On 8/13/26 15:26, Sergio Pastor Pérez wrote:
Hello!

Vagrant Cascadian <[email protected]> writes:

I think this is a great idea. I believe this could reduce the
frustration from an "unexpected" outcome. For example, a requisite for
"disapproving" (rejecting?) could be to have 1 standing issue/pull
request which has not been closed.

There is also a problem at the start of the GCD process that makes creating such issues/PRs difficult.

GCD008 started from a very extreme position (compared to similar documents in other projects), and as a result it feels we got into a 'local maximum' that the GCD process couldn't get us out of.

The final GCD falls short in both directions: it allows plenty of LLM use that I consider harmful, while it also disallows plenty of LLM use without proper justification.

It would have been easier to reach consensus the authors and sponsors of the GCD would represent a wider range of opinions and ideas to begin with.

My improvement suggestion would be to have smaller discussions before submitting a GCD to ensure that a broader group of people feel represented already in the initial proposal.

Note that I believe this process could also lead to a more strict LLM policy, because it would create a less hostile environment that would let us figure out what we actually care about. I believe we are all pretty close in our views actually.

At times the discussion felt more like 'trench warfare' than consensus finding. I feel we were mostly talking past each other, not understanding each others viewpoints. I still don't think there is enough mutual understanding. Consensus is hard to reach without understanding.

I find the result quite unsatisfactory, and agree with Gabber's analysis.

The asymmetry of the GCD process works both ways: even a single disapproval represents multiple people that share the sentiment. (Also in GCD 007; I don't know about the others.)

There are plenty of points we didn't even get to discuss. But the points we did manage to discuss already cost so much energy. I think we can, and should, do so much better on such an important policy.

Hugo




Reply via email to