Because to me, it's entirely fair to care about the process when you're the one working with it, but if you've never looked at the code and it works well, it seems odd to care how 10% of it was produced.

I still wouldn't quite use the loaded "tribalism" term here myself though, because there are still a few valid reasons to be skeptical of LLM-generated code, but for many people, their hard anti-LLM stance appears to be more of a shibboleth than a principled stance from my perspective.

Sure, and that's the motte.

This is a direct quote from the linked website:

> Any software whose code, documentation, binaries, or other assets include, in any part, the output of a generative AI tool such as an LLM, is considered “tainted” for our purposes.

... and that's the bailey.

One could also argue this is a temporary condition - driving the world to better, more efficient power-supplies in the long term (like nuclear).

It also allows engineers to more quickly arrive at efficient solutions for their problems - enabling more efficient systems being deployed earlier - possibly saving energy in the long run.

I don't think this is as clear cut yet as some people make it out to be.

> We also see a lot of AI-based pull requests to open source projects that completely overwhelm maintainers.

I agree this is a negative. Open-source maintainers have limited capacity to follow up work - and externally AI-generated PRs only make this problem worse.

> AI is harmful for small libre and open source projects.

Not if the projects themselves decides to use AI to more quickly diagnose bugs - and fix issues or features they want to implement. There are several OSS projects which are going this route, to the point that they don't want PRs. Instead they want bug-reports which are accurate enough for an LLM to create the PR for them, in accordance with the projects coding and process standards.

How is this harmful?

> AI is forced on people because of capitalist incentives

This sounds like a very weak argument.

Everything in a capitalist society is imposed on people based on capitalist incentives. That's how and why capitalism works (compared to other, lesser successful models).

Why should AI be treated differently?

> People who care about their craft are labeled as old-fashioned

Not really. Develop what you like, how like!

But people who see AI as the big-bad-wolf and go on a war-path against it, are being labeled old-fashioned yes. Same as people who were fighting against IDEs, linters, type-checks, compilers, etc.

> the way the software was produced might have been a very joyless process, because AI was doing all the fun work.

Some people find craft at a higher level - creating a specification for what would be the perfect piece of software.

And with AI, they can use that spec to magically materialize real working software from that.

That brings me REAL joy - even as a software craftsman.

> AI in programming also leads to deskilling and makes it less likely that junior programmers become as good as the current senior programmers

This is true.

Also: My predecessors were probably better at fixing their car should it break down than I am. Despite that society seems to have moved on.

> And in the meantime AIs will run dry of new high-quality learning material.

This is a real risk and a real possibility. AI going brain-damaged over training or "reproducing" on its own generated output. Digital incest.

But it's not a proven outcome yet either.

But in a capitalist society - we'll choose the model which works best - and this should self regulate.

> Free software licenses are based on copyright. These licenses are supposed to protect the freedoms of the end user (because that's the ethical thing to do).

Software licensning has long been impossible to enforce.

If somebody copies GPLed code to stack overflow without the license - what means do we have to prevent people from using that in closed-source products, or permissively licensed OSS projects?

AI doesn't solve this at all - it just makes the impossibility obvious to all.

One could argue this is a net benefit (a reality check - over idealism) or a problem. I see both cases.

> AI-contributions to software projects might be incompatible with the license of that project

So could any non-AI contribution as well. See above.

In short - I don't find your argument convincing, but at least I can see what angle you are coming from. Thanks for taking the time to answer at thoroughly.