smol bean agency forgot to do records retention as required by federal law 🥺👉🏻👈🏻

"Then you lie to the computer about what's in the data to make it learn better because if it can see the actual data it doesn't learn as well"

Sometimes I see a pre-print and pray to the science communication gods that it isn’t reported in the news as though it is a peer reviewed study.

Sometimes I see a pre-print and can only pray that the peer review gods will bring it to me.

It's weird that we're well into the Gilmorenaissance yet haven't really enjoyed a Sam Phillipsaissance yet www.youtube.com/watch?v=7aB-...

The New York Times is going through some things, although not, regrettably, a woodchipper.

This really hasn’t gotten enough attention this week. @wired.com #ICE www.wired.com/story/fbi-wa...

Apropos of nothing, this is how I found out that the version of Star Wars I grew up watching was the 1997 special edition and NOT the original.

We can only conclude that Trump is a participant in the UK Biobank imaging substudy.

Oh my god

Do these sound like the actions of a man who had "all he could eat"? www.youtube.com/watch?v=CqT-...

*kanye voice* b-b-b-b-b-bound by I/O

I haven't read this yet but I really hope it was well done because I would love it if this works. aacnjournals.org/ajcconline/a...

As an elder millennial, if there's two things I'm going to do on the first Tuesday after the first Monday in November, they're vote and pop my collar.

I've made the Adam Ragusea roast chicken dinner more times than I can count. It is very good. www.youtube.com/watch?v=G8Q3...

putting samples from the training set in the validation set, as a treat

No matter what happens, im pretty likely to confirm my priors

Other people have parasocial relationships, I have a friend who tells me about the economy while I cook dinner [Kai Ryssdal]

The more deep learning I do, the more I like logistic regression

To me, the embarrassing thing about "Wood" is the fact that it references Travis's podcast

When the BERT drops

submitting your AI generated position paper to the arxiv should get you immediate ticket to the icy pit of hell

Critics of AI, also known as “paperclip minimizers,”

There isn’t an incentive system alive that can enforce “be a lunatic about the quality of your results”

Calling it "altitude generation" instead of "oxygen removal" is brilliant marketing, to be honest. hypoxico.com/products/sie...

No grab bars, huh?

“my wife left me for the male form factor robot”

OK I wasn't going to post about the Penn 'hacking' situation but I need these script children to stop sending me emails.

Um, Betteridge's law? www.nytimes.com/interactive/...

What's the deal with econ working papers - do the authors really want feedback? And how can you tell who the corresponding author is if they're not designated and the authorship is alphabetical?

Ask your doctor if grim determination is right for you

please 😭 stop 😭😭 reporting performance metrics on the training set 😭😭😭

I use Claude (an AI) to scan my posts for spelling and grammar errors. On the most recent post (about the East Wing, link below), I got a really amazing addendum. www.pbump.net/o/yes-most-a...

I can't see an "in conversation with" talk advertisement without thinking of "Women on the Wall: An Exploration of Gender in Text and Media: Barbara Kruger and Jenny Holzer in Conversation with Helen Molesworth"

With diffusion of #medsky #healthpolicy across platforms, at times miss hearing about work posted by people on other media. Want to share this post & study from Justin Berk @brownmedicine.bsky.social on discrimination in post acute care re: incarceration: www.nature.com/articles/s41...

My name is Ozymandias, King of Kings. Look on my works and let me know if you have any questions! 🤗

This might be the best RCR training I've ever taken.

Why in the world is Matthew Yglesias in this WaPo "man on the street" video about the new US citizenship exam?? youtube.com/shorts/tUt6W...

Yeah Outlook I totally wanted to randomly capitalize the word "don't" in the middle of this sentence. Thanks for providing Copilot, a tool whose performance fails to exceed a rules-based algorithm? Apparently?

This instrument makes me want to cry

Literally so hype to hear Amy Finkelstein speak about one of my biggest pet peeves, the decedent analysis.

Some disorganized thoughts about this trial will follow here.

📚 Development and Validation of the Sequential Organ Failure Assessment (SOFA)-2 Score: Presented today in the #LIVES2025 Hot Topics session. 🔗 Free to read online: ja.ma/4qzouAz

💬 Editorial: SOFA-2 improves the measurement of organ failure in intensive care by integrating current treatments and data-driven criteria, but further research is needed to address persistent challenges and optimize bedside utility. #LIVES2025 @esicm.bsky.social ja.ma/47pxgZb

I have to give it up to The Onion's graphics artists, this is a more realistic ETT setup than most movies and TV shows manage.

not to brag but i can generate slop with natural intelligence

We wrote an article explaining why you shouldn't put several variables into a regression model and report which are statistically significant - even as exploratory research. bmjmedicine.bmj.com/content/4/1/.... How did we do?

It’s astonishing how many researchers seem to believe that a cluster analysis is such a sensible analysis that it needn’t even be justified through a coherent research question. Just cluster analysis go brrrrr