New blog post: LLMs are going to summarize your papers whether you like it or not โ and they're going to drop your limitations every time.
Here's a proposal for how to fix that (and help with the reproducibility crisis at the same time!):
paulgp.com/2026/03/10/l...
Yes, great example.
Excited to share a new @nber.org paper w/Blonz/Hossain/Mulder/Weill where we show credit scores impact homeowners insurance premiums as much as disaster risk.
Hereโs what happened in WA when they banned credit scoring:
I was able to do this with python code that Claude code helped me write, but so help me Wiley did everything in their power to stop me
there's so much bad in the world but there's also people who lift up and carry an elderly bat around every day so he can pretend he's flying again, and that's the part of the world I think is worth fighting for
LLMs are finally good enough at latex that codex helped me port over my gigantic (150+pg 2col) grad school methods notes into an obsidian vault that makes upkeep considerably easier.
Read, refer relevant sections to your favourite clanker, enjoy!
apoorvalal.github.io/lalgorithms/...
New paper in press at Biometrics by PhD Candidate Esteban Fernรกndez-Morales
1) Develops Bayesian spike & slab and horseshoe models for causal inference under spatial spillover
2) Analyzes Philly's 2017 beverage tax accounting for cross-border shopping
arxiv.org/pdf/2501.08231
Back in December, I presented at a "Fast Computing" workshop hosted by the Bank of Portugal. The videos + materials are now all available online. github.com/BPLIM/Worksh...
(Also including cool talks by @s3alfisc.bsky.social, @sebkrantz.bsky.social, and others.)
As someone who routinely mocks permabullish clickbait oil forecasts, I want to be exceptionally clear:
Crude WILL go to $200/bbl, en route higher, unless traffic through the Strait resumes.
Not clickbait, but rather brutal physics and necessary economic incentives.
it's like these people didn't have group chats smh
I have to admit that Hormuz is a very clever design that keeps things interesting. Kudos to Slartibartfast.
You can't convince me that it was an accident.
I gave Zoรซ Kravitz & Penn Badgley a lunch recommendation in Paris (they asked) & then a bunch of people around us asked to take a photo with them & my mom didnโt know who they were & misunderstood the social situation & so asked if those people also wanted to take a photo with me & my now wife.
NY is proposing to "Impose liability for damages caused by a chatbot impersonating certain licensed professionals." nysenate.gov/legislation/... How does a chatbot trick you into thinking its a doctor? Senators: If you forgot you were conversing with AI, you need a doctor.
So true.
Hereโs a full draft of the upcoming second edition of my โData Visualization: A Practical Introductionโ: socviz.co
Conversation on NPR right now about how people don't have access to live music anymore because ticket price are too high and shows are always sold out
LOL amazing
things i use a lot more often in our agentic age
- uv init + uv add + uv sync everything
- quarto markdown instead of jupyter - hidden state is especially cursed now
- interactive viz libraries that generate tooltips
- tmux splits with different agent instances playing generator-discriminator
Maybe so, maybe not. We'll see
ๅก็ฟๅคฑ้ฆฌ๏ผ็็ฅ้็ฆ
Also this. Buddy Guy is a legend www.youtube.com/watch?v=m5Xx...
๐ฅDouble celebrations this week๐ฅ
Our new JDE publication and the notification of my promotion to FULL professor of economics!
Updating my CV ๐
๐คฉ๐๐๐
Believe in your work. Stop ending papers with โMore research is neededโ and start concluding with โNo more research on this topic is needed.โ
In early 2024, researchers were already heavily using AI for work - Survey of 816 verified authors via Semantic Scholar - 81% of researchers reported using LLMs in their workflow - Top uses: information seeking & editing - Rare for data tasks: 69๎73% never use LLMs for data cleaning or generation
The measurement problem LLM content has risen sharply in both review and non-review papers. Review papers do have a higher prevalence rate. But non-review LLM papers outnumber review papers ๎ฃ6x. CS.CY ๎Computers & Society) faces potential 50% cuts compared to CS.CV (Computer Vision) would only face 3%
Interdisciplinary researchers โ who move between cultures and write in the โborderlandsหฎ โ are experts at adapting their writing. LLMs currently are not.
Private information can appear in unlikely prompts
I gave a short talk at Cornell yesterday on my science-of-science work investigating how AI is being used by researchers and how we should go about crafting policies in response.
Blanket policies are hard, privacy is important, we need more measurement.
Slides: drive.google.com/file/d/1gNTK...
This is fascinating
Are AI models effective collaborators, or mere assistants awaiting your next command? (Preprint: arxiv.org/abs/2602.24188)
To find out, we make AI collaborate with itself, in private information games: tasks that require sharing private information, like this chess board ordering task.
I hadnโt seen it until @cohodes.bsky.social shared it with me! Definitely worth comparing.. Iโll try to do that
Toddler: *gestures at 200m-year-old crustacean fossils at Natural History Museum* How did they die?
Me: I donโt know. Maybe there was some kind of disaster, or they got really oldโ
Toddler: *stares at me, tears welling up* they died because they got old?
