User:Durin/Thoughts on consensus

On April 16, 2007 I nominated Matt Britt for administrator on en.wikipedia using an RfA format that was akin to an RfC. See Wikipedia:Requests for adminship/Matt Britt. The resulting discussion regarding this format has been, to say the least, rancorous. Discussion on this can be found at

A point that has been raised at least a few times has been that it is impossible to gauge consensus using this format. I disagree. Or, in the least, I offer a counterview. Here's why:

Consensus is not about the percentages of people who agree overall with a concept. RfA has been operating as a vote for a considerable period of time. As a result, a great many contributors are possessed of the idea that voting super majority = consensus. As has been noted by some countering that notion, training in consensus points directly away from super majorities equating to consensus.

Before continuing further, please read the section "Example RfA" below.

A different way to evaluate consensus[edit]

If a bureaucrat were to come to that RfA in today's climate at RfA, it would almost definitely be closed as no consensus to promote. At 10-4-2, it's only achieved 71% support. That's outside of the normal discretionary range of 75-80% in which bureaucrats say they exercise judgment (I say this because the active bureaucrats virtually universally state that is their intention in their RfBs, and because all records of practice show this to be the case, except for a very small handful of exceptions).

But, there's another way of looking at this RfA. From this view, there is in fact support for this candidate. Consider; there are two people who opposed the candidate for the answer to Q1. In the supports, there are 7 people who opposed that view, and in the neutral section we find a person saying it isn't relevant to whether or not to promote the person. Taking 7-2 as a measure, 78% of the people who voiced a vote based on Q1 felt it was good, and just 22% felt it was bad. That's a pretty strong indication that the opposition based on Q1 is wrong. Discount these two votes, and you get 83% support.

Now, the people who insist on voting are probably screaming at this point "You disenfranchised me! That's not fair!" But, RfA is not about voting; it's about consensus.

In discussing any complex topic where a group wants to form consensus on what to do, a considerable number of conflicting views have to be discussed, considered, and weighed. As a group, there may be insufficient support to move in X direction. But, the reasons for not moving in X direction can be scattered, inconsistent, and strongly opposed by the vast majority of the group. If a consensus can not form on why moving in X direction is bad, and there's a strong majority indicating that direction is good, one take of this is that consensus has formed that X direction lacks significant opposition, and that only extreme minority opinions have been raised in objection.

This is how the Matt Britt RfA supports this process. There's a considerable number of views presented. For each view, there is discussion on that view. Is the view relevant? Does it matter to being an admin? Is there evidence to the contrary and how does the group as a whole feel about that view? If a view stands in objection to a candidate, and a super majority forms that says "yeah, this is a reason not to promote" you've got something meaty to grasp to conclude that consensus has not formed to promote the candidate. If instead, you have a number of do-not-promote views weakly supported and strongly opposed, the opposition is incapable of forming consensus that the candidate is bad.

Undue weight of a vote[edit]

Under the current "I must vote and I must be counted" system, extreme undue weight can be placed on a single vote. One vote can literally cast an RfA into failure, even if the vast majority of people on the RfA would agree that the view of that one vote is absurd. There is a (flawed) perception that a voting system makes everyone equivalent, and everyone's views count the same. That isn't true. In our super majority system for RfAs one opposition vote equals four support votes. You generally need 80% to pass in this system. Thus, if one user says "Nom shouldn't be promoted because of X", X carries more weight than the Y in "support because of Y against X". Y has to have four times the support in order to wipe out the effect of X. Yet, the current system prevents focus on X and Y and instead instantly values X as four times more potent.

Result; RfAs are frequently sunk because of a small minority opinion being given undue weight.

Timing of a vote can place undue weight[edit]

In our example below, opposition vote 4 comes in 15 minutes before the RfA closes. During the last 24 hours, there's been virtually no activity. The RfA hovers just above 75%, in the range for bureaucrats to exercise discretion and promote. 15 minutes before close, a voter comes along and says "Oppose: This user performed very badly in the dispute at Z" Prior to this, nobody made any mention of Z. Nobody's had any chance to respond, and no chance for people to consider whether it's relevant or accurate. It's a subjective opinion that can not be readily discounted on objective grounds, such as "blocked for 3RR on March 16" which could be readily proven wrong (if it were in fact wrong). No, in this case the bureaucrat would almost certainly consider it a valid "vote". This vote sends the RfA below 75%; no consensus to promote.

Result; this last vote carries enough weight to invalidate every single supporting comment made in the RfA before it. In this case, one person...one vote, being cast very late in the process sinks the RfA before there's any chance for the community to express displeasure with the vote. This one vote IS the RfA.

Similarly, if we treated opposition vote 4 in our case below as the vote that came in 15 minutes before close, opposition based on Q1 is insurmountably strong; it has sunk the RfA from 77% to 71%, well below the normal discretion range. even though there's strong opposition to this particular view, it was enough to sink the RfA by itself.

Example RfA[edit]

Let's offer an example:

ZippityDooDaa
Voice your opinion (10/4/1) Scheduled to end 17:39, 20 April 2007 (UTC)

ZippityDooDaa (talk • contribs) is a great contributor to Wikipedia lorem ipsum dolor sit amet, consectetur adipisicing elit, sed do eiusmod tempor incididunt ut labore et dolore magna aliqua. Ut enim ad minim veniam, quis nostrud exercitation ullamco laboris nisi ut aliquip ex ea commodo consequat. Duis aute irure dolor in reprehenderit in voluptate velit esse cillum dolore eu fugiat nulla pariatur. Excepteur sint occaecat cupidatat non proident, sunt in culpa qui officia deserunt mollit anim id est laborum.

Candidate, please indicate acceptance of the nomination here: Accept. --ZippityDooDaa

1. What admin work do you intend to take part in?

A: lorem ipsum dolor sit amet, consectetur adipisicing elit, sed do eiusmod tempor incididunt ut labore et dolore magna aliqua. Ut enim ad minim veniam, quis nostrud exercitation ullamco laboris nisi

2. What are your best contributions to Wikipedia, and why?

A:lorem ipsum dolor sit amet, consectetur adipisicing elit, sed do eiusmod tempor incididunt ut labore et dolore magna aliqua. Ut enim ad minim veniam, quis nostrud exercitation ullamco laboris nisi

3. Have you been in any conflicts over editing in the past or have other users caused you stress? How have you dealt with it and how will you deal with it in the future?

A: lorem ipsum dolor sit amet, consectetur adipisicing elit, sed do eiusmod tempor incididunt ut labore et dolore magna aliqua. Ut enim ad minim veniam, quis nostrud exercitation ullamco laboris nisi

Please keep criticism constructive and polite. Tally: (10/4/2)

Support

Support: Seems a good contributor, trustworthy. --Fraggle
Support: Been here long enough, 3000 edits, no blocks. What's not to like? Q1 answer is encouraging. --MeatyBop
Support: I've interacted with this user a great deal and found them wholly dependable --FreezerKing
Support: Answer to Q1 is great. Awesome! --SurpriseMeKnot
Support: SteinFrank's opposition is absurd. The response to Q1 is exactly what we want. Everything else looks great. Why not? --BalboasPunchingBag
Support: Found this editor's work on contentious debate outstanding. Level-headed. --AppaBappaMountainTop
Support: User's page has been vandalized 73 times. Must be doing something right to anger all those vandals! Also per SurpriseMeKnot. --LateNightDoubleFeatureCreature
Strong Support: Really enjoy her image uploads. Fantastic work, and great performance on properly tagging them. Article work is great. Q2 answer a bit weak, but Q1...wow! Go Zippity! --FarthingThrow
Weak support: The answer to Q1 doesn't trouble me; quite the opposite. But, the Q3 answer...not enough reason to go neutral, but hope it improves. --Beach Comer
Support: Echo BalboasPunchingBag, and also the same in response to StableSlug's oppose. --PoolBoy

Oppose

Strong oppose The answer to Q1 is repulsive. There's no way I could support with an answer like that. This editor is obviously the wrong choice to be an admin. --SteinFrank
Oppose Interacted with this user over a speedy delete, and expect her to abuse the tools in that arena. --Extravaganza
Oppose - If this editor becomes an admin, I quit. --Fire Truck Monkey
Weak oppose The Q1 is troubling, and I can't find it in myself to support because of that. --StableSlug

Neutral

Neutral: Can't make heads or tails of the rationale behind and for answers to Q1, but I don't find Q1 relevant anyways. Don't know enough about this editor otherwise, so neutral. --DF987
Neutral, leaning support. Can anyone provide a reason why we should oppose based on Q1 other than hating the answer? Help me out here... Insanity Plea