Wikipedia talk:Levels of consensus

Presumed consensus

Should we say something about presumed consensus when defining "consensus" in Wikipedia:Levels of consensus#Defining some terms? - Butwhatdoiknow (talk) 15:46, 3 July 2024 (UTC)[reply]

Thank you for your improvements, and for the suggestion. Makes sense, I've added it in the assumptions section, I don't think it needs to be defined as a term in the context of this essay. Scribolt (talk) 08:28, 16 July 2024 (UTC)[reply]

Regarding your add -

The concept of presumed consensus is policy (Wikipedia:EDITCON). So maybe "This can be true." should be "This is true."
I'm not sure what is being said by "This can also be said to apply to content which has gone through a peer review process, e.g. FAC."

Butwhatdoiknow (talk) 17:01, 16 July 2024 (UTC)[reply]

Tweaked and updated. You're right that it is technically true, but in terms of the purpose of the essay, the level of consensus is so weak, it's effectively no consensus. I wanted to mention peer reviewed stuff, because by it's nature the content of the silent edit gets a bit more approval because we know someone has actually read it as opposed to missing it in their watchlist. And this also interacts a bit with BRD (which I know is not policy but is reflective of how a lot of editing actually goes) in that you generally need a good reason to remove or change long standing content. If you don't want to discuss, the original consensus on what the content should be stands. I don't want to get too into the weeds here, as this wasn't really the direction I wanted to go in for the essay (which was identifying and contrasting levels of consensus and how they arise), but feel free to further tweak. Scribolt (talk) 08:47, 22 July 2024 (UTC)[reply]

Three things

@Scribolt, I've been thinking about this recently, and I think that the name "CONLEVEL" is causing an WP:UPPERCASE problem. People are guessing what it means from the name:

The level of consensus – that's how strong or weak the consensus is, right?
The level of consensus – that's the scope of the consensus, like whether it applies to one page or to the whole site, right?
The level of consensus – that's about how likely it is that the consensus represents the whole community's view, right?

It's the latter, of course, but I've seen editors claim that it's both of the other two. I don't see anything in Wikipedia:Consensus about the first two, and perhaps the absence of anything real is why people assume that it "obviously" is part of CONLEVEL. WhatamIdoing (talk) 16:50, 30 October 2024 (UTC)[reply]

Interesting point, I think I agree with you with regards the name. CONLEVEL as-is contains four elements (not entirely in order here). First off, and most importantly, it establishes the principle you describe in your third bullet point (i.e. that it is possible to distinguish between a consensus held widely in the community and one that has more limited traction). Second and closely related is what this means in terms of behaviour and editing. If something can said to represent the wider communities view, then something that is less representative should not supersede it. Then we have an example of what this then means in practice; that policy with a high level of community consensus should not be disregarded due to an agreement formed within wikiproject representing a sub-set of the community. Lastly we have some content related to policy creation and maintenance which personally I think is unrelated and out of place.

With regards the first "misconception" bullet point. I've got some sympathy for those who use CONLEVEL as a shorthand for it, and one potential issue with the current name is that "levels" implies a vertical hierarchy (which could relate to size, strength or weight), whereas in fact we're actually talking about something more horizontal; how far the consensus reaches throughout the community. I don't like applying the hierarchy approach here in general, it can ignore important context. The current text already includes "widely" and "broadly", which is the right language. However, the misconception is amplified by the fact that in my view the current content doesn't adequately define on even high level terms what makes a consensus wide or broad. If it did so, it might also assist with people using it in the correctly and was one of main things I wanted to discuss in this essay. Should it be covered elsewhere in CONSENSUS? Hm, I'm not sure there are any real community norms that derive from a "strong" consensus versus a "weak" one, beyond possibly not trying to immediately re-litigate something that had recently and decisively settled. And a strong narrow consensus does not override a weak wide one as per CONLEVEL ;-).

The second bullet point you mention I think is less legitimate and simply people not reading things properly (either CONLEVEL when they quote it, or the closure of the discussion they are referring to). As it may be a common misconception it might be worth mentioning it explicitly somewhere but I'm not usually in favour of writing "this doesn't mean this" in such documents. If it goes anywhere, it's probably more guidance about closing RfCs or the implementation of discussion outcomes, it's not really a consensus related topic in my view.

Hope this rambling maybe helps your thought processes somewhat, thanks for raising it. Scribolt (talk) 08:49, 31 October 2024 (UTC)[reply]

@Scribolt, I'm very interested in your thoughts, but I'm going offline for a while. A couple of links: location is dispositive and RFC results applied to other articles. WhatamIdoing (talk) 02:20, 1 November 2024 (UTC)[reply]

I hope you feel better soon, I hope that my wall of text didn't contribute to it. Thank you for the links. I'm still doubtful as to how relevant / helpful the consensus policy itself can be in the situation Nikkimaria described; there is a consensus that a source is reliable for claiming X in article Y, and that then is is claimed this consensus applies use of the source in general. To extrapolate this, it would be the same as claiming that because there is a consensus not use non-MEDRS sources to support certain information in medical related articles we can't use non-MEDRS sources anywhere. Or to take this still further, there's consensus for AGF, so that means I can't write anything negative about a living person. None of this is really consensus related, although it's adjacent and people use the word consensus. It comes from people misreading discussions, poorly framed RfCs or inaccurate closes. Now this doesn't mean we can't have some text somewhere to try and counter this, but it's not going to be about consensus per se.

With regards to the location comment, this was the core of what I was getting at in the essay. The level of consensus (as we agree, the extent to which the community holds the view) can be expressed as the sum of the bullet points in the "so what does this mean" section (with the exception I guess of the explicitly wide scope identified, which was an attempt at least cover the above concern), with each element multiplied by a weighting of how importantly someone sees it. Some (for example BM based on the link and other comments I've seen from them e.g. here) would weight the page space / policy or guideline aspects far higher than I think you would, or indeed me, and would weight the "correctly advertised" parts far lower. From that perspective, their comment regarding the prospect of a wide consensus emerging from a user talk page is consistent with a belief that the wiki's standards are policy and rule derived, which I think is quite widely shared. I personally consider the page space / policy side of things to be more an aspect of the making sure the right stakeholders are involved, but it would have been negligent not to break it out separately as I wasn't intending the essay to be my soapbox. Scribolt (talk) 14:19, 1 November 2024 (UTC)[reply]

Been mulling this over a bit more, a few thoughts. If I could re-write the section entirely, I'd do the following:

Rename it Level of Consensus. A minor tweak with potentially only minor benefit. Has the advantage of not breaking the continuity of the well established shortcut and also reflecting that we're referring to the level of consensus that something has an not an arbitrary tiered structure.
Keep the first paragraph pretty much as-is, delete the rest.
New para, give some criteria for assessing the level of consensus for something, without trying to get too technical. The level of consensus for something within the community is primarily determined by how many editors could be said to be aware of and agree with it. Long standing content in policy or guidelines, well attended discussions on central noticeboards or well advertised RfCs can all usually be said to result in a high level of consensus. In all these cases, interested editors would have a reasonable opportunity provide input. A more local consensus would be one that a significant portion of the relevant part of the community would not necessarily be aware of.
Next para, provide some caveats and warnings. Don't want to draft something here atm, but should probably include something about Nikkimaria's concern, (i.e. respecting the scope of the discussion) and not to act contrary to existing policy and guidelines unnecessarily. Scribolt (talk) 18:06, 1 November 2024 (UTC)[reply]

@Scribolt, I realized that this is so important to me that I've been postponing replies until I can do it perfectly and completely. I apologize for appearing to ghost you. In the spirit of not letting the perfect be the enemy of the good, here's one thing on my mind:

In re the situation Nikkimaria described, I wonder if the difficult situation might be less a case of a consensus that a source is reliable for claiming X in article Y and more of a consensus unrelated to sources. For example: We decided that Alice shouldn't have an infobox, and that proves that Bob shouldn't either. We decided to use this painting as the lead image for sensitive article A, and that proves that we should also use the same painting in the lead of the article History of A. To name one example I've seen recently, we have a stable first sentence in Trans woman, and that is "proof" of consensus to use the same wording in other articles. WhatamIdoing (talk) 15:30, 12 November 2024 (UTC)[reply]

No need to apologise at all, I wasn't offended in the slightest. I have many pages watchlisted and I've seen you busy in what seem to be more urgent discussions. I'll try and use the "woman" situation as an example of few things, not the specifics as there's way too much background I don't want to dive into, but to try an illustrate different ways I see I can see those kinds of discussions playing out. So, in these hypotheticals we have article A and article X. Both touch on the same concept (woman). Article A describes it using wording B, and article B uses wording Y. Both wording forms are sourced.

There is a discussion involving several editors on article A talk page to say that a woman is B. There is no consensus for any B on article X at this point. It is legitimate to refer to the discussion and the arguments made to develop a separate consensus for B on X, but the context and scope of article X may mean that a different framing may be more appropriate. This is not CONLEVEL related, and is only in my view CONSENSUS policy related to the extent that a consensus is still required to see whether a consensus formed elsewhere for something similar but different applies.
There is a well attended discussion involving several editors on article A talk page that woman is B and should be referred to that way in both articles A and X. There are notifications placed on the talk page of article X and a couple of related wikiprojects There is a consensus to say that woman is B on articles A and X. A subsequent discussion only involving a discussion on talk page X to change back to woman is Y would fall foul of CONLEVEL.
There is a well attended discussion on the RS noticeboard that determines the sources currently used to say that woman is Y are unreliable. Talkpage discussion on X can't determine that the status quo remains (CONLEVEL). There is no consensus for woman is B on X. There is no consensus that it cannot be stated on X that woman is Y with more appropriate sourcing.
There is a well attended discussion on the Fringe noticeboard that determines that defining woman as Y is a fringe viewpoint among reliable sources. A non-advertised talkpage discussion on X can't decide that woman is Y (CONLEVEL). There is no consensus for woman is B on X.

That's probably enough examples. There are very legitimate concerns as to how some of these play out in practice. If the discussion on fringe involved 3 editors, is this sufficient to say that a wide community consensus exists on the topic even if it was in the right place? And in some divisive subject areas, I'm sympathetic to the view that after considerable effort has been expended to develop some kind of consensus on how to say something, having to repeat the process from square one in the face of what may be ideological opposition in every related article may have a negative effect on editors (I think this came up in IP articles recently, and of course this is very much the direction INFOBOXES went). But, ultimately I think that unless someone can point to something without procedural flaws that explicitly indicates that this applies here then you go again.Scribolt (talk) 09:10, 13 November 2024 (UTC)[reply]

With numbers:

I agree with this. Editors may point to precedent, but precedent elsewhere is not binding here.
I'm not sure about this. Wikipedia:Consensus can change, and one would not expect a desire to change a sentence in article B to permanently/indefinitely require a discussion at article A. However, in the short term, we could safely assume that the consensus hasn't changed yet. In the near-term, one would expect that change to only happen as a result of another well-attended, well-advertised discussion.
I mostly agree with this, though I can think of counterexamples. One might, for example, get an RSN vote saying "The Daily Drivel is horrible and should never be used". But if that discussion didn't mention its use in the article The Daily Drivel, then I think editors would agree that the RSN discussion wasn't intended to apply to WP:ABOUTSELF statements in the article about the source, and that the generally applicable RSN decision therefore does not apply in that specific case. Also, what if the "Talkpage discussion on X" is an RFC, and it gets more participation than the RSN discussion? A general discussion with 10 people at RSN is more people than typical for RSN, but it doesn't overrule a subsequent, narrowly framed decision among 20 people at the article.
I mostly agree with this. In terms of wordsmithing, I think this one mixes "well-attended" with "advertised". Discussions on some talk pages will be well attended even if they're not advertised, and some advertised discussions get no responses. Also, FTN has a reputation for not being a completely neutral noticeboard. Still, if they say that WP:FRINGE rules apply, then that can usually be trusted (though that determination alone doesn't necessarily provide a clean answer about what the next edit should be).

WhatamIdoing (talk) 04:49, 14 November 2024 (UTC)[reply]

2. I don't think we disagree. The intention wasn't to say that nothing will change without an equally scoped discussion. Anything born of a wide consensus is going to get re-written at some point, either by someone who knew the history (and ignores it, or who thinks that things have changed with time), or by someone who completely oblivious. The application of CONLEVEL is for when someone else sees the change, disagrees with it and needs a policy based rationale to returning to the prior version pending further discussion.

3. For the first part, I don't think it's reasonable for closers to list every policy exception where their close might not apply. In these cases, and even if the closer did intend to say never use it anywhere ever, the fact that ABOUTSELF is defined in policy would have a demonstratively wider level of consensus than the RSN discussion should be enough to allow the it's usage in that case. Of course this also goes a bit in the direction whether policy follows editing but in this case, the RSN discussion isn't article space, it's just a discussion which is arguably the wrong place in the general WP:V related set of pages.

The second part of this is much harder, because this comes down to trying to compare two similar but different consensus's formed in different places, and as per my essay this becomes subjective. I'm certainly a believer that the RfC on the talkpage you describe is a meaningfully wide consensus, but so is the other. Is the audience that is aware of the talkpage RfC (TP watchers, those who actively monitor RfC adverts) a "better" audience to decide than those who watch the RSN noticedboard? I don't know, in order to decide the community would need to agree the weightings I mentioned. I think in this situation, what actually exists is a "no consensus" for the usage of the source, and we'd need to proceed accordingly (i.e. a discussion that includes at least both sets of participants)

4. You're right re advertised vs attended, but this was a specific example to make a couple of points. The first would be that 50 accounts who turn up on the magic crystal healing TP to decide that it's not FRINGE without telling anyone else (i.e. advertising) don't get to overrule 5 people (plus the silent watchers) in the original discussion on FTN (at least without telling them all about it). The second point I wanted to make was exactly the fact that loudest voices in FTN (and RSN as well) are not always perceived as neutral arbiters. That said, they are the people who speak up and make their views heard, and even if they aren't always representative of the community, they will continue to the be ones who define a lot of what the official WP consensus is on these topics.

Another related discussion might also be; is the model of informal central noticeboards on policy areas the right one for getting an accurate view of the community's view on things? And if it is, are the scopes of what and how they do things really transparent and with endorsed by enough of the community behind them to let them run the way they do? As I said in another discussion with you a while back, it's not so hard to imagine a version of WP where policies are more content area oriented. I'm not advocating that, but it might be worth also considering whether or not it might be better to have discussions about formalising the processes involved in declaring a concept as fringe, or a source as unreliable or deprecated. This might set to rest some of the arguements as to the extent to which a sparsely attended noticeboard discussion applies elsewhere. Scribolt (talk) 14:05, 15 November 2024 (UTC)[reply]

In re the "New para" bullet point ("The level of consensus for something within the community is primarily determined by how many editors could be said to be aware of and agree with it. Long standing content in policy or guidelines, well attended discussions on central noticeboards or well advertised RfCs can all usually be said to result in a high level of consensus. In all these cases, interested editors would have a reasonable opportunity provide input. A more local consensus would be one that a significant portion of the relevant part of the community would not necessarily be aware of."):

Is "it" that editors are aware of and agreeing with supposed to be a discussion, a decision, or a rule? Like: We all agree that copyvios are bad, but we don't need a lot of editors to be aware of any specific decision, and we don't need a well-attended discussion or anything advertised. We have a high degree of consensus for rejecting copyvios and do not need anybody to have any sort of discussion about it. One person with an Undo button and an edit summary of "Blatant WP:COPYVIO" is the highest possible level of consensus.

As a side note, I think the fundamental idea of consensus is "We agree", so "Level of consensus" should be reasonably synonymous with "Level of agreement". I have found this word swap to be useful when I'm trying to differentiate between WikiConsensus™® and plain old consensus. WhatamIdoing (talk) 00:21, 18 November 2024 (UTC)[reply]

The 'it' could be any of the things you list, depending on if we need to compare how strongly the community agrees or feels about something. For the copyvio example, I agree with your reasoning re the revert, and CONLEVEL would say that WikiProject Frogs couldn't decide on a more local basis to allow copyvios. However, when someone reverts something this might trigger a discussion of "is the description of the frog in the article actually a copyvio?". This might involve multiple discussions in multiple places in terms of how or if we apply the general rule here, and resolving those individual cases might also involve CONLEVEL (I don't know enough of how we manage copyvio issues to make this more concrete). In the same way, that crying BLP on it's own doesn't bring the full weight of the consensus of the BLP policy behind you, it actually has to be a BLP violation in the eyes of whoever happens to be around.

Yes, this is another area where our words diverge from what might be expected. When "consensus" is invoked it's usually in the context of people who vehemently disagree, and there needs to be a way to find out who "wins". I think you're right in framing it as an agreement (possibly also a how does the community feel or think about something?) A RfC outcome can certainly be seen as the community agreeing to do something in a certain way, rather than there being a consensus between the participants that it's the right thing to do. Compromise and agreement can be the result of the structured discussions and other methodologies described in the CONSENSUS policy, but they definitely aren't pre-requisites to getting on with things. Scribolt (talk) 15:53, 18 November 2024 (UTC)[reply]

I think it's important to figure out what the "it" is, and to pin that down in the policy (to a reasonable extent), so that when editors invoke a given sentence, they're all talking about the same thing. Otherwise, we have the wiki equivalent of a BrEng/AmEng conflict: "I'd like to table (talk about) this subject" – "What?! This is far too urgent to table (postpone discussions on) it!". WhatamIdoing (talk) 17:26, 18 November 2024 (UTC)[reply]

Ah, I see what you mean, fair point. I think the issue is more with the something, rather than the it. How about for that first sentence: The level of consensus within the community for the application of an editing practice (whether policy, guideline or other informal guidance), or for implementing a particular decision, is primarily determined by how many editors could be said to be aware of and agree with it.

I was going back and forth on the reference to "editing practice". I landed here because I didn't particularly like "behaving in a certain way", using "rule" will be a red rag to some, and I think it's helpful to include the fact that informal "that's the way we do it" practices can and should be in scope of considering how many people actually agree with it. Scribolt (talk) 11:45, 19 November 2024 (UTC)[reply]

I like "editing practice".

But I think the rest is a little bit wrong. Consider the case of blatant vandalism: I implement "a particular decision" to revert it. (Miraculously, I get there before ClueBot does.) How many editors are aware of the "particular decision"? Basically none. But there's strong consensus behind my "particular decision", because it complies with the written policies and, if they did know about it, I can be confident that everyone would agree with it. WhatamIdoing (talk) 16:56, 19 November 2024 (UTC)[reply]

I think this is covered. If what you revert is objectively (I'll come back to that in a moment) vandalism as per VANDTYPES, then not only is the editing practice you're engaged in (reverting vandalism) seen as a good thing, it can be already be said the community agrees with how the you've applied the policy in the edit in question, regardless of who sees it because it's already been described. The people who are aware of and agree with the revert is everyone who has read and agreed with VANDALISM. This is one of the reasons it's a good reason to have enough detail in policies or guidelines to give as many kind of edits a seal of approval before you even make them. Uploading an image of genitalia at the top of an unrelated article is pretty much described word for word in the policy, so you know that even if no one sees the decision you've made, you know that there is agreement for your "decision" even before you've made it.

Now, not all such edits can be quite definitively linked back to community endorsed statements. There are plenty of cases where things are reverted, often correctly, but where vandalism is cited as the reason when it really isn't (at least not how WP defines vandalism). Until we have totally infallible and honest editors, or policy documents that cover every eventuality so that policy application can never be questioned, anything that isn't explicitly described is always going to be subject to community assessment after the edits been made. So, you're correct in a sense that a "decision" you made to revert vandalism that isn't explicitly defined as such in policy is going to be lacking a degree of consensus until other editors have seen it and agreed with it. A slightly less strong form of consensus also exists for reverting edit types similar that have consistently been reverted as vandalism in the past. Community precedent / norms also grants a certain amount of pre-emptive consensus for the revert. I'd be a bit reluctant to refer to this here though, because if it's really such a norm, the purpose of policies and guidelines is to capture them.

But in a sense, I don't think it even matters in the context of CONLEVEL. At present anyway, CONLEVEL is an entirely retrospective concept. Going back to it's basic purpose (a subsection of the community can't choose to reject something more widely agreed), by definition we're talking about something that's occurred in the past. If your reversion of vandalism was undone by someone saying "WikiProject Frogs say vandalism is OK", they aren't questioning the decision you made, they're challenging the wide community consensus on vandalism and CONLEVEL can be used to refute this. If they question the decision you made in the edit re the photo, you can point to VANDTYPES and say an inappropriate explicit image is widely agreed to be vandalism and community is with you on that and they can't decide differently per CONLEVEL. If you are questioned of whether the uncited addition of the phrase "frogs like honey" is an example of silly vandalism, they aren't challenging the consensus behind the policy of reverting vandalism, they're challenging your particular application of it. The level of consensus for that particular edit/decision you made is really only established after you've made it, not before.

Now, it might be useful for other reasons for us to develop a concept of saying "how much consensus do I have for the edit I'm about to make or have just made", but I think there aren't immediately practical applications I can think of, beyond telling us whether we're actually improving the encyclopedia when we hit publish. If we go in that direction, we need to revisit the 1st para which sums up the purpose and when CONLEVEL is applied. Scribolt (talk) 13:35, 21 November 2024 (UTC)[reply]