On Sunday, February 20, 2022 10:13:03 PM EST Russ Allbery wrote: > Sam Hartman <hartmans@debian.org> writes: > > Figuring out how to accomplish requesting a statement is a little > > tricky, but I think it is worth the effort. DAM takes membership > > actions (including warnings) by consensus. It's fairly difficult to get > > all the members of DAM together. > > > > I don't think it would work in practice for the request for a statement > > to be a consensus action and to be followed shortly there after by > > another consensus action to take a decision and to write it up. That > > would require DAM to get together as a group twice in short succession; > > given how hard it is to schedule DAM action, that would not work. > > I think Debian is in danger of a degenerative spiral, both here and in > other places. We make fewer and fewer decisions, slower and slower, which > raises the cost of reversing a bad decision because it requires a second > decision, which will also be slow. This raises the stakes of each > decision, so they have to be made more carefully. This makes the decision > take more effort, and thus we make even fewer decisions, and those > decisions then carry even more weight. That in turn leads people to want > them to be made even more carefully, and the spiral continues until we > talk endlessly and make no decisions at all. > > We need a careful and slow process for kicking someone out of the project > because that's a big deal. Having a careful and slow process for issuing > a warning is faintly absurd, and I think we've only arrived at that state > because it's so hard to decide to ever do anything that they've reached an > unrealistic level of apparent importance. > > I think the solution in many, many places across Debian is to make more > decisions, faster, and allow some of them to be wrong. Lower the stakes > and consequences of a bad decision, and lower the perceived weight of a > single decision, rather than trying to make every decision perfect. > > Anyway, to be more concrete, what your description of the process says to > me is that ideally DAM would be much larger and would deal with more minor > things, such as warnings, in panels. Have a rotating "on call" or > something similar, empower them to make decisions on anything that comes > up while they're on call, and if someone thinks their decision is > profoundly unfair (I still think people are making far too much out of > warnings), or if some more serious issue comes up, it can be reviewed by a > different panel, a larger panel, or by DAM as a whole, but that would be > rarer. > > Having more people empowered to make decisions faster would also lower the > perceived significance of each decision, since there's going to be some > minor human inconsistency and I think that's actually healthy. The goal > of warnings is not to precisely measure and describe exactly what someone > did wrong to some nonexistent objective standard. It's to say "hey, this > is making things shitty for other people, you need to knock it off." > People can grumble about that all they want; the grumbling doesn't require > a response. If they think twice about doing the thing that was making > things shitty for other people, mission accomplished. If it turns out > that what they were doing was fine in context, great! It was a warning; > no one did anything. If that was the first warning someone got for > something they didn't actually do, they've led a way more sheltered life > than I have, and my life has been pretty sheltered. > > I dunno, I realize I may be being too cavalier here, but see the point > above about making more decisions, faster, and accepting a few mistakes. > If we end up with a rash of bogus warnings, we can reconsider. But right > now warnings are about as frequent as Papal encyclicals, and I think > partly as a result people have gotten really weird ideas about them. > > I guess the other possibility is that people really want warnings to be > way more serious than any meaning I personally would ascribe to the word > "warning" and are thinking of them as formal project censure or something > akin to that. In that case, my argument is that we need a warning that's > actually just a warning, and the thing we've got is much too strong and > the real problem is that we don't have something lighter touch. Currently a DAM warning is a suspension/expulsion with deferred execution. I think every non-government job I've had had a discipline process that went: 1. Verbal warning. 2. Written warning. 3. You're fired. No, Debian isn't an employer, but I think the sense that DAM warnings are used is similar. I agree with the idea that more feedback with a lighter touch would be a good idea, but I think anything with a lighter touch doesn't need DAM to do it. We are all empowered to provide other developers feedback when we see concerning behavior. People just need to do it. It doesn't take any new rules. Scott K
Attachment:
signature.asc
Description: This is a digitally signed message part.