thefourthchime

6 ms·

61.
by thefourthchime 6mo ago
It's like they TRYING to get people to move to GPT 5.4, which I trust far more at coding. It's just slightly more annoying as an agent with codex.
62.
by thefourthchime 6mo ago
Wow, that sounds like you have a astoundingly terrible EVP.
63.
by thefourthchime 6mo ago
5.4 is smarter than Opus when it comes to really figuring out a problem. Codex agentic stuff takes forever though.
64.
by thefourthchime 6mo ago
hahah. It's slop all the way down.
65.
by thefourthchime 6mo ago
I didn't get into the details too much, but I kept thinking, why isn't he just having an agent discover things from various data sources? I've had much better success with that.
66.
by thefourthchime 6mo ago
The workflow that got me into Cloud Code was instructing it that whenever I create a new feature or bug, it should make a new git worktree. And then when I'm done, merge that back to main and delete the worktree. That enables me to ope
67.
by thefourthchime 6mo ago
Maybe some more distant future. For me, I'm still struggling with the hallucinations and screw-ups that the state-of-the-art models give me.
68.
by thefourthchime 7mo ago
I won’t use anything less than the SOTA. It tried using Opus 4.6 medium and immediately regretted it. High messes up enough.
69.
by thefourthchime 7mo ago
I had a recent experience with the Lowes agent today. It was pretty decent! Until I asked "how many of that item is available", and it didn't know how to answer that (It was a clearance item). At least when I asked to talk to
70.
by thefourthchime 7mo ago
This is very cool!! but a test video I did and I played it back on Safari, the video playback was very, very choppy (m2 air). Is this a known issue?
71.
by thefourthchime 8mo ago
Out of curiosity, how long did this take to write and what AI stack did you use?
72.
by thefourthchime 8mo ago
The 45k is a myth for now. The vehicles that have been reviewed so far are going to be $60k+ performance models. We'll see if they actually get down to 45k. From the analysis I've seen with that drag coefficient, the 45k vehicle i
73.
by thefourthchime 8mo ago
"The miracle is not that the bear can dance well, it's that the bear can dance at all." - Old Russian proverb.
74.
by thefourthchime 8mo ago
Grok is pretty good too!
75.
by thefourthchime 9mo ago
From what I've heard, what few interviews there are for software engineers these days, they do have you use models and see how quickly you can build things.
76.
by thefourthchime 9mo ago
Software development hiring is terrible right now, but hiring has been pretty slow in general. We gained 2 million jobs in 2024 and only 500,000 in 2025.
77.
by thefourthchime 9mo ago
Autopilot / FSD was a mess. Autopilot is very old tech and people confusing "self driving" with it, which it's not. We'll see how many pony up for FSD, but I think the play is to force people to try it.
78.
by thefourthchime 9mo ago
Tesla is going to stop selling FSD outright on Feb 14th. It will be subscription only.
79.
by thefourthchime 9mo ago
I can tell you the light rail in Austin is a complete failure. There was some ridership before the pandemic, but after a few years, the numbers are dismal. They've covered the windows with ads, so you can't even tell how empty the
80.
by thefourthchime 10mo ago
I completely agree. I can't even imagine using a local model when I can barely tolerate a model one tick behind SOTA for coding.
81.
by thefourthchime 10mo ago
I've used all of these tools and for me Cursor works just as well but has tabs, easy ways to abort or edit prompts, great visual diff, etc... Someone sell me on how Claude Code, I just don't get it.
82.
by thefourthchime 10mo ago
My father, who never did any public speaking, and as much an introvert as you'll find, did this for my wedding rehearsal. I was amazed at how naturally and well he did. All he wrote down were 6-7 topics to talk about. He got a huge app
83.
by thefourthchime 10mo ago
This is exactly how I write code. I never engineer anything until I have to. I don't try to get rid of code duplication until it works. And I try to be as "least clever" as possible.
84.
by thefourthchime 10mo ago
I've worked on a playout system for broadcast television. The software has to run for years at a time and not have any leaks, We need to send out one frame of television exactly on time, every time. It is "C++", but we also f
85.
by thefourthchime 10mo ago
My test of a new model is always: "Generate a Pac-Man game in a single HTML page." -- I've never had a model been able to have a complete working game until a couple weeks ago. Sonnet Opus 4.5 in Cursor was able to make a ful
86.
by thefourthchime 10mo ago
Isn't that what GPT 4.5 was?
87.
by thefourthchime 10mo ago
The price of inference has been dropping like a rock. I wouldn't expect that 2c to be true in a couple of years.
88.
by thefourthchime 10mo ago
As a serial DIYer, I respect the engineering depth here, especially the custom vector index, but I disagree on the self-hosted ML approach. The innovation in embeddings is just too fast to keep up with locally without constant refactoring.
89.
by thefourthchime 10mo ago
The article glides over the fact that FMVSS 226 is a performance standard, not a materials mandate. Manufacturers can stick with tempered glass if they beef up the side curtain airbags enough to prevent ejection, which is exactly what happe
90.
by thefourthchime 11mo ago
I like to ask "Make a pacman game in a single html page". No model has ever gotten a decent game in one shot. My attempt with Gemini3 was no better than 2.5.

More ›