Today the Ladybird Browser project announced changes to how they’re developing Ladybird. Basically, they’re going to remain open source but not be open for externally created pull requests. For those of you reading this in 5 years, free open source software projects used to be written by humans, and because writing code took a lot of time, people that wanted to make larger contributions to a project would contribute code. It was sort of the next step up from contributing bug reports, one that involved actual effort on the contributor side, rather than most of the effort being on the maintainer side.
The economics have changed this year because writing code is effortless with modern AI agents. You could have argued the value of a product is never in the code, it’s in the culture and the tests, but that’s a topic for another day. Spending hours finding a bug, learning the project’s code structure, making a fix, prettying up the commit history, figuring out how to use git send-email (heh), and having the maintainers of the software review it took a long time. Now doing all that could take someone a minute, but the maintainers of the project still need to spend a good amount of time reviewing the code. Instead it just makes more sense for them to type the sentence or two to their AI agent and then they only need to trust their AI agent to do the right thing.
It’s inevitable, and this will only be more common in the future. There’s currently a lot of projects in the early phases of AI who are still trying to figure out if AI-generated source code is copyrightable or if they should ban it because they don’t agree with the environmental footprint of the technology. If you’re using the crawl/walk/run terminology, that’s pre-crawl. It’s unfortunate, and their future prospects could be described by some Darwin metaphor.
Continuing the analogy, you have crawl when some of the contributors are using AI. Walking is when the maintainers use it near 100% of the time on coding and reviews. Running is when AI’s are triaging issues, creating PRs, and reviewing PRs. I think it’s only a matter of time until a lot of projects have cron + triggers for all the maintenance work and a human doesn’t need to be in the loop. I would imagine projects like openclaw are the cutting edge of what you might see publicly.