Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Generalizing AC Appeals and using this procedure for recall. #888

Draft
wants to merge 5 commits into
base: main
Choose a base branch
from
Draft

Generalizing AC Appeals and using this procedure for recall. #888

Changes from 3 commits
Commits
Show all changes
5 commits
Select commit Hold shift + click to select a range
acf7faf
Refactor AC Appeals
frivoal Jun 18, 2024
9881333
Add recall procedures for AB and TAG members
frivoal Jun 18, 2024
95013e2
Limit AB/TAG recalls by the AC to collective votes of no confidence
frivoal Dec 6, 2024
ebb1fba
rephrase
frivoal Feb 1, 2025
9818823
improve phrasing
frivoal Feb 7, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
118 changes: 98 additions & 20 deletions index.bs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1309,15 +1309,66 @@ Verifiable Random Selection Procedure</h5>
The incomplete terms are assigned in result order.
</ol>

<h5 id="AB-TAG-recall">
Recalling AB or TAG members</h5>

<h6 id="individual-removal">
Individual Removal</h6>

Individual participants of the [=Advisory Board=] or the [=Technical Architecture Group=]
can be <dfn export lt="remove|removal">removed</dfn> from those groups
if they are found by their peers
to be grossly neglecting their duties,
or to be acting in a way that seriously hampers the group's ability to function normally.

A chair of the [=AB=] or [=TAG=] <em class=rfc2119>must</em> hold a hearing
on the potential [=removal=] of a participant
if requested by at least three of the participants in the group.
After giving the individual in question
an opportunity to defend themselves,
a vote on the proposed [=removal=] is held.
If at least three quarters of the participants in the group,
excluding the individual who is the subject of such vote,
then vote in favor of the proposal,
the individual's seat on [=AB=] or [=TAG=] is [=vacated=] immediately.
Comment on lines +1330 to +1333
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I would say that a lower threshold is sufficient. A simple majority might suffice in this case.

Consider this: if there are three people who are willing to raise this, then that alone is indicative of dysfunction in need of remediation. Removal should not be the first option there, but if you need it, then there is a chance that some people will be forced to recuse or some will be absent, making it quite difficult to reach 3/4.

Copy link
Collaborator Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I disagree. I think it should take a strong supermajority to eject someone. We want to stay clear of the possibility that someone gets kicked out (or simply fears getting kicked out) simply because half the group disagrees with them (and lacks restraint).

Making an allowance for people who need to recuse could make sense, but I don't think it should be possible to kick out someone just because some of their supports didn't show up that day. It is an extraordinary measure, and I find it appropriate for it to need an extraordinary level of support.

Copy link

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

What might be more appropriate is that for the vote to succeed, all members of the elected body must participate (avoiding the issue of "my supporters didn't show up), and thus for the proposal to carry a simple majority is sufficient. Not requiring every member to participate is where risk is introduced. To be additionally clear, abstaining is acceptable.

Copy link
Collaborator Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Exclusion by simple majority seems inappropriate to me for something that amounts to a disciplinary procedure. In cases where it is warranted, reaching a significant super majority (everybody but a friend or two) should be doable.

Simple majority seems far too easy to reach for mere differences of opinion, which is not what this should be about.

Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I can make a different argument: those who initiate this process risk the wrath of the AC (and the other process whereby the AC loses faith in all the participants). They do so because they believe that they cannot continue to work with the identified person. They have to convince a whole bunch of other people -- half the body -- that this extraordinary disciplinary action is warranted. The result -- success or failure -- is then made public.

I've seen some dysfunction in groups, but never at this level. If it were the case that half the body wants someone gone, that's enough for me.

I would instead say that removal on the basis of convincing half a body is already too high a bar to clear.

Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I would object to a simple majority (even with all voting; in practice, those tend to mean the same thing). I could be agreeable to a 2/3rds supermajority (instead of 3/4), but simple majority is not sufficient, in my strongly-held opinion.


<h6 id="collective-removal">
Collective removal</h6>

The [=Advisory Committee=] <em class=rfc2119>may</em>
hold a <dfn export>vote of no confidence</dfn>
in the [=Advisory Board=] or the [=Technical Architecture Group=].

An [=Advisory Committee representative=] initiates a [=vote of no confidence=]
by sending a request to the Team, and <em class=rfc2119>should</em> also share this request with the Advisory Committee.
The request <em class=rfc2119>must</em> identify which of the [=AB=] or [=TAG=] is targeted,
frivoal marked this conversation as resolved.
Show resolved Hide resolved
and <em class=rfc2119>should</em> also include the rationale.
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I think that you want a three member threshold for this too. Otherwise, this is open to trolling and DoS.

I'd be OK with a higher threshold than three, but not a lower one.

If that takes the form of one AC member initiating an override that has to be seconded by two other members in all cases, that would be ideal. I know that this mechanism hasn't been activated, but it's an organizational vulnerability.

Copy link
Collaborator Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

It already is a 3 stage thing:

  1. someone calls for a vote of no confidence
  2. we check if >=5% of the membership agree that we should run a vote of no confidence (within a time limit of 1 week)
  3. if so, we run the actual vote of no confidence

Adding an "at least two people need to agree" seems redundant with step two.

Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I think that the second step is not feasible (you allow more time for the final vote, which only needs to clear the same threshold). So I am suggesting a replacement for that stage.


Within one week, the [=Team=] <em class=rfc2119>must</em> initiate an [=Advisory Committee Override=]
on the proposal.

The conclusion of the [=Advisory Committee Override=] is final,
and cannot be the subject of a [=Formal Objection=] nor of an [=AC Appeal=]:
if the [=Advisory Committee Override=] proposal passes,
all seats on [=AB=] or [=TAG=] are [=vacated=] immediately;
if it fails, it cannot be invoked on the same body
sooner than 6-month since their previous invocation.
frivoal marked this conversation as resolved.
Show resolved Hide resolved

Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I think that in both cases, you want to include some sort of reporting responsibility. That is, there is an announcement that the seat is vacated.

There is no way to hide what is happened here and you don't want to have people learn via rumor mills. Information should be publicly communicated to the AC by the chairs or Team as promptly as possible.

Copy link
Collaborator Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Does this help?

Suggested change
The [=Team=] must announce the conclusion of the [=Advisory Committee Override=] to the [=Advisory Committee=].

Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

That would be fine for the latter, but there is no requirement on the TAG/AB to announce when they act.

<h5 id="AB-TAG-vacated">
Elected Groups Vacated Seats</h5>

An [=Advisory Board=] or [=TAG=] participant's seat is vacated when:
An [=Advisory Board=] or [=TAG=] participant's seat is <dfn lt="vacated|vacant">vacated</dfn> when:

<ul>
<li>
the participant resigns, or

<li>
the participant is [=removed=], or

<li>
a [=vote of no confidence=] in the body they are part of is successful, or

<li>
an Advisory Board or TAG participant changes affiliations
such that the <a href="#AB-TAG-constraints">Advisory Board and TAG participation constraints</a> are no longer met,
Comment on lines +1360 to 1374
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I can only suggest through line 1374, but similar changes should be made through line 1384.

Suggested change
An [=Advisory Board=] or [=TAG=] participant's seat is <dfn lt="vacated|vacant">vacated</dfn> when:
<ul>
<li>
the participant resigns, or
<li>
the participant is [=removed=], or
<li>
a [=vote of no confidence=] in the body they are part of is successful, or
<li>
an Advisory Board or TAG participant changes affiliations
such that the <a href="#AB-TAG-constraints">Advisory Board and TAG participation constraints</a> are no longer met,
An [=Advisory Board=] or [=TAG=] participant's seat is <dfn lt="vacated|vacant">vacated</dfn> when any of the following occurs:
<ul>
<li>
The participant resigns.
</li>
<li>
The participant is [=removed=].
</li>
<li>
A [=vote of no confidence=] in the body they are part of is successful.
</li>
<li>
An Advisory Board or TAG participant changes affiliations
such that the <a href="#AB-TAG-constraints">Advisory Board and TAG participation constraints</a> are no longer met.
</li>

Expand Down Expand Up @@ -1368,6 +1419,12 @@ Elected Groups Vacated Seats</h5>
and the maximum number corresponds to all unoccupied seats.
Except for the number of available seats and the length of the terms,
the <a href="#AB-TAG-elections">usual rules for Advisory Board and Technical Architecture Group Elections</a> apply.

<li>
If seats are vacated due to a successful [=vote of no confidence=],
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I think that you want to set a threshold for triggering an election, not just have this for the no confidence vote. The bodies will likely function with a few gaps, but there is a critical point at which the groups are non-viable. I might suggest 2/3 of the target size as that threshold.

Suggested change
If seats are vacated due to a successful [=vote of no confidence=],
If vacancies reduce
the number of [=AB=] participants to 6 or fewer
or the number of [=TAG=] participants to 7 or fewer,

Copy link
Collaborator Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

That seems like a separate discussion. The previous bullet point in this list already deal with the case of seats being vacated for reasons other than a successful vote of no confidence, by letting the chair decide whether to run an special election or not. Your proposal would make it no longer a choice by the chair, which may or may not be a good idea, but is a separate idea. If you want to pursue that, could I ask you to open a separate issue?

Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I sort of agree, but I don't think that you can separate that out completely. You are introducing new ways in which the membership of these groups can drop to critically low levels. Though we don't expect this process to be used often, in combination with existing stuff, you have increased the risk of the groups being understaffed. Staffed to zero is just one special case.

the [=Team=] <em class=rfc2119>must</em> organize an election,
under the same condition as for individually vacated seats,
Comment on lines +1425 to +1426
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Suggested change
the [=Team=] <em class=rfc2119>must</em> organize an election,
under the same condition as for individually vacated seats,
the [=Team=] <em class=rfc2119>must</em> organize an election.
Vacated seats are filled for the remainder of the term of the vacancy,

unless the next regularly scheduled election is fewer than three months away.
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I'm not clear on the three month choice. That means that you might win an election for a 3 month and a day term. That's pretty pointless, I'd go with six.

Suggested change
unless the next regularly scheduled election is fewer than three months away.
unless the next regularly scheduled election for the vacancy is less than six months away,
in which case the remainder of the term is extended by two years.

Copy link

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Strong +1 to this, three months is just too short a time to get involved.

Copy link
Collaborator Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Works for me. I'll let chairs decide if there's consensus on this point or not.

</ul>

<h3 id="GAGeneral" oldids="ChapterGroups, WG-and-IG">
Expand Down Expand Up @@ -2887,24 +2944,50 @@ Appeal by Advisory Committee Representatives</h3>

An [=Advisory Committee representative=] initiates an [=appeal=] by sending a request to the [=Team=],
and should also share this request with the [=Advisory Committee=].
The request should say “I appeal this Decision”
The request <em class=rfc2119>should</em> say “I appeal this Decision”
and identify the decision,
and may also include their rationale for appealing the decision.
and <em class=rfc2119>may</em> also include their rationale for the [=appeal=].

Note: See [[DECISION-APPEAL inline]] for a recommendation
on how to communicate an appeal request to the [=Team=] and the [=AC=].

Within one week the [=Team=] <em class="rfc2119">must</em> announce the appeal process
Within one week, the [=Team=] <em class="rfc2119">must</em> initiate the [=appeal=]
in the form of an [=Advisory Committee Override=]
on the proposal to overturn the decision.

The conclusion of the [=Advisory Committee Override=] is final;
the same decision cannot be appealed more than once.

If the [=Advisory Committee Override=]
approves the proposal to overturn the decision,
those who had initiated the proposal <em class=rfc2119>may</em> revise it
to address the causes of rejection
and follow the ordinary applicable process
to submit the revised proposal.

<h3 id="ac-override">
Advisory Committee Override</h3>

An <dfn export>Advisory Committee Override</dfn> is an exceptional two-step procedure
used to resolve certain matters where neither the usual [=consensus=]
process nor its [[#addressing-fo|escalation path]] is sufficient.

Note: Currently, this only applies to [=AC Appeals=]
and [=votes of no confidence=] in the [=AB=] or [=TAG=].

First, the [=Team=] <em class="rfc2119">must</em> announce the proposal for an [=Advisory Committee Override=]
to the [=Advisory Committee=]
and provide a mechanism for [=Advisory Committee representatives=]
to respond with a statement of positive support for this appeal.
to respond with a statement of positive support for holding a vote.
The archive of these statements <em class="rfc2119">must</em> be [=member-only=].

If, within <span class="time-interval">one week</span> of the Team's announcement,
5% or more of the [=Advisory Committee=] support the appeal request,
the Team <em class="rfc2119">must</em> organize an appeal vote
5% or more of the [=Advisory Committee=] support holding the vote,
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This is basically like having a vote to have a vote. I don't see the point.

It reduces some of my concerns about DoS, engaging this mechanism is exactly the sort of thing I'd like to avoid, but it also makes it nearly infeasible to use this process. Even though 5% seems small, that's a pretty significant movement given the activity levels in a pretty large consortium. I suggested one objector and two seconds, but you could make the two into five or something like that.

Copy link
Collaborator Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This is the pre-existing mechanism and threshold used for AC appeals. This attempts to reuse this system, not to change this.

And you're right, I don't think a 5% threshold is small. It is actually a non trivial barrier to meet. But I don't think it should be trivial to launch votes of no confidence (or AC appeals), so I'm personally comfortable with that.

If you do want to insist on a lower threshold, could you clarify if you want that change both for AC appeals AND votes of no confidence (in which case I'd suggest opening a separate issue), or only for votes of no confidence (in which case this is indeed the right place to discuss it)?

Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I'm more concerned about the time limit. Not only do you have to muster 5% of the membership, you have to do so within a week.

I was viewing this with a lens of the vote of no confidence, but I can't think of a case where this wouldn't be a problem for other such appeals. I don't want this sort of thing to be easy to trigger, but right now it seems to be akin to impossible. A healthy system might have this appeal run once every few years, but I can't remember a single occurrence, partly because this requires the moving of mountains. An accountability system that does not operate is probably one that cannot function, so it might as well not exist.

the Team <em class="rfc2119">must</em> organize a vote
asking the [=Advisory Committee=]
“Do you approve of the Decision?”
together with links to the decision and the appeal support.
whether they approve of the Advisory Committee Override proposal,
including details of the proposal
and links to support for holding the vote.

The ballot <em class="rfc2119">must</em> allow for three possible responses:
“Approve”,
Expand All @@ -2917,24 +3000,19 @@ Appeal by Advisory Committee Representatives</h3>
(including explicit “abstain” ballots)
by [=Advisory Committee Representatives=]:
* if fewer than 5% participate,
the vote fails.
the proposal is rejected.
* if at least 5% but no more than 15% participate,
and the number of “Approve” ballots exceeds three times (3x) the number of “Reject” ballots,
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This seems pretty silly. I understand the goal, but it encourages tactical voting due to it being non-contiguous. Opponents can withhold participation to get a better outcome.

Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

It might be worth an example. For simplicity, let's assume 100 members.

At the 3x to 2x threshold (15%), four "reject" votes blocks any number of "approve" votes up to 11, but one more "reject" vote causes the motion to pass. This is because at 16 participating, five "reject" votes is overridden by 11 "approve" votes.

Worse, prior to the 2x to 1x threshold (20%), a motion is blocked by seven "reject" votes at 13:7 in favor. Up to six more "reject" votes causes the motion to pass. If opponents want to have their say and retain the same outcome, they need to find seven more "reject" votes (with no more "approve" votes).

There are ways to address this, but it involves math. I wonder if this goal is worth that.

Copy link
Collaborator Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

We're not defining this mechanism here, just reusing it (and adjusting the phrasing slightly to make it work in new contexts). I suggest filing a separate issue if you dislike the mechanism itself. For context, this was adopted through PR #901, based on issue #886 (the prior state was a simple 50% majority, independent of the level of participation).

Now, as to the substance of the question, in case you do decide to pursue it, here's a little extra information / opinion:

  • It is true that this is discontinuous, and has slightly weird properties because of that, but I don't think it actually matters in practice, because it's very hard to operationalize: it's only reliably advantageous for "no" voters to refrain voting when close to the threshold if they can (a) know what the current tally is, and (b) count on no more "yes" vote coming in. Depending on how we set up the vote, (a) might be true, but we could easily guard against that by deciding that this must always be a secret ballot. And even if we don't, (b) won't be true, so people cannot really count on staying below the threshold.

  • I think there is value in having something of that nature in general to avoid consequence-heavy decisions being taken by accident just because not enough people were paying attention. Especially in the case of votes of no confidence, I consider them more likely than other kind of votes to be at risk of being invoked repeatedly by disgruntled parties. I think it would be unfortunate if the nth instance of a vote of no confidence passed despite wide opposition, simply because people got tired of answering largely the same question over and over again, or even on the first try because we're just past an election, and too many don't bother answer because they think it's obvious. So having a high bar when participation is low seems appropriate.

  • Though it is indeed a little odd, I think there's value in matching the same system used by the bylaws rather than coming up with something similar but different, even if better. Or we should change both to match.

the vote passes.
the proposal is approved.
* if more than 15% but fewer than 20% participate,
and the number of “Approve” ballots exceeds twice (2x) the number of “Reject” ballots,
the vote passes.
the proposal is approved.
* if 20% or more participate,
and the number of “Approve” ballots exceeds the number of “Reject” ballots,
the vote passes.
the proposal is approved.

If the vote passes,
the decision is overturned.
Following such rejection,
those who had initiated the proposal may revise it
to address the causes of rejection
and follow the ordinary applicable process
to submit the revised proposal.
Otherwise,
the proposal is rejected.

<h2 id="Reports">
W3C Technical Reports</h2>
Expand Down