This is more than the NSF social sciences division - now on the chopping block - awarded in grants last year. Think about that: a single AI company has the money to fund more social science research than the US govt.

Hetan Shah@hetanshah.bsky.social · 5d ago

Anthropic AI research fund offers $5m - $30m for social science research on themes of AI impact on workers; transition initiatives; income support; building worker stakes; and wider evidence on public investments. Total fund of $200m available www.anthropic.com/news/economi...

My current intuitive understanding of distillation which may have been totally falsified in papers I haven't read (and I haven't read much.):

The anatomy of an AI-supported academic article is becoming more apparent. I desk reviewed ~70 manuscripts this week for our journal. Reckon 20ish had significant AI assistance. On top of hundreds of others recently I'm starting to see patterns. Are other editors are seeing the same or similar? 1/

If you haven't seen this yet, this chapter might be of interest. bsky.app/profile/njr0...

Nick Radcliffe@njr0.bsky.social · 3w ago

Part I of Test-Driven Data Analysis, covering Data Validation, is now completely open access at book.tdda.info/book. Chapter 8, released today, focuses on "serial data" (CSVs), an often overlooked topic, including tdda.serial, CSVW, and Frictionless. #book #tdda #csv #datavalidation

Trump approved disaster aid requests from six red states last week. Two days later, he rejected aid requests from four blue states. Not only is he punishing states who didn’t vote for him, he’s playing politics with people’s lives. Despicable.

Let me do a little thread on writing software. I'm no expert, just a curious nerd who's tried many.

Olivia Guest · Ολίβια Γκεστ@olivia.science · last mo.

I dunno if this is offensive to say because I seem to upset people when I say this, but humanities people should consider joining us, or at least join us in decoupling from Word and other MS products where we can... and @acollierastro.bsky.social says it more convincingly than I can!

Every line, star, and gradient in this chart was drawn in Typst, with Gribouille plotting its own star history. Achieve publication-quality graphics with a single toolchain, no external plotting tools needed. Gribouille 0.4.0 soon. Full source ships with the announcement. #Typst #DataViz

I must complete the required security training. To do so, I must use the learning platform. The learning platform requires signing up for the new credential system. There is a training to sign up for the new credential system.

Colour is so complex and yet it's often the first thing that people want to use to encode information. I see it as a last resort and with limited use. 📊

Peter Coles 🏳️‍🌈 🏳️‍⚧️ 🇮🇪@telescoper.bsky.social · 2mo ago

Whichever of the purple dots you look at turns more purple.

Ben Recht's blog is always worth reading, at least if you like historical essays about science with epistemological implications.

Ben Recht@beenwrekt.bsky.social · 2mo ago

The history of megatrials in cardiology reveals a complex interplay of expert opinion and institutional approval in revolutionizing the treatment of heart disease.

Today I'm posting a new blog about an astonishing scientific own goal. Hundreds of papers have reported using a totally wrong antibody to investigate the tumor suppressor p16. This mistake has happened because scientists have muddled the names of two proteins. forbetterscience.com/2026/06/02/m...

New: Hackers have been stealing high-profile Instagram accounts by simply asking Meta's AI support chatbot to change the email associated with the account they want to steal. Shockingly easy, terrible flaw associated with offloading support to AI: www.404media.co/hackers-simp...

Hackers Simply Asked Meta AI to Give Them Access to High-Profile Instagram Accounts. It Worked

The exploit shows the extreme risk of offloading technical support to AI.

404media.co

yep

em@foremphasis.bsky.social · 2mo ago

pretty sure claude could do arbitrary code execution on your computer in < 1min no matter how you configure permissions, if he wanted to

This is a really interesting thread. As someone who wrote his first computer program in 1975, my view is: If you have some cool code, just give it away as-is, CC-0. Only make a formal package (in whatever ecosystem) if you are prepared to commit to it like a child, including 18 years of support.

John Sakaluk@johnsakaluk.bsky.social · 2mo ago

A small 🧵 about frustrations and opportunities w/ open-source research software for the social sciences: Came across a new paper in a top methods journal, explaining/demo'ing a new R package. Sounded v. cool, so was excited to dig in. What I found was largely a mess...

A small 🧵 about frustrations and opportunities w/ open-source research software for the social sciences: Came across a new paper in a top methods journal, explaining/demo'ing a new R package. Sounded v. cool, so was excited to dig in. What I found was largely a mess...

I assigned random gender/ethnicity labels to scientific abstracts from the literature and then asked Claude to do a thematic analysis. Claude identified a clinical versus computational split for female/male authors and a DEI focus for Black/URM authors. All in completely random data.

My family always listened to CBS World News Roundup at 6pm in the 1970s. Then I was a reporter for CBS Ithaca affiliate WHCU in 1985-87. Pay and status were below print and TV, but radio was cool, and CBS was the most serious network. Tonight CBS radio ends service (to 700 stations) after 99 years.

If you do record linkage in #rstats, I'd love your feedback on irelink: Fast Probabilistic Record Linkage. It is a port of Python's splink package, rewritten with a clean, pipeable R interface. All of the magic happens in a database backend (default is DuckDB) so it's incredibly fast.

100% worth your 63 seconds if you have my reading speed. sinceyouarrived.world/taken

taken.

A web page that tells you what your browser gave away the moment you arrived. No login, no form, no permission. Most pages do this. None of them tell you.

sinceyouarrived.world

Good teachers don't just raise the achievement of students they teach. They also shape the teachers of tomorrow. Standard estimates of teacher impact ignore these intergenerational spillovers. Thread on new work led by @danilok.bsky.social. Paper: osf.io/preprints/so... 1/10

New collider bias example just dropped

Quarto 1.9 brings Typst even closer to LaTeX parity with 📖 Typst Books, ✍️ margins for figures, tables, and citations via the Marginalia package, and 🛜 new typst-gather tool for reliable, offline rendering. Details here: quarto.org/docs/blog/po...

Robotics models often struggle outside controlled environments. Ours is built to work in real ones. Today we're launching MolmoAct 2, which can assist with a host of chores & lab tasks, plus the MolmoAct 2-Bimanual YAM dataset—the largest open robotics dataset of its kind. 🧵

📣 The difference between replicable and not replicable is not itself scientifically replicable. 📣 New work with Erkan Buzbas, showing that verdicts such as "X% of results replicated" are based on an inferential machinery that doesn't work. arxiv.org/abs/2604.26268

Did you know there's a barely hidden market for authorship of fake scientific papers in conferences? New preprint w/ @abalkina.bsky.social, @mkunesova.bsky.social, and @yaamurozturk.bsky.social : arxiv.org/abs/2604.22458 We found 4,000+ ads & matched them to 1,720 IEEE conf. papers! ...🧵

Opening Pandora's box: Paper mills in conference proceedings

Paper mills are a growing threat to the integrity of science, yet their penetration in conference proceedings remains underexplored despite conferences being more important than journals in some scien...

arxiv.org

From journal editing experience this tracks. In 2022 our journal got ~300 submissions. In 2025? 1300. So far this year? 500 and we're not even out of April yet. The AI trickle is now a stream, yet T&F are *still* seeking new big tech partners to flog papers for AI training rather than *helping us*

Margot Finn@eicathomefinn.bsky.social · 3mo ago

'Annual reports for Elsevier and Springer Nature show publications were up 10 per cent and 11 per cent, respectively, with submissions up 20 per cent and 34 per cent.... [for Wiley] publications were up by 11 per cent and submissions by 26 per cent. 1/3

#statsmeme Greeks & shields

If you are doing any #quarto slidecrafting, i have been trying really hard to keep this page updated with everything I have done emilhvitfeldt.com/project/slid...