Behind the products I love are those who pour their hearts into what they do.
The most beautiful thing is that both @dhh and @mitchellh have kids, families, and interesting hobbies. They build great things while living a whole life.
Thank you for great products & inspiration.
This is the feeling I'd not been able to express until @mitchellh made it so clearly.
Sometime I feel lucky that I built every single line of code for my products before AI coding is a thing.
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.
@bcherny A feedback
After months with Opus 4.7, xhigh. I see that Claude Code has very high level of self-reflection. It think, and self-criticize to fix its thought process. Or it builds then self-reflect (after my complains) to bring better version.
So, in 4.6: Initial ideas/prompts -> Faster to final product -> quality 60-70%
with 4.7: Initial ideas/prompts -> Very long chain of thoughts -> Some things not good -> Very good re-thinking process with minimal complains -> 90-95% quality.
So ultimately I can have better product. But it's hard to say I'm totally happy. It's the feeling that "Claude you should made it right from beginning". My feedback/complains in the loop are minimal, I feel that Claude should be able to ask those question itself and have better results during first iterations.
It's not conflict with my feedback on "high level self-reflection" - 4.7 did a very good job, it's just (I feel) missing some tiny thing to trigger that self-reflection and self-correction.
It's time for interfaces to break out of flat land.
Lately, I have been experimenting with spatial experiences supported by foundational gestures.
Truly feels magical.
And advsertising you can send these details all in one tap. No it's not "in-one-tap". It's 3-5s when you move focus from the road to review that address details.
It's true, 100%
The only problem is that I feel boring. I can have my jobs done. But with 50% of my day facing digital products, it makes my day so boring.
Stuff like Shadcn make building UI.UX so easy. Just need a bit more taste from builders.
Unpopular opinion: I don’t care if most web apps look the same. All I care about is whether it does what it says and does it fast.
Make it fast. Make the UX obvious. Put the right things in the right place and little to no animations.
Update: Socket has found 121 more compromised npm package artifacts across 84 package names, including 64 UiPath artifacts.
Combined w/ TanStack, the current known total is 205 affected npm package artifacts across enterprise automation, AI/MCP, auth, workflow, and dev tooling.
When I was at Apple, I loved working on micro interactions that you see all over the OS. Now that I’m not an apple I still like to solve for these little problems that really annoyed me. In this case, I designed a backspace button with a speed controller, so by just pressing it you can delete by letter and then immediately by word as you stretch it, without having to wait (like it usually does on the OS) and then if you stretch a little more, you can speed delete through words… I’m also working on another one where you can repair the words if you over-deleted it by accident 😜 (it also has haptic feedback, which makes it really fun)
🚨 Bitwarden CLI 2026.4.0 was compromised as part of the ongoing Checkmarx supply chain campaign after attackers abused a GitHub Action in Bitwarden’s CI/CD pipeline.
We’ll continue updating our coverage as more details are confirmed.
https://t.co/G0aakn8swq
Do you even understand what this means?
An open source model just released that is:
• Outperforms models 20x its size
• Can run on a base model Mac Mini
• Is AMERICAN 🇺🇸
If you have a base model Mac Mini you can have unlimited super intelligence on your desk. For free.
Sonnet 4.5 was released 5 months ago
In 5 months that level of intelligence went from frontier to free on your desk
And not only that, can run on any basically any computer out there
If you have even a remotely modern computer, do the following immediately:
1. Download LM Studio
2. Go to your OpenClaw and ask which of these new Gemma 4 models is best for your hardware
3. Have it walk you through downloading and loading it
4. Build apps with it knowing you are using your own personal, private super intelligence on your desk
The people denying this is the future are so beyond lost.
🚨‼️ BREAKING: Adobe has been breached by threat actor Mr. Raccoon, leaking 13 million support tickets with personal data, 15,000 employee records, all HackerOne submissions, internal documents and more.
Mr. Raccoon gained access through an Indian BPO, first deploying a remote access tool on an employee, then phishing their manager.
Mr. Raccoon told us: "They allowed you to export all tickets in one request from an agent."