Comment by mattmaroon
1 month ago
The codebase is old and really hard to work on. It’s a game that existed pre-iPhone and still has decent revenue but could use some updating. We intentionally shrank our company down to auto-pilot mode and frankly don’t even have a working development environment anymore.
It was basically cost prohibitive to change anything significant until Claude became able to do most of the work for us. My cofounder (also CTO of another startup in the interim) found himself with a lot of time on his hands unexpectedly and thought it would be a neat experiment and has been wowed by the results.
Much in the same way people on HN debate when we will have self driving cars while millions of people actually have their Teslas self-driving every day (it reminds me of when I got to bet that Joe Biden would win the election after he already did) those who think AI coding is years away are missing what’s happening now. It’s a powerful force magnifier in the hands of a skilled programmer and it’ll only get better.
I agree that code is being written in exactly the same sense that Teslas are driving themselves.
Yes, in both cases it takes someone to steer. It’s not a complete solution. Someone who can’t drive can’t just ride a Tesla around town and someone who can’t program can’t vibe code anything complex.
But if it can do 90% of the work for you, it is a serious force magnifier.
> But if it can do 90% of the work for you, it is a serious force magnifier.
Well, we could characterize a Tesla as doing 90% of the work but it's not at all a force multiplier. Your "10%" supervisory contribution takes just as long as doing 100%.
You didn't mention it was rewriting the codebase from scratch. That's the consensus, that AI is only good at scaffolding.
Oh it can't do 90% of the work for you. It CAN type 90% of the work for you, but someone still has to read the code and know what the best course of action is, supposedly...... I suppose some people never learned to use their IDEs or to touch type so as to find LLMs such a crazy productivity boost.
Do you have tests at least? Seems reckless to yolo the codebase if you don’t or can’t test easily.
He’s simply building an entirely new app. The old codebase plods on untouched.
Millions? Of people in self driving teslas?
The actual number of such vehicles produced is two orders of magnitude less.
No it isn’t. They sold 1.6 million vehicles last year and every one of them can self drive. At least once a year they give out free access to it and I don’t know a Tesla owner who hasn’t tried it.
You’re right that it doesn’t happen every day but that doesn’t change the point, you all are debating whether something can happen after it already happened.
What has already happened is that you unfortunately have manipulated the facts. How long were you planning to not reveal what you really mean: that some drivers have an oportunity to use self-driving once year and that they allegedly tried it, althought that's really just opinion based on your own preference. That's much different than claiming millions of such drivers roaming everywhete at any moment of time.
Plus, the idea that every single Tesla has this option is just your rought estimate, right?
> you all are debating whether something can happen after it already happened
My goalposts have always been on levels 3 and 4.
Elon Musk has been promising level 4, sometimes level 5, this entire time.
It has not already happened. The cars are level 2.
Sounds like a good reason to rewrite. And sounds like a rewrite just would not happen by any other means. Thanks for sharing the details.
When I say I want a self driving car I mean one that actually drives itself so I don't have to be involved other than setting the destination.
What Tesla is selling now is the worst of both worlds. You still have to pay attention but it's way more boring so it's really hard to do so. Well until it suddenly decides to ram a barrier at highway speeds.
Wake me up when I can have a beer and watch a movie while it's driving.