(cc'ing Charles since I'm not sure if he's reading all of debian-vote; let me know if this is annoying and I should stop.)
Sam Hartman <hartm...@debian.org> writes: > It sounds like what russ and Charles are talking about is the following: > * You as a proposer want to accept an amendment > * A sponsor objects, and so you can't even though you would have been > able to if you had fewer sponsors. > I have an alternate proposed fix: > If the proposer accepts the amendment and there are k sponsors who have > not objected, the amendment is accepted. > (I think it's okay even if we end up counting new sponsors to get to k > who have not objected) This and Timo Röhling's similar idea were very helpful for thinking through this. Thank you! After pondering this for a couple of days, I'm going to advocate for not making a change here, even though it looks a bit restrictive. I think the key point is this analysis: > But under Russ's approach, the whole amendment process is really just a > convenience to make it easier to update an option with less withdrawing > and re-proposing. I arrived at a somewhat different conclusion given that point, though, because I think there's a principle of ballot stability here that's worth maintaining. Here's my argument: Once a proposal has been sponsored and added to the ballot, we, as a general social convention, stop sponsoring it unless it feels particularly important to be listed as a sponsor. That means that any given option currently on the ballot usually has "hidden" support in the form of Developers who intend to vote for it but haven't sponsored it. It seems likely that in some situations those Developers may think "okay, the opinion I really cared about is on the ballot so I can vote the way I want" and then may tune out the subsequent discussion. In other words, I think once a ballot option makes it onto the ballot, the rules are attempting to capture the sense that it no longer belongs just to its proposer, but now represents some unknown number of people who want to vote for it. Given that, if the proposer changes their mind and wants to propose a substantially different ballot option, I think the default should be that the proposer do that as a separate ballot option and get sponsors for that new ballot option (and possibly withdraw as the proposer for the original ballot option). This reflects the fact that just because the proposer changed their mind, that doesn't mean that other supporters of that ballot option also changed their minds. As Sam says, the ability to make substantive changes if all sponsors agree is essentially an optimization. It's tedious to propose a new option, have everyone sponsor it again, withdraw as proposer of the old option, and confirm that no one else is stepping forward, so for changes that everyone agrees make the ballot option better, we should have a way to allow those to be made more easily. But we want some sort of check that this is really true and not just take the proposer's word for it, so we in essence draft the sponsors of the ballot option as referees to decide whether this change does make sense in the context in which they originally supported that option. Given that, I lean away from setting the bar for modifying a ballot option the same as the bar for proposing a new option on the grounds that, once on the ballot, I think options should be stickier than that and the bar for changing them should be higher. The sponsors are, in effect, representing that unknown group of people who were satisfied by that option and may not be paying attention, so it should be hard to change an option in a way that may be incompatible with those preferences and, in the event of any conflict, the default should be to draft a new option. (But we don't go as far as letting any Developer object, like we do for the trivial changes provision, because involving people who would never have voted for the option anyway is less likely to be constructive.) I believe there is a real bug in the existing constitution in that it at least implies that someone could sponsor a ballot option solely to then object to a proposed change to it, which I don't think we want. I mostly fixed that bug by requiring people have already sponsored the option before the change was proposed if they want to object. But after thinking about this some more, I don't think we want to go farther. Once people who have sponsored the original ballot option start objecting, I think we should take that as concrete evidence that the new option is sufficiently different that it may not represent the people who were supporting the old option, and we should therefore default to adding the new option via the normal mechanism. Does that make sense to everyone? -- Russ Allbery (r...@debian.org) <https://www.eyrie.org/~eagle/>
