@kzui
iAccount based inArgentina
About this account
- Account based in
- Argentina
- Connected via
- United States Android App
Account-level information from X, not a live location or the device used for a specific post.
Christian, husband, dad, creator of Moq and other developer tools and libraries. Also known as 'kzu' (like 'kah-zu' rather than 'kzoo' :))
Joined November 2007
- Tweets14.7K
- Following919
- Followers3.4K
- Likes16.3K
Pinned Tweet
I think every company should have a #fridathon policy, to foster innovation, energize teams and improve morale cazzulino.com/fridathon. I'm realizing the term might actually stick, so I'm giving it a permalink, hehe :)
I think one of the greatest (and under appreciated) features of the @GitHubCopilot harness is its ability to automatically check complex work with a more powerful model automatically (aka 'rubber duck's). So you can implement code with Sonnet and auto-review with Opus when complexity deserves it.
Sorely miss this when using @cursor_ai Cursor agent or @SpaceXAI Grok Build CLI
A new subscriber can get @SpaceXAI SuperGrok Plus for $100. An existing SuperGrok Heavy subscriber cannot downgrade to it, only $30 and $10 options.
I find asymmetric offerings for new vs existing customers egregious. Might as well cancel the subs entirely, and "become new" in a while? Makes no sense.
Daniel Cazzulino 🇦🇷🗽 retweeted
FX: Premios Nobel, billonarios y vendedores de humo.
Todas las semanas aparece alguien diciendo que el tipo de cambio deberĂa estar a tal o cual nivel. Más de 40 años de investigaciĂłn econĂłmica, financiera y de práctica en mercados indican que esa persona no sabe de que está hablando.
Lo que digo no es mi opiniĂłn. Es el resultado empĂrico más robusto de la macroeconomĂa internacional de los Ăşltimos cuarenta años.
En 1983, Meese y Rogoff ("Empirical Exchange Rate Models of the Seventies: Do They Fit Out of Sample?") mostraron que ningún modelo de tipo de cambio le ganaba a lo que se llama una caminata aleatoria. O a un chimpancé tirando dardos contra una pared donde están anotados distintos tipos de cambio.
El resultado fue confirmado una y otra vez: Cheung, Chinn y Pascual (2005, "Empirical Exchange Rate Models of the Nineties: Are Any Fit to Survive?"), Rossi (2013, "Exchange Rate Predictability", Journal of Economic Literature), Engel y West (2005, "Exchange Rates and Fundamentals").
Obstfeld y Rogoff (2000, "The Six Major Puzzles in International Macroeconomics") lo incluyeron entre los seis grandes enigmas de la macroeconomĂa abierta. La economĂa no tiene herramientas para predecir el tipo de cambio ni para calcular su nivel de equilibrio.
Y no es sĂłlo la academia. En los 80 y 90 habĂa una industria entera de fondos dedicados exclusivamente a operar monedas. Prácticamente desapareciĂł. Gastaron decenas de millones de dĂłlares en modelos, datos y sistemas para predecir el tipo de cambio, y desaparecieron.
FX Concepts, el más grande del mundo, manejaba USD 12.000 millones en 2008; cerró en 2013.
QFS, fundado por Sanford Grossman, cerró en 2014 tras años de pérdidas.
Grossman no era un operador cualquiera: fue uno de los académicos de finanzas más importantes del mundo. En 1987 ganó la Medalla John Bates Clark, que se otorga al mejor economista estadounidense menor de 40 años y cuyos ganadores, en cerca de la mitad de los casos, terminaron recibiendo el Premio Nobel. Fue presidente de la American Finance Association y profesor de finanzas en Wharton.
Con esa formaciĂłn, ese prestigio y cientos de millones de dĂłlares de inversores sofisticados, no pudo predecir el tipo de cambio.
Lo mismo ocurrió con casi todos estos fondos. Los profesionales mejor pagados del planeta, con la mejor información, la mejor formación académica del mundo y los mejores sistemas, no pudieron predecir el tipo de cambio con dinero propio en juego.
Muchos de estos fondos contrataron a los mejores acadĂ©micos y cientĂficos del mundo, y terminaron perdiendo plata y cerrando.
Y eso es en economĂas "normales".
Argentina no lo es: está cambiando radicalmente su estructura de precios relativos, su polĂtica fiscal y su rĂ©gimen monetario; se está integrando al mundo comercial y financieramente, y facilitando la incorporaciĂłn de nueva tecnologĂa, desregulando.
Además está desarrollando de manera exponencial sectores exportadores —gas, petrĂłleo, litio, cobre, economĂa del conocimiento— que van a cambiar dramáticamente la oferta de dĂłlares por dĂ©cadas.
Pretender saber el tipo de cambio de equilibrio en Argentina no es serio, para no usar adjetivos más precisos.
El gran error
Uno de los errores más grandes que podrĂamos cometer es tratar de hacer polĂtica cĂclica manipulando el tipo de cambio: dirigir recursos a sectores que no serán viables en el futuro, mediante un cambio temporal en el tipo de cambio real.
Es importante recalcar: ningun gobierno del mundo controla el tipo de cambio real.
Argentina tiene gran escasez de ahorro; usarlo en sectores que no son viables y que a futuro quebrarán es un desperdicio que no nos podemos permitir.
Si un sector sĂłlo es viable con un dĂłlar más alto que el que determinan los flujos futuros de divisas, ese sector no es viable: es un subsidio que pagan todos los demás vĂa precios, salarios reales e inflaciĂłn.
En definitiva, si te encontrás con alguien que sabe cuál es el tipo de cambio real de equilibrio, y sabe a donde va a estar en el futuro, estás frente a un futuro Premio Nobel, un futuro billonario, o un vendedor de humo.
Daniel Cazzulino 🇦🇷🗽 retweeted
The British Empire around 1833
Was the seizure of the Malvinas Islands an isolated event? Look at what Britain was doing around the world:
🇮🇳 1803 – Maratha territories — war and territorial conquest
🇿🇦 1806 – Cape Colony — invaded and captured
🇦🇷 1806 – Buenos Aires — invaded and captured; subsequently retaken by local forces
🇩🇰 1807 – Copenhagen — attacked; Danish fleet seized
🇺🇾 1807 – Montevideo — invaded and captured
🇦🇷 1807 – Buenos Aires — invaded again; the second British attempt was defeated and the British forces surrendered.
🇲🇶 1809 – Martinique — invaded and captured
🇬🇵 1810 – Guadeloupe — invaded and captured
🇷🇪 1810 – Réunion — invaded and captured
🇲🇺 1810 – Mauritius — invaded and captured
🇮🇩 1811 – Java — invaded and occupied
🇺🇸 1812–15 – United States — invaded; Washington captured and burned
🇳🇵 1814–16 – Nepal — invaded; territory ceded
🇮🇳 1817–18 – Maratha territories — conquered; British control massively expanded
🇦🇪 1819 – Ras Al Khaimah — attacked and captured
🇲🇲 1824–26 – Burma — invaded; territories ceded
🇬🇶 1827–34 – Fernando Po — British occupation/presence
🇦🇷 1833 – Malvinas Islands — Argentinians were removed and British administration established
🇮🇳 1834 – Coorg — invaded and annexed
🇿🇦 1834–35 – Xhosa territory — invaded; Queen Adelaide Province proclaimed
🇾🇪 1839 – Aden — attacked and captured
🇦🇫 1839 – Afghanistan — invaded; Kabul occupied
🇨🇳 1839 – China — First Opium War begins
And we are supposed to believe that what happened in the South Atlantic in 1833 should be understood completely outside this context?
Britain was one of the world’s dominant imperial powers and was repeatedly using military force to acquire, occupy or control territory during this period.
1833 wasn’t happening in a vacuum. It was happening in the age of the British Empire.
Jesus christ
Some data we recently assembled on entrepreneurship/compute in Europe: eudata.vercel.app.
We hope that one of the useful roles that Stripe can play is in collecting and publishing empirical data pertaining to entrepreneurship and industry in Europe. There's growing appetite to get Europe on a better footing, and cross-sectional comparisons can often shine light on where opportunities lie. If you're interested in this kind of thing, we publish more at stripeeconomics.substack.com.
Daniel Cazzulino 🇦🇷🗽 retweeted
Google is starting to build its Dyson Swarm AI supercluster and you're bearing anon??
What's keeping @SpaceXAI and @cursor_ai behind Claude, Codex and even AWS in implementing the MCP 2026-07-28 spec?
blog.modelcontextprotocol.io…
claude.com/blog/bringing-mcp…
codex.danielvaughan.com/2026…
aws.amazon.com/blogs/machine…
Daniel Cazzulino 🇦🇷🗽 retweeted
Replying to @GergelyOrosz
And @JensenHuang is right: coding was the *task*, it was never the (sole) *job* of software engineering.
Daniel Cazzulino 🇦🇷🗽 retweeted
El compromiso del presidente Trump y el liderazgo del presidente Milei se traducen hoy en algo concreto: Estados Unidos y Argentina lanzaron el Corredor Andes-Atlántico, la iniciativa que conectará la matriz productiva argentina con los mercados de Occidente a través de inversión de EE. UU. 🇺🇸🇦🇷.
ar.usembassy.gov/es/declarac…
Daniel Cazzulino 🇦🇷🗽 retweeted
Replying to @NYGBertie
"Si no fuera por los crayones, estarĂamos en el horno". - Fabricante de velas (~1900)
@grok explica la cita ilustrativa
Daniel Cazzulino 🇦🇷🗽 retweeted
Replying to @Jason
Adjusted so all numbers have 1980 purchasing power....
Is it more -- yes but most of the $800,000,000,000 - is inflation....
BTW -- Tells you how useless and crappy the @federalreserve is.
Daniel Cazzulino 🇦🇷🗽 retweeted
We ran fresh Next.js evals. The tally:
â‘ Opus 5.5 [đťźżđťź˝%]
② GPT 6 Sol [𝟿𝟽%]
③ Fable 5.1 [𝟿𝟽%]
④ Grok 4.7 [𝟿𝟺%]
Notably, Grok is 2x-7x cheaper
GPT-6 Sol debuts at 97% on Next.js evals, tying the highest success rate.
Claude Opus 5.5 also reached 97%, joining Claude Fable 5.1. Opus ranks first among the three on average cost.
View the leaderboard ↓
nextjs.org/evals
This is the way. There's nothing more egregious than sharing a pay-walled article to someone. It's like rubbing in their faces that they aren't paying and are therefore losing on content you think is interesting.
Paid subscribers can send gift articles, giving friends and family free access to any story, no subscription required. thefp.com/p/you-can-now-shar…
Daniel Cazzulino 🇦🇷🗽 retweeted
C#/.NET being pushed hard running infrastructure at Ethereum's scale.
Nethermind runs on ~3k nodes and accounts for ~33% of staked nodes - supporting a ~$334B native asset, ~$146B of stablecoins and ~$54B deposited in DeFi apps.
Nethermind 2.0 has some amazing performance work in it: up to 2.55x block throughput, p99 block processing from 1.65s to 208ms, eth_call RPC 15-27% faster and Keccak-256 ~30% faster with platform intrinsics.
Nethermind Client 2.0.0 is out. It changes how the node stores Ethereum's state: archive nodes drop from ~30 TB to ~2 TB, you choose the history you keep, and blocks process 2–3× faster than the previous release on a fresh node.
Nine years since the first commit. The fastest execution software we've ever built.
Breaking changes; read the notes before upgrading. Release: github.com/NethermindEth/net…
Daniel Cazzulino 🇦🇷🗽 retweeted
Javier Milei – The Proof of Concept.
The long march took a century to build. Milei has been dismantling its Argentine expression in real time, with a chainsaw, on television, while calling the central bank a criminal organization. This is not a metaphor. The chainsaw was a campaign prop. The central bank assessment was government policy.
1. The biography is the argument. Born in 1970. Economist. Rock musician – he played in a Rolling Stones tribute band, which is either irrelevant or the most relevant detail depending on your theory of what the counter-march needs. Self-taught in Austrian economics – Mises, Hayek, Rothbard – at a time when Argentine academia was entirely captured by the Keynesian consensus that had been producing Argentine economic crises with metronomic regularity for seventy years. He became a television economist – the secondhand dealer in ideas, in Hayek’s precise formulation – explaining on cable news why everything the establishment was doing was wrong. He was right. Nobody in the establishment listened. The audience did.
2. Argentina in 2023 was the laboratory result of a century of Peronism – the specifically Latin American variant of the long march. Inflation at 211 percent. Poverty at 40 percent. The state consuming everything and producing nothing. The establishment consensus – across parties, across media, across academia – was that the solution was more state, more spending, more management. This consensus had been producing the same results for seventy years and drawing the same conclusions.
3. He ran on the complete rejection of the consensus. Not the modification – the rejection. The central bank is a criminal organization that steals from citizens through inflation. The political class is la casta – extracting value from Argentine society for generations while calling it governance. These were not positions calibrated for electability. They were the Austrian school applied directly to Argentine politics, stated without diplomatic softening, by a man who clearly enjoyed saying them. He won with 56 percent of the vote.
4. What followed was the Hayekian pipeline producing its intended result. Decades of work by libertarian think tanks in Argentina had produced the intellectual infrastructure, the vocabulary, and the trained personnel that made a Milei administration possible. He didn’t arrive in office without a plan. He arrived with economists who had spent years preparing exactly this plan for exactly this country. The pipeline that produces Obama also produces Milei – when run in reverse. The difference is which ideas were in it.
5. The chainsaw was not performance. In the first year: the fiscal deficit eliminated – the first balanced budget in over a century. Government ministries cut in half. Hundreds of regulations abolished. Inflation falling from 211 percent toward 50. The establishment that said it was impossible watched it happen and explained it wouldn’t last. It is lasting.
6. The reaction was the complete toolkit. Lawfare. Media coverage calibrated to delegitimize rather than report. Street protests organized by the unions whose privileges were being dismantled. The Peronist establishment discovering that the institutions it had captured could not stop a government willing to act faster than the institutions could respond. The ratchet being run in reverse. It turns out the ratchet can go both ways. It requires someone willing to turn it.
7. Milei reposts this series. The intellectual supply chain Hayek described is working in real time. Ideas produced in the Polish-French register, distributed on an American platform, reposted by an Argentine economist with a chainsaw and a rock band – reaching people who need the vocabulary to name what is being done to their countries and what can be done instead. The counter-march does not look like the march. It is louder, faster, less elegant, more honest, and considerably more fun.
The proof of concept is running. Watch Argentina. 🇦🇷
What a beautiful story! 🇧🇷
With his Dad censored, silenced, and imprisoned, Flávio Bolsonaro is now the frontrunner to become the next President of Brazil.
To beat the socialist Lula with his demoniac Chief Judge Alexandre de Moraes…
Jair Bolsonaro – the Father – censored, silenced, passport confiscated, prosecuted, a political prisoner – watches from the sidelines, banned from running.
The system that tried to erase the Bolsonaro movement has produced its successor.
This requires some context for those who haven’t been following.
Lula is the corrupt socialist who runs Brazil. Three times president, once convicted of corruption, imprisoned, then freed by a Supreme Court justice who subsequently became his political instrument and the most powerful unelected man in the country. Alexandre de Moraes – appointed to the Supreme Court, now operating as a one-man censorship apparatus. He ordered the shutdown of X in Brazil. He froze accounts. He prosecuted political opponents. He banned platforms. He ran criminal investigations against the movement that had defeated his patron’s party in 2018. A Supreme Court justice doing the work that, in a functioning democracy, no court should be able to do – and doing it openly, without embarrassment, because the institutions that should have constrained him had already been captured.
This is what the Long March looks like in Latin America. Faster, louder, less elegant than the European version – but the same mechanisms. Elite capture. Lawfare. Manufactured consensus. The ratchet. All of it, compressed into a decade, in a country of two hundred and fifteen million people.
Jair Bolsonaro won in 2018. Lost narrowly in 2022 – in a contest whose integrity his supporters disputed and whose resolution was managed by the same judiciary that subsequently prosecuted him. Passport confiscated. Banned from running. Multiple criminal investigations. The full toolkit deployed against one man, by one institution, in the service of one political project.
And yet.
His son is the frontrunner. The movement didn’t die – it deepened. The persecution that was supposed to destroy the Bolsonaro phenomenon produced instead exactly what persecution always produces when it is applied to something genuine: it separated the opportunists from the believers, clarified the stakes, and gave the cause a martyr it hadn’t needed to manufacture.
This is the counter-march working in real time. Not through the institutions – around them, despite them, against them. The preference falsification cascade beginning: the people who had been silent discovering that they were not alone, that the system that had told them the debate was over had been wrong about that too.
The son carries the name. The father watches from the sidelines. The system that tried to erase them is discovering what every system that has tried to erase something genuine eventually discovers.
You cannot imprison a majority.
The Murderbot Diaries is finally done for me. In the latest book the AI protagonist is more emotional than a teenager. What a joke.
Platform Decay by Martha Wells on Audible. audible.com/pd/B0GJ6MQLKJ?so…
Nuevo dato del que seguramente no hablará más de 5 segundos (si lo mencionan) ningún "periodismo serio".
Buen momento para recordar que el peor de los polĂticos de hoy tiene mejor imagen que el periodismo:
> la confianza alcanza mĂnimos histĂłricos con apenas el 26%.
https://reutersinstitute .politics.ox.ac.uk/es/digital -news-report/2026/argentina
El Índice de Confianza del Consumidor (ICC) elaborado por el @CIF_UTDT sube 2,4% respecto a agosto de 2026, sobre la base de las encuestas realizadas entre el 1 y el 9/9. Descargá el informe completo acá: bit.ly/3WTPs81
Daniel Cazzulino 🇦🇷🗽 retweeted
Remember how eagerly the press swallowed stories about how Twitter was going to crash after Elon bought it? How could they have believed that someone who could run rocket and car companies couldn't run a forum?
How it started, How it's going.
20x+ later, he still thinks he's got an insight.
The dead cat continues to bounce.
What is Bitcoin, 17 years later?
It’s not great for transactions or smart contracts; the user experience is still intimidating for most, and it’s no longer capturing the public’s imagination.
If it comes up at the dinner party, it's followed by a hearty “remember that!”
It’s a boring store of value, trusted as that—which is its biggest value.
It's boring.
Folks expect bitcoin to be stable and that it’s no longer a way to get rich quick.
Institutions embracing and accumulating it took $btc from punk rock to Muzak.
Advocates went from pirates to suits in orange ties, awkwardly sharing cringe memes — just like the cool kids do!
If Bitcoin were going to reach mass adoption and an important use case, it does better than anyone else, it would have by now.
Bitcoin feels like the CD in the age of Spotify, the DVD in the age of Netflix — a bridge technology that folks who invested in “their collection” keep saying is the better experience.
Perhaps bitcoin is vinyl, cool memories and fun to drag out and put on display, but a technology that served us to get to where we are now.