Claude Code and I wrote a very sharp looking cli (doi2pdf) that works approximately 1/20 times from my desktop at work. Including failing on several of my own papers...
08.03.2026 04:12
👍 0
🔁 0
💬 0
📌 0
Claude Code and I wrote a very sharp looking cli (doi2pdf) that works approximately 1/20 times from my desktop at work. Including failing on several of my own papers...
So you ported it to typst?
When the vibe is acute political instability, social fragmentation, and looming fascism, choose cb2
An "AI Assistance Disclosure Generator"? But how do I disclose that I used AI to generate my disclosure?
Shout out to the NIMBY neighbors who managed to kill the original proposal because they felt their aesthetic prejudices, and access to street parking, were more important than having affordable housing. Hope they enjoy the resulting even-taller building!
Economists have solved this problem by being awkward af and slightly unfriendly with everyone
We live a block away from Newton North. Based on the quality of driving around the high school, the fines from red light cameras and school bus cameras could fund a nice chunk of our schools budget.
And no money down!
The timeline for a state that "cares" about protecting kids taking school buses from idiots in cars
Jan 2025: Gov. Healey signed a bill allowing cities and towns to install bus cameras
Feb 2026: Chicopee is the first town to deploy
www.wbur.org/news/2026/02...
I guess there's something for NIMBYs who don't understand counterfactuals in there too www.boston.com/news/local-n...
NIMBYs who don't understand that housing markets are regional: "sEe, BuIlDiNg HoUsInG dOeSn’T wOrK"
Every couple of months or so, someone puts out a piece of bad social science meant to undermine the case for more housing. For Roosevelt Institute, I wrote a blog post responding to the two most recent specimens. rooseveltinstitute.org/blog/there-i...
Would love to see how different chatgpt makes the graph if you prompt with "make it for an econ journal" vs poli sci vs sociology vs etc... Which discipline gets more stata default aesthetics
Perfect. Graph makes me think we're safe a few more weeks from these things taking all of our jobs
Did chatGPT decide to use labels on the points rather than just using country codes as the points?
Congratulations DHS Secretary Stephen miller on continuing to be DHS Secretary
Blood *lead* levels? Just aggregate data at those race x age cuts? Sampson and Winters have a few review chapters that might point the right direction?
@billingsecon.bsky.social thoughts?
Consistently, in survey after survey, housing availability and affordability is THE number one issue facing Massachusetts voters.
A ‘status quo’ approach to housing from any elected official, of any political party, should be considered malpractice today.
Good post by @joshgans.bsky.social on why the scientific publishing system will not collapse from the rise of AI-supported research. open.substack.com/pub/joshuaga...
Love and hate:
"The use of those AI tools will be expected and the level of analysis will be something quite different from our eeking out of little burrows to find nuggets of identification."
Devil's advocate: it makes refereeing at top5s both easier and harder. Easy to reject because something isn't totally new because some other previous top5 already exhausted the topic. Harder because it is a lot of appendix reading...
The fact that anyone got arrested and convicted of anything (irl or in movies) before the 1970s is astonishing. I know, the past is another country and all, but just wild
There is/was a tradition at Wesleyan econ seminars to introduce speakers as trained at [PhD] and educated at [Undergrad].
Easy to be proud of where you were educated when the university prez writes like this
Plus post kids, I hate to travel much
I like in person verification a lot, but the paper mill factories might just slop up the submission portal and plan to deal with the presentation later.
Arguably why job talks are such high stakes. If we think you don't get your paper 1000%, you are dead.
Tbf I also had a slide titles "Robustness Checks" with a /pause and then a Coming Soon banner...
But I love proof of work as a talk concept
Economists and other social scientists aspire to our mental model of engineers and physicists. No time for the messy (aka real) versions
I like it. You're probably right that there are some flaws and it probably just increases the value of prepub stamps like the NBER WP Series (as an insider, I'm not complaining)
But also the main way I read or encounter new papers is by refereeing them. So idk what a good paper discovery process is
I mean most journals still literally print things, rather than just posting pdfs online
