Milan Račić

En 2004 la revista médica BMJ hizo un experimento incómodo. Metió errores graves a propósito en tres ensayos clínicos y se los mandó a más de 600 revisores. De media, detectaron el 27% de los errores importantes. Esta semana un grupo de investigadores repitió la prueba con 15 modelos de inteligencia artificial. Encontraron entre el 53% y el 96%. Para descartar que se supieran las respuestas, los probaron también con 58 ensayos publicados este año, y el mejor, Claude Opus 5.5 de Anthropic, llegó al 99%, en minutos. Todavía es un preprint sin revisar y las máquinas también ven errores donde no los hay, pero la distancia es enorme. Lo mismo empieza a pasar en economía. El miércoles Tyler Cowen, economista de la Universidad George Mason, difundió en su blog una auditoría hecha con IA por un lector anónimo sobre 138 trabajos de Gabriel Zucman, el economista francés detrás del impuesto mínimo del 2% a las grandes fortunas, y de su coautor en Berkeley Emmanuel Saez. La auditoría dice haber encontrado 147 problemas capaces de cambiar una cifra principal, y que en el 97% de los errores serios con una dirección clara el resultado exagera lo que el estudio quiere demostrar. Son acusaciones, no una verdad comprobada, y habrá que ver qué responden. En España esto no es un debate de académicos. Hacienda firmó un memorando con el Observatorio Fiscal Internacional que dirige Zucman, y el ministro Pablo Bustinduy defiende que su tasa recaudaría hasta 5.200 millones al año. Ese cálculo sale del propio observatorio de Zucman. ¿Cuántos de esos 5.200 millones sobrevivirían a la misma auditoría? Antes de cobrar un impuesto basado en un estudio, que lo revise una IA. Sale más barato que equivocarse.

Zagreb gets more rain than Amsterdam, Dublin, Vienna, Paris and Berlin. 🌧️ Mean annual precipitation 1991–2020, weather station (WMO) data: 🇸🇮 Ljubljana 1369 mm 🇭🇷 Zagreb-Grič 890 mm 🇳🇱 Amsterdam 851 mm 🇮🇪 Dublin 784 mm 🇫🇷 Paris 622 mm 🇩🇪 Berlin 569 mm Zagreb is the 2nd wettest EU capital, with 43% more rain than Paris. ERA5 reanalysis (black lines) agrees for Zagreb: 882 mm. Data: NOAA GHCN-Daily, ECA&D, WMO/CHMU normals; ERA5 (Copernicus) via Open-Meteo / Computation: Grok

"The most diverse cabinet in history…" 88% Black. Virginia is 20% black. 0% White men. Virginia is 30% White men. Friendly reminder when these people say "diversity", they just mean anti-White.

This chart is eye-opening, German carmakers’ market share in China has roughly halved since 2020, from nearly 24% to around 12%. It marks the end of a business model. For years, China was the cash cow that subsidized high-cost German structures, rigid union agreements, elevated wage and energy costs, slow decision-making, and layers of bureaucracy. Profits from the world’s largest car market papered over an industry that had grown too expensive, too protected, and too slow to adapt. Those profits are now disappearing as Chinese brands dominate electric vehicles, software, and cost leadership. Not only in China, they start to dominate in other markets as well. The response at home has been telling. Instead of accelerating new models and software platforms, German industry is stuck in union negotiations and spending a fortune on severances. Volkswagen is looking at cuts of tens of thousands (up to 100,000 across the group). BMW is offering eye-watering packages, reported well into high six figures in euros, to remove roughly 8,000 white-collar roles. Mercedes is restarting voluntary exits. Billions are going toward shrinking the existing structure rather than building the next one. Overregulation and administrative drag hinder innovation at the speed Chinese competitors have shown. High fixed costs and protected labor frameworks reduce flexibility. Companies react rather than lead. Arrogance played a role. German engineering excellence was long treated as a durable moat. Many underestimated how quickly Chinese firms would close the gap in battery technology, software-defined vehicles, and manufacturing scale, and how willing Chinese buyers would be to switch once local products were competitive or superior. The same pattern of underestimating focused, cost-efficient competitors has appeared in solar, batteries, and many other sectors. The German business model is finished. With regulations, bureaucracy, and powerful unions, Germany won’t be able to react and adjust fast enough to keep its manufacturing edge and standard of living. Most Germans don’t grasp yet what is coming.

What you see happening in France is essentially what will happen throughout the whole of Western and Southern Europe at some point Europe regulated away all new and existing businesses so they with the money and the talent moved elsewhere (like US and Asia) When money flees a continent, the businesses, then the people and the government run out of money Meanwhile cost of living keeps growing → people get poorer and → people need more government welfare → meaning the government has to borrow more money to pay for that → borrowing for worse and worse rates → which pushes down the Euro → meaning less spending power for its people → making Europeans even poorer

This is eye-opening. Europe’s share of global tech R&D has collapsed from 23% in 2005 to just 8% today, while the US has climbed to 62% and China has risen from near zero to 15%. South Korea now outspends Germany, France, the UK, and Italy combined in tech R&D. Yep, you read that right. This is the predictable result of policy choices that systematically favor control and redistribution over growth. Layered regulation, data rules, AI restrictions, labor rigidities, and precautionary frameworks, raises the cost and slows the pace of turning ideas into products, forcing companies to spend scarce resources on compliance rather than experimentation. High taxation and expansive redistributive systems extract capital that would otherwise fund private R&D and risk-taking, while shrinking the after-tax reward for building something new and scaling it. The broader left-leaning mindset that prioritizes equality of outcome over dynamism treats entrepreneurial success as something to be managed, taxed, or shared rather than unleashed and rewarded. Places that protect the returns to innovation and keep barriers low continue to pull ahead, attracting talent and capital that Europe increasingly loses. Europe’s relative decline is the cumulative outcome of treating dynamism as a problem to be regulated rather than the engine of prosperity. Without a shift toward lighter rules, lower burdens on enterprise, and greater tolerance for unequal outcomes from unequal effort, the gap will keep widening, and Europe will become an open-air museum for wealthy Americans and Asians.

I’ve been a member of my local Catholic parish, Saint Andrew by the Bay, for 16 years. We’ve never had more than six or eight adult converts at the Easter Vigil. They are currently 56. 🤯 AmAZing. What is happening?

The way @MIT’s bloated DEI bureaucracy was weaponized to serve an imaginary feminist crusade and naked personal vendetta against a tenured male professor over unrequited love must never ever ever happen again. Federal agencies @NIH and @NSF fund and oversee a large share of MIT’s research (yes - your tax dollars!) to the tune of hundreds of millions annually, which means that they have enormous leverage through grant conditions, compliance reviews, and enforcement. It’s up to the government to throw their weight and pressure MIT to maintain fair processes that do not destroy top scientific talent. The obvious move is to withhold or condition funding on gutting the rot. @DOJ: Enforce civil rights statutes (including Title IX-related conditions on federal funding) and investigate potential patterns of discriminatory, retaliatory, or abusive HR/DEI processes that undermine due process or scientific productivity. Sabatini’s was a major career criminally terminated resulting in untold losses to science and to humanity. I’m not even being hyperbolic. The stakes are high and they demand accountability and widespread reform across all institutions of import @HarmeetKDhillon @DrJBhattacharya @regardthefrost

Ab heute Abend (29.10.) gibt es die nächste Dunkelflaute in Deutschland. Windkraft und Solar können um 17 Uhr nur insgesamt gut 2,2 GW an Leistung bereitstellen. Den Bedarf zu dieser Zeit prognostiziert das Fraunhofer ISE jedoch auf 61,2 GW, also Wind und Solar decken nur 3,7% des Bedarfs ab. Den Rest müssen überwiegend fossile inländische Kraftwerke sowie Kraftwerke in Nachbarländern liefern. Von Erdgas kommen rund 14,5 GW, von Kohle 17,2 GW und aus dem Ausland knapp 11 GW.