Hello, I think it is difficult to add anything after a number of good contributions, especially the well thought through messages by Pinoaffe and Christine, and the suggestions to go forward by Janneke and Noé. So I will try to be brief (I know, famous last books...).
On Mon, Jul 27, 2026 at 10:09:13PM +0200, Gabriel Wicki wrote: > Hi Christine > > the status quo is assumed that we are at free rein to accept genAI > > contributions > I don't think this is an adequate assumption of the current status > quo—or did I miss something? This is indeed how I interpret things. Right now there is no regulation. You may call this fearmongering, but what is not forbidden is allowed, so to me rejection of GCD 008 has opened the floodgates. While as far as I know up to now we have not seen many LLM contributions, it all depends on decisions of individual committers. I have once pushed an LLM made contribution (which, while we were still discussing the topic in GCD 008, we agreed to revert and replace by an alternative commit handcrafted by Florian). During the GCD discussions, some committers were more in favour of LLM use (at least as far as I understood); what would prevent them from pushing such commits in the future, now that the efforts to regulate LLM contributions have been rejected by the Guix project? What would entitle me to revert these commits? In the current state of affairs, I might then have to accept GCD 009: I have read through it and realised that it actually proposes more limits to LLM use than the current de iure (not de facto) situation. The comments by Liliana on the Codeberg pull request illustrate this. > Why would we suddenly opt for non-ideal "solutions" when (as I perceive) > most people are not completely satisfied with the outcome? Here I can only paraphrase Pinoaffe. We cannot expect to reach an ideal solution for each contributor individually. I personally preferred Ludovic's original proposition to the final one, since it suggested much stronger regulations. Someone else prefers laxer regulations than the final proposition. It is materially impossible that the two of us come out of this process with our ideal solutions (and there are 80 more team members). However, there is something like an ideal solution for the project as a whole. And as Pinoaffe, my impression is that we had roughly reached the optimum within the community of contributors at a point of the spectrum where most people would feel comfortable to continue contributing, and that their essential needs would be met. > To go forward, I think we agree on many essential things discussed in > the process: The problem is that technically, after the rejection of GCD 008 we have agreed on nothing; even people who supported the compromise lined out in the document are not bound by it now. So essentially we need to start from scratch (where of course the final form of GCD 008, the discussions we have had, or a completely different proposal such as GCD 009 can be the basis of a new discussion). But who is going to spend another three months on another GCD? I remember my own GCD 006 (on a rather minor question in comparison) as very taxing and exhausting. I admire the effort Ludovic has put into GCD 008, including interacting with comments and trying to find the ideal solution for the project as defined above. One cannot expect the author of a rejected GCD to start over. Then shall we rehash the same arguments? Compared to GCD 008, which already balanced out so many rather different points of views, where do you expect another equilibrium to be found? In a sense, the only person who could start another GCD is the one who disapproved - they know what disturbed them and should have a clear view for a possible compromise that was overlooked by the author during their work on amending the proposal. So assuming from what you wrote that the current non-regulation is not your desired state of affairs, as Noé I am waiting for you to draft GCD 010 with a proposal that could form a compromise based on what you consider we agree on. If you start now, we might have a policy towards the end of September, at worst the end of October. I suppose that people would be willing to keep the de facto moratorium on LLM generated contributions until then. Andreas
