Vibe coding kills open source - https://news.ycombinator.com/item?id=46765120 - Jan 2026 (285 comments)
Just because some things suck, for now, doesn't mean open source is being killed. It means software development is changing. It'll be harder to distinguish between a good faith, quality effort that meets all the expectations of quality control without sifting through more contributions.
Anonymous participation will decrease, communities will have to create a minimal hierarchy of curation, and the web of trust built up in these communities will have to become more pragmatic. The relationships and the tools already exist, it's just the shape of the culture that results in good FOSS that will have to update and adapt to the technology.
* no longer any pressure to contribute upstream
* no longer any need to use a library at all
* Verbose PRs created with LLMs that are resume-padding
* False issues created with LLM-detection by unsophisticated users
Overall, we've lost the single meeting place of an open-source library that everyone meets at so we can create a better commons. That part is true. It will be interesting to see what follows from this.
I know that for very many small tools, I much prefer to just "write my own" (read: have Claude Code write me something). A friend showed me a worktree manager project on Github and instead of learning to use it, I just had Claude Code create one that was highly idiosyncratic to my needs. Iterative fuzzy search, single keybinding nav, and so on. These kinds of things have low ongoing maintenance and when I want a change I don't need to consult anyone or anything like that.
But we're not at the point where I'd like to run my own Linux-compatible kernel or where I'd even think of writing a Ghostty. So perhaps what's happened is that the baseline for an open-source project being worthwhile to others has increased.
For the moment, for a lot of small ones, I much prefer their feature list and README to their code. Amusing inversion.
I'm also not particularly fond of the other extreme of toxic positivity where any problem is just a challenge and everybody is excited to take them on.
Once seems to understate the level of agency people have and the other seems to overstate.
The world is changing. Adapting does seem to be the rational approach.
I don't think Open Source is being killed but it does need to manage the current situation in a way that provides the best outcome.
I have been thinking that there may be merit in AI branches or forks. Open source projects direct any AI produced PRs to the AI branch. Maintainers of that branch curate the changes to send upstream. The maintainers of the original branch need not take an active involvement in the AI branch. If the AI branch is inadequately maintained or curated, then upstream simply receives no patches. In a sense it creates an opportunity for people who want to contribute. It produces a new area where people can compartmentalise their involvement without disrupting the wider project. This would lower the barrier of entry to productively supporting an open source project.
I doubt the benefit of resume-padding will persist long in an AI world. By the very nature of their act, they are showing what they are claiming to do is unremarkable.
The AI-forgery attacks are highly polished, complete with forged user photos and fake social networking pages.
The legitimate code contributions are from people who have near-zero followers and no obvious track record.
This is topsy-turvy yet good news for open source because it focuses the work on the actual code, and many more people can learn how to contribute.
So long as code is good enough to get in the right ballpark for a PR, then I'm fine cleaning the work up a bit by hand then merging. IMHO this is a great leap forward for delivering better projects.
Knowing how to write a database could make one fabulously rich. Now the person who knows how to make and promote a simple crud app backed my MySql becomes the rich one, while the db people beg for donations.
Linux killed Sun/Solaris and SGI Irix
Developers have voluntarily moved further down in the chain of value - now just describing themselves as primarily a business liaison who can translate to code. All the computer whispering necessary to do all this is freely available and digestible for free.
LLMs are just the expected endpoint of this.
Another article written by someone who doesn't actually use AI. Claude will literally search "XYZ library 2025" to find libraries. That is essentially equivalent to how it's always worked. It's not just what is in the dataset.
But the general purpose machinery, the substrait we work on? That's hugely open source today, and will gladly accept and make use of that platform innovation that you can offer up.
The authors talk about it being harder to get traction. And that's both true because of LLMs, and also, has been the case for a while now. Theres so much open source already, so many great tools, that it takes real effort and distinction to stand out & call attention to yourself.
> The LLM will not interact with the developers of a library or tool, nor submit usable bug reports, or be aware of any potential issues no matter how well-documented.
Arn't these interactions responsible for the claimed burn-out suffered by open-source maintainers? If you want interaction then, I don't know, go to a conference? Again, I don't get the issue. Seems like a good thing! Users are able to find answers and solutions to their quesitons more efficiently--all the while, still using the open-source library. The usage chart is still seeing tremendous growth! Developers are still using the library to solve their problems. It seems like exactly what open-source was intended for.
The issue to me is that, the incentives for investing in open-source have changed for some maintainers in such a way that they're no longer in alignment with their return on their investment. Maybe there are fewere people interacting with them and so fewer people to discover how "great" they are. Maybe fewer eye balls on their resume. The point is, open-source was a means to an end. And, so, frankly, I don't give a shit.
LLMs are making open-source technology accessible to more people and that's a good thing.
I’m convinced that GitHub and GitLab will eventually stop offering their services for free if the flood of low-quality, "vibe-coded" projects—complete with lengthy but shallow documentation—continues to grow at the current rate.
The trend of rewriting existing programs ("vibe-coding" a rewrite of $PROG in Rust, for example) threatens to undermine important, battle-tested projects like SQLite. As I described in this comment: https://news.ycombinator.com/item?id=46821246.
I’m quite sure developers will increasingly close-source their work and black-box everything they possibly can. After all, source code that cannot be seen cannot be so easily "rewritten" by vibe-coders.
The small libraries will be eliminated as a viable solution for production use, but that’s a good thing. They are supply chain risk, which is significantly amplified in the LLM age.
It may happen and it will be great if it happens, when open training datasets will replace those libraries to recalibrate LLM output and shift it from legacy to more modern approaches, as well as teaching how to achieve certain things.
There are also a lot of open source projects that are simply one-man shows. And llm should be massively helping those and I really don't see that so far.
I would say they should be a massive gain to the open source community cuz let's face it. The people that do open source are simply going to be different than the people that just feed on it.
Llm should be a massive enabler to open source. It should permit easy porting between architectures, programming languages and interfaces to a degree that simply wasn't possible before
Again, I'm not really seeing that.
I used to think open source was something I'd focus on when I retired. Instead I'm doing it now. I built Clippy (https://github.com/neilberkman/clippy) from scratch, a much better bcopy, that fixes a real gap in pbcopy, and it hit 200 stars and Homebrew core in seven months. Neither of these would exist without AI assisted development. My DocuSign Elixir SDK sat basically unchanged from 2021 to 2025, other than a few external contributions. I knew it needed a full migration of the OpenAI code gen layer, LiveBook examples and a million other tweaks. With coding agents I implemented everything useful I could imagine.
Am I the only one? There must be other devs out there resurrecting stale projects or shipping tools they never had time for.