Comment by antirez
9 hours ago
I believe we will see a new huge wave of useful open source software. However don't expect the development model to stay the same. I was finally able to resurrect a few projects of mine, and many more will come. One incredible thing was the ability to easily merge what was worth merging from forks, for instance. The new OSS will be driven not much by the amount of code you can produce, but from the idea of software you have, how the software should look like, behave, what it should do to be useful. Today design is more important than coding.
The real question is how much of the new wave of vibe-coded software will be able to graduate from pet project to community-maintained project.
It feels that vibe coding may exacerbate fragmentation (10 different vibe-coded packages for the same thing) and abandonment (made it in a weekend and left it to rot) for open source software.
I believe the process of accumulation of knowledge / fixes / interesting ideas will be still valid, so there will be a tons of small projects doing things that you can replicate and throw away, but the foundational libraries / tools will be still collaborative. But I don't agree with the idea of fragmentation, AI is very good at merging stuff from different branches, even when they diverged significantly.
I don't trust software that has .claude in its GitHub repo.
You won't have to ignore this stuff for long. Pretty soon it'll be mandatory to keep up.
I've been a senior engineer doing large scale active-active, five nines distributed systems that process billions of dollars of transactions daily. These are well thought out systems with 20+ folks on design document reviews.
Not all of the work falls into that category, though. There's so much plumbing and maintenance and wiring of new features and requirements.
On that stuff, I'm getting ten times the amount of work done with AI than I was before. I could replace the juniors on my team with just myself if I needed to and still get all of our combined work done.
Engineers using AI are going to replace anyone not using AI.
In fact, now is the time to start a startup and "fire" all of these incumbent SaaS companies. You can make reasonable progress quickly and duplicate much of what many companies do without much effort.
If you haven't tried this stuff, you need to. I'm not kidding. You will easily 10x your productivity.
I'm not saying don't review your own code. Please do.
But Claude emits reasonable Rust and Java and C++. It's not just for JavaScript toys anymore.
- - - - - - - - - - - -
Edit:
Holy hell HN, downvoted to -4 in record time. Y'all don't like what's happening, but it's really happening.
I'm not lying about this.
I provided my background so you'd understand the context of my claims. I have a solid background in tech.
The same thing that happened to illustration and art is happening here, to us and to our career. And these models are quite usable for production code.
I can point Claude to a Rust HTTP handler and say, "using this example [file path], write a new endpoint that handles video file uploads, extracts the metadata, creates a thumbnail, uploads them to the cloud storage, and creates the relevant database records."
And it does it in a minute.
I review the code. It's as if I had written it. Maybe a change here or there.
Real production Rust code, 100 - 500 LOC, one shotted in one minute. It even installs the routes and understands the HTTP framework DSL. It even codegens Swagger API documentation and somehow understands the proc macro DSL that takes Rust five minutes to compile.
This tech is wizardry. It's the sci fi stuff we dreamed of as kids.
I don't get the sour opinions. The only thing to fear is big tech monopolozation.
I suppose the other thing to worry about is what's going to happen to our cushy $400k salaries. But if you make yourself useful, I think it'll work out just fine.
Perhaps more than fine if you're able to leverage this to get ahead and fire your employer. You might not need your employer anymore. If you can do sales and wear many hats, you'll do exceedingly well.
I'm not saying non-engineers will be able to do this. I'm saying engineers are well positioned to leverage this.
I'm not saying that you shouldn't use AI.
There was a submission to a blog post discussing applications of AI but it got killed for some reason.
https://news.ycombinator.com/item?id=46750927
I remain convinced that if you use AI to write code then your product will sooner or later turn into a buggy mess. I think this will remain the case until they figure out how to make a proper memory system. Until then, we still have to use our brains as the memory system.
One strategy I've seen that I like is using AI to prototype, but then write actual code yourself. This is what the Ghostty guy does I believe.
I agree that AI can write decent Rust code, but Rust is not a panacea. From what I heard, Cursor has a lot of vibe-coded Rust code, but it didn't save it from being, as I said, a buggy mess.
2 replies →
> Holy hell HN, downvoted to -4 in record time. Y'all don't like what's happening, but it's really happening. > > I'm not lying about this. > > I provided my background so you'd understand the context of my claims. I have a solid background in tech.
There are lots of people claiming this. Many of whom have a solid background. Every now and then I check out someone's claim (checking the code they've generated). I've yet to find an AI-generated codebase that passed that check so far.
Perhaps yours is the one that does, but as we can't see the code for ourselves, there's no way for us to really know. And it's hard to take your word for it when there are so many people falsely making the same claims.
I expect a lot of HNers have had this experience.
2 replies →
> "The same thing that happened to illustration and art is happening here"
What are you talking about? Illustrators and artists are not being replaced by AI or required to use AI to "keep up" in the vast majority of environments.
> "I don't get the sour opinions."
The reasoning for folks' "sour opinions" has been very well-documented, especially here on HN. This comment reads like people don't like AI because they think it's slow or something, which is not the case.
5 replies →
> Holy hell HN, downvoted to -4 in record time. Y'all don't like what's happening, but it's really happening.
I gave you an upvote FWIW, after all, I mean, my job's codebase is already a buggy mess, so it doesn't hurt to throw AI on it, which is what I do.
> You might not need your employer anymore. If you can do sales and wear many hats, you'll do exceedingly well.
Wasn't this the case before AI as well?
1 reply →
> I've been
so not now, then?
3 replies →
We need a new git. (could be built on the current git)
> One incredible thing was the ability to easily merge what was worth merging from forks, for instance
I agree, this is amazing, and really reduces the wasted effort. But it only works if you know what exists and where.
More we need a new GitHub.
Also this.
But IMO the primitives we need are also fundamentally different with AI coding.
Commits kind of don't matter anymore. Maybe PR's don't matter either, except as labels. But CI/hard proof that the code works as advertised is gold, and this is something git doesn't store by default.
Additionally, as most software moves to being built by agents, the "real" git history you want is the chat history with your agent, and its CoT. If you can keep that and your CI runs, you could even throw away your `git` and probably still have a functionally better AI coding system.
If we get a new Github for AI coding I hope it's a bit of a departure from current git workflows. But git is definitely extensible enough that you could build this on git (which is what I think will ultimately happen).
May I recommend SourceHut (https://sr.ht/)
2 replies →
There are a pile of alternatives which have similar UIs.
Jujutsu?
Jujutsu has been the tool that actually got me into making full use of version control software. Before, through multiple attempts at grasping at the deeper fundamentals, I only learned the bare minimum git commands I needed to make commits and branches, and very careful merges. Jujutsu maps to a much clearer and simpler mental model. Blockchains are nifty and all, but awfully inconvenient to work with as meatbags.
That sounds a little extreem, why not just a new auto merge feature?