Wikiversity talk:Canvassing

Notifying the subject of a CR
Someone whose testimony or reputation is either being invoked or impeached in a discussion should be notified and invited to participate, either to affirm any testimony attributed to them or to defend their reputation or testimony if it is impeached.

This is simple matter of ethics, and is also necessary to avoid unseemly trials in absentia.

Caprice 11:05, 1 December 2010 (UTC)


 * Doesn't "in cases where the individuals being contacted are the subject of a discussion or are likely to be affected by the outcome of the discussion" address that already? --SB_Johnny talk 11:36, 1 December 2010 (UTC)


 * I'm going beyond "encouraged" to say it's required as a matter of ethics. —Caprice 14:04, 1 December 2010 (UTC)


 * I think that's too strong, unless of course the discussion is specifically and explicitly about a particular person. --SB_Johnny talk 14:10, 1 December 2010 (UTC)


 * CRs are almost always about specific people whose conduct is being reviewed. It's important to avoid problematic hearsay evidence (e.g. "Moulton said Abd is the exact opposite of a scientist.") and to avoid trials in absentia.  —Caprice 14:39, 1 December 2010 (UTC)


 * Contacting the subject of a review or complaint or the like is not canvassing at all, it's "notice." It's is universally permitted, and often required. Since the is a page on general policy re canvassing, it does not need to be mentioned when notice is required, and notice, again, is not canvassing.
 * Canvassing refers to the solicitation of comment. As pointed out, well, that solicitation can be beneficial. It only does harm if decisions are being made by pure vote. It does become, then, necessary to allow pointing to possible canvassing for individual comments, where canvassing is a reasonable explanation for a !vote. Comments should not be removed because of allegations of canvassing, and the opinion of one commenting does not become invalid because it was canvassed.
 * The content of comments, as to support or opposition for a proposal, is useful only for judging the "sense of the community" as to a proposed action. For this part, a closer will want to consider the degree to which the one commenting is "part of the community," i.e., has been active in areas relevant to the proposal. If it's about a desysopping on Wikiversity, for example, that someone posts a comment opposing desysopping, or supporting it, who has no conributions on Wikiversity, is really not relevant to the "community view. But the person might have valuable opinions or evidence to present, which can and should be considered.
 * I do suggest that a closer, if possibly canvassed opinions are relevant to a conclusion, state this, making the basis of the decision clear.
 * I did see a recent example where canvassed !votes made a difference, but the usage of those !votes was not stated by a closing 'crat. A custodial candidate got, with the canvassed votes, 66%. Without just one of these, the level would have been considered to be passing. As a result, we seem to have lost someone who was enthusiastic and who had not committed offenses here, but only elsewhere, and who had admitted and acknowledged those earlier errors. So better policy is in order, I'd say, so that we make fewer errors -- assuming that this was an error, which could certainly be controversial.... --Abd 15:46, 1 December 2010 (UTC)
 * If there is a notification requirement, it should be specified in the policy for a particular process. This proposed policy is about canvassing in general. --Abd 16:51, 1 December 2010 (UTC)


 * Five years later. Ah, Moulton/Caprice was such a card! The Joker, obviously. Some great Atrocious Song Parodies. (Actually, some quite clever.) About that "exact opposite of a scientist." At the time, I had zero credentials, only some training, fifty years earlier, i.e., physics from sitting with Richard P. Feynman, an inclination and full acceptance of the scientific method, and a lot of writing. What I did on Wikiversity had an impact
 * I learned, and so did some others, and we continue to learn. I still don't call myself a "scientist," but:
 * Google scholar for Abd Ul-Rahman Lomax
 * Google scholar for Barry Kort. Impressive. But how many of those are actually scientific papers? Many of those are not by our Barry Kort, i.e., Moulton. I encouraged Moulton's positive contributions here, even enabled it when he was blocked, but he was completely intransigent, revert warring with highly objectionable outing, and his science ... well, he was a science educator (at least as a volunteer) and I'd want to keep him away from my kids.... He had no concept of how to actually deal with children, other than "obey my authority." Had no kids himself, I think. But ... I didn't read those papers, except some years ago when he was active here. All I would want if he were to want to come back would be some simple commitments. He did make one, one time. He broke it. But people, even oldies like Barry and myself, can learn. --Abd (discuss • contribs) 18:29, 2 September 2015 (UTC)

On the topic here, there is a huge defect in our discussion process. Polls should not be set up initially. There should be consensus on the question to be asked, or at least an attempt and an impass (in which case alternate questions could be asked). Instead, very commonly, questions are set up ab initio, and people who do show up immediately express their position. Which is obviously not based on complete evidence, which may arrive later. Yet they most commonly will not look back at the page. After all, they already voted. So discussions become warped toward what is "quick." Basic wiki problem. It's remediable, but there has long been high resistance to any improvements that will address this. So, given a broken process, with polling or effective polling ab initio, canvassing, then, becomes a problem! It still should not be a problem, if closes are based on arguments, which is generally pretended and which is sometimes true.

If there is just a raising of an issue, with solicitation of commentary and evidence, canvassing is not a problem at all. It should be encouraged, and, then, before a poll is set up, the discussion can be refactored to summarize it. That summary should have consensus. How to do that? It takes some skill. However, the best way to learn that skill is to do it. Then see complaints. Respond to the complaints. If someone does not like your summary of what they wrote, offer them the option of writing something better. In real-world deliberative process, the question goes through, sometimes, many votes. If we want "quick," we can have it, but it creates shallow decisions that often do not resolve controversies, but decide that one side or another is "bad and wrong," hence ... Wiki studies/Wiki disease.

So, notifying editors who might be impacted by a decision is obviously desirable. In early process, anyone can be invited, and numbers of votes do not matter, rather, the initial task is the collection of evidence, analysis, and opinion. There are wikiprojects on Wikipedia, where editors with special interests participate. These are routinely notified when articles of interest to them are AfD'd. Does that warp process? Damn straight it does! But the problem is not with those notifications, those people will be experts with the topic. The problem is with AfD process. What is missing is an evidence and argument phase before the voting begins. The early Wikipedians were mostly naive about standard organizational decision-making. People who arrived with experience with it, and who pointed out these issues, were rejected as outsiders who just did not understand wikis. In fact, some of these understood wikis very well! --Abd (discuss • contribs) 18:29, 2 September 2015 (UTC)

"Recent history
I think the word "recent" should be added back because I think having little or no recent history is an important factor in weighing canvased individuals input. What do you think? -- dark lama  17:36, 1 December 2010 (UTC)


 * Actually I don't think that would always be the case. If the person being notified is commenting in the role of someone who contributed to the item under discussion, their opinion seems just as important. --SB_Johnny talk 18:03, 1 December 2010 (UTC)


 * When do you think it would be the case, or did you mean to write, "Actually I think that would never be the case"? -- dark lama  18:26, 1 December 2010 (UTC)


 * I can't really think of an example offhand, but never say never... :-). --SB_Johnny talk 18:45, 1 December 2010 (UTC)


 * I don't see any reason why "recent" would correlate at all with the desirability of comment. On the one hand, if a particular subject requires knowledge of recent events, we might speculate that someone who has not been active recently might not have enough information. Or we might speculate that this very lack of involvement will lead them to a less biased view. We might assume that inactive users might not know "current community standards," but this is a bit of self-contradiction, because they are part of the community, and "community standards" may drift away from wisdom as well as toward it. We might, in the other direction, suspect that current users are not familiar with the founding standards, which sometimes are deeper in understanding than what comes later, as new people bring in differing concepts of what the wiki should be and how it should operate.
 * Further, someone who has been, say, working on their personal educational projects, may have a great deal of recent activity, but no understanding of other issues on the wiki, and I've seen users, on Wikipedia, with years of contributions, who, when they encounter a problem, are about as clueless as a newcomer, and get blocked in a flash, with administrators assuming that, with such a contribution history, they "must have known" that the behavior, such as revert warring, would not be accepted.
 * Basically, "recent activity" is not useful at all for judging cogency and value of comment. Overall activity might be, though that could certainly be difficult as well. --Abd 20:51, 1 December 2010 (UTC)

What's the problem?
The problem with canvassing is that it can increase participation bias, by attracting some unrepresentative segment of the community to a discussion. Long-term solutions to this will require structure that we are not yet prepared to create.

Canvassing only causes a problem when a decision is altered by it, whereas an even larger and more representative sample of the community would decide differently. Since participation bias can happen without any canvassing at all, canvassing is not the primary problem; with proper closing process, canvassing would have almost no effect except a possible appearance, sometimes, of "ignoring consensus," or of "claiming consensus when there is disagreement." Generally wikis have some layering to decision process that allows correction of participation bias, but this can greatly reduce efficiency. I.e., I just gave my arguments in an AfD on Wikipedia, and now I have to give my arguments again in a DRV? A little creative thinking about this would lead to proposals that would increase efficiency and not require that duplication of effort, at least with some kinds of discussions and debates.

In any case, I'm grateful that this process is under way, because we may be able to set up guidelines that will reduce unnecessary argument over "canvassing." We should look to some definite statements:


 * What kinds of notifications are clearly permitted and even encouraged -- or, perhaps in some cases, required?
 * What kinds of canvassing should lead to immediate warning, and block if continued? (If any, I'm not necessarily arguing for blocking as a response.)
 * What should be done about apparent canvassed votes? And about those who voted and those who canvassed?

We do not need to enable 'crats to consider the issue of factors that might deprecate !votes, it is already part of the basic job description for a 'crat, but it doesn't hurt to briefly describe the principles, and question 3 is about what we, as a community, being responsible for maintaining a decision or discussion process, do while collecting discussion to provide the 'crat. --Abd 20:38, 1 December 2010 (UTC)

The problem with canvassing on Wikipedia is related to the problem of decision-making based on !votes, and more subtly even when !votes aren't the source of a decision, because a large number of canvassed people can present a plethora of possibly irrelevant or defective arguments, and any decision-maker can be influenced by the number of arguments, and investigating all arguments can be a burden. So, I've seen even ArbComm on Wikipedia be influenced by the sheer number of people appearing on one side of an issue. How could so many be wrong?

Easily, they can, and it can happen even without explicit canvassing. Rather, if a group of people share a biased viewpoint, and are accustomed to cooperating toward favoring that viewpoint -- and this can be perfectly legitimate, in itself -- they may have watchlists that inform them of a discussion, and if they outnumber editors "on the other side," a large number of editors can appear and argue for a position that would not be accepted by consensus if carefully considered in detail.

The problem is, in fact, the lack of "consideration in detail." Canvassing cannot affect a discussion that is closed by a breakdown and detailed examination of arguments. The long, threaded discussions that wikis tend to foster are not conducive to this, it takes specific work to create it. I suggest that whenever a decision seems truly controversial, that full-on consensus process be used. Rather than simply closing a discussion, an admin or 'crat would create a page to collect in an organized fashion, evidence and arguments on the issue, and this executive would manage this, or delegate management of it. I will call this person the "facilitator." Some people are quite good at this, in fact, and we should notice and enable and empower such skills. The goal is consensus, full consensus, but the process will terminate when the facilitator is satisfied (and has checked with participants) that the collection of evidence and arguments is complete. At this point, if consensus is not obvious, community vote would be solicited, and it would be openly solicited that such a vote was being taken. Frankly, I wouldn't allow the use of the banner until such a "decision page" was ready. Nobody should unilaterally add their discussion of interest to the banner, it's prejudicial toward the arguments present at that moment.

The structural point here is to confine discussion of a controversy to a relatively small group, facilitated by someone who has the power to control participation, but the motivation to do so only to facilitate consensus, not to support one side or another. This is efficient; consensus frequently takes a great deal of discussion, and it wastes the time of the community to involve everyone in it. Rather, before a final decision is made, everyone will have an opportunity to review it and make any necessary choices. Even to make corrections by suggestion to the facilitator. The whole process would be transparent, but not necessarily fully open at all times. By default, it would be open, but a facilitator could "topic ban" anyone from it, requiring that person to filter their contributions through an unbanned person.

This is, in fact, standard deliberative process, the "decision page" is a committee, created to address a question in a small group rather than on the floor of a deliberative body. Committees then present reports on the issue, they do not make the final decision.

How would the final decisions be made? That's a separate question, actually. Initially, the decision would be made, I'd suggest, as they are now, by a closer who reviews the evidence and arguments, considers the votes -- they are relevant because we do want to know if a decision will be accepted by the general community, if not, it's possible it is not completely expressed, or that participation bias has somehow come up with an "error" as to true consensus, and decides. This person should be neutral, if possible. The facilitator should not ordinarily make the final decision, but in some cases might; a facilitating 'crat might close a desysopping request, for example, or a controversial Request for custodianship. The time to object to the decision of a 'crat to take up the process as described would be, then, at the beginning. Given that only 'crats can promote, if there is objection, I'd assign the decision to the (small) 'crat community.

"Canvassing," through this consideration, can be seen as not a problem in itself, it is only a problem in the presence of defective or deficient decision-making process. The adhocracy of Wikipedia seemed efficient, but, in fact, under some circumstances, it can be highly inefficient, and, I suggest, many of Wikipedia's long term problems have been a consequence of not setting up deeper and more thorough process for finding community consensus when controversy arises. The inefficiency there is horrific, but is mostly not noticed, because the adhocracy seems to work on a large scale. Most articles do grow and improve. It's only from a very long-term perspective, and especially with articles where real-world controversy exists, that the inefficienty becomes blatantly obvious. Wikipedia will need to move into a hybrid system, that uses ad hoc contributions to build basic materials, but that then filters them and maintains them through ... horrors! ... bureaucracy (or something like it; many ad hoc elements can remain, participation can be quite open, but must become restricted at some points, or else we have ... mob rule or chaos. Democracies figured out how to do this long ago, even though what was found is very imperfectly implemented in real politics.

I consider Wikiversity very important because the community size is still small, so there is an opportunity to develop structures that will handle scale. This bucks a trend: "If it ain't broke, don't fix it." I.e., don't set up structure until its needed. Good principle, in part. The problematic part is that by the time the problems of scale become obvious, it can be too late to fix it. We already have problems, in fact, with long discussions that go nowhere. That can be fixed! And what will fix it, done properly, will also prepare us for the rapid expansion of the Wikiversity user base, which I forsee and intend to facilitate. --Abd 18:27, 22 December 2010 (UTC)

Break into bullets?
The language seems to be getting a bit ramblish. Maybe it would be easier to break it into an intro (the first paragraph as it is would be fine), a section on "acceptable situations", and then a section on "unacceptable situations"? Bulleting also makes the policy easier to modify in the future if needed. --SB_Johnny talk 00:50, 3 December 2010 (UTC)


 * Yesterday I split the page into Examples, Weighing input, and Professionalism sections with the intro (the first paragraph) at the top, and added a See also section which currently only includes the navigation boxes for the official and proposed policies. Right now each example is its own paragraph and there are only 3 examples. Do you still think it would be better to break it up into bullets? Can you include how you wish to bullet it in your reply, if you still do?


 * It's definitely better with the headers, so maybe the bullets aren't as necessary. I'm a bit busy with a stretch of good weather, so not a lot of brain power to offer otherwise ;-).
 * One other thing that I was thinking about the other morning was the use of second person address in the policy... something about that just strikes me a bit strange. --SB_Johnny talk 11:09, 4 December 2010 (UTC)


 * I think a lot of the policies at Wikiversity use second person grammar. "When you can", "You'll quickly notice", "If you happen to offend someone", "If you are unable to", "If you have", "You can use Wikiversity to", etc. What is a bit strange about it here? -- dark lama  13:16, 4 December 2010 (UTC)


 * Huh, ok. I think it looks more "professional" to substitute "contributors should" for "when you can", but as long as it gets the point across I suppose I shouldn't be anti-semantic ;-). --SB_Johnny talk 17:05, 4 December 2010 (UTC)


 * I occasionally wonder if the use of second person to address people had intent behind it as a way to humanize Wikiversity's processes more, which can be associated with a level of professionalism. I think the only way Wikiversity could get any more professional, would be to addressed people by their username or real name, which would require or  be implemented ;-)
 * Of course the more we actually do use second person to address people the more it does become a real intention. -- dark lama  18:03, 4 December 2010 (UTC)


 * Perhaps. I'll try not to adopt it as a pet peeve. I have too many pets already :-). --SB_Johnny talk 00:53, 5 December 2010 (UTC)

The present lede
It's important that we find consensus on this as a policy or guideline, because I was just warned for "canvassing," when I wasn't canvassing, as defined. Unclarity in this proposed policy probably caused that. Wikipedia is a little clearer. See discussion at my User talk page and the Colloquium.

This is the current lede:
 * Canvassing is contacting like-minded individuals to coordinate efforts for a pending decision. In general, notifying people of ongoing discussions is perfectly acceptable, provided the intent is to improve the quality of the discussion by broadening participation to more fully achieve consensus and provided it is done indiscriminately (i.e. when done without regard for expected biases). However canvassing done with the intent to influence the outcome of a discussion is considered dishonest because the normal consensus decision-making process may be compromised, and therefore it may be considered disruptive behavior even when other intentions are good.

This bristles with unclear and loosely-defined concepts. "Like minded." "coordinate efforts." "Pending decision" could seem clear (I'd say an open poll would be such), but in practice, every discussion can be thought of as a "pending decision. It is required to judge intention. If I know that someone is expert on a topic, has studied it or has been impacted by it, inviting them can be expected to "improve the quality of the discussion." However, there is an implication that "consensus" is determined by who shows up and how many show up. In fact, consensus is created by discussion, and full consensus -- always desirable -- is shown when all participants agree.

What is the "normal consensus decision-making process"? In fact, it is misnamed. On Wikipedia, it's called "rough consensus," for good reason. It can be very, very rough!

Selective invitation can improve the quality of discussion by inviting knowledgeable participation, which is valuable in the stages of discussion before voting begins. It can improve the quality of decisions if they are made based on quality of arguments, not quantity. "Biased canvassing," even if truly biased, can only harm outcomes if the closer is blinded by numbers, if the closer does not actually consider arguments, which happens all too often. It is very obvious on Wikipedia, with articles for deletion. Some closers will close and explain the close, with reference to evidence. Some just say "consensus is," when it's obvious that there are unresolved disagreements, many votes that did not consider evidence, and, in fact, many votes before crucial evidence was presented. But an admin who closes based on some majority is almost always safe. I have never seen an obviously poor close result in flak for the admin.

Notice the usage of inflammatory and judgmental language: "considered dishonest."

Wikiversity should dump Canvassing as a sanctionable problem. Rather, when it is known that participation was invited, any user should be able to add a note pointing to evidence or stating belief. I've done that. It was ignored by the closing bureaucrat, even when the canvassing was evident. And the canvasser was not sanctioned. So ... it depends on whose ox is being gored. That canvassing was not on-wiki, it was by IRC. Canvassing on-wiki cannot be "dishonest," unless the canvasser lies!

Where to go with this? I'll be making suggestions. --Abd (discuss • contribs) 18:56, 2 September 2015 (UTC)

Examples
The present text:


 * As an example, notifying all contributors of a page being considered for deletion is legitimate, even if all contributors are expected to be predisposed to keeping it. However if only people who are expected to support deleting the page are notified, that is discriminative notification and is considered inappropriate canvassing. While moderate canvassing of a few individuals may be acceptable in some cases, you are strongly discouraged from doing so, as it may increase tension and cause accusations of inappropriate behavior. You are also strongly discouraged from canvassing a group of people based on their age, philosophy, academic standing, or level of participation at Wikiversity, except when a group is itself the subject of discussion.


 * People can provide unique insights and perspectives that can prove to be invaluable in improving the quality of discussions when they are affected by the outcome of pending decisions or when they have expertise on the topic. Requests for people's input is encouraged in these cases.


 * Neutral public notices that may attract many people indiscriminately may also be a legitimate way to notify people of ongoing discussions.

"While moderate canvassing of a few individuals may be acceptable" is vague. On Wikipedia, people are blocked for canvassing. I've seen it be done for a very small number of invitations. Some Articles for Deletion discussions have very low participation. One person may make a difference. "Strongly discouraged" because it may "cause accusations of inappropriate behavior." "Inappropriate" is a mild term. Ordinary criticism is a claim that behavior is "inappropriate." What I saw come out of my being warned over canvassing was a custodian reluctant to try to increase participation in a resource and discussions of how to handle the resource. This causes actual harm!

There is a difference between rough consensus and genuine consensus. Genuine consensus is always increased by the encouragement of all involved to express themselves -- even if their expression is uninformed. Or biased. Consider what happens when someone with a concern and knowledge is not informed. "Consensus" is found, and, then, the person, if they raise an objection later, is told that they are "against consensus." A consensus that they had no opportunity to participate in. Or were not watching their watchlist for the short period of time when some notification might have popped up. Better if all who might reasonably have opinions are informed, so they can comment if they choose. If they have had full opportunity for expression, and the decision goes another way, and the process shows their their opinion was considered, they are far more likely to accept it. This is genuine conensus process.

But genuine consensus process famously takes extensive discussion. It's considered disruptive on the wikis, often, out of an assumption that everyone must read it. Then, rules against canvassing suggest inviting "neutrally." Which then is highly likely to bias the discussion toward ignorance. If there is full consensus in a small group, that has had open participation, but also has "participation bias," that is, by wiki traditions, and soundly, a "provisional consensus." If it's important to broaden that, then this small group may prepare a report for the consideration of the full community. They may create a ratification poll. This is how to improve the quality of decision-making, and it requires disregarding what has been called "canvassing" in the early stages, and canvassing in a final poll is only a problem if numbers count, and "canvassing" is not disclosed.

There is no way to prevent off-wiki canvassing, nor, usually, to detect it. The result has been, on en.wikipedia, highly erratic enforcement of rules. The Eastern European Mailing List case was raised because a mailing list was hacked. In fact, that list was mostly harmless, there was no serious coordination. There was an admin on the list who was informed of a situation on the list, and who handled it conservatively, later confirmed, much more strongly, by other administrators. He was attacked, and he resigned. He was a professor who had written glowingly of Wikipedia process. He was about to see the other side. The Arbitration Committee -- using hacked evidence that violated copyright and privacy -- threw the book at the members of that list, and it's obvious why. With the defective Wikipedia structure, factions can dominate, easily. There is another faction active, they have not been secret, but they are not sanctioned. And it's obvious why: they are supporting activity that is approved by a major administrative faction. Antifringe, antipseudoscience, pro "Scientific Point of View." Which actually violates neutrality policy!

We should actually encourage users to solicit participation in discussions by Wikiversitans with related expertise or interest. We should encourage solicitation of participation here by non-Wikiversitans, and we only need note that someone showed up and voted with few or no other edits (and this is routine).

Now, examples of canvassing that is routinely allowed that might be a problem.

I was recently a Candidate for Custodianship. Given my history, I expect to be more approved by experience non-custodians here than by custodians. A custodian considered my candidacy dangerous. So he posted a note to WV:Notices for custodians, which, by placement, was specifically inviting custodians, as distinct from Wikiversitans, to comment. Ostensibly the purpose was to discourage my mentor (I had an acceptance) from continuing. The place for that was on the mentor's user talk page! This was, in effect, canvassing. A note on the Colloquium would not have been. My point is not that this was "wrong." My point is that rules against canvassing depend for application heavily on what can be subjective judgments. Nothing wrong with gathering opinion from custodians. It should be understood, though, that this is not the consensus of the community.

The sentence about "People can provide unique insights and perspectives that can prove to be invaluable in improving the quality of discussions when they are affected by the outcome of pending decisions or when they have expertise on the topic" contradicts what is in the lede.

Then, Neutral public notices that may attract many people indiscriminately may also be a legitimate way to notify people of ongoing discussions. Is it legitimate or is it spam? Broad public notice has been called "spam." The other side of this is not covered: indiscriminate notices can lower the quality of a discussion. Further, when such notices become common, people stop responding to them. This problem is avoiding, where possibly, by only inviting those likely to be interested, and, as this proposed policy is being interpreted, that was considered canvassing, and wrong and harmful.

This policy never gained consensus on the text. I'm pointing out some of the causes for that. We need to stop sweeping these things under the carpet. It's getting lumpy. --Abd (discuss • contribs) 19:32, 2 September 2015 (UTC)

Weighing input
We presently have:
 * Canvassed individuals may not know all the facts involved in a discussion, and may voice an opinion without reading discussions, and may be predisposed or have been persuaded to believe a POV is correct prior to their participation in discussions. For this reason the input of canvassed individuals who weigh in are generally given less weight when reading community consensus, particularly if they have little or no history of involvement with the Wikiversity community. Canvassed individuals should be treated with respect, while at the same time, the nature of their involvement in discussion should be made clear. New people joining Wikiversity as a result of canvassing should be welcomed like any other newcomer.

This is the best piece of business on the page. It's not quite right. What does "weight" mean? This is a reference to "weighted voting." When the goal is to assess overall support/opposition to some proposal, that's reasonable. There are many methods of weighting, though. There are actually two aspects to a close: arguments and a measure of consensus. It's clear on Wikipedia: closes are to be by weight of arguments, not by weight of participants. But, in fact, administrators also do not want to face the disruption that can happen when they close against a majority, I've only seen a few willing to take the heat for that. I remember one very contentious Wikipedia Miscellany for deletion discussion. Strong majority for deletion. Which violated long-standing consensus against deleting rejected proposals, but rather deprecating them as "failed." A very experienced Wikipedian closed as keep, based on arguments. You could practically hear the screams. They went to Deletion review. It was then noticed that the user was no longer an admin, he'd resigned the bit (with no "cloud"). Naturally, that was raised. For whatever reason, it was relisted. Closed the same. Huge waste of time, cause by people reacting hysterically to a proposal. If anyone wants to see it, I'll link.

A closer should be aware if a close is not "majority opinion," and exercise caution. In that, consideration of canvassed votes is appropriate. However, in fact, the real issue would be inexperienced votes, visible in contribution history. And when we are seeking maximized consensus, a few votes don't really make a difference. --Abd (discuss • contribs) 19:45, 2 September 2015 (UTC)

One more point. I have invited many people to participate on Wikiversity. Sometimes they show up, and it has happened that one or two of them commented in a process involving me. There were claims of canvassing. There was a request for checkuser. I have never off-wiki canvassed participation in a specific process here, and I don't recall canvassing on-wiki; what I did the other day wasn't canvassing, we were in a discussion phase. It had been claimed that the proposal was only my own opinion, so I could have had a motive to show this was not so. However, what I had in mind was people who had been affected in the past by the speedy deletion reason I was proposing to remove. That was one of the founders of Wikiversity, and the issue of the speedy deletion criterion had just come up on his talk page, and that, in fact, was what had led me to act on our guideline page, a custodian who had clearly been affected, and a Commons administrator with high expertise on the subject. The discussion was improved. No decision has been made, no poll has been started, yet. (Even though the custodian used support/oppose templates.) The question had not been asked that way. --Abd (discuss • contribs) 19:52, 2 September 2015 (UTC)