hazn

@trq212 @badlogicgames @mitsuhiko i would translate it as "the age its art, the art its freedom"

great naming, i love the vienna school of economics and the vienna secession movement: en.wikipedia.org/wiki/Secession… the vienna secession movement dropped the all-time banger quote of: "To every age its art, to every art its freedom" (German: Der Zeit ihre Kunst. Der Kunst ihre Freiheit). @badlogicgames @mitsuhiko

People of pi. BIG NEWS. I've sold out. Let me know how you feel about this in the comments below. mariozechner.at/posts/2026-04-…

@badlogicgames double congrats then!

@hazn_com yes, openai officially supports using codex subs in pi.

@Sirupsen just buy them my man, we are allowed to have nice things

my dream in life has always been to have coloured plates at home I bought one pair and could not get myself to buy anymore. colours are soooooo expensive I am enjoying them very much for some light morning cleans though 🥲

@michael_nielsen in a messed up way, jihadists. in a less messed up way, some tribes in indonesia: mooremisadventures.com/blog/tana-tora…

Curious: which cultures are most comfortable talking openly about death? Which are least comfortable? Pointers welcome, especially when accompanied by evidence!

@GrantSlatton alwys with an open mouth, impressive

@elliotarledge i use claude code, for it's banger defaults for JUST GOING, i use pi if i want no defaults/pure autism/super specific task mode, with gpt 5.4 xhigh

ive thought more about this so called "harness/model moat" where openai (codex-cli) and anthropic (claude code) can build a model around the harness and vice versa, rendering those tools as the best for those models. you would assume droid, cursor, and others to be behind on general performance. the thing is, what if anthropic isn't hyper-optimizing the harness constantly, and they have other focuses instead? droid for instance, might be constantly changing their tool such that developers have to adapt super fast to new workflows, but the harness team at factory is faster and more tasteful than anthropic's harness team. this is just food for thought. im getting some good vibes from @0xSero but i will have to see for myself. i love all the tools and hate them at the same time. ill stop yapping as i do want to hear what YOU are using and why.

On intelligence Knowledge is knowing your discarded toenails contain keratin, which in turn, contains nitrogen. And that your living room plants need nitrogen. Wisdom is not putting your toenails in the plant pot. Intelligence is making the leap from knowledge to wisdom. Or rather, it’s having the sensibility to put the toenails there, but having the wisdom to bury them deep in the pot, so people don’t see what you did. And intelligence is still not doing this in front of your friends, or talk about this in front of your friends. Maybe intelligence is hard-to-define, but it for sure entails the skill to not talk about this publicly. “Intelligence is what AI is not able to do,” they say ... but even AI will tell you to not talk about your toenails in your flower pot, so ... on a second thought, if you have hippie friends they might see what you do as resourceful, and it might be genuinely status increasing and inspiring that you put your toenails into flower pots. Ah well, this isn’t the neat conclusion I wanted to reach. Probably intelligence is heavily context dependent, and path dependent. And dependent on if you actually put your toenails into that pot or not, and if you behaved wisely while doing it. I guess intelligence is more of a verb? hazn.com/intelligence

Bucket list: 40 for 40 Two months until thirty. Ten years until forty. Forty things I want to experience in between: 1. Hold a talk or an event at the Colosseum 2. Win a BJJ match 3. Attend a Sufi dhikr ceremony 4. Play an official soccer match again 5. Live in California again, specifically the bay area 6. Attend a jhana retreat 7. Learn how to sharpen knives properly 8. Be in the best shape of my life (the leanest or the most effective I have ever been) 9. Host at a technical conference 10. Learn a Latin language 11. Have someone come up to me and tell me something I wrote or said inspired them 12. Learn Arabic, read books from the golden age of Islam 13. Hold Swiss, German, and Turkish passports 14. Have equity in something, acquire land 15. Be married 16. Have kids 17. Leave something that gave me enormous success because the marginal loss was greater 18. Achieve enlightenment/fana, only the beginning 19. Maintain a narrated chronicle of my life, week by week, marking what happened 20. Complete a multi-day water fast 21. Record my parents telling their life story 22. Do improv at an open mic session 23. Run one marathon every year 24. Build a network where any two phone calls can solve almost any problem 25. Write my dad a letter in Turkish that makes him cry 26. Go a full year without missing a single day of publishing 27. Try transcranial ultrasound neuromodulation 28. Go back to Dortmund and give a talk or workshop for kids there 29. Start doing videos 30. Visit Korea, China, Tibet, Nepal 31. Learn my mom's recipes 32. Do a major contribution to a technical project I love 33. Have a real, extended conversation with someone whose watcher also never sleeps 34. Fund someone else's dream with no strings attached 35. Keep a powerful secret and never leverage it 36. Learn three chords and write one song and record yourself singing it 37. Reclaim something abandoned (a neighborhood, a company, a tradition, an idea) and make it alive again, and make it outlive you 38. Build an artifact that is simultaneously machine, art, and prayer 39. Learn dancing on a longboard(youtube.com/watch?v=ynakjV…) 40. Increase liberty and happiness in the world hazn.com/40

@m_ashcroft one pet peeve of mine is that the zettelkasten/second brain people only write about ... zettelkasten/second brains

@nickcammarata don't get me started at test-time inference, which actually runs .... not at test-time!

i like how we've split things into pretraining and post-training such that it sounds like the model never actaully trains

@come__softly life imitates art s16home.com/products/antiq…

This was always very romantic to me

@eternaltxts leave your read receipts on: hazn.com/read-receipts

I’m 21. Give me oddly specific life tips. No general ”surround yourself with positive people” tips. I want the most random, specific advice possible.

@come__softly monarch butterlfy 🦋

Mourning cloak is such a beautiful name for a butterfly

I know in my heart that anthropic is messing with the claude plan limits

how can you not wake up everyday amazed at what ML has become - AI is really real. the wildest dreams of a million nerds came true. i just cannot understand the naysayers.

@swyx stock claude code/codex is almost always the best deal, pi is fine as well

@_wilfredh jj, lean 4, PB testing method, bun

Rust and RISC-V both feel like they've reached critical mass and I'd guess that they'll be used more in 5 years than they are today. What other technologies fit this description?

@viemccoy scan for signs of contempt the person is extruding, contempt is high signal for fucked-up-ness

When you meet someone who is smarter than you, what signals do you use in order to decide if you can trust them? Designing experiments around this, and loss of control, is how we can ensure that humans have autonomy over human-shaped things deep into the future. Open question!