Economist. I'm constantly updating my priors. 🍷🍾 WSET certified. 🌊⚓️ Master Scuba Diver.
Perla del Pacífico, Ecuador
Joined August 2011
- Tweets13K
- Following496
- Followers699
- Likes715
Freddy García retweeted
I wrote about the state of AI, why I’m concerned about the next few years, and the choices we need to make to keep the future in humanity’s hands.
An Alien Mind: openai.com/index/an-alien-mi…
Freddy García retweeted
We will give one banked reset for every day you don't have access to Astra on your paid ChatGPT plan, starting today. Team is moving mountains to give access as fast as we can.
First one will land in ~ 3 hours. There is still time to create your account if you don't have one.
Freddy García retweeted
#OPINIÓN | "La Vía a la Costa exige presencia policial y sanciones al transporte de carga hoy pues la seguridad vial no puede esperar por la carretera Nobol-Progreso", escribe Mónica Cassanello.
Lee su columna completa 👉 expreso.ec/opinion/transport…
Freddy García retweeted
Students are rational forward-looking agents who already account for low future exam grades when deciding to use AI in current homework.
Freddy García retweeted
Interesting chart, but the thing is... it doesn't prove forecasts were suboptimal.
In fact: any mean-reverting process produces a similar figure. To illustrate, here is one. You can play with its parameters yourself!
Play 🎶😉Details and link below.
NY Fed President John Williams today:
"My forecast personally is for inflation to come down in the second half of this year and come down further next year."
This is also Wall Street's inflation forecast.
As this chart shows, Wall Street has had this exact inflation forecast for 64 straight months, and it has been wrong for 64 straight months.
Question for Williams: why is this forecast suddenly going to go right now? What is about to change that is going to make this 0 for 64 forecast now correct?
---
Explaining the chart
The black line is the year-over-year change in core PCE.
The colored lines are median results from roughly 70 Wall Street economists. Each color grouping is a different calendar year. This survey was conducted by Bloomberg.
These economists forecast inflation for the next six quarters.
Prior to COVID in 2020 (the shaded recession area), almost every forecast is ABOVE the black line, meaning every forecast was that in 18 months the inflation rate would be back at 2%. It never did.
Since COVID (now six years ago), every inflation forecast has been BELOW the black line. Every forecast is that in 18 months the inflation rate will decline to 2%. For 64 months now (and counting), it has not come close to 2%, and now Core PCE has been trending higher for the last year and a half.
Isn't this just another case of the basic definition of insanity: forecasting the same thing over and over again, expecting a different result?
Freddy García retweeted
La docencia como cachuelo y la rectoría como dirigencia le han hecho daño a la academia.
Producto de aquello la gente piensa que los académicos deben opinar de la coyuntura en vez de divulgar sus investigaciones (las cuales ya de por sí usualmente contradicen a los políticos).
Freddy García retweeted
Efecto placebo. Claro que hoy todo “parece fluir”: después de meses de cierres, recorridos de 40 minutos, accidentes y un tráfico convertido en pesadilla, cualquier alivio se siente como una maravilla.
Pero la verdadera prueba no es un video en horas sin tráfico. La prueba es en horas pico, cuando vuelvan a coincidir miles de carros. ¿USD 15 millones para una “solución vial” que solo funciona cuando casi no hay tráfico? ¿Dónde gradúan a estos planificadores?
Y lo peor… esto no ataca el problema de raíz. Cuando crezca el parque automotor y sigan creciendo las ciudadelas de Vía a la Costa, volveremos al mismo punto… más congestión, otra obra millonaria y el mismo círculo vicioso. Sin transporte público eficiente, planificación urbana y seguridad vial, esto es apenas un parche carísimo con fecha de caducidad.
‘Uno se demoraba el doble del tiempo, hoy bajó casi a la mitad’: #Guayaquil abre el nuevo paso elevado de la avenida del Bombero para beneficiar a 120.000 vehículos ow.ly/sNlw50Zq8gC
Freddy García retweeted
Con 5 meses de información de actividad económica en 2026, se ve complejo llegar al 2.5% de crecimiento anual esperado por el BCE, FMI y otros multilaterales. Esperemos que no se presenten obstáculos en la ruta. A mayo crecemos 1.9%. Un año atrás 3.9% y dos antes caíamos 1.9%
VAR is all you need
🚨🚨🚨NEW PAPER 🚨🚨🚨
Attention-Based VARs
The Attention Mechanism at the core of Transformers and LLMs has an interpretation as a reduced-rank time-varying VAR in time series.
Tune in today at the NBER Summer institute or follow on YouTube: youtube.com/nbervideos
Freddy García retweeted
Yes. The old idea of documenting market failure before regulation seems to have evaporated among these esteemed economists. Questioning regulatory capacity even more so
How did we get to a place where so many people assume you must regulate anything new if you can imagine any problem coming of it? Our new, mad, norm is preemptive regulation of imaginary potential harms without waiting to see if a) any harms actually emerge, or b) if the market can solve the harms, as they so often do, since no one wants them. Guaranteed stagnation if we allow it to happen.
Freddy García retweeted
Yesterday I came across a paper, “Minimum Viable Scale: Extinction and Escape under Increasing Returns,” by Javier Milei (@JMilei) and Demian Reidel (@dreidel1):
drive.google.com/file/d/1u8i…
I downloaded it because it sits close to something I am working on (and I am actually sympathetic toward the point the authors are trying to make).
As I read, I had the clear sense that a large language model had produced the text: the formatting, the style, the kind of assumptions, the way the derivations were laid out, the papers cited.
So I took the abstract and four paragraphs at random and ran them through Pangram. In all five cases, it returned a 100% probability of AI generation.
Now, I don’t mind economists using LLMs in their research. I do it all the time, for many purposes. Nor do I mind economists whose first language is not English using them to polish their prose. I do that too, and I have been open about it (this very same post has been tested for grammatical accuracy).
But there is a line between “I use LLMs to help with my research” and “I asked an LLM to write a paper and put my name on it.”
I may be wrong, of course, and I will gladly correct this assessment.
Freddy García retweeted
We’re sharing the next major milestone in our non-invasive brain-to-text decoder research: Brain2Qwerty v2.
Building on v1, which was published today in @Nature, Brain2Qwerty v2 is the highest-performing end-to-end pipeline capable of real-time sentence decoding from raw brain signals. It advances beyond character-level performance to decoding words and semantics, enabling accuracy for overall communication.
We believe this research has the potential to make a real difference for the millions of people who suffer from brain lesions or disorders that prevent them from communicating.
🧵👇