Thanks to @jabaluck.bsky.social from @yalesom.bsky.social for coming on the Make It Make Sense to explain why Trump's Rx price plan is a bad idea. Find it wherever you get your podcasts!
There are three types of social scientists: Pornographers, plumbers, and social justice warriors 🧵
All economists should read AI 2027. Slow adoption and diffusion is a correct prediction for all past transformative technologies. But AI is different. The essential difference is the capacity for self-improvement w/ superhuman software engineering.
It remains a civilizational failing that you can't challenge people like @oren_cass and Peter Navarro to proper intellectual duels to reveal them as the complete charlatans they are.
Replacing the H1B lottery with something based on wages makes sense, but we need exceptions for brilliant young people who often are underpaid due to career stage dynamics (in econ lit: "career concerns"). If you are under 30 and score high enough on the GRE, we should let you in.
My take is "Foucault's Pendulum." Trump believes in the narrative he and his followers create for themselves. He believes tariffs are good because that's what he said (for reasons beyond his recollection). His followers built rationalizations. He can't change course.
If you don’t think economics nomenclature is important, think about how GDP growth would be 1 percentage point higher this year if “trade deficits” were called “trade differences.”
Just heard from a family friend who is literally a prince from Denmark, first-rate fencer, skilled theater director, and can talk to ghosts -- rejected at all the Ivies, waitlisted at Stanford, accepted at Duke (they should rename it "Commoner"). Ridiculous!
An underdiscussed alternative to pre-registration for observational data analysis: research assistants should keep a log file for all data analysis they do, and this file should be posted publicly along with any published papers.
There is a common cognitive error I call, "The Bridge Builder's Fallacy": the idea that you can easily tell from observation whether something works. If you build a bridge and it doesn't collapse, it's a good bridge. Many successful people believe everything works this way.
Good AI writing (coming soon) shows how much authors draw on literature rather than life. It proves right "The Anxiety of Influence": writers want to express dimensions of experience, but instead they are imitating and recombining writing they have seen before.
Academics (including myself) infrequently update their personal websites after they get tenure. What happens when you win the Nobel Prize? This is a current screenshot of Paul Milgrom's personal website:
Jason Abaluck and I are organizing the Cowles Labor and Public Economics Conference at Yale, June 2-3. Submit your papers by March 24: cowles.yale.edu/conferences/... @jabaluck.bsky.social
Cowles Summer Conference Paper Submission Form
cowles.yale.edu
Sure I'm a few minutes late, but do you understand that in order to get here I needed to endure several seconds of being uncomfortably cold, from when I stepped out the shower until my body was sufficiently dry?
In the interim before we have extremely dexterous robots, many tasks can be automated by low-skill humans receiving elaborate step-by-step instructions from AI via headphones. What economic consequences will this have?
Doctors should stop prescribing antibiotics to treat bacterial infections. All the existing evidence showing health benefits has serious shortcomings, and the data show that antiobiotics massively increase mortality. (Read to the end before commenting)
My view of drugs & med devices is that you want: a) Easy to try unproven stuff / easy access for patients b) Strong incentives for firms to evaluate what works c) Penalties for firms that role out stuff that doesn't work
We may be only a few years away from broadly available, brutally honest feedback about your life from an entity that knows better: "Your work is useless, change jobs" or "You will be happier with a different partner". A lot of people might start living very different lives.
I suspect the world is about to become a lot more explicitly transactional with the rise of AI agents. Money will need to be used to resolve competing demands on your time that are currently resolved by norms.
A parameter that shifts over time is the relationship between wealth and skill of various kinds. This seems understudied by economists: different institutions reward different skillsets, and we want to reward skillsets correlated with pro-social preferences.
Anecdotally I would have guessed economists had shifted much more heavily to BlueSky but it seems like Twitter still has almost 10x the # of economists (might have shifted somewhat since the survey was done).
The question of when a generalized car can get out of an n-dimensional parking lot feels like something you would read in a Quanta article trying to explain the next Fields Medal in non-technical terms.
Going to be hard to disentangle shifts in political opinion due to Trump taking office from the impacts of 50 million people no longer getting the majority of their political information from TikTok.
People overestimate the degree to which research is about new ideas and underestimate the degree to which research is about correctly identifying which ideas are important and deserve more attention and why they deserve it.
When non-researchers read studies, they often say things like, "GDP doesn't include HUMAN DIGNITY." This is not a constructive point if you cannot define or measure dignity. The relevant question is -- what is the best available thing we can *measure* to evaluate any policy?
I suspect we are about to enter an interim period where AI exceeds human performance on many cognitive tasks, but this is not common knowledge, and so most people and institutions act like this is not generally the case.
A lot of AI-skeptics say things like: sure, AI can solve math problems, but that doesn't mean it can manage a business. That's true -- but we also don't know that o3 can't do this! We haven't done the labor market turing test.
A book review of Tomorrow and Tomorrow and Tomorrow by Gabrielle Zevin. This is probably the first book review you will read that has absolutely no spoilers and that you will appreciate equally whether you have read the book or not (at least if you read to the end).
