Martin

@mengxi Banger

Turns out AI recreated the oldest org structure in existence: one expensive person who plans and never touches the code, and cheap labor that does everything. $1.3k vs $20k for the same passing test suite. Middle management stays undefeated.

@IterIntellectus Ironically ”after” has one count of plastic and before has none ;)

the beauty of a place is inversely proportional to the amount of plastic you can see

@PaulSkallas @firstnamelastga American exceptionalism. Soccer wasn’t invented in the US. It’s even in the name ”soccer” - which is different to from what it’s called everywhere else

@MPorkholm @firstnamelastga Yes but why can't you? Go deeper

@PaulSkallas @firstnamelastga You can’t become rich playing soccer in the us. Sports is a way out of poverty for them

@firstnamelastga But why dont they play soccer? Where does that bias come from?

@reed_rawlings @ericjang11 I do. It’s great

@ericjang11 Never met a snowflake customer that likes streamlit

very smart for Streamlit to have sold to Snowflake in march 2022 for $800 million

@nikitabier Not surprising. Best part of X is the surprising high entropy content. Revenue maximizers are not that

In a 3% experiment, removing the Top-30 highest paid revenue share accounts from the For You timeline increased both time spent and daily active users on X.

@__tinygrad__ Why do sales peak in may?

The new personal computer revolution is just beginning.

@giffmana @paulg Nailed it

@paulg You could have stopped at the first comma :)

I strive to make my writing unsummarizable, in the sense that it has so little fluff left in it that if you take any words out, as summaries by definition do, you lose a lot of interesting ideas.

@petergyang We could call it README.md 👀

It's pretty annoying that Codex uses agents.md and Claude Code uses Claude.md. There should be some industry standards to this stuff?

@yacineMTB Bait

Just let your baby cry it out. It's fine

@MrsCMFrancis This is a good one. Bookmarked

I teach my kids from about age 3 to come put their hand on my arm or leg if they want to speak to me while I’m speaking to another person. I then put my hand over theirs to show them I see them and I’ll get to them soon. Manners are so important to instill young.

@andrewmccalip I think u need a tiny laser here

Immaculate vibes from the deck of Project Bob So far the score is: autonomous boats: 0 ocean: 3

@___Harald___ It’s not like they process video like u guys, so doubt they need that much compute?

Surprisingly little GPUs. We have ~700 GPUs in our cluster, only one OOM less. And we pay for it by selling cool products instead of skimming off the economy!

@ID_AA_Carmack These kinds of posts is why I am on this platform and not linkedin

Space launch was a clear case where there was a large difference in efficiency between what was possible and what was done in practice before SpaceX. A large part of that was due to everything being locked in to what (just barely) already worked, with huge risk aversion. WIth national prestige or a half billion dollar geosync satellite on the line, speculative engineering ideas that might result in a public debacle were not welcome. When failure is not an option, success can stay very expensive. You need to experiment to improve, and that fundamentally means being comfortable with failure. If you know it is going to work, it isn’t an experiment. I have long believed that nuclear power today is in precisely the same state as space launch two decades ago, but the even more pressing question now is if semiconductor fabrication might also be. On the one hand, Moore’s Law has been a sequence of heroic miracles of technology at the wafer fabrication level, grinding out hundreds of compounding small improvements. On the other hand, fabs are “too big to fail”, and there are elements of extreme conservatism at play. Intel’s “Copy exactly!” fab development exemplifies that mindset – instead of every new building being an opportunity to explore and optimize processes, it was deemed more valuable to just replicate. While each individual machine may be straining against physical limits of technology, it is possible that the systems orchestrating them all together could be far from optimal. The explore / exploit axis is fundamental to all decision making, but human risk avoidance probably biases away from optimal exploration.

@___Harald___ She looks just like you!

I now understand all the people who were inspired by their kids to do great work. This is my muse, and I want to build an exciting future for her.

@yacineMTB To be fair. Most people probably didn’t notice. They are optimizing for the mean of the normal distribution, not the outliers

"we changed the default from high (expensive for us, better for you) to medium (cheaper for us, worse for you, but we don't think you are smart enough to notice) because we wanted to reduce the time it takes for you to get an, albeit worse, answer"

@giffmana Dinner with dad is always pleasant

Long weekend back home, working hard on getting that cholesterol level back up.

@levelsio Me!

Okay and who of you couldn't reply to this? Please please only if you couldn't okay?

@effectfully Higher interest rates baby

>The equity doesn't pay anything unless you sell it We went from "why would you go after revenue" as a meme to tech founders seriously not comprehending that businesses can make money and pay it out as salaries/bonuses/dividends.

@justalexoki Yes

i 100% guarantee anthropic are routing requests to dumber models at certain times