People still fall for the same old pattern: Launch a shiny new thing >> leave building strong foundations for later >> masses arrive >> foundations can't handle the scale >> Shiny new thing breaks. I prefer a different path: Build strong foundations >> bring in the masses >> live happily ever after.

On a long enough time horizon, nothing can beat building from First Principles. Sure you can take shortcuts short term and try to go with the "easy" path. I prefer building it right, with the correct choices, even if it means taking longer and facing hard challenges. This is one of the main reasons why I believed in StarkWare vision back in the days. I always felt this mindset of building things from first principles, not trying to fit a square peg into a round hole. This is why Starknet is unique, this is why Cairo, this is why STWO. It took much more time building everything from scratch, but now it means a path of full sovereignty over our roadmap and infra. Results are already visible and now that we have those extremely solid foundations and battle tested tech the fun will begin. We will be able to demonstrate truly the power of STARKs. My personal obsession is about leveraging STARK proofs to help accelerating hyperbitcoinisation of the world and also about empowering freedom tech (like Nostr, I am extremely excited about the perspectives of building permissionless marketplace of digital services with Nostr DVMs) and privacy with zero knowledge proofs. It's about building the Integrity Web, a world where you trust math and not middlemen. Bitcoin will fix the money. Nostr will fix the communication. STARKs will fix the problem of integrity of computation globally.

Thoreau, what a line

Nice short reading: "What Is Geometry?" 1) Axioms (Euclid) 2) Coordinates (Descartes, Fermat) 3) Calculus (Newton, Leibniz) 4) Groups (Klein, Lie) 5) Manifolds (Riemann) 6) Fiber bundles (E. Cartan, Whitney) "A property is geometric, if it does not deal directly with numbers"

The sole purpose of leisure travel after age 30 is to remind yourself that you’ve lost all childlike joy and the only thing that excites you is aggregating land and resources and watching number go up.

RT @kukreja_abhinav: Eshan Chattopadhyay is the latest Godel prize winner!

*puts on nerd glasses* umm actually, she only shown that it's SO(2)

We created a canvas that plugs into an image model’s brain. You can use it to generate images in real-time by painting with the latent concepts the model has learned. Try out Paint with Ember for yourself 👇

unemployed starter pack

Today @GoogleDeepMind released AlphaEvolve: a Gemini coding agent for algorithm discovery. It beat the famous Strassen algorithm for matrix multiplication set 56 years ago. @Google has been killing it recently. We had early access to the paper and interviewed the researchers.

So you’re telling me that Deepseek with private funds can release an open source model, but govt awarding Rs 220 crores of public funds to Sarvam isn’t asking for the same? This is tax payers money, so the full pipeline ought to be open source!

No. Physics is about finding the shortest *encoding function* that maps from concrete physical states of the universe onto abstract representational states of our internal models. To see trivially why @beffjezos's definition doesn't work... (1/2)

Realising that very few folks in India actually understand how frontier AI research works. Hence, @smallest_AI is considering having community research discussions every weekend in Indiranagar. Here's how it will operate: 1. We will float a problem statement 2. We will come up with 4-5 solutions (you can bring your own here) 3. We will have a debate on which is the best solution to take up 4. You can join in, bring your own solutions, or just take notes 5. We disperse with strong action items to be discussed next week All discussions of course go on YouTube as well. Would you be interested in joining something like this? If there are over a 1000 people interested, then we'll make it happen.

you're telling me you can vibe code this?

It's hard to believe, but due to H100 restrictions, DeepSeek was forced to train R1 manually, with thousands of Chinese citizens holding flags to act as logic gates.

The issue with CS is that people think the degree is a web dev degree.

DeepSeek Engineers

China releasing a MIT license model that is on par with o1 and 30x cheaper was not on my bingo card

As usual I will give my x earnings to someone who likes this post. Must follow this account to dm info. Will choose winner randomly Sunday am. Good luck and thanks for following!

Coming from an Indian tech founder