Pinned Tweet
Life update: I'm currently unemployed due to an acquisition (decided not to join the acquiring company)
I've been working on a DAW for the web. Like a baby Garageband that runs in your browser. It's starting to take shape!
nitter.gq/aykev/status/172456658…
Apple thrives post-pop on hardware prices, reasonable capex and the slow but steady rise of OSS local models
17
really interesting how some Claude sessions are very good, on top of everything, super effective, etc
while others are just not great, make more mistakes, get blocked or fail to notice things, etc
1
3
171
Kevin Chavez retweeted
i'm excited to share a new free train building game! it runs in your browser.
it's very optimized, and can handle 2000 stations and 100M daily riders smoothly, so you can build a continent-scale network.
here is one of my saves:
Gringos will make up all these obtuse ways analyze elections
Guys; there’s two buckets. Working class and bourgeoisie. Everything else is correlation
2
71
Esto no es cierto. Un centro de datos produce muy pocos empleos permanentes. Los mas grandes entre 100 y 300 personas, y en su mayoría nisiquiera son empleos de informática
Una proporción terrible de empleos/m^2, empleos/MWH, etc. Por algo no los quieren en EUA
2
56
Kevin Chavez retweeted
Una década perdida y años de dolor innecesarios para que este sinvergüenza salga a admitir tan campante que él orquestó el fraude de 2006. Que la derecha de este país nunca se atreva otra vez a vender cuentos sobre su compromiso con la democracia.
building a native spotify client bc I got tired of loading states and whitespace
(very) work in progress
1
99
hardware ceo: central inference is a terrible allocation of important resources. from a global allocation perspective, you're putting all your eggs in one basket, using internet traffic. you want maximum decentralization for a technology this important, you get better latency too
sam: local inference is a terrible allocation of scarce resources
“a lot of people want local inference, and somehow they assume that the problem is solved now because we have open-source models, we're going to be able to run them on our computers, and so we're not going to need all the stuff in the data centre
but from a global allocation perspective, actually it's quite terrible to have everybody, like, have a bunch of the world's limited supply of DRAM sitting in their own computer and being used five percent of the time
you want maximum utilisation in the cloud if you're trying to deal with this very supply-constrained environment
somehow many people's intuition is the opposite of that”
1
4
305
































