The Zig project's rationale for their anti-AI contribution policy

Why don't you stop avoiding the actual question? Or perhaps you have nothing of substance to add?

anything unpredictable is inherently untrustworthy and requires extra effort to review.

Lack of determinism is not a practical concern.

it is to me. it's a knockout criteria. it is the only reason that keeps me from using LLMs for coding. nothing else is as serious an issue to me as this.

here is why: i tell the LLM to build something with requirements A B C D and E. it builds, i review and i find A B and D are good, C and E are broken. i tell it to fix them, it does, so C and E are fixed, but now A is broken. i tell it to fix that, and i have to keep iterating until i find a combination where everything works. in every iteration any part can randomly break, so for every iteration i get changes all over the place. they never are confined to the issue i pointed out. i have to review the whole thing every time. that's what i mean by lack of determinism, and that is a serious practical concern because instead of getting done in two or three iterations it requires dozens of them. see my related replies elsewhere. i just don't want to work that way.

The compiler is being actively worked on by Adam and his team at Nectry, but unfortunately those developments are not currently being backported to the open source repo. I'm fairly confident this will happen eventually.

I maintain my own private fork with some small modifications which I started polishing up this week to release it for a talk that I'm preparing.

The project I'm using this on is an ecommerce site [0] written in 100% Ur/Web with a hand-rolled backend ERP system written in PHP (not by me) which I am slowly replacing bits of with new Ur/Web code. As of today, we have 22223 lines of Ur/Web code, weighing in at 701 KiB.

[0]: https://liepelt.design

> What does it take to "earn your trust"?

multiple successful contributions of increasing complexity, among other things.

>> LLM contributions are non-deterministic, which means they can never be trusted.

> Provably incorrect. LLM contributions can be reviewed, tested, and understood like any other contribution. There's nothing "special" about LLM contributions.

read this comment to see what i mean: https://news.ycombinator.com/item?id=47968180

> Contributions authored by human brains are also non-deterministic, perhaps if the author was feeling in a slightly different way they'd have formatted the code a bit differently.

i can tell a human to focus on a certain issue. they will either listen and follow my instructions, or i will reject their contribution. the LLM is almost guaranteed to not follow all my instructions and make changes i didn;t ask for. see my comment above.

>> therefore, if you use LLM to contribute, you can not earn my trust.

> The premise is wrong.

how so?

>> if you believe that you can not create a meaningful contribution without the use of LLM then you are realizing that you are not skilled enough to understand the code that you contribute

> What if I believe I can do so without an LLM, but that it could be even better with an LLM?

what you believe is not relevant. only what you can convince me of. you'll have to first show that you actually can work without an LLM before i will consider your contribution.

> What if I'm great at understanding code, but terrible at writing it?

your problem not mine. if you are terrible at writing code but good at understanding it then it's your choice to only do code reviews. you can still make a meaningful contribution that way. i'd even let you write code so you can practice that, but i am not interested in your LLM generated code.

> Again, this is a premise that you just decided to take as truth, without proof.

i don't need proof. i need trust. you need to convince me that your code can be trusted.

>> because if you could understand it, then you could write it yourself.

> False. I can understand a novel algorithm by reading and studying it, but perhaps I could have not come up with it myself.

that's called learning. once you learned it, you can write it. but in order to effectively learn you also have to practice. if you let LLM write all your code then you are not practicing, so you won't improve.

>> i want you to earn my trust by showing me that you understand what you are doing

> I can easily do that even if my contribution involves LLM assistance.

it depends on the level of assistance. i am not ruling out use of AI to do research and learn, just don't let it write the code for you.

>> i want you to grow your understanding of my project

>> none of this happens when you use LLMs

> False. Why do you think so?

as i said above, if you don't practice writing the code yourself you are not learning. not enough at least to satisfy my expectations.

>> if you are unable to make a contribution without the help of an LLM then you are not ready to contribute.

> Again, this is your opinion and you have no way of proving it. I can prove the opposite.

whether you are ready to contribute to my project or not is not something i need to prove. it is a choice based on my preference which depends on the amount of trust you have earned. you can not prove to me that you are ready to contribute. this is not a standardized test that if you pass you automatically qualify. you can only convince me by earning my trust. this is a human decision, based on feelings.

Why would it be pre-approved ? Code is code, whether it's bad quality LLM code or meatbag code it shouldn't matter.

The entire problem is that before the meatbag code was either not submitted at all (developer knowing they are not competent enough to do the fix) or the volume of it was low.

With LLM people not competent to even review, let alone write are emboldened to just throw shit on the wall at rapid pace. So the wall is entirely covered by the shit