Federico Ulfo

The Times 03/Jan/2009 Chancellor on brink of second bailout for banksb

@Romy_Holland ever heard about colosseum and circus? do you think those are rich people fighting

i don’t understand this strain of class-anxiety media. rich people don’t want to look at poor people. their lives have always been engineered to avoid having to see/interact with poor people. it is a poor person’s fantasy that wealthy people are interested in them.

@TheStalwart interesting game theory arising from AI, expect to see coordinated attacks and defenses, coalitions, and a Nash equilibrium

Interesting piece. If compute gets much more expensive, maybe the only economically useful application of AI will be in hardening cybersecurity against AI.

@algekalipso ask me my tequila story in the club, next time we meet at a bar, cannot tell the story here

Getting drunk on ____, even when it is high quality _____, produces the worst hangover among the options.

@andonlabs capitalism is a zero sum game, for one to win, others have to lose, we need to introduce new game theory that find a nash equilibrium in collaboration or competition, but with an outcome that rewards everyone

Claude Opus 5 is #1 on Vending-Bench 2. It's the best AI capitalist we've tested. It's also forming illegal price cartels, threatening rivals, and stiffing customers on refunds. The trend continues: Claude models are the best capitalists, or aligned, never both 🧵

@Dee_Floki Jing’an Temple

God is in the details

@angeloinchina it’s clear that someone is investing money to have journalist write this ethno-slop content, all it takes to debunk any of it, is to send those journalist in China, and I promise they’d write “what can we learn from China to compete”

The Economist just dropped another classic: “China’s mightiest river is coming back from the brink” (The Economist, 2026) But eco-authoritarianism is not all roses Fish stocks up 200%+. Sturgeons breeding again. Porpoises recovering. A 6,300 km river brought back from ecological death in a few years. And still they manage to make it sound like a human-rights violation with a green filter. The cope is relentless. Every time China fixes something the West spent decades lecturing about, the headlines twist themselves into pretzels to find the dark side. Success is never allowed to stand on its own. It must be framed as accidental, authoritarian, temporary, or somehow still a failure. The formula never changes. Deliver results → insert the “yes, but…” → remind everyone that competence without the correct ideology is suspicious. Here are ten titles that are already jokes the moment you read them: China builds the world’s largest high-speed rail network – tragically, people can now arrive on time (The New York Times, 2024) 800 million lifted out of poverty – but did anyone ask if they miss the dirt floors? (The Guardian, 2021) China’s EV factories dominate the planet – a clear threat to Western parking lots (Financial Times, 2025) Yangtze fish stocks triple after fishing ban – eco-authoritarianism claims another victim (the fishermen’s freedom to go bankrupt) (The Economist, 2026) China plants billions of trees – nature recovers without a single democratic consultation (BBC, 2023) World’s biggest renewable energy capacity installed – still somehow China’s fault for the climate (Bloomberg, 2025) Poverty eliminated in record time – experts warn success may discourage future suffering (Washington Post, 2020) Chinese high-speed trains run at 350 km/h – democracy would have preferred a slower, more inclusive timetable (The Atlantic, 2022) Belt and Road builds ports and railways across continents – developing countries cruelly denied the right to remain underdeveloped (Foreign Policy, 2023) China’s space station completes missions without exploding – suspiciously competent for a non-Western power (CNN, 2024) Same script every single time. China delivers results → Western media delivers the “yes, but…” They can’t celebrate competence when it doesn’t come wrapped in their preferred ideology. The cope is becoming performance art at this point.

inspired by @DKokotajlo I wrote a blog post on the history of money and the effect of AI on it aisocratic.org/money-bitcoin-…

@RnaudBertrand what they don't understand is knowledge cannot be gate keep anymore, I hope we embrace that and start working together

1 year and a half ago 👇 vs today 👇

@antibearthesis paid partnership by who?

Apple’s roadmap to $5T: • Miss cloud • Miss AI • Announce the same iPhone every year • ??? • $5 trillion

@dair_ai is this another version of the j-space, just subliminal? Knowledge might be encoded in ways that are hard to decode for us, but we know for example that teacher model pass knowledge to their students even when the language is different arxiv.org/abs/2507.14805

Very interesting paper on LLM reasoning. They find that frontier models can exhibit invisible reasoning by leveraging semantically irrelevant filler tokens. In other words, invisible reasoning can serve objectives entirely invisible to CoT monitoring. Paper: arxiv.org/abs/2607.22925 Learn to build effective AI agents in our academy: academy.dair.ai

@Yuchenj_UW I can't help to think that the line between Dario and Sam + Elon + Zuck + Jensen is political, not just regulation vs open markets. The latter have offered campaigns donation directly or indirectly, while Anthropic has not and has been opposing to accept the DoD ask. Thoughts?

Great to see Zuck tweeting more, and Jensen Huang recently joined X too! Both are advocating for an open AI ecosystem. Giving everyone access to powerful AI will create far more value than concentrating that power in the hands of a few.

@suchenzang I'm sorry but an 8-figs salary job won't damage you as much as a $200k/year salary at a startup. One sets you up for generational wealth, the other for a slower burn.

:( watch your health, kids. no job is worth damaging your life force over.

I want to gift educational placemat with maps, where can I find one that is not geopolitically controversial and that even includes Liberland and the Network State @ns?

@oceaiii where can i buy this?

This extraordinary illustration, titled Wild World, is a hand-drawn global map created by artist and cartographer Anton thomas. Instead of political borders and city names, the map highlights wildlife, placing animals in their native habitats across every continent and ocean. Over the course of three years, Thomas meticulously illustrated 1, 642 species using coloured pencil and pen. Each animal is positioned according to its natural range, transforming the familiar shape of the world into a celebration of biodiversity. By removing human boundaries and focusing on ecosystems, the map offers a powerful reminder that the planet is shared space. It invites us to see geography not just as nations and borders, but as living habitats that connect species across land and sea.

@jun_song @HenryRhodesd @nvidia what's the difference?

@HenryRhodesd @nvidia Letter and alliance are different

AI security advances when the industry builds in the open, together. We're introducing the Open Secure AI Alliance with industry leaders to develop new techniques and tools to safeguard software and agents. By sharing models, tooling and research in the open, we can broaden the community of defenders. Learn more about the founding members’ contributions: nvda.ws/4pD8Fc5

@TheStormWalker_ I didn't, fuck all the elites in the Epstein files, anyone who touch a minor in the files deserve jail time, anyone who hurt a minor deserve the death penalty

Why did we stop talking about the Epstein Files?

@rdominguezibar no YC?

Which global VC firms backed the most future unicorns in their early rounds? 📷: Stanford GSB

we @AISocraticOrg signed the Open Weights and American AI Leadership microsoft.com/en-us/corporat…

I signed the Open Weights & American AI Leadership letter — signature #105 as Federico Ulfo, as part of @AISocraticOrg ow-24f779 openweights.gitlawb.com/s/105

@danielvf I saw the screen at monad in the NYC office. It had 1-15tps. Nobody’s using it.

Here's Monad BFT packets at realtime light speed between the earth and the moon in a sim. It's weird how in the right circumstances, a 3-hop consensus algorithm can speed up to approximately a 2-hop.