This is more true than we want to believe. Even plausible sounding advice is often blatant falsehood.

AI slop. I gave this talk Tuesday. My abstract is available online. I never once mentioned Rio Puerco glass (I've never heard of it). I never mentioned Chaco Canyon supernovas or eclipses. Zuni Salt Lake? No mention whatsoever. Groundwater surveys I mentioned had nothing to do with craters. Crazy BS

People misunderstand how a small court ruling in the US can have GLOBAL consequences far bigger in reach than some wars.

Still remember the crushing moment as - a young law student trying to find a path to change the world for the better - when I realised that no bigger project or idea would become infrastructure or common practice unless it was politically useful to believe it. Also applies to popular philosophies.

@lawoffairness.bsky.social is utter word salad.

it's awesome that money is speech but only sometimes and for some people

From now on, I'm going to write papers where I take an arbitrary performance metric, restrict the sample to top performers on that metric, and then hunt for negative correlations with everything else in the truncated sample. I'm at a business school, so this should be extremely lucrative.

Yes, there's an AI-generated rat with enlarged testicles involved. But behind the laugh, @elisabethbik.bsky.social raises an alarming point — fake AI images are passing peer review undetected. New OOTL episode out now: open.spotify.com/episode/2zJT... 🐀 #science #AI #fakenews #podcast #ootl

Paper mill content growing faster than legitimate research, according to study by @reeserichardson.bsky.social & colleagues — which @elisabethbik.bsky.social mentioned in her talk at #WCRI2026: cen.acs.org/policy/publi...

People misunderstand how a small court ruling in the US can have GLOBAL consequences far bigger in reach than some wars.

The Boston Globe agrees: Citizens United did not create SuperPACs, and it is time for the First Circuit to fix that absurd mistake. www.bostonglobe.com/2026/04/26/o...

What would you do to the media landscape if: you were big tabacco one of your tobacco (insert other major company who plays the playbook) compounds that just can not be removed, has hormonal effects and simply makes newborns have a less clear gender and maybe all kinds of other mental effects?

Just for the sake of training those models to the highest accuracy one should buy a degree for a couple hundred bucks, publish with paper mills to no end and take one of those remote job to train AI. Then only cite mdpi and researchgate or everything retracted.

What was this "landmark study" that showed the addiction risk of opioids was low? It was this. A one-paragraph letter to editor, about how things looked in their files. A one paragraph letter, upgraded to "landmark study" because, I have to think, no one bothered to read the original.

🚨 For nearly two years - and even now, depending on the day and model - AI models could tell you all about bixonimania, a totally made-up disease in a paper funded by “the Professor Sideshow Bob Foundation". My @nature.com story on AI's fallibility in medicine www.nature.com/articles/d41...

Hey @natureportfolio.nature.com is this some sort of joke?? Please act ASAP. This is an abomination! 👇How the hell did this ever pass peer review, editorial processes??? doi.org/10.1038/s415...

You might think your smoke detector's false positives are annoying, but let me tell ya...

"In the most extreme case, our model achieved the top rank on a standard chest X-ray question-answering benchmark without access to any images." Well this isn't good 😮 arxiv.org/abs/2603.21687

How do the "death fold domain" proteins in the innate immune system react so quickly and thoroughly, but without needing any driving energy like ATP? A weird but quite believable answer:

It needs this creative destruction of Landlordism, like rent freezing and decaying housing. Anyone who hoards a home and doesn't use it themselves had a bit too much of the perfume called capitalism and forgets that he should actually try to bring value to people and not extract it.

The research integrity community didn't move forward since imagetwin. Retractions bla. Ben goldacre does infrastructure. James heathers maybe a thing. Nobody knows what Peter Doshi does. Government agencies are not bound to a scientific process but money. LLM's make raw data evidence unreliable.

I am a normal person with job and family. The way the world is right now: I have to keep my invention for myself until I have time in 20 years to make it happen on mostly my own then patent it and hope nobody thinks of it before. What a beneficial system for humanity.... ... ........ ..............

We are constantly reminded that the most promising ideas still have to be proven in the clinic - and that the first trial is almost never enough. If you've never heard of TIGIT, it's yet another cautionary tale:

"We only accept clearly AI written papers. No computer can understand what you humans mean. "

Next up: mandatory not sharing data

Ed, stop it. Without them defrauding everyone on the way they would never have been able to build the rails or those neat data centers. We exchange new high tech infrastructure for a max 10 year financial crisis. It could be worth it in the long run. Forget ROI.

Thousands of scientists inflate their CVs with self-published studies that cost millions of dollars of public money 1 in 8 'special issues' is filled with articles written by the editor, particularly at the publisher MDPI @manuelansede.bsky.social english.elpais.com/science-tech...

It took a special type of shamelessness, directly from the gutter into their face, mixed in the results of their own actions, to break the political ice.

Britegration

That's just inflation catching up. So the big ones don't fall so far.

Patent Law International Tax optimisation strategies Chemistry Law Good Laboratory Practice Investigative Journalism

Super interesting Webinar by Consilium Scientific: "The Role of Litigation in Exposing Challenges in Medical Research and Healthcare" by Reuben Guttman, JD and Daniel Aaron, JD, MD. January 29th 2026, 5pm London Time mailchi.mp/consilium-sc...

This is my go to country if the west is going into world war 3

“Uruguay did what most nations still call impossible: it built a power grid that runs almost entirely on renewables—at half the cost of fossil fuels. The physicist who led that transformation says the same playbook could work anywhere—if governments have the courage to change the rules.”

John Cage, submitting the sheet music for "4'33"" to his publisher: What do you think? Publisher: No notes.

www.nature.com/articles/s41... Even as outsider I know that many of those trials are nothing to look out for. Loratadine for hard to define new chronical illness? Stem cell injections? Single RNA inhibition for complicated disease? Is nothing else happening? @dereklowe.bsky.social a better list?

An ugly "stealth correction" at @tandfresearch.bsky.social's Hum Vaccin Immunother of a paper from authors at the Uniformed Services University of the Health Sciences, Bethesda. They just removed the overlapping image panels 😱 #ThisImageIsFine #ImageForensics pubpeer.com/publications...

Social media will ultimately be as lame as LinkedIn because whatever you say could be one day construed to lose your job or whatnot, even if it was just a joke. It is such a downside. A permanent job interview. More anonymous social media with moral leeway will prevail.

How would they lose control?

I read diversely enough that I allow myself to say that most brains go into Amazing ideas that are iterative on improving a technology or product. Even seemingly low hanging fruit are super cool, just read the tech behind max solubility of sulfur, meta optics solar sails & their fab, FPGA's, etc.

@bengoldacre.bsky.social Hey I once left this Naive comment on your turf about using LLM's to add new functions to Opensafely. I read about prompt injections rn and realised it could work pretty well if you button it up: Limit it to single useful functions that are pressed by a button= No prompts.

Come on man

@dereklowe.bsky.social Now i got an RSS reader only because I could have easily missed your Reverse Proteolysis Blog post from only rrading Bluesky. This thing was insane news. It made me think of prions as well..

This "looks like" you could just throw it into the AlphaGo @googledeepmind.zpravobot.news.ap.brid.gy engine and change the rules in how the boars looks a bit to get a potential improvement.

Random colleagues in other disciplines now commonly say "the code's on OSF". It's just so great to see an academic org like COS, that goes beyond advocating for change, and actually builds the tools and services that make science work better. More!

Scientific error-checking, difficulty: easy Find some issues in this table of results from an RCT of vitamin D for COVID-19. There's a couple of obvious issues, and some that are less easy to spot.

Hey @nature.com, have you got an explanation for how the hell THIS happened? & especially why you accepted a paper with such a bizarre piece of genAI slop in it?! & more to the point, why we should take you seriously at all going forward? www.nature.com/articles/s41...

@lessig.bsky.social youtu.be/-NuXpwB2DV4?.... Johnny Harris finally did a pretty good piece on this. You probably already saw it.

From the authors of the argument that will end SuperPACs: buff.ly/mZe8Jm5

lads i found ANOTHER Table of Impossible Summary Statistics in a "green economics" paper 🚩 Can you see the problem?