I’m now formally 4 weeks by my eight weeks in Europe. This final week, I spoke on the European Fee in ispra which is a small space outdoors of Milan. It was fantastic. I met new individuals who have been additionally fantastic hosts. I hope to come back again.
However I stayed in Arona, a small city on Lake Maggiore. Ask me why. As a result of I had Claude select the place I stayed primarily based on the correspondence and my descriptions of the place I’d be. Now ask me if I used to be on the proper facet or the unsuitable facet of the lake to simply journey each morning to the EC. That’s proper I’m on the unsuitable facet of the lake.
However there was a silver lining. And the silver lining was it’s completely lovely right here within the sleepy city of Arona. Lake Maggiore is among the lakes on this space like Lake Como, which is the subsequent lake over from the place I’m. The lakes are gorgeous and I fell in love with the place Claude obtained me, together with a room on the Lodge Giardino, which finest I can inform is a resort that’s the absolute closest to the water you will get, and the room Claude obtained me had a balcony and it too was the closest to the water. So I ended up extending my keep right here and go away Sunday for Pisa, after which after Pisa, I’m coming again right here for 3 nights to loosen up.
However the workshops have gone moderately properly. I’ve taught steady remedies 4 instances now and that’s been in style. The best way I’ve been instructing all the things is like our JEL, the place we rooted diff-in-diff within the 2×2 and emphasised forwards engineering in the direction of mixture causal parameters that signify attention-grabbing and related analysis questions that you’ve.
However tomorrow I go away for Pisa — I’ll take a shuttle to Milan prepare station, then a prepare to Pisa, to the Sant’Anna Faculty of Superior Research. Right here’s that schedule — it’s best to come! Me and Daniel Lewis are instructing.
However sufficient about me. Let’s see if we are able to’t shut a few of these tabs why don’t we! I noticed Ethan Mollick publish this to LinkedIn, displaying rising employee productiveness at Anthropic on account of Claude code. It’s from this text by Anthropic on AI constructing myself.
Talking of Ethan Mollick, he has a brand new e-book popping out known as Co-Existence, a sequel to his bestseller, Co-Intelligence, which has nearly 3,600 opinions at Amazon and a median rating of 4.5.
NeurIPS has a brand new coverage simply enacted requiring that the manuscript be “largely” written by you, the human, and never the chatbot.
Jennifer Leight confirmed me this text from Plos-one the place researchers did an RCT to see if people had any biases towards AI writing. I need to maintain seeing extra experiments on this. I believe we have to study extra about repugnance on this space.
How is Starlink doing as of late and over the past 4 years. It appears like typical Elon Musk — in absolute efficiency, it’s thoughts boggling how briskly and dependable it’s gotten. However how does it examine to the guarantees? It’s not fairly on the stage of what he’s now promising though it’s precise efficiency is unreal. Form of attention-grabbing that it’s grow to be so secure and dependable and freaking quick.
I obtained this within the mail from Harvard. It’s about twenty school tasks awarded grants for doing impactful tasks for the general public good by the Frontiers of Innovation for Social Affect Fund.
This new marriage market research obtained a protracted factor too within the every day digest I get from the college. It’s an attention-grabbing sounding article and I need to learn it. Principally, with males decreasing school enrollment relative to girls, there’s an efficient scarcity of faculty educated males within the marriage marketplace for heterosexuals. School educated girls are extra probably now to match with larger incomes non-college educated males. However this has put a squeeze within the marriage marketplace for non-college educated girls. Right here’s a protracted dialogue with the authors and a hyperlink to the paper.
Harvard additionally has a dialogue on the worldwide “loneliness epidemic” in the event you’re desirous to examine that. That’s a good suggestion for dissertation I believe, as there’s been a mixture of things which can be thought to have contributed to it like Covid, earn a living from home, political polarization, numerous commonplace explanations about telephones and social media, however I believe causal proof is missing.
Brodeur, mostly related to replication and discovering p-hacking in economics, additionally has a paper on the origins of the intercourse business in Thailand in JEEA. I believe this will have been his job market paper? Undecided. A brand new pal on the European Fee talked about it to me.
Martin Huber on when adjustments in remedies determine therapy results.
Hackers for Meta chatbots to present them Obamas outdated White Home account on instagram by some good quaint intelligent social engineering blended with exploiting safety vulnerabilities. I ponder what Mythos goes to do.
The Brenton Lodge in Newport Rhode Island.
A brand new research from the Fed thinks the blame on AI for hurting latest school grads job prospects could possibly be misplaced; it might be earn a living from home. The NPR article is an effective intro to the paper.
A Disney govt proclaims that his chatbot is his son. I learn it twice and nonetheless didn’t perceive what he meant. However that is type of what I meant yesterday about AI and repugnance — regardless of how insightful and wonderful you assume your convos with AI is, I don’t assume anybody else will agree with you. In truth they’ll most likely assume the precise reverse. Nobody cares about your desires except they’re in them.
Kevin Tune at UBC and his attention-grabbing work. Right here’s considered one of his papers tracing out the connections between synth and diff-in-diff that I hope to learn.
Oh by the best way. Right here’s the code for replicating our Medicaid ends in the JEL I discussed earlier. Stata and R.
A meteor landed within the ocean outdoors Boston and it was apparently fairly loud.
Housing costs and the impact on the wedding of youthful folks in China.
The earlier January financial index from Anthropic.
Should you haven’t been following the tales with Mythos, it’s a brand new mannequin from Anthropic that has emergent properties of being glorious at discovering safety vulnerabilities in giant code banks. Firefox story right here. Anyway, they shared it with some companions, like giant monetary companies, tech companies, and so forth, to assist them get forward of it earlier than it went public. They’ve only recently begun permitting monetary companions to report on it. The report can be out July sixth apparently. Right here’s a shorter dialogue of it and the “Challenge Glassing” element it’s part of.
But it surely appears like we must always all change our passwords, and positively change our safety questions, earlier than Mythos comes out.
Try this cool pizza oven. Most likely bread too. I noticed it on Kottke. It’s offered out for now.
My shiny app for illustrating the decomposition of the TWFE estimator for steady therapy in diff-in-diff into weighted averages of unit-specific doses (primarily based on Frisch-Waugh-Lovell).
Artificial management utilizing excessive dimensional information and LASSO by Hollingsworth and Wing.
Stability in matching markets with peer results by Anna Bykhovskaya.
Guido Imbens in an outdated interview I did with him and the way folks responded to his and Josh Angrist’s LATE theorem paper.
This new paper by Brantly Callaway, Derek Dyal, Pedro Sant’Anna, and Emmanuel Tsyawo tries to transcend parallel traits and reliance on panel information by figuring out teams of models much like the therapy group within the pre-treatment interval after which utilizing them for estimation of the ATT. It apparently connects to quite a lot of estimators together with diff-in-diff, changes-in-changes, and extra.
Greater than talent, I believe checklists for workflows are most likely the route I’m going to be going for utilizing Claude Code for analysis. Right here’s one thing about that.
Elsevier’s AI coverage for its journals.
Some proof that faculty bans of good telephones has useful impacts from Matt Gentzkow and others.
Carolina Cattaneo and Brant Callaway’s paper “Diff-in-Diff when Parallel Developments Holds Conditional on Covariates” has been, I believe, conditionally accepted at Restat. Congratulations! As I’ve mentioned about ten million instances on this summer time tour in Europe, covariates are a giant deal in diff-in-diff!
There’s a few new entries within the “omitted variable bias” papers. First, there’s a brand new AER entitled “The Impact of Omitted Variables on the Signal of Regression Coefficients” by Masten and Poirer. After which there’s one other one on diff-in-diff by Juejue Wang, Pedro Sant’Anna, Carloa Cinelli, and Victor Chernozhukov. All of those papers are within the wealthy historical past of sensitivity evaluation papers began by Altonji, Elder and Taber by Emily Oster, Carlos Cinelli and Chad Hazlett, the brand new Restat Carlos once more, Whitney Newey, Victor Chernozhukov, Amit Sharma and Vasilis Syrgkanis.
“Medical Innovation and Racial Well being Disparities: Proof from a Breakthrough Therapy” by Mark Anderson, Kerwin Charles and Dan Rees.
I’m re-reading “What the Residing Do: Poems” by Marie Howe.
“Discovering That means within the Second Half of Life” has good opinions. I’m sadly within the second third of life, and even it I solely began the second third final yr on my fiftieth birthday. However I could verify this e-book out simply to see.
Additionally once more in the event you didn’t see this, you can be banned from Arxhiv in case your paper had hallucinated cites. Presumably your complete coauthorship crew can be.
5 causes to go to the city of Monza, Italy.
I discussed this the opposite day, however right here is somebody debating which is healthier — Lake Como or Lake Maggiore (the place I’m proper now). I nearly was going to run over to Lake Como for the tail finish of my keep after the European Fee however determined to remain and get work accomplished right here.
Nicely that may be a lot! Want me luck as I end my work at present, movie an episode of the podcast, and do some laundry. Tomorrow I head for Pisa which I haven’t visited since school. Trying ahead to it!
