Hi Vagrant, qua 12 ago 2026 às 16:42:55 (1786563775), [email protected] enviou: > [...] > > * several cycles of calling for consensus > > In all the discussion, sometimes it is not clear what issues are severe > enough concerns to block consensus, and it is helpful to have one or > more calls for consensus (e.g. the "deliberation" phase > support/approve/disapprove), iterating until the issues that are > blocking consensus are resolved or rejected. > > [...] > > The process should never end with an abrupt surprise, but should ideally > be a gradual realization to everyone... >
IMO, the issue is not the "surprise" factor per se, but the fact that we are acting as if the deliberation phase was an end of line all or nothing situation and, at the same time, it seems (at least to me) that, we are also acting as if the burden of reaching consensus is mostly (solely?) on the original proponent. So, I think it would be an improvement to have some provision stating that: - when one disapproves, they should state as clearly and concisely as possible their reasons for doing so and (if possible) propose a patch that would address the stated concerns; - after that, we may start a new (maybe shortened) discussion period where we address specifically the reasons given for the disapproval followed by a new deliberation. I would be in favor of having yet a third possible iteration for when this last cycle addresses the concerns of the one(s) disapproving but then creates new issues for those who didn't have them before. More than that, IMO, would be unproductive and possibly lead to an infinite loop. I also think that it would be good if we formalized the possibility of "amending" (aka patching) older GCDs through a new one. So, in the current situation we wouldn't invalidade GCD-1 in favor of GCD-X, but rather we would propose a patch to GCD-1 and have some header which stated that it was amended by GCD-X. This would make it clear that any prior consensus can be revisited, modified, improved or superseeded. > > * Every GCD as a separate project > > I envision each GCD as a new project for the initial draft, which is a > reference from which to submit issues and/or pull requests, and also > gives the ability to compare changes across revisions, see when a change > was merged, etc. we can track the primary author(s) of a given change > using git metadata, and use various *-by git commit traileer > conventions. > > We can keep the whole change history archive as a reference, and > possibly only merge the final result into the repository of official > policies... > My only doubt here is if that would scatter the discussion on many smaller PRs and make it harder to have general, transversal, in-depth discussions on what's being proposed. So if we were to do this, I think it would be good to have an "General Discussions" issue for each GCD where we can have a more unstructured flow where ideas can pop and float around (similar to what we're doing here). > > * word choices > > I think the "deliberation" phase in the current GCD process is very > confusing to me as a native English speaker, as it more commonly is a > synonym for "discussion" rather than the tallying the final results of a > decision. Inevitibly, social creatures that we are, we will want to > discuss all the way through, but at some point it should refocus. > > I find "disapprove" to be far too weak of a word given the severe > implications for the process. Again, with my native English speaker hat > on, the difference between "disapprove" and "accept" is just a matter of > degree, whereas the function "disapprove" holds in the process is much > more severe than anyone's individual opinion. Not entirely sure of a > better choice, but perhaps "reject" is more honest? > > Also using "I accept" "I support" "I disapprove" centers "I" more than > the issues raised, which again, it should not be about the individual > community members, but about the issues and concerns raised by the > community. > Beyond terminology, I also think that instead of a ternary deliberation a quaternary could be an improvement: - "Welcome GCD", instead of "I support": meaning one thinks the proposed course of action is mostly welcome; - "Acceptable GCD", instead of "I accept": meaning one is not sure the course of action is needed or desired, but does not have any significant issue to oppose it if others think it to be welcome; - "Doubtful GCD", instead of being silent: meaning one has significant concerns with the proposed course of action, but does not know how to address them in a satisfactory fashion to them or others and is willing to live with those concerns for the time being if the rest is welcoming the GCD; - "Harmful GCD", instead of "I disapprove": meaning one has significant issues with the proposed course of action and can explain what is harmful, why and what would be a better course in their view. One "Harmful" manifestation would be enough to make us go back to a renewed iscussion period. Ten percent of actual "votes" saying the GCD is "Doubtful" would have a similar effect, even if there's no clear path to resolve the issues raised. This way, someone stating that a GCD is "Harmful" knows that they have a particular burden to the rest of the group. Since they are willing to "block" solely, they should also be willing and capable of steering the consensus in a new direction. People not willing or unable to assume that role would still have a way to express their concerns and, if they are in company, this would be enough reason per se to halt or go back to renewed discussions. *** Cautionary Measures Besides that, I also think it would be good if we had a way to apply cautionary measures. What I imagine is: if there is some course of action that is deemed urgent we would have an inverted GCD process: Someone proposes a GCD with urgency and calls for Deliberation. If ten percent of members agree with the urgency and proposed cautionary measures, they go into effect. After that the discussion period is started and the cautionary measures remain in place until the end of the GCD process (with possible iterations at the end). If the GCD does not achieve consensus the cautionary measures are also undone. *** Personal preferences I also think that it could be good to formalize that personal, arbitrary preferences are not acceptable reasons to block the consensus. Neither are appeals to authority, secret meetings, fears, insecurity and uncertainty. Saying a GCD is harmful or doubtful comes with the duty of at least showing that those harms and doubts are shared concerns/goals to the project. Thanks for caring!
