L von Mises sobre matemáticas en economía: "en definitiva no es más que un pobre pasatiempo que no proporciona ilustración alguna" y "En el terreno cataláctico adviértase, por doquier, los perniciosos efectos del análisis matemático" (La Acción Humana, 1949/1980, pags 388 y 535)
The oddest intellectual position I have encountered over the last decade is the refusal of so many to believe that the current version of Pangram is incredibly good at avoiding false positives (i.e., flagging a human-written text of some length as AI-generated). It is not terribly hard to trick Pangram into false negatives (i.e., classifying an AI-generated text of some length as human-written) with “humanizing” tools and some careful editing. But the other way around? Nearly impossible.
For two weeks, I have been running experiments in many different situations for a new paper I will circulate tomorrow (and because, as an editor, I care a lot about the answer). Not in a single case did a human-written text of reasonable length score above 10% for AI generation. And, yes, many the texts I used (including mine!) are from ESL writers.
Frankly, the counterarguments I have read over the last week here on X (and the papers linked to “back them up”) are childish in the extreme.
If you think that false positives at Pangram are, as of November 16, 2026, a concern, you are lying to yourself.
There are quite a few people on X who, in the face of stunning technical achievements, belittle the achiever. Rockets caught with chopsticks? "That's just a guy who can get money from Wall St." AI solving conjectures? "It's just a stochastic parrot". Deploying computing at unheard-of scale? "It must be stopped". The modern financial system owes a lot to quants. Well, "they are misallocated resources."
These criticisms would be slightly more credible if the random X guys were nor professional advocates of their own self-serving interests. More than that: human unreasonableness in the face of impossible odds is one of the few things that make progress possible. And progress is a *very good thing*. Irrespectively of their very human flaws, I think we should celebrate the ones who take the risks and bet their entire lives on an idea. At least, I do. And you can learn a lot about someone by seeing how they react to unreasonable humans.
This rare rant was inspired by the following post by @paulg .
claude trace: The bridge works, and the chariot has flown by night; next I'll light the lantern.
gemini trace: All necessary data will be identified and utilized to ensure comprehensive consistency
gpt trace: give codebase me give eat codebase me eat codebase give me eat codebase give me you
This is a very textbook phenomenon in the market called “rotation out of all the names you specifically own and into names you don’t”. This happens when all the market participants have a meeting you aren’t invited to and decide they hate you.
The real AI psychosis is thinking "this is just another tool" or "it's just a stochastic parrot" and "humans still need to review every line". You gotta snap out of it.
You know what I'd really like?
Tell me.
An RBC model.
…We can do an RBC model.
A real RBC model. No financial frictions, no Bayesian estimation, no sticky-price bullshit.
I'll make you a very good RBC model. Log-linearized, calibrated, a productivity shock, and everything
This effect doesn’t seem to be due to low effort. One analyst spent 40+ minutes reading the materials and writing up their own analysis plan.
The agent built it, but then argued for a different approach on two decisions. The analyst replied: "Okay, let use you recomendation"
Holy shit. I just got pwned. iPhone suddenly crashed, rebooted, locked. Checked router stats. Total exfiltration ongoing. Panicked, tried to shut off network. Router also locked. Unplugged router. Didn't matter. Router harvesting zero-point energy. As a last-ditch resort, tried to smash my iPhone. All the atoms in the hammer quantum-phased right through. Hackers got everything, all of it. All my penis pictures.
Excited to share our new working paper: how much did the US standard of living really rise over the twentieth century — and when? With @bhwittenbrink
Using 5.1 million Sears catalog listings, we build a quality-adjusted price index for consumer goods, 1900–1990. Two headline findings: growth was much larger than official statistics imply, and its peak came before WWII, not after. (1/9)
El 30 de octubre conversaremos sobre instituciones monetarias. estabilidad y desafíos para Argentina.
Un gusto participar de este encuentro, moderando la conversación que mantendremos con Vladimir Werning, Vicepresidente del BCRA.
¡Los esperamos en UdeSA!
AI is increasingly used to find errors in research. But which ones actually matter and deserve our attention?
Our new paper with @ellliottt and @sergallet maps the logic of economic models into claim graphs to assess how much an error changes their conclusions.
Very happy to share that our paper with Rafael Di Tella and Ramiro Gálvez (@RamiroHGalvez) has recently landed in JEBO!
In this paper, we study whether varying access to Twitter/X during a presidential debate affects political polarization.