Berna Devezer

without assuming this is how anything works in real life, without turning researchers into caricatures with a single-minded focus on a given external carrot, without trying to establish distrust in the scientist as a structural norm in science. Theoretical models can be better used for exploration.

specific assumptions. But ultimately, I get frustrated that so many metascientific works assume the most strategic, externally motivated, dishonest researcher behavior as the default and then make recommendations to avoid that worst-case scenario. The model is interesting to explore on its own

I've been thinking about this paper a little bit the last couple of days. I have an issue with the author's claim "Our new paper shows that is true empirically and makes it precise". The paper cannot and does not show whether this is true empirically because it is a very theoretical model with very

A few people have asked for the syllabus from my grad seminar on Generative AI for social science -- just posted it here:
statmodeling.stat.columbia.edu/2026/03/10/n...

i have tolerance for so much nonsense and so many people being wrong on the internet. but this, i don't know how to tolerate so i had to part ways with so many mutuals over it. if you're standing at the gate deciding whom to let in, i don't want to play in your backyard anymore.

That doesn't sound like it would necessarily achieve what it's meant to. Typically when units of different sizes/resources merge, the underdogs lose their voice altogether. Good luck!

Indeed, the higher ed admin is divisive enough. We don't need to do it to each other either.

i'm sick of gatekeeping based on stereotypes about any discipline. walk on campus, meet someone from a different college, talk to them. many of them care about the students, care about the scholarship, care about the world.

art belongs on university campus. medicine belongs on university campus. so do law, engineering, agriculture. social and life sciences and humanities belong on university campus too. so does business. they all make up the world, culture, society we live in. they're all worth studying and teaching.

The Danger of Silence When Academic Freedom Is Under Threat Inaction from rank-and-file workers enabled government censorship during the Red Scares.

You ever write something and there are so many parts you hope people engage with that you can't figure out which ones to quote?

I'll do a wee thread on the 3rd, and final, essay from my Red Scares series

www.insidehighered.com/opinion/colu...

In case you were wondering about the state of free idea exchange in academia, this is where we are.

I appreciate colleagues reaching out. But I wish they'd say it publicly, especially if untenured. That's the only way to change this insanity where experts are afraid to speak up.

Feeling some social reproach from colleagues for doing something those colleagues think of as morally problematic is not even in the same ballpark as the actual free speech issues facing academia right now

That's quite possible. And you may remember having read about Al-Biruni.

Way before the Islamic golden age, there was Galen of Pergamon (2nd century AD).

Very interesting related observation from Schickore:

"The word control became common in the experimental sciences when it turned out that experimental situations were too complex to be completely controlled."

Trivia: While the practice of comparative experimentation goes back to antiquity, many historians view J.S. Mill's A System of Logic as the first philosophical conceptualization of the narrow methodological notion of experimental control/controlled experiments, though he didn't use the word control.

Ironically (or not), that, too, would be very Feyerabendian 😂

The rhetorical function is non-negligible and so very Feyerabendian

A shift from viewing the reader as an active participant in the research process toward positioning the reader as a passive recipient of research conclusions. And a shift in the researcher's role, from explorer to expert.

One interesting shift from the 18th to 19th century was in the focus of scientific writing: From wanting to demonstrate the discoverability of the findings (giving the audience a sense of thrill and awe) toward a rhetorical function—making it easier to appreciate the significance of the findings.

"In the first half of the 19th century, experimental research was often vehemently criticized for being speculative, mystifying, and thus unscientific." Supporters of lab-based research and medical reforms had to distance themselves from the medical community. RCTs wouldn't have seemed inevitable.

A century later in a different continent experimentalism was hardly the main scientific approach in medicine. American physician Silas Weir Mitchell struggled to secure himself a position since lab work wasn't supported in the profession. Observation-based medical practice was the scientific norm.

And it's not just snakes. They needed pigeons, rabbits, chicken, etc. in even greater quantities because they needed 'research subjects' which tended to expire rather sooner than the snakes 😬

I think the framing of the question assumes it's a tool whereas AI could generally be better thought of as a technology (so not necessarily designed with a fixed and well-defined problem or use in mind).

🤫🫢

I don't actually suggest that anyone do the study proposed in the OP or any of the specific meta-metastudies implied but it is a low hanging fruit so I'm sure someone will write one up soon. And once again we'll be expected to act shocked that different model assumptions yield different inferences.

I'll take the prose

companion post here

meta-meta-metascience study meta-analyzing comparisons between multiverse analyses on full specifications retrieved from many analyst datasets to many-analyst results when

Never used reed rush. I have a reed geek but I'm pretty sure I'm ruining my reeds with it because idk what I'm doing 😅