638. It s terrible law, based in bigotry, that hurts all girls who want to compete in school sports, requiring invasive sex testing. If you’re in WA state or know someone who is, please spread the word: Vote NO on I-638. www.seattletimes.com/opinion/edit...

seattletimes.com

Seattle Times recommends 'No' on I-638, ban on trans girls in sports | Editorial

If this were a “Washington problem“ it would be worth studying as a pressing issue to be resolved by the electorate, informed with science. But it is not.

8379173

Reposted by Martin Modrák

Matt Makel @mattmakel.bsky.social · 31/08/2026

I made a mistake. And because of that mistake, I no longer have confidence that results that I reported in several research papers are an accurate reflection of replication rates. 1/

19430

Reposted by Martin Modrák

Berna Devezer @devezer.bsky.social · 28/08/2026

OK I have to point this out: So he does know how experiments work because last I remember, he (and colleagues) tried to sell the story that reforms work using a design that did not have a corresponding treatment or control about the reforms, and insisted that there was strong evidence of causation.

Dr. Nosek said that Transfyr would have to present detailed results to prove that this system really does make science better. “Take 50 labs, track half the experiments with this stuff and don’t track the other half — then look for differences in progress,” he suggested.

2313

Reposted by Martin Modrák

Berna Devezer @devezer.bsky.social · 28/08/2026

We need to stop this nonsense of insisting on viewing scientific error as a problem to be eliminated. Scientific error is the fuel we need for scientific progress. Without uncertainty, we don't need science. With forced certainty and required faith, we get religion

311025

Reposted by Martin Modrák

Demetri @phdemetri.bsky.social · 26/08/2026

Bayes doesn't make experiments run faster. Stop saying this dpananos.github.io/posts/2026-0...

dpananos.github.io

Bayes Does Not Make Experiments Faster On Its Own – Demetri Pananos Ph.D

5226