Rohan

I don’t know anything about AI. Carnegie Mellon. Computers. Healthcare. Products. Economics. Travel. Politics. History.

[Rohan @ygivenx]

@elonmusk @RapidResponse47 @POTUS I am against any war but travelling across Europe this summer, I saw a lot of Ukraine license plates and they were all luxury cars. Not what i expected from a war torn country.

[Kiran Bedi @thekiranbedi]

Fantastic and National Gratitude.
May you continue to succeed.
Key is integrity of purpose with determination.
#SwarnPrabhat IPS and your team.

Fight against drug mafia: Bihar IPS Swarn Prabhat’s two-year crackdown yields 100 convictions, 52 crore drug seizures https://t.co/ijSrMytykp

[Mitchell Hashimoto @mitchellh]

Claude Code v2.1.295 supports OSC 7501 (program status). Incredible speed @AnthropicAI supporting this protocol which will help hundreds or thousands of downstream applications! They had access to Rex and verified through Rex it works. 🥰

[Rohan @ygivenx]

@karpathy

[Brian Lui @brianluidog]

I think programmers kind of underestimate their vibe coding advantage versus non-programmers. I know a guy who spent 4 billion tokens to make a chess webapp.

[Andrej Karpathy @karpathy]

Cool eval. Simply ask an LLM “Land or Water?” and give it a latitude and longitude coordinate as text. Ask 16,200 times, plot as image. The models know. From compressing the internet.

[Sidharth @Cloudwatch199]

🇮🇳 REMEMBER THIS MAN: CAPTAIN SMIT MACHCHHAR.

An Indian captain flying hundreds of people from Dubai to Tel Aviv was stabbed in his own cockpit.

Wounded, bleeding, with the aircraft plunging thousands of feet, he fought back.

Then passengers and crew did something extraordinary:

They stormed the cockpit.
They restrained the attacker.
Other pilots took control.
And that plane LANDED.

Hundreds of families got their loved ones back today because ordinary people refused to freeze when everything went wrong.

Captain Smit Machchhar — India and the whole WORLD should be proud of you. 🇮🇳

And to the passengers and crew who ran toward the danger instead of away from it:

Heroes. Every single one of you.

This is what courage looks like.

[Garry Tan @garrytan]

Personal AI is in its Homebrew computer club era but it won’t stay there

Truly personal AI will mean you need to own your own skills and memory

[Rohan @ygivenx]

@aravind Even DS wants to do politics in clean air and water. You’re closet government.

[Anand Ranganathan @ARanganathan72]

We have an 8000 year-old culture to showcase and yet while one city - Paris - attracts 25 million foreign tourists a year, an entire country - India - gets 11 million. We haven’t plucked the lowest-hanging fruit. We are watching it rot.

My views on tourism, on World Tourism Day:

[Rohan @ygivenx]

@aravind @jarawayama You are one fing idiot. Focus on what India does well not how everything is one big scandal.

[Rohan @ygivenx]

@p_sahibsingh GenZ should slap him back #noslapsunanswered.

[Mitchell Hashimoto @mitchellh]

I've got an agent in a loop optimizing a renderer with the goal to minimize frame times (and tests to measure). It got times down from 88ms to 2ms and allocations down from ~150K to 500. Sounds good, right? Wrong. This is exactly why agent psychosis is a big fucking problem.

As an experiment, I rewrote the Ghostty core render state in Go, with access to identically laid out data structures as Ghostty and the exact same validation tests. I made a purposely naive renderer (simple, correct, but slow). 88ms per frame with 150,000 allocations (horrendous, lol)!

I then kickstarted a Ralph loop to bring the frame times down. I told it it can't modify input data structures or the public API or tests (they're correct), but it can do anything else it wants. It got to work.

It has worked for about 4 hours. I've spent around $350 on this experiment so far. The results?

88ms => 1.5ms
150K allocs => ~500 allocs

Incredible right? Nope.

My hand-written renderer I ported has frame times (same benchmark) of ~20us (0.020ms) and 0 allocations in the update path.

This is the problem with psychosis and lacking systems understanding. If you don't understand the system, you're going to accept that this is an incredible result. If you understand the system, you'll see better solutions immediately and can do roughly 75x better on throughput.

The people who blindly trust agent output are in the former camp. They're sheeple, overdrinking from a fountain of mediocrity.

Standard disclaimer: I use AI all the time. I like AI. The point I'm making is to not blindly accept results. Think. Analyze. Learn.

[Rohan @ygivenx]

@sabeer Income tax commissioners will become billionaires 😂

[Rohan @ygivenx]

@cszhu Sleeping 💤

[Rohan @ygivenx]

@sabeer India has to create a biasless society is a better framing.

[Rohan @ygivenx]

@CuteAaruhi4 It is simple math.

[Rohan @ygivenx]

@trq212 @ycombinator Never join politics Thariq!

[Rohan @ygivenx]

@trq212 *stable version

[Rohan @ygivenx]

@trq212 When will claude code get the next stable for opus 5.5?