Feeling thankful that economists don't have to put chartjunkviz in our papers, with or without AI assistance
huggingface.co/papers/2601....
@jamesfeigenbaum.bsky.social
Assistant Professor of Economics at Boston University | Economic History, Labor, Inequality https://jamesfeigenbaum.github.io/
Feeling thankful that economists don't have to put chartjunkviz in our papers, with or without AI assistance
huggingface.co/papers/2601....
Massachusetts isnβt just moving too slowly on housing, itβs becoming a national outlier in its lack of urgency to fix the systemic barriers that made it one of the hardest, most expensive places in America to build homes.
06.02.2026 16:26 β π 33 π 12 π¬ 3 π 2Restrictive immigration quotas in the 1920s reduced intergenerational mobility for US-born white men, and had imprecise but positive effects on intergenerational mobility of Black men, from @jamesfeigenbaum.bsky.social, Yi-Ju Hung, Marco Tabellini, and Monia Tomasella www.nber.org/papers/w34775
06.02.2026 18:03 β π 8 π 6 π¬ 0 π 0BU day at Harvard-BU Economic History Friday. Our student Yicheng Chen presents on the GI Bill and farmers at the workshop and my colleague Martin Fiszbein presents on rural electricity and directed technical change at the seminar
06.02.2026 15:03 β π 5 π 0 π¬ 0 π 0Same vibes
06.02.2026 14:03 β π 6 π 0 π¬ 0 π 0Potential grad student coauthors in shambles rn
06.02.2026 12:35 β π 5 π 0 π¬ 0 π 0This x1000
06.02.2026 12:33 β π 6 π 0 π¬ 0 π 0Fair. But it doesn't need to be robust to... What if the input data format changes or someone renames an upstream file? We're doing social science replication not SWE future proofing here, Claude
06.02.2026 11:41 β π 3 π 0 π¬ 1 π 0Capable but error prone research assistant vs oracle sums up the bsky vs X views of Claude Code for social science divergence really well
06.02.2026 11:11 β π 26 π 2 π¬ 2 π 0It saved that to one final skill md. There's a little yaml front matter to make a proper skill, but you can just point Claude to the file to be sure it reads it in the future
06.02.2026 11:04 β π 3 π 0 π¬ 0 π 0Then I gave another Claude the project specific style guide mds and asked it to interview me in obsessive detail about discrepancies (eg one project was old so my pivot verbs were out of date). Love the poll feature
06.02.2026 11:04 β π 2 π 0 π¬ 1 π 0Ready for tmi?
I started by asking for a style guide in md for each of like 6 recent projects. Told it to just read R files and ignore my coauthors code if any.
Great (@gmcd.bsky.social) and mediocre (me) minds think alike
I didn't even teach it project flow yet, just super simple comment prefs, package stuff, etc. Basically enough to write one R file like me, not a whole directory of them. Maybe not ready to relinquish that much control yet
Maybe this is telling on myself, but LLM written code is constantly checking if files or variables exist, writing intricate fallback plans, etc. Maybe marginally useful in the dataprep phase, but annoying af in analysis step.
I just need analysis code that runs so @lars.vilhuber.com will okay it
Huge win for economic historians who failed to learn cursive in middle school
06.02.2026 02:47 β π 19 π 2 π¬ 0 π 0Probably also a good idea to teach it house style for stata because there's so much less stata than R or python in the training corpus
06.02.2026 02:30 β π 5 π 0 π¬ 1 π 0Spent an hour yesterday trawling through my old replication packages with Claude Code making a skill so it could write R like me. Weird naming conventions, idiosyncratic formatting preferences, etc. Strongly recommend.
06.02.2026 02:30 β π 9 π 0 π¬ 1 π 0TFW you are asked to referee an R&R for which you were supposedly a first round reviewer but you have absolutely no memory of the paper or your report. But one of the reports sure looks like stuff I'd write
05.02.2026 22:04 β π 9 π 1 π¬ 2 π 0My 5-year-old would kill to live in an MBTA maintenance yard, but I'm not sure how many others would want to combine the two
05.02.2026 16:52 β π 4 π 1 π¬ 1 π 0There are only two good uses for former industrial sites in Greater Boston: housing and MBTA infrastructure
05.02.2026 16:52 β π 0 π 1 π¬ 1 π 0"The big triangular chunk of land in the heart of Newton Centre has been a parking lot for so long that itβs hard to remember it didnβt start life that way."
RETVRN
cc John Henry's Red Sox post 2018
05.02.2026 12:36 β π 2 π 1 π¬ 0 π 0Pedestrianize Beacon Street from Charles to Washington (the whole length seen here) as well as Park St
So many streets in downtown Boston have throngs people walking & biking crammed into a few feet of space, so a handful of cars can have lots of room to sit gridlocked & occasionally kill people
tbc Susan Albright is doing the right thing here to charter this (delay it), but the drive to kill this lovely little plaza to save a handful of parking spots in a sea of parking in Newton Centre was supported by 6/8 councilors on the transportation committee
05.02.2026 11:48 β π 5 π 1 π¬ 1 π 0We can't have nice things in Newton, part infinity, because boomer NIMBYs are car-brained and obsessed with parking
www.newtonbeacon.org/not-so-fast-...
New answer for when my students ask "how long ago does something have to be before an economic historian can study it?"
05.02.2026 11:34 β π 14 π 1 π¬ 1 π 0"Sidewalk incompletely shoveled, not passable for a stroller" usually leads to a warning and then a fix on the Newton 311 app, not sure about Boston or elsewhere
05.02.2026 11:27 β π 2 π 1 π¬ 2 π 0New coauthor management system just dropped
05.02.2026 11:03 β π 7 π 2 π¬ 1 π 0Good news, the bat is no longer in the stairs. Does that mean he left or he's hiding in someone's office? Tune in next week, same bat time, same bat channel
30.01.2026 16:50 β π 15 π 1 π¬ 3 π 2