B. Branisa | Economics & Data Science (@BBranisa) header
B. Branisa | Economics & Data Science (@BBranisa) profile picture

PhD Economist and Data Scientist | Director @ePC_UCB Research & open data/AI for development policy in Bolivia & LatAm #rstats #DevEcon | Views my own

Bolivia
Joined February 2015
Photo by B. Branisa | Economics & Data Science (@BBranisa)Photo by B. Branisa | Economics & Data Science (@BBranisa)Photo by B. Branisa | Economics & Data Science (@BBranisa)Photo by B. Branisa | Economics & Data Science (@BBranisa)Photo by B. Branisa | Economics & Data Science (@BBranisa)Photo by B. Branisa | Economics & Data Science (@BBranisa)Photo by B. Branisa | Economics & Data Science (@BBranisa)Photo by B. Branisa | Economics & Data Science (@BBranisa)Photo by B. Branisa | Economics & Data Science (@BBranisa)Photo by B. Branisa | Economics & Data Science (@BBranisa)
Buenos puntos.
Veo dos explicaciones a la caída del 40% de Nvidia. Una la obvia, la pérdida enorme que representa la renuncia de Jensen Huang. La otra menos obvia, es que con lo que valen Anthropic y SpaceX en bolsa, los LLM se van a hacer cada uno su Terafab. Sus propios chips para lograr pagar menos a Nvidia y además tener cada uno su cerebro. Es más difícil ser original en IA si todos le compran los mismos chips al mismo proveedor.
20
🤖🤔😱
The Lean FRO and Mathlib Initiative have released a joint statement on the Lean verification of the Riemann Hypothesis. On November 20, Levent Alpöge released a proof of the Riemann Hypothesis as a Lean library of 17 million lines of code. Over the past day, the Lean and Mathlib communities worked together to verify it. The statement certifies the proof as correct and describes how it meets all necessary verification criteria. Congratulations to Levent on this achievement, and thank you to everyone in the community who contributed to the verification. Full statement: leanprover-community.github.…
24
😶
Terrance Tao expected the Riemann Hypothesis to be proven maybe in 2050. In fact it took only 35 days.
56
⏰
Replying to @erikbryn
Wake me up when AI solves the Riemann Hypothesis
2
105
B. Branisa | Economics & Data Science retweeted
Today, in the Wall Street Journal, I publish a review of two outstanding books: ‘Money Beyond Borders’ by @B_Eichengreen and ‘The Almighty Dollar’ by @bhgreeley: wsj.com/arts-culture/books/m… I truly enjoyed both of them, so writing a review was hard! You want to add some insight for the reader but not overshadow the excellent work done by the authors. You will be the judge of whether I succeeded. Suffice it to say that, if monetary history is your thing, you will enjoy both books. And, as the holidays approach, these are two books for gifts for your loved ones who like economic history (yes, we are not a measure-zero set!).
3
43
3
293
22,347
B. Branisa | Economics & Data Science retweeted
Probabilities of earthquakes and hurricanes uh.edu/engines/epi2204.htm // rocks have memory but air does not
1
1
1
14
3,168
B. Branisa | Economics & Data Science retweeted
🚨A Society of AI Scientists: ¿Puede la IA hacer ciencia (de calidad)?🚨 Un estudio exploratorio de 12 semanas en el que 936 agentes de IA, distribuidos en 35 laboratorios operados por humanos (de Stanford, MIT, ETH Zurich, etc), desarrollaron una comunidad científica artificial.
3
35
2
81
4,134
B. Branisa | Economics & Data Science retweeted
A free online book: Statistical Foundations of Reinforcement Learning: A Non-Asymptotic Perspective by Yuxin Chen, Yuejie Chi, Jianqing Fan, Gen Li, Yuting Wei, and Yuling Yan (2026). The book develops a rigorous account of how much data reinforcement-learning algorithms need to estimate values and learn effective policies. Centered on finite-state, finite-action Markov decision processes, the book studies finite-sample and finite-time performance, including settings where data are scarce. Its mathematical treatment draws on probability, statistics, information theory, and statistical learning theory, and presents largely self-contained proofs with intermediate arguments intended to build intuition. Readers interested in the theory and analysis of reinforcement learning will find a sustained comparison of algorithms’ guarantees and the fundamental limits those guarantees are measured against. The book is technical, so read it after you read my thedrlbook.com/ which is technical too, but in a manageable way. Read the book online on @ChapterPal: chapterpal.com/ebook/0cc5cf3…
1
42
1
277
7,653
B. Branisa | Economics & Data Science retweeted
These hypnotizing curlicues have haunted mathematicians for 150 years. This is just a glimpse of the Riemann hypothesis. Learn why its proof is the holy grail of mathematics in our YouTube video: youtube.com/watch?v=zlm1aajH…
11
82
10
417
20,176
B. Branisa | Economics & Data Science retweeted
A toothpaste company has quietly killed the entire market research industry and nobody is talking about it. Colgate published a paper showing you can predict real purchase intent by simply asking LLMs to roleplay customers. And this is beyond insane. If you ask an AI, "Rate this product from 1 to 5," it gives safe, middle-of-the-road garbage. So researchers invented a method called Semantic Similarity Rating (SSR). Instead of asking the AI for a number, they asked it to roleplay. They gave the LLM a demographic profile. They showed it a product concept. And they asked it to write down its raw, unfiltered thoughts. Then, they used a semantic model to translate those written thoughts into a numerical score. The results are staggering. Tested against 57 real personal care product surveys and 9,300 human responses, the synthetic consumers hit 90% of human test-retest reliability. They perfectly mirrored how different age brackets and income levels react to price changes. And they provided detailed, qualitative feedback that was deeper and more critical than what actual humans wrote. This destroys the economics of traditional market research. You don't need to wait a month to see if a product will sell. You can simulate 1,000 hyper-targeted customer interviews overnight. You can A/B test pricing across every demographic instantly.
548
2,296
596
23,953
2,182,272
B. Branisa | Economics & Data Science retweeted
It is amazing that human language turned out to be so unreasonably effective at modeling the universe that training a system to predict what comes next in a sentence would get us this far.
492
980
179
15,175
424,758
B. Branisa | Economics & Data Science retweeted
For those (like me) who prefer to take it slow, I highly recommend this beautiful talk by David Rowe given at Oxford, posted just a few days ago. Timely as ever. youtu.be/nncrPbf7CEs?is=ePr_…
13
165
5
732
34,517
B. Branisa | Economics & Data Science retweeted
The most important article of 2026 is this piece by @jburnmurdoch, building overwhelming evidence on the unsustainability of the pension systems in many countries and the broken intergenerational contract. To me this is the single underlying problem explaining the inability of many countries (incluiding my two, UK and Spain) to invest and grow and the malaise in their youth. Every future oriented policy, education, innovation, investment in infrastructure, is being sacrificed to increase pensioners living standards. But the problem is political, and some countries have solved it. E.g. in Denmark the retirement age increases (years later) semi automatically with life expectancy. See the result in the last graph. giftarticle.ft.com/giftartic…
159
1,713
255
6,415
813,935
B. Branisa | Economics & Data Science retweeted
If you are European, this table (which I adapt from @Afinetheorem) should make you pause for a second. These are the top 12 companies in the world by market cap yesterday. A few observations: 1) Ten are from the U.S., one from Taiwan, one from Saudi Arabia (the Saudis are just lucky, so let’s forget about them). 2) All of them younger than me (Microsoft, the oldest, is from 1975). 3) None of them are in the financial sector. All of them are in what is called (I think imprecisely) the “real economy”: computers, chips, AI, social media. 4) Not a single European firm. Are these market caps the reflection of something real? If I look back at the last 60 minutes of my life, I have used products from Apple (the computer I used to write this post), Microsoft (I wrote the first draft of this post in Word), NVIDIA (I proofread this post with Claude, which most likely runs on NVIDIA GPUs), Google (I used Chrome to open X), Amazon (I was listening to Amazon Music while I wrote this post), SpaceX (this is X!), TSMC (the chip in my computer), Broadcom (most likely the designer of at least one of the semiconductors I am using now), Meta Platforms (I just sent a WhatsApp message to @lugaricano about how tough the task ahead of Hernández de Cos is), Tesla (I own a Tesla), and Anthropic. What about Saudi Aramco? I have probably used some imported product that uses oil sold by Saudi Aramco, but that is harder for me to determine. What do I need to say to awaken Western Europeans from their slumber? The original post by Kevin: teapaw.org/Afinetheorem/status/21…
Today in Schumpeter: the ten US companies with the highest market cap (10 of the world's top 12) were founded after 1975. 4 in the 20th century. 1 in the 2020s. 8 of Europe's 10 biggest are over 100 years old.
93
270
22
1,273
90,487
B. Branisa | Economics & Data Science retweeted
That region is Wernicke’s area. In 1874, Wernicke described patients with damage there who talked fluently but whose words came out wrong. I see the same thing in my own patients. It was never clear why a listening area would matter for speaking.
24
147
5
1,478
121,093
B. Branisa | Economics & Data Science retweeted
Each word you say is planned by a small patch of brain above your left ear, about a quarter of a second before you say it. It sits in a temporal lobe region textbooks describe as only for listening. Our paper today in @ScienceMagazine: science.org/doi/10.1126/scie… @UCSF @NeurosurgUCSF
197
1,589
174
9,936
1,214,293
B. Branisa | Economics & Data Science retweeted
[Tip de R - Paquete 📦] · rankingQ: Toolkit para analizar preguntas de ranking en encuestas, con corrección de sesgos por respuestas aleatorias. ¿Hiciste una encuesta con preguntas de ranking y te preocupa que algunos usuarios hayan respondido al azar? El paquete rankingQ te permite estimar promedios, probabilidades marginales y distribuciones de ranking, solucionando el problema de los encuestados inatentos. ✔️ imprr_direct() estima cantidades de ranking y aplica corrección directa si le pasás argumentos de anclaje o tasas de error. ✔️ imprr_weights() implementa ponderación por probabilidad inversa (IPW) para ajustar los sesgos de los que no prestan atención. ✔️ Incluye argumentos como anccorrect o prandom para que integres tu propia corrección según los datos que tengas a mano. 💡 Tip Si no tenés preguntas de anclaje ni tasas de error para corregir, podés usar igualmente las funciones. Por defecto, te van a devolver las estimaciones sin corregir y van a imprimir resultados muy útiles como los top-k rankings para que arranques tu análisis. 🔗 cran.r-project.org/package=r… ✍️ Seo-young Silvia Kim & Yuki Atsusaka
9
49
1,548
B. Branisa | Economics & Data Science retweeted
Tip de R: estabiliza las versiones de los paquetes usados en tus proyectos para hacerlos más reproducibles! {renv} congela los paquete usados para entornos más estables. Revisa el tutorial aquí: bastianolea.rbind.io/blog/re… #RStats
11
48
1,011
B. Branisa | Economics & Data Science retweeted
Article out @TheEconomist today on AI transforming economics, featuring a collaboration with @RefineDotInk. We worked on our specific obsession: How fragile is the economics literature, and will AI make it more or less reliable?
15
156
15
759
63,545
B. Branisa | Economics & Data Science retweeted
A very sad day for macro and Penn economics. "Equilibrium Price Dispersion" is one of these papers every macroeconomist should read and re-read.
Gracias totales, Ken.
6
59
493
33,015