Time limits on AC Appeal process

[martinthomson]

https://www.w3.org/policies/process/#ACAppeal describes a process for an AC member to appeal a decision. This is similar to the formal objection process, except that it can force an AC vote.

The time limits in place on that process are pretty unreasonable. It's hard to imagine why anyone would choose that process over the formal objection process.

1. The appeal needs to be filed within three weeks of the decision in question. Note that formal objections have no such statute of limitations clause.
2. The Team needs to announce the appeal within one week. This is reasonable, but it's worth including.
3. The appellant needs to gather support for their appeal from 5% of the AC in one week. Note that this isn't a vote on the appeal, just a process to ensure that a vote isn't initiated frivolously. This period is absurdly short. 5% is a pretty high threshold to reach in a week.
4. The AC then votes on the motion. No time limit is specified, but we might assume that the Team would do something sensible.

To make this more concrete, if there is a bad decision that warrants appeal, an appellant needs to identify the problem, make the decision to act, muster 18 other members to support their decision, and do that within four weeks. That sounds like a lot of time, but there are many cases where a decision isn't even noticed for some time.

This might seem rash, but I would propose that the AC appeal process be removed entirely. The formal objection process is just easier to use overall and it would work for most things.

There are discussions of things that would rely on this process, in particular the removal of AB and TAG members, which would not be able to use the formal objection process. I propose that those be resolved by a simpler process. For that, I would personally prefer that any decision be open to reconsideration via a vote if some small, fixed number of members (five, say) opened the question. That should be enough to defend against frivolous use of the process.

That three week period does serve a purpose: it ensures that a whole chain of decisions that the consortium makes cannot be unwound. The council in the FO process can be expected to handle that sort of complexity. As long as there is adequate time to explore the consequences of overturning a decision, I don't see the AC as being especially prone to making that sort of bad decisions

Even if this doesn't take the suggestion, the three week and one week pre-vote periods should be extended. I'd also be OK with the five member petition idea I suggested above.

[frivoal]

This is similar to the formal objection process, except that it can force an AC vote.
The time limits in place on that process are pretty unreasonable.
It's hard to imagine why anyone would choose that process over the formal objection process.

I think there's a little misunderstanding here. An AC Appeal cannot be filed over any decision, but specifically only over "W3C Decisions", which are defined as those taken as the result of an AC Review.
This is not a process you chose instead of filing a Formal objection. Instead, it can be used when one or more formal objections were filed in an AC Review, and the Council (previously, the Director) overrode them, but you think they got it wrong and that there's a reasonable chance that a majority of the AC would agree with you. It's the last level of escalation.

The appeal needs to be filed within three weeks of the decision in question. Note that formal objections have no such statute of limitations clause.

Right, the idea is that this is a question that has been litigated already. If there's massive support for the idea that the decision was problematic, then it should be revisited, and there's not need to wait, and if there isn't massive support, it should stand and people should be able to move on, without delay. If one would need many more weeks or even months of rallying support for their appeal, then it isn't that contentious in the first place, and shouldn't be reopened. At least that's the idea.

The appellant needs to gather support for their appeal from 5% of the AC in one week. Note that this isn't a vote on the appeal, just a process to ensure that a vote isn't initiated frivolously. This period is absurdly short. 5% is a pretty high threshold to reach in a week.

• it's one week from the announcement that we're checking if there's enough support to reach the 5%. But that's on top of 3 weeks since the decision, during which people can already talk to each other and attempt to garner support.
• I agree a week is very short. I wouldn't mind if it were two weeks instead, for example, but I think it does make sense for the period to be short. As discussed above, this is only supposed to be used for hot controversial topics which have already been objected to and overruled. That earlier phase gives ample time for anyone who cares to become informed about the matter, discuss the topic, etc. At this point, we're not giving people a short amount of time to rally support, we're using a short amount of time to check whether there already is a high enough level of discontentment with the decision that it's worth running an appeal vote.

This might seem rash, but I would propose that the AC appeal process be removed entirely. The formal objection process is just easier to use overall and it would work for most things.

Given the explanation above, that it is not a process to be used instead of FOs, but as a further and final appeal path after FOs have been ruled over, do you still hold this view?

AC Appeals are used extremely rarely. I believe it has been only invoked a handful of times in the history of W3C:

• Once, where it cleared the 5% participation threshold, but did not get >50% support for overturning the decision: https://lists.w3.org/Archives/Member/w3c-ac-members/2017JulSep/0055.html
• Three times where it did not clear the 5% participation threshold https://lists.w3.org/Archives/Member/w3c-ac-members/2023JulSep/0013.html, https://lists.w3.org/Archives/Member/w3c-ac-members/2022AprJun/0035.html, https://lists.w3.org/Archives/Member/w3c-ac-members/2013JanMar/0053.html

Maybe we could say that this is rare enough to show that it is not useful, but I would be tempted to think otherwise, and to say that even if rarely used, it is good that there exists a final level of escalation and that the decision rests directly in the hands of the membership.

(I'm setting aside the discussion of reusing AC Appeal for recalls or not for now, as it seems useful to get on the same page about the above first)

[martinthomson]

I think there's a little misunderstanding here. An AC Appeal cannot be filed over any decision, but specifically only over "W3C Decisions", which are defined as those taken as the result of an AC Review.

No misunderstanding, just a useful point of comparison. You can't FO decisions where AC Appeal applies, but you can FO all around those decisions.

Given the explanation above, that it is not a process to be used instead of FOs, but as a further and final appeal path after FOs have been ruled over, do you still hold this view?

Yeah, I still think that the process is heavyweight and unlikely to ever be successful. The information you shared on history only cements that view. I do think that having a means for the AC to overturn any decision, including FO Council decisions and recalls) is useful, but the process should be much simpler. Right now, AC Appeals are narrowly applicable, time limited, and cumbersome.

How does this sound as a simplifcation:

1. FO can appeal any decision (without the present exceptions, instead excluding a short list of things where FO doesn't make sense: FO council decisions and AB/TAG recalls seem most relevant).
2. A small number of members (as I mentioned, five seems reasonable) can request an AC overturn ANY decision, invoking the "Requisite Member Vote" process in bylaws, or the copy of that process as you augmented in Generalizing AC Appeals and using this procedure for recall. #888 to include the 5% participation threshold.

In neither case do I think that we need a time limit in process, other than to ensure that the process is conducted in a reasonable time period. A specific time doesn't need to be specified (a decision not to act is a decision that can be appealed; the resulting constitutional crisis might be welcome if that needs to be repeated).

[fantasai]

@martinthomson That would turn W3C into a voting body rather than a consensus body, which is not consistent with our history or our vision.