Cavalli

Surfer during the day | A purple Machine Learning Engineer and Economist on the off hours | A completely personal (and decidedly unprofessional) profile

[Cavalli @dancavlli]

Thread com as apostas idiotas que eu fiz, pra facilitar a fechar o acordo
https://t.co/dRCvYNstdu

[Cavalli @dancavlli]

Deixo aqui registrado que apostei 5.000 unidades monetárias brasileiras com o meu pai caso o homem chegue em marte até 2020.

[AAADRIELLL @AdrielAAB]

Gente vcs de esquerda Rio de janeirense vamos ter noção que a candidata do PSOL nao tem o mínimo de chance

É melhor dar o segundo voto pro Pedro Paulo do que deixar um random bolsonarista entrar

[Cavalli @dancavlli]

@artchad It's the same thing OpenAI did and requested with GPT-4 when they were compute constrained.
Sounds a lot like regulatory capture on their part

[Cavalli @dancavlli]

@davidad "you are doubly correct"

[Cavalli @dancavlli]

é isso que o IE UFRJ forma galera, cuidado

[Cavalli @dancavlli]

Top coisas mais engraçadas que eu já li no twitter kkkkkkkk

[João Pedro Garcia @garciacjao]

O Pessoa está certo. Não porque seja efetivamente inviável. Vejam, a questão do metrô do Rio de Janeiro é a seguinte:

A operação do metrô é cara, e é cara por DETALHES operacionais relativamente simples. O maior deles é o fato do metrô ser uma linha reta, o que exige que, ao final da linha, o trem tenha que manobrar para mudar de direção. Essa é uma manobra caríssima que aumenta muito o custo e diminui muito a eficiência.

A prioridade nº 1 do transporte metroviário tem que ser finalizar a estrutura anelar para acabar com esse problema, reduzir custos operacionais, e assim conseguir caixa e eficiência suficientes para FAZER SENTIDO mais um investimento bilionário no setor.

Nós sabemos o que fazer. O seu pré-candidato sabe?

[swyx @swyx]

"Developers can update Claude’s instructions mid-task without breaking the prompt cache or routing the update through a user turn"

wtf? how??

[✿ damir @otomir23]

people are so used to their data being sold and ads being shoved down their throats, that they cant fathom that all services actually cost money to run

[Cavalli @dancavlli]

@VictorTaelin @willdepue The only issue is that you assume this is possible. If AI can't be made 1000x cheaper due to architectural reasons, it doesn't matter how capable AGI is

[guilérme @DesturbioMental]

to tao viciado nessa serie a forma que eles misturam terror com uma leve dose de comedia é muito boa! apple pra mim nao para de se consolidar com o melhor catalogo de series

[Daniel Jeffries @Dan_Jeffries1]

The most revealing thing about this AI leadership paper is that it reads less like a vision for innovation and more like a glossy whitepaper for a 21st century East India Company.

Every generation of incumbents discovers a new moral vocabulary for why they alone should control transformative technology.

In the 90s it was cryptography. We were told strong encryption was too dangerous to spread because terrorists, rogue states, chaos, dual-use, etc. So the US crippled exports, weakened products, slowed adoption, and kneecapped parts of its own software industry. Right up until reality steamrolled the policy and we woke up to its stupidity and then eCommerce, secure communications, software signing, and the modern internet exploded and gave us tremendous benefits.

Now the exact same priesthood has returned with AI.

- “Dual-use.”
- “Strategic advantage.”
- “Model distillation.”
- “National security.”

- “Responsible access.”
A few different nouns but mostly the same ones. Same instinct:

Centralize control, gatekeep compute, fuse state and corporate power, and call it safety.

The funniest part is that this strategy is almost perfectly designed to accelerate the thing they claim to fear.

You do not stop a rival superpower (who happens to be the absolute best at scaling energy and manufacturing and who has a choke-hold on rare Earths refinement) from building domestic capability by permanently attempting to strangle them.

You create the economic and political incentive for total self-sufficiency.

We have already done that as Jensen warned. We went from 100% market to nearly 0%. Huawei is now manufacturing millions of chips. DeepSeek v4 trained on them. They have more energy than the rest of the world combined. Meanwhile, we have activists and anti-economic fools like AOC and Bernie pushing for data center moratoriums and we can't build a single bullet train in 20 years and folks fighting to not expand the energy grid here and new nuclear plants getting tied up in environmental regulation for a decade.

The sanctions did the exact opposite of what the hawks wanted. They jumpstarted a moribund, dinosaur of a Chinese chips industry. We basically said to the people who happen control the most powerful manufacturing engine on the planet "we intend to squeeze you."

They rightly saw it as an existential threat.

The sanctions become the industrial policy.

Huawei. SMIC. Domestic lithography. Packaging. Memory. Entire Chinese supply chains that did not exist at serious scale a decade ago now exist precisely because Washington convinced Beijing they had no choice.

Brilliant work.

So the endgame here is what exactly?

1) Push China into a Manhattan Project for chips and AI.
2) Increase the strategic value of Taiwan even further.

3) Once China reaches self sufficiency that can invade Taiwan and choke off our own super advanced chips where are made there exclusively (and no we don't have even close to enough TSMC factories in Arizona or anywhere else in the world).

That's every NVIDIA chip. Every Google tensor chip. Every Apple chip. Every chip in you iPhone and Android phone. Every Amazon chip. The chips in your car and truck and hair dryer and washing machine.

4) Escalate a cold tech war into a permanent civilizational bloc conflict that is likely to turn into a shooting war at one point.

5) Fragment the global software ecosystem.

6) Create American AI aristocracies protected by regulation and compute licensing.

And somehow call this “open innovation.”

Meanwhile the actual history of software keeps screaming the opposite lesson:

Knowledge diffuses, open ecosystems win, developers route around gatekeepers, and attempts to permanently contain computation usually fail.

What really jumps off the page is the assumption that a tiny cluster of frontier labs should become quasi-sovereign actors, deciding who gets intelligence, who gets compute, who gets models, and which countries are permitted to participate in the future.

Not elected governments.

Not open markets.

Not open-source communities.

A handful of corporations sitting beside the national security state, insisting that concentration of power is necessary to protect democracy.

You almost have to admire the audacity.

[Cavalli @dancavlli]

@cassinaooo Já até comecei a alinhar a hora do café com quando bate o limite de requests simultâneos

[Robert Graham @robertgraham]

Utah is not building a 40,000 acre data center.

They are using 40,000 acres of water rights for the power generators that will power a data center.

Pretty much all outrage you see about data centers is deliberate misrepresentation designed to fuel outrage.

[Daniel @leittedaniel]

EU QUER VER A INDÚSTRIA NACIONAL TÊXTIL FALIDA

EU QUERO VER A MAGALU NA SARJETA

EU QUERO VER A ANFAVEA SEM CONSEGUIR COMPETIR COM CARRO CHINÊS E VIVENDO DE ESMOLA

EU QUERO SIM A QUEBRA DA INDÚSTRIA NACIONAL

EU AUTORIZO SHEIN

EU AUTORIZO BYD

[Shamanic Depressive Gymnosophist (book in bio) @sigilante]

I am once again advocating for no one shipping software AT ALL. Ship test suites defining an interface and let users′ autonomous agents write necessary software on the fly. A compliant instance is a real instance.

[João Pedro Garcia @garciacjao]

Meus amigos, vocês não têm noção do tamanho do problema.

Na UFRJ, há anos o currículo está sendo "adaptado", isto é, facilitado, devido às taxas de reprovação em matérias de exatas.

As matérias de exata lá são uma piada. Professores passam questões idênticas às das listas de exercícios. Estudantes não sabem somar vetores nos últimos períodos.

Não estudamos função trigonométrica em cálculo. Na verdade, a matéria foi tão decepada que agora se chama "matemática".

Em algum momento vou comçar a expor os podres do currículo universitario.

[Cavalli @dancavlli]

4 days and it's actually good

[Cavalli @dancavlli]

Codex 200x is amazing

[Cavalli @dancavlli]

@lauriewired @HSVSphere With the right tooling around it K8S is way easier to maintain then the options

And there are a lot of tools for one to choose from, it's just that mature of an ecosystem

[Cavalli @dancavlli]

@joaosenzi Po, é R$ 2.59 um novo...
Eu preferi não arriscar, os aqui de casa eram de 2025 tmb