User talk:Jimbo Wales

(Manual archive list)

Announcement about Pending Changes

I have been studying the end of the Pending Changes poll and was asked by the Wikimedia Foundation to help interpret and to advise them about the meaning of the poll and what should be done going forward. My feeling is that, as usual, none of this is up to me to decide as a matter of finality, but that I can and should play a strong role in communicating community desires to the Foundation.

Based on the poll result, I’d like us to take a path forward that’s in the spirit of consensus driven decision-making. It’s clear from the poll that there’s a strong majority in favor of continuing to use some version of Pending Changes (65% / 35% of Support/Oppose voices, plus a considerable number of people who opposed the straw poll). It’s therefore clear that there’s absolutely no consensus for simply turning the system off and walking away.

On the other hand, there’s substantial, vocal, and articulate opposition to using a system of this kind at all, or to using it in its current form. I’ve looked carefully at the feedback, and want to address some of the most common concerns:

Openness: I believe Pending Changes, used properly, can make Wikipedia more open if used on pages that would otherwise be put under some other form of protection. We’ve been very thoughtful in using this system where it makes sense, and I believe that we can and should continue to exercise thoughtful discretion about its use.

I want to share here a personal lesson that I learned through the trial. When we first talked about this system, I believed we could use it to open up pages like George W. Bush and Barack Obama. At least in the current incarnation of the system, that’s not the case, I was wrong: these particular pages get flooded with so much vandalism, that the workload of vetting edits and the polluted edit history weigh greater than the benefit of the system.

But, I also think that this represents a useful future direction for the development of the system -- the development of more effective bulk review and edit management, to ultimately make it possible to unlock these pages. So, it continues to be my belief that Pending Changes represents a useful strategic direction to open up problematic pages rather than locking them down, and that this is in fact one of the most important topics to talk about as we refine and tweak it.

Effectiveness: Many who opposed continued use of Pending Changes expressed doubts regarding the effectiveness of the feature. I’ve looked over the results at Wikipedia:Pending changes/Metrics, and I think there’s still much to learn. As noted above, it doesn’t work well on highly trafficked pages yet -- but there are some places where the tool seems to be effective, such as articles relating to current events (e.g. Toy Story 3 or Spain national football team). Supporters of the system have also pointed out that it can be usefully applied to determine if a page is ready to be moved from semi-protection to unprotection, without exposing it to visible vandalism. All this argues for continued careful exploration of the tool.

Complexity: Many believe that the feature is too complicated, and that the user interface is too difficult. I agree that the current iteration is rough around the edges, but I’m hopeful that this feature will become a more integrated part of the user experience over time, just like traditional page protection. I'm asking the Foundation to fix the more problematic parts of the Pending Changes interface. Based on my initial discussions, they know of several things they can do to streamline things, and plan to continue to improve and integrate the interface. If I didn't feel that I had a firm commitment from them to do this in a timely fashion, I would feel differently about the whole thing. But I think they'll do fine - and volunteer developer input would be massively appreciated.

In my traditional leadership role in the English Wikipedia, I am therefore also asking the Wikimedia Foundation to keep Pending Changes enabled, and to increase the hard-coded limit of pages as the performance characteristics of the system allow it. We will discuss on various policy pages where the use of the system is appropriate. I would ask all those opposing the system to use the coming weeks to provide feedback (or code!) that will reduce the complexity of the system, and help us to make Wikipedia more open, not less. And I suggest that 6 months from now, in mid-March, we have a deeper conversation about determining success or failure of this system, on the basis of the data available by then. My hope and belief is that, working together, we can build a system and policies that have even broader support, similar to page protection and user blocking today. --Jimbo Wales (talk) 19:11, 10 September 2010 (UTC)

Discussion

Thank you for this statement Jimbo. Might I ask what you are asking the developers to raise the hard-coded limit of pages to? NW (Talk) 19:19, 10 September 2010 (UTC)

As the performance characteristics of the system allow it. My view is that limits on the use of the tool should be up to us, through policy, not imposed by the software, but if there are performance reasons to limit it for now, I'm ok with that as well. I don't know what they will do first, but probably doubling the limit would be reasonable and then again if there are no problems as we increase our usage.--Jimbo Wales (talk) 19:23, 10 September 2010 (UTC)

I was hoping that we might start trying out Scott MacDonald's proposal, with a few thousand low-edited BLPs at a time. Perhaps we could reserve a few thousand spots for the higher-traffic articles, and add pending changes to maybe 5 or 10,000 pf the BLPs as a test. What would you think of that? —Preceding unsigned comment added by NuclearWarfare (talk • contribs) 19:26, 10 September 2010

I am all for it, "as the performance characteristics of the system allow it".  :-) I think it would be unfortunate if slowness that should be fixed in the software causes people to shy away from using the system. But once we identify a class of articles (low-edited BLPs, particularly those which are not well-watched?) where it is useful, we should go ahead. I think that one problem with the test to date is that with only 2,000 slots, we had to be careful about where to use it... so lots of people (myself included) didn't yet add pages that will obviously benefit. Still, we must defer to the techies on the performance characteristics.--Jimbo Wales (talk) 19:28, 10 September 2010 (UTC)

Lets get this straight -- there is no consensus to turn it off? It was off in the first place! It is being turned on, and the concept of community discussion is being squelched. This is a very unfortunate decision. If this is so critical, perhaps some WMF employees might consider doing a half decent statistical analysis. It is clear that this could be potentially alluded to in your comments above. User A1 (talk) 19:38, 10 September 2010 (UTC)

I'm sorry, A1, I don't understand your reasoning. It has been on and in use for some time now. There was a 2/3 approval to continue it. I wasn't happy with 2/3 - I think we should shoot for 80% or more, so I'm asking the Foundation to improve it based on community feedback so far, so that we can discuss an improved version. How is any of that "the concept of community discussion being squelched"?--Jimbo Wales (talk) 19:43, 10 September 2010 (UTC)

It was a trial, remember? It was supposed to be turned off a while ago, at the end of the trial, but wasn't - Kingpin¹³ (talk) 19:45, 10 September 2010 (UTC)

No, it wouldn't have made sense to just shut it off at the end of the trial - and then vote - and then turn it back on. Why bother with all that?--Jimbo Wales (talk) 19:58, 10 September 2010 (UTC)

Jimbo, if you're not happy with a 2/3 approval, why is it still on? Stop its use here until such time as the developers feel they've got it fixed on the trial implementation which was going on in the labs, then let's have another go here. Keeping it switched on and enabled when you've said yourself you're not happy with the figure reached is not, at least to me, acceptable. Send it back to labs, let them test it there, then let's test it here when the devs have worked on it. BarkingFish 19:53, 10 September 2010 (UTC)

I'm ecstatic that we got to 66% with version one of the feature, but that's not good enough to simply stop and say it is done. It is plenty of support to continue, but to insist on improvements.--Jimbo Wales (talk) 19:58, 10 September 2010 (UTC)

@Kingpin: Part of the trial was to reach consensus once it expired to continue or shut it off. As Jimbo said, there is currently no consensus in support of the latter. Tyrol5 [Talk] 19:55, 10 September 2010 (UTC)

But part of the reason the trial started in the first place was due to editors supporting a limited trial to see what it was like, and then review. So the consensus should be gained to switch it on now. We shouldn't have to gain consensus to switch this off, when we were promised it was just going to be a trial... - Kingpin¹³ (talk) 20:01, 10 September 2010 (UTC)

@Tyrol5 - Per Jimbo's own quote "I wasn't happy with 2/3 - I think we should shoot for 80% or more" - that shows even he isn't accepting that 2/3 is enough to keep it enabled. I don't see that we HAVE reached consensus to keep it on. We haven't, not by a long shot. BarkingFish 20:06, 10 September 2010 (UTC)

BarkingFish, that is not my view. 2/3 is enough to keep it enabled. I'm not happy with stopping here, though, which is why I'm requesting that the Foundation address the specific concerns so we can get to 80%.--Jimbo Wales (talk) 20:30, 10 September 2010 (UTC)

• (edit conflict)I think the Pending Changes system is just a huge assumption of bad faith. If somebody vandalises, there are always 4 or 5 people on Huggle ready to revert it instantly. I don't see how makes Wikipedia more open; it does just the contrary IMO. The Accept/Unaccept system is confusing too. And PLEASE, no matter what you do, don't follow de.wiki's approach of putting Pending Changes on EVERY SINGLE PAGE. Access Denied 19:44, 10 September 2010 (UTC)
  • (edit conflict) My thinking, and from Jimbo's comments I gather Jimbo's thinking too, is that PC is good for pages that would otherwise have semi protection (or worse). In the case of semi protection no IP edits; in the case of PC IPs can edit. Of course that needs policy to ensure it happens - I'll concede that without policy enforcing that, then PC could well be used instead of no protection. TFOWR 19:53, 10 September 2010 (UTC)
  • @Access Denied:Yes, but the feature is intended to keep questionable material, that is not blatant vandalism, from appearing to anonymous viewers before being reviewed by another, more experienced, editor. Tyrol5 [Talk] 19:50, 10 September 2010 (UTC)
  • @Access Denied: I don't get your point. de.wp is my home-wiki and when I compare the recent-changes page here and there the difference is obvious. Here the diffs are full of kiddie's entries like "LOL" and "Hellooooh!" whereas such entries are almost entirely missing in de.wp. Changes by IPs in the German wiki are not visible to them after they have edited the page. So most of them simply give up. The system is working efficiently for more than two years; why do you think I does not make sense implementing it over here? I agree that the interface needs improvements. I'm wondering whether the developers talked to their German colleagues? →Alfie^±Talk 11:45, 11 September 2010 (UTC)
• Jimbo, let me also say thank you for this analysis, which in my opinion you have gotten exactly right. Where you discuss your experiences with it working somewhat differently than expected on some pages, I had a similar observation, and I'm inclined to think (as I've said on pages where the trial and poll have been discussed) that PC isn't so much a substitute for semi-protection as it is a substitute for no protection, on some but not all pages. --Tryptofish (talk) 19:47, 10 September 2010 (UTC)
• @Access Denied: Set Huggle to show oldest edits, keep 10 minutes of diffs, otherwise default settings. See how much slips past the aggressive RC patrollers. You'd be surprised. I usually see several blatant vandalism that are 5+ minutes old when doing it that way, and a lot of sneaky vandalism that becomes obvious if you spend more than 2 seconds looking at it. Huggle is the best tool right now, and it does an ok job, but it's so far from perfect that it's not funny. The big flaw IMO is there's zero coordination, so we all have to look at the same diffs. As a result, the only coordination possible is "if it was reverted, it was vandalism". That, and the competitive spirit of RC patrol (not itself a bad thing, but some bad outcomes), mean that RC patrollers try to work with as much volume as possible, when what we need to be doing is spreading out the workload to make sure that every edit gets a high confidence decision as to vandalism, good faith but harmful, not vandalism, or is tagged "not sure" and left for someone else to take a more thorough look at. PC is the beginning of one system to do that, but it's not the only possible way forward. The review model needs to extend even to edits that aren't under PC. That means, when we find an edit to be constructive, we sign off on it - taking responsibility not only for what we revert, but also for what we let through. Triona (talk) 06:14, 11 September 2010 (UTC)
• Jimbo did you read the discussion on the talk page and in particular this section here? - Wikipedia talk:Pending changes/Straw poll#Alternative phrasing - where almost unanimous agreement was being reached on a way forward which is different to what you are saying/proposing/dictating and I am to understand you are over-ruling that discussion? Davewild (talk) 19:51, 10 September 2010 (UTC)
  • That discussion was started after the time when I was reading a static (downloaded so I could read it offline) version of the discussion... and it only progressed a short time. I am not going to ask that the feature be removed from any articles where it is currently active, as there appears to be absolutely no positive reason to do that. If there are articles where it is causing problems, then it can be removed from those articles - but why remove a popular feature that is working well in most cases, just to have more discussions? It works well enough to keep - the community has spoken loud and clear on that point (65% in favor!). The best energy looking forwrad is to assist with comments and code on how to improve the feature.--Jimbo Wales (talk) 20:01, 10 September 2010 (UTC)
    • I couldn't agree more. There simply isn't a positive reason to remove it from where it is unless it is to solve a problem. @Davewild: I don't think that Jimbo necessarily contradicted the contents of that discussion in his closing comment. Tyrol5 [Talk] 20:08, 10 September 2010 (UTC)
    • I wanted to be careful here, so I just went and checked the original votes of those who supported this proposal (which came to late) - their votes are mostly split between (3) - keep with gradual limited expansion - and opposing the vote on principle. I think the question of what do to while we wait for the foundation to release version 2 is one that isn't hard for everyone to agree on, at least in principle: remove it where it is causing a problem, keep it where it is working, and use it more in cases like the ones where it is working ("gradual limited expansion").--Jimbo Wales (talk) 20:14, 10 September 2010 (UTC)

I've asked for a while that proposals derived from the straw poll accommodate the significant majority support shown. Jimbo's suggestions here very clearly do so, IMO. BigK HeX (talk) 20:15, 10 September 2010 (UTC)

So once again, we're walked on. The 35% of community who have spoken out against this, for varying reasons including bureaucracy, complication, ineffectiveness and general all-round annoyance at its introduction are being discounted. It doesn't work well enough to keep unless there are significant improvements to the system and its effectiveness. If it does get kept (and it looks like its going to, whether we like it or not) can we at least have your assurance, Jimbo, that after the 6 month discussion you propose in mid-March of next year, that if the community says "switch it off", you'll abide by that? BarkingFish 20:16, 10 September 2010 (UTC)

I'm sorry that you feel that the community has "walked on" you. That's what happens in voting situations. Fortunately, we aren't in a pure voting system, we try to proceed based on consensus. You can have my assurance that I will, at the appropriate time, again recommend to the Foundation that they follow the wishes of the community, and that they listen really hard to dissent to try to find a way to make things more universally appealing for everyone. I don't really understand your attitude here, as if someone is doing you wrong, or I'm overriding the community. The numbers just don't support you.--Jimbo Wales (talk) 20:19, 10 September 2010 (UTC)

Thank you for the response, allow me to clarify my position, please. I don't feel as if someone is doing me wrong, but you quoted yourself only a small number of lines above here, that you felt two thirds or 65% wasn't enough, yet you still decide to keep it on. Where is the reasoning in that? You're not overriding the community, you're overriding your own judgement in my opinion. Can we at least get some fixes in place off-wiki, i.e in the labs, before it carries on here? Maybe if some of the issues were sorted first, I'd be a little more open to accepting its continuation. BarkingFish 20:26, 10 September 2010 (UTC)

You have misunderstood me. 2/3 support is plenty strong enough to keep it on for now. When I said that I'm not happy with 2/3, I meant it in the sense of: I want it to be even better. That's why I'm asking the Foundation to fix the known problems with it, improve it, and do so with energy and vigor!--Jimbo Wales (talk) 20:33, 10 September 2010 (UTC)

My question was about workload as I found it fiddly - to see how that impacted, another 'trial' of a few thousand BLPs is not going to tell us much. If we make it on all BLPs for three months we are actually looking at something different WRT workload and something along the lines of the original intention of the thing. Hence we either try it fully or not, not continue in this piecemeal fashion. Casliber (talk · contribs) 20:22, 10 September 2010 (UTC)

Hence the concept of gradual expansion (if that proves to attain consensus at some point in the future). Tyrol5 [Talk] 20:24, 10 September 2010 (UTC)

(ec)I think this makes sense. Pending changes is a feature that will be selectively employed on some articles. It didn't work well because of implementation issues but we all know that implementation issues are better sorted out in the real world than in a lab. So, we have a feature that can be used where necessary, which will only get better over time, and that 66% of editors apparently already find useful. Continuing it makes sense and is in no way a squelching of discussion or a trampling on consensus. IMO, of course. --RegentsPark (talk) 20:27, 10 September 2010 (UTC)

I will elaborate on my comments from the top of the page. There is obviously going to be a lot of talk here.

• By keeping this enabled, and I am sure you can see this, you create the impression that no matter the outcome of the trial, this was going to be kept on. This is regardless of the reality, the impression will always be generated by this sort of action -- surely this devalues the validity of community discussion.
• From a purely political perspective if you had made this "announcement" with some statistics, this would have been a better political move. Rather this is being continually made with the same "gut" response.
• PC does not alter the relative openness of wikipedia, rather how users apply pending changes does. If we rolled out PC to the entire encyclopaedia, this would surely be closing wiki. So PC itself does little here. What it does do is alter the workload that users have to deal with with respect to vandalism. We are very good at keeping vandalism under tabs -- the success of WP is itself a testament to this. What we are not good at is writing articles, and keeping the number of policies, concepts and ideas that users need to keep in their head at once in order to be part of the WP community.
• After my first edit I wrote "Did I do that right?" -- I had no idea. I still find policies being bandied about that I have never heard of before. WP:PEND does little to help new users, and does a lot to confuse the basic WP model.
• Comments have been made like "lets roll this out to all BLPs", this is precisely the opposite of open, and represents the horrid closed, censured approach to article writing. There is certainly some of that 65% who believe that we should lock up a large subset of articles, purely for convenience. This is not what one would expect of a dynamic community, and precisely what one would expect of a stoic, slow cautious beauracratic process.

In summary, if it was clear that PC was going to be kept running, calling it a trial was not a good idea, it should just have been switched on, and saved us all a lot of wasted discussion. Our time has been used here in this debate, if nothing else. Now the discussion is going to morph from "should we have this on or not" to "what ever happened to consensus", this is a unwelcome development. 81.110.184.105 (talk) 20:36, 10 September 2010 (UTC)

None of this makes any logical sense whatsoever. How exactly does accepting the communities overwhelming support (65%!) amount to a signal that it was always going to be kept on? That is not only false, quite obviously, it is also not even remotely plausible in term of giving that impression. If I asked that it be turned off, that would be a horrible violation of the community, it would be ignoring what the community has said loud and clear. I just don't know what you mean.--Jimbo Wales (talk) 20:46, 10 September 2010 (UTC)

65% won't appoint an admin, which is "no big deal". How did it get to be "overwhelming support" on a major change to Wikipedia's operation? Septentrionalis PMAnderson 22:49, 10 September 2010 (UTC)

There's a vocal minority who hate pending changes, and they're the ones who are going to pop up here griping as the decision went against them. Some people have this bizarre idea that "consensus" means "everyone agrees", which is obviously not the case. Just by numbers, 2/3 approval is a supermajority, so this is hardly the imposition of something without community support. And I say this as someone who finds pending changes to be frustrating. Fences&Windows 21:26, 10 September 2010 (UTC)

And there is plenty of good faith opposition to a system which completely rewires our core tenet (assertions from Jimbo to the contrary). It's not fair at all to just announce that the consensus was "overwhelming" (it wasn't) or that opposition is knee jerk/inflexible (they aren't). And the general concern that a huge number of editor-hours were wasted in a trial/discussion/post-trial discussion is valid. If jimbo is going to step in at each veto point and assert his authority, then he ought to just cowboy up and declare there will be no veto points. Wastes everyones' time otherwise. Protonk (talk) 21:31, 10 September 2010 (UTC)

Unlike others, I do not feel that this is a disaster. I opposed PC, but felt that its implementation was inevitable. I only wish that you Jimbo, had trusted the process and the community and allowed us to reach a decision by consensus, which we were on the point of doing before you stepped in. I understand that you may not have agreed completely with the decision that was emerging from discussion, but that is an inherent characteristic of operating on consensus. It requires compromise, even from you. Revcasy (talk) 21:41, 10 September 2010 (UTC)

I personally believe that PC could never get the 80% support that standard consensus requires, but PC is a good thing for Wikipedia, it can help us keep an eye on low traffic BLPs and other articles prone to low profile vandalism. Perhaps in the future, it can be used in some improved form on more contentious pages.

I can see the arguments both ways on using pending changes, but I do think that the consensus that had pretty-much been reached at Wikipedia talk:Pending changes/Straw poll#Alternative phrasing was a better way forward than that suggested by Jimbo. Yaris678 (talk) 22:36, 10 September 2010 (UTC)

Yes, it is; people of both sides agree to it. Septentrionalis PMAnderson 22:44, 10 September 2010 (UTC)

(edit conflict) I'm sympathetic, and I appreciate the work being done on the Straw Poll talk page, but the alternative Phrasing proposal has 12 support !votes - including struck !votes and weak supports. I can't accept that 12 !votes a consensus makes. I'd !voted in an earlier proposal or two, and that page is littered with proposals: I'm not convinced that the Straw Poll talk page is the right venue for any serious attempt at consensus building. TFOWR 22:47, 10 September 2010 (UTC)

We won't have another two-week poll on interpreting the result of the poll. This is the best way to go and reflects consensus accurately. Cenarium (talk) 22:51, 10 September 2010 (UTC)

It has the voice of everyone who discussed it, with two exceptions. One can be dealt with by tweaking the wording, the other is irreconciably opposed to any future commitment to even discuss PC - and therefore to this decree also. Considering that what's being discussed is a poll closure, generally done by one admin without discussion, that's pretty good - and more consensus than here. Septentrionalis PMAnderson 22:56, 10 September 2010 (UTC)

I would be in support of PC were this the mythical Utopia of Perfection. Unfortunately, human nature is such that PC is fatally flawed. I've seen enough of my fellow man to know that any implementation of PC is going to end up, despite all its good intentions, anathema to Wikipedia. —Jeremy ^{(v^_^v PC/SP is a show-trial!)} 00:59, 11 September 2010 (UTC)

Mr. Wales, given that a minority of reviewers requested a way to decline an edit made to a PC protected page if it was blatant vandalism, patent nonsense, unsourced/unreferenced etc. I was wondering if the WMF is currently working on developing such a way of declining such changes to an article. Many thanks, --James Lu/Ғяіᴆaз'§Đøøм | Champagne? 11:30, 11 September 2010 (UTC)

Break

Jimbo, did you even look at the talk page ? I would like you to read this and respond point by point to my statement (also copied below). Before anyone starts shooting at me that I'm an anti-FlaggedRevs, know that I am the creator, proposer and (by far) principal organizer of this trial, without me there would have been no trial. And I have not spent hundreds of hours devising this trial proposal and trying to get a real consensus for it, to see it destroyed like this. The trial was supposed to end after the two months, so that we can have a constructive discussion on the results of the trial, the merits of PC, the problems raised and then that we make a thoughtful, careful decision, as a mature community and in serenity. What's the point of having a trial if we don't analyze it properly, try to find solutions to the problems identified, etc, so we can make an informed decision ? It's because the devs said that this wouldn't be practical to turn it off then back on that we've got this imbroglio - and that poll was badly designed by a single person (too bad I was absent at that time). It was not a consensus to turn it off that was necessary, but the opposite. A compelling consensus was necessary to divert from the default deactivation, what we've got is nowhere near that. And you are compromising the future of PC by not respecting this, because people will be less willing to compromise if the implementation is kept active on frail grounds (and more, see statement).

Erik Moller had assured here that 66% were needed to approve a FlaggedRevs implementation, we've got 60% here (you forgot the other responses), so PC ought to be removed from articles on that sole basis. We have found a compromise closure here, making the implementation dormant, discuss thoroughly in serenity (which is not possible in the state of permanent controversy that would result of a forced extension) and come up with new implementation proposals, which was roughly what we expected to do after the trial from the beginning. Now that would be showing that we are a strong, mature community.

I copy the text below of my analysis, please address each point:

My interpretation of consensus

Initially, the implementation was supposed to be deactivated at the end of the two month trial, since as the name suggests this was supposed to be a trial limited to two months, we would then analyze the results, the merits of PC, and discuss potential new proposals for implementation. However, the WMF team in charge of the trial having indicated it would be more convenient not to have to turn off then back on the implementation, if a new implementation were desired; the community was given the opportunity, if supported by a compelling consensus, to extend the duration of the implementation beyond the two months, provided a decision would be reached within a month. So, unless a compelling consensus exists to continue the active implementation, it should be halted, though not necessarily entirely turned off.

There has not been any consensus on how the discussions regarding the extension of the implementation beyond the two months trial should be organized, input on this question was requested at different occasions but were not sufficient. Thus, it resulted in incertitude, and this poll being set up by essentially one user. While the poll initially specified how the results should be interpreted (6-4 majority in keep would result in keep, and the specific keep option chosen would be the one with most supports), those specifications were proposed by a single user and not established through consensus, so can't be used to determine the outcome. During the course of the poll, no proposed interpretation of the poll was agreed upon, and considerable objection to the poll structure, from all sides surfaced, with no resolution (other responses, archive). Therefore we should per general practice determine if, based on all discussions, including the poll, there exists a consensus which is compelling enough to continue the active implementation.

Numerically there is almost 2/3 60% (correction counting 'other responses') in support to keep, but it doesn't mean much in terms of consensus, moreover several are conditional, and whether they should be taken in conjunction or separately is disputed, and doesn't appear satisfactorily resolvable. To determine consensus, we should analyze the reasons given for and against extending the implementation and their level of support. That the poll offers different possibilities for keeping without provisions confuses the two different questions of whether we should extend the implementation beyond the two month trial and what kind of implementation we want in one. This creates several problems, it makes interpretation more difficult, and it has the consequence to disenfranchise those opposed to continue an active implementation at this point but who want to have a say on which kind of implementation the community could adopt in the future. Therefore this poll cannot form a basis to valid a specific keep option (2, 3 or 4), and should only be considered as a potential justification for a provisional extension of the implementation under the same terms or in a more limited form.

Moreover the keep options of the poll were offered unconditionally, and while the community can change its decisions, it could be argued that, assuming one of the keep options had been approved, in order to later deactivate the implementation, a consensus for doing so would have to be achieved, thereby reversing the burden of consensus building. The level of consensus here is nowhere high (and informed) enough to justify adopting such unconditional options, particularly as this is a major shift from traditional editing practices, for which an even higher level of consensus than normal seems required (see below) (to handle this, a future proposal could include the provision that if contested, a consensus is required to keep PC - as opposed to a consensus to remove it, at least for some time).

A further justification that the poll cannot support any of the keep options presented is that no comprehensive analysis of the trial occurred, by lack of time. This implies that any non-provisional decision at this point would be uninformed, and hasty. Thus the discussions could only legitimate a provisional extension of the trial, during which we would analyze its results, merits of PC, then decide of a course of action. The question is thus reduced to: is there a compelling consensus to extend the implementation provisionally ? We could theoretically start a new poll explicitly asking this, but there would be no time in the deadline of one month given by developers to organize such a poll.

Considering the various issues raised at the closure page, here and with the statistical data obtained, I think there is enough material to conduct an analysis of the trial, and to form consensus on various points of policy such as the extent of the application of pending changes or the use of 'level 2' protection, which can help to form a consensus on the question of adopting a form of PC or not, so as not to require an extension of the trial. To sum up, the trial served its purpose in giving the community an apercu of what they could expect of PC and will help the community in reaching a more informed consensus on the question of using it or not. And we can always make new trials if there's consensus to do so.

In addition, an extension of an active implementation could lead to considerable controversy, as shown by the above discussions, and it would probably be adverse to a proper and calm discussion on the trial and merits of PC. At worst it could turn into an intractable sitewide dispute, which we don't want. The wide differences of opinion show that there is a need to reconcile them and find a general consensus on a myriad of points. Building consensus on such an issue is complex and should not be precipitated, continuing an active implementation of PC wouldn't bring much in this endeavor and would on the contrary detract from it, due to it being controversial. So in the circumstances, I think that keeping an active implementation would be hardly tenable and counter-productive, but we should still see if there are more substantial reasons not to do so:

Due to the lack of in-depth discussion of the merits of PC, the opinions given have been mostly based on personal experience, or superficial analysis. Both sides gave diverging viewpoints but no consensus developed, for example on the important issue of whether PC is an effective barrier against vandalism, BLP violations, and other, compared to semi-protection. Many arguments of similar strength were given for and against, but for the above reasons, they were hardly ascertainable (such as effect on new users). However those against extending the implementation have given many more arguments, many listed here, and they indicate that various problems need to be addressed, with some clearly identified and recognized, including disturbance of usual editing, increase in the need for vandal-fighting and inefficiency in cases of sockpuppetry or specific article issues; examples which directly affect the quality of articles and work of editors, so must be given particular weight. Since the default was to discontinue the implementation, the burden was on those in favor of extending the implementation to find counter-arguments to those or provide solutions, which in general didn't happen. Accordingly, this is sufficient to show that there is no compelling consensus to continue an active implementation. This should not be seen as a victory or defeat for anyone, because it is what was supposed to happen, it was a two month trial as agreed; and the final outcome of those discussions do not prejudice future discussions of PC by the community.

I would also point out that the WMF explicitly stated that in a poll, a two third majority was necessary in general for deploying 'FlaggedRevs' [1], being understood that the default was deactivation, this applies here and the poll didn't reach 66%, only roughly 65% 60% (correction counting 'other responses'). Though purely numerical, it reinforces the proposition that an implementation of PC is a substantial modification of our editing practices which requires a strong consensus in the community. The very nature of pending changes, by relying on human review, makes it also necessary that it be adopted by a large majority in order to work efficiently.

Therefore, I think that the discussions do not demonstrate a consensus compelling enough to divert from the default option which was to halt the implementation, the active part in any case. Now comes the question of whether we should keep the implementation in a dormant state as opposed to turning it off completely, I think the former would be more convenient for developers and the community, as testing for example could be available in project space, and tackling the numerous usability issues and other technical problems identified in the trial is important. However it should be disabled as an option in mainspace, so as to avoid unnecessary interface load for admins and prevent further uses of PC, and a sunset clause should exist.

With regard to the proposal for gradual shutdown, per my assessment that no consensus existed for extending the implementation, I do not think that the gradual shutdown could satisfy the consequences of this - if uninvolved admins agree that the mandate for the trial is over, we should remove PC from mainspace without delay, which can be done in a matter of days or even hours with a script (guided by those metrics); and I do not find substantial gain in a gradual shutdown, for the same reasons as above. In addition, the one month limit makes that a new poll could not conclude in time. This applies to the other proposed polls, which appear very unlikely to generate a consensus. Now is well past time for determining consensus, not building it. In conclusion, in my opinion:

1. Pending changes should be removed from any mainspace page and replaced with protection if appropriate, which shouldn't take more than a few days.
2. If feasible, developers should remove the PC interface from the protection interface; but keep the usergroup, logs, and PC in project space for testing.
3. In the coming months, the community can calmly analyze the trial, the merits of PC, and discuss of possible new implementation proposals which would have to acquire consensus for adoption, including any trial of any order of magnitude.
4. If no implementation proposal is accepted after a reasonable period, of say one year, pending changes should be entirely turned off.

Please note that this is not a proposal, but an effective assessment of consensus. In order to move on and as the deadline is getting close, I will seek the opinion of uninvolved admins. Cenarium (talk) 06:41, 8 September 2010 (UTC)

I do not want to see the hundreds of hours I spent on this destroyed. Cenarium (talk) 22:51, 10 September 2010 (UTC)

Cenerium, I don't see any possibility that anything is going to be destroyed. We're going to get improvements to the software that respond to objections, and have another poll. There is no crisis here, there is no reason to panic. Your work is successful.--Jimbo Wales (talk) 00:49, 11 September 2010 (UTC)

There is if we don't suspend use of PC until we have worked out the multiple issues which have been raised (of whom the "don't use PC on heavily vandalized articles" was the most predictable and obvious) and made some in depth analyzis of the trial. There is simply no consensus to continue with PC until we haven't made this work, it shouldn't take more than a few months, and then we'll better know what we want of PC and be able to make an informed decision. Cenarium (talk) 01:05, 11 September 2010 (UTC)

I'm sorry but I don't follow you. I do agree that it shouldn't take more than a few months. And when we get the revised version from the Foundation, we can test that. It's an iterative process. That's all right. But I don't see why on earth we should turn off a popular feature - against the wish of the community - in the meantime. Let's use it in the cases where it works well, and not use it in the cases where it doesn't work well. What's the problem you are trying to solve?--Jimbo Wales (talk) 01:48, 11 September 2010 (UTC)

Popular feature yes, but there ain't no consensus to keep it up and running at the moment, and it's pretty obvious. It's not against the 'wish of the community' to turn it off, people of both sides agree it's the right thing to do now. Read the discussions. There's not just the technical side of things, I've written on this at length, given plenty of arguments, I won't repeat them here. You're giving a dimension to this poll that it just doesn't have. We needed consensus to keep it, not consensus to deactivate it - it's expressly stated in the poll, it was on this promise that we launched the trial. We make the trial, we think about it, then we decide. People didn't want that we make the decisions hastily, and there are too many issues to keep it running now even as an extension of the trial. The thing is, that we should not reverse the burden of consensus while the initial consensus was for a trial limited in time, the burden of consensus building is on extending it, not the reverse, otherwise the trial is nothing more than a backdoor to implement PC in a more definite way without supporting consensus to do so. Cenarium (talk) 03:37, 11 September 2010 (UTC)

(In response to you post on SV's talk page) This is simple. The consensus was for a two-month trial, full stop. If we want to extend it, we need consensus. We don't have consensus, so we stop using PC. With regard to the devs' concerns of the difficulty in turning the feature off then back on, we have reached a compromise: keep the implementation dormant. Cenarium (talk) 04:01, 11 September 2010 (UTC)

Given that nearly 2/3 of the community disagrees with you, what compromise are you willing to accept and support? What problems is your proposal designed to solve? Are there particular pages where you think it should be turned off? Or is it your view that a minority of the community can - and should - block progress forever?

We know the feature is popular. We know that it is working extremely well in the vast majority of cases. Going backwards is not an option. You can try to redefine the terms so that change requires near-unanimity, but that's not the deal and never was the deal.

I listen to the community... all the community, and we have strong support for this. We also have some legitimate concerns, and so we will address those concerns and have another poll. Iteratively. In the meantime, turning the feature off is impractical and would be outrageous, given the demands of the community.--Jimbo Wales (talk) 04:21, 11 September 2010 (UTC)

We already have a compromise: keep the implementation dormant (we keep the extension enabled, just don't use it for now). My trial proposal was for two months, not indefinite, PC should be removed from all articles until further consensus because there is no consensus to extend it. Consensus is not near-unanimity but clearly this is not consensus. We can reach consensus on a permanent implementation if we do it properly. It's not going backwards, it's being faithful to our decision-making process. Out of principles, I cannot endorse that.

Do you recognize the problem in the unexpected reversal of the burden of consensus ? In your proposal, does the next poll require consensus to keep PC on or the opposite ? Cenarium (talk) 05:09, 11 September 2010 (UTC)

There is no reversal of burden. In future polls, if there is consensus to keep it permanently, it will be kept permanently. If there is consensus to remove it permanently, it will be removed permanently. If there is, as in this case, an overwhelming majority to keep it, but not quite consensus, then it will be kept temporarily and improved.

There has never been any tradition in any of our decision-making processes that in the middle of an iterative process of improvement, we should completely delete or remove something that has near-consensus and start all over.

What problem are you trying to solve with temporarily removing it from all the articles where it is working well and very popular, only to re-add it when the next version comes out? That seems entirely pointless and a grave violation of the will of the community, which is 65% in favor of keeping the feature.--Jimbo Wales (talk) 12:41, 11 September 2010 (UTC)

Kept temporarily until what ? Under such terms there may never exist a consensus supporting the permanent implementation of PC, but it would be kept indefinitely regardless. In fine, the trial would have been used as a backdoor for an implementation of PC indefinite in time, conveniently bypassing the need to establish a consensus for a permanent use. This was a great concern among users at the time of the proposal, and I assured that this would not happen.

If there is no consensus it's not for nothing. There are numerous, substantial issues which need to be addressed. It's not working well in many cases, that's the point. The efficiency of PC to prevent malicious editing isn't as good as we could hope for various reasons, it can have a detrimental effect on editing, etc, see the poll comments. They are not deal-breakers but need to be addressed, and until addressed they harm the encyclopedia.

It makes total sense to pause the implementation, we have a trial, it's done, then we analyze it, we address the multiple issues which come up, before going ahead with something more permanent. If we keep it up regardless, there will be little effort from 'supporters' to analyze it and help in addressing issues because they'll think they have 'won' so it's done, and no more from 'opposers' because they'll feel disenfranchised. If there had been consensus here and consensus would be needed for the next poll, it would be totally different because people would be concerned and willing to compromise. Cenarium (talk) 14:09, 11 September 2010 (UTC)

60% of the hundreds who voted is pretty substantial. Before serving a minority, I would recommend another poll. but with a more substantial effort to get more !votes. Ronk01 talk 22:56, 10 September 2010 (UTC)

By present policy, Wikipedia is not run by majority vote. Those who want to change that policy had better get much, much more than 60% support. Septentrionalis PMAnderson 23:11, 10 September 2010 (UTC)

This (Mr. Wale's decision) is good. There has been floundering around the 'what is consensus?' (51%? 60%? 80%? 90%) question and this ends that. Obviously with a split vote not everyone could be happy. Now the WP community can focus on how to improve the shortfalls of PC. Gerardw (talk) 02:38, 11 September 2010 (UTC)

No, it's actually not good. A lot of people agreed to a limited trial to see if this thing works, including me. A substantial number of people now (35% by Jimbo's numbers, higher by some others) have now stated "Alright, it didn't work satisfactorily." 65% would not be enough in favor to turn something as major as this on, and that's the standard we should be applying here. 65% isn't even enough for someone to be promoted to admin, which is a far less major change than this. 65% in favor to delete on a highly-commented AfD would be teetering right between delete and no consensus, and again, deleting a single article is orders of magnitude less of a change than making a structural change to the entire site.

What really needs to be done here (since even the guy doing this says he isn't happy with 65%!), is to turn the thing off, make the improvements, and then seek consensus after, not before, the improvements have been made to trial it with those improvements. At the end of that trial (if there is consensus to allow it), see if we get consensus to turn it on—that's the standard after such a trial, not "no consensus to turn it off". We agreed to try this thing, not to accept it. It didn't work, and a substantial portion of the community has said it didn't work. If it needs fixed, turn it off until it is fixed, and then we'll try it again. Seraphimblade ^{Talk to me} 02:56, 11 September 2010 (UTC)

• Comment. This is unfortunate. There was a trial, after which it was meant to be switched off. Then there was a straw poll, in which it failed to gain consensus. Now it's still not being switched off. SlimVirgin ^{talk|contribs} 03:17, 11 September 2010 (UTC)

Objections removed

As a result of an offwiki discussion, the contents of which I will not reveal here without consent, between myself and Jimbo Wales, I have decided that my interpretation of the issues surrounding PC and its usage here are widely off the mark, due to misinterpretation and misunderstanding of quite a lot of the issues. As such, I've decided to withdraw all of my objections to PC, its implementation and usage here, and hope that what I have said in the debates up to now will be taken as read that it no longer applies (other than the UI issues, but these can be resolved). BarkingFish 23:47, 10 September 2010 (UTC)

Sounds great~can someone alert the media? 69.180.160.77 (talk) 00:58, 11 September 2010 (UTC)

Sarcasm is the lowest form of wit :) BarkingFish 01:16, 11 September 2010 (UTC)

While a media alert is probably unnecessary, striking through the comments which you are withdrawing would be useful. Gerardw (talk) 02:34, 11 September 2010 (UTC)

I don't really know how I could do that without breaking conversations, since Jimbo has replied to me more than a couple of times. I'd rather gain approval from him to remove the entire exchanges, rather than have him talking to himself :) BarkingFish 03:13, 11 September 2010 (UTC)

You can just use the <s></s> to strikethrough your comments you no longer mean. Casliber (talk · contribs) 04:22, 11 September 2010 (UTC)

BarkingFish: Given your significant previous opposition, it seems that such an epiphany might be useful for the rest of us. Any chance you could get Jimbo's permission to disclose your discussion? —UncleDouggie (talk) 07:58, 11 September 2010 (UTC)

PC and other issues

I'd say #1 concern about pending changes, all the tools that we are so used to using to police live changes, are useless for pending changes, so we are going to see backlogs. We need a huggle-esque tool for PC for the human element of PC to be able to scale, and we need it before we have large numbers of pages under PC. Honestly, I think the foundation needs to become a part of the tools business in general, so that better software for "advanced" users (rc patrollers, newpage patrollers, admins, etc) is developed alongside MediaWiki itself, and so that those tools acccurately reflect the vision of the community and foundation. We've had a number of people working to develop tools over the years, and some very good ones, Huggle, VandalProof, Igloo, AutoWikiBrowser, and others over the years, but they've all been developed in a vacuum, to scratch an editor's itch, by editors that also happened to be competent developers. As such, they perform well, but not as well as they could if the development was structured towards things the community needs, rather than whatever that tool author needed at the moment. This needs to change for long term success. Part of keeping an open community and open editing environment successful is making sure that the people working to keep it open are better equipped (and thus, more efficient) than those who would abuse that openness. With vandalism in particular, we want the difference in effort between some random vandal working to undermine our work and the RC patrollers working to stop them to be as great as possible - where it takes them anywhere from 15 seconds to 5 minutes to destroy, it should take a second or two at most to protect, keeping a "balance of effort" in favor of those who create, improve, and protect, versus those who destroy and corrupt, all the while keeping us open to anyone who wants to contribute constructively. Triona (talk) 06:41, 11 September 2010 (UTC)

I agree with everything you have said. Check out EzPR on English Wikinews as an example.--Jimbo Wales (talk) 12:32, 11 September 2010 (UTC)

• Yes, we need to have Pending Changes easier to use here. I'm a reviewer on Wikinews, and Easy Peer Review is great. I also think the Pending Changes should be activated on all the Wikipedia pages, just like German Wikipedia. Cheers, --Diego Grez (talk) 20:21, 11 September 2010 (UTC)

How's de's edit rate doing these days?©Geni 20:23, 11 September 2010 (UTC)

Well, the bad part is that no one is gonna ever review 3.000.000 articles. PC should be only on Good and Featured articles, that are supposed to be reviewed in a way or another. Diego Grez (talk) 21:08, 11 September 2010 (UTC)

Short and sweet

The consensus was for a two-month trial of PC. After two months, the trial was not stopped, and now it is being indefinitely continued without consensus. Since the last promise of it being a temporary trial was completely ignored by supporters, what reason is there to believe that this promise of it being a temporary trial will be honored? "I'll only put it in a little bit" is not a strategy that should be effective on Wikipedia, and I think it is shameful that it was deployed here.—Kww(talk) 14:41, 11 September 2010 (UTC)

I don't agree with your interpretation of what we were doing all along, nor what we are doing next. I don't know where we ended up with this strange idea floating around that nothing can change without getting "consensus" (which is undefined, but apparently must be a lot higher than 2/3 support) for anything new.

My view is that we should drive for consensus, that we will get consensus, and that in the meantime we should pursue an iterative process of improvement, as opposed to allowing a minority to block progress.

It is also worth contemplating the level of screaming that I would *quite justifiably* face, were I to say "65% are in favor of this, so I'm asking the Foundation to turn it off".--Jimbo Wales (talk) 15:16, 11 September 2010 (UTC)

What on earth did "two month" mean in "two month trial" if it didn't mean "we'll stop in two months and evaluate the results?"—Kww(talk) 15:53, 11 September 2010 (UTC)

It meant that we would vote after two months to figure out what to do. So that's what we did. And the results are in: the extension is popular, with 65% support, but has some problems which led to 35% opposition. The next step? Revised software from the Foundation, a new trial, and a new vote. Iterating until we get strong enough support to call it stable and keep it permanently.--Jimbo Wales (talk) 16:31, 11 September 2010 (UTC)

That's not a two-month trial, and I strongly suspect that you know that. That's an indefinite-length trial with a review at the two-month point. That's a substantially different thing, and I don't think you ever would have gotten consensus for an indefinite-length trial. The terms are being changed after the fact to suit the supporters, and that's immoral.—Kww(talk) 17:54, 11 September 2010 (UTC)

No one's asking to turn the extension off, we just remove it from articles until we have properly reviewed the effects of PC and address to a satisfiable extent the serious problems that came up. That's a compromise supported by people of both 'camp'. Cenarium (talk) 16:08, 11 September 2010 (UTC)

It is also a compromise opposed by people of both 'camp'. Convince me that it's a good idea? Why should we stop using it in the cases where it is working perfectly well? It's popular and it works! Removing it from articles where it is doing a good job makes no sense to me. What problem are you trying to solve?--Jimbo Wales (talk) 16:31, 11 September 2010 (UTC)

Consensus isn't the same as a vote, but generally, the closer of a discussion does evaluate the numbers along with the arguments. Firstly, there was no consensus to a particular keep position, and I know at least some people said they would rather remove it entirely than see it expanded past a certain point. There were nuances in the !votes that don't seem necessarily to have been taken into account. Not everyone supported an expansion, per se.

Regardless, however, I had very much the same idea that Kww did—that a "trial" meant "We'll test it for a couple months, shut it off, and then see if there's strong consensus to enable it permanently." Honestly, if you already consider this "progress", and those who oppose to be "blocking progress", you've taken a side and probably shouldn't be closing the debates—someone who doesn't care much either way should be. Those who opposed considering it are not trying to "block progress", they disagree that it is progress. I think that position has some merit. As you said yourself—it has bugs, even at a small sample size it's created backlogs, it's not fulfilled one of its major promises (namely, to let high-profile articles be reduced to this rather than indefinite semiprotection), and it's opposed by a sizable portion of the community. If I'd known the concept was going to be "We're going to try it and then put continuing it to a majority vote (and count three of the four positions in the debate as one lump toward what constitutes a majority)", I never would've supported the trial. It honestly sounds like you're trying to ramrod this, and whether or not that's true, if it's your intent that it's going in no matter what anyone says, just say that. If it's going to be up to the community, that's fine—but a massive structural change like this should require a higher, not lower, degree of consensus than what it takes to make a single person an admin (or even bureaucrat, for that matter—and that's even higher). It's not like the devs' work will go to waste if we don't use it. Many other Wikimedia wikis, with a different scale than we have, use it and are happy with it. But at this point, the community's response basically amounts to "Maybe if some things are fixed, but at this point no consensus to enable". Seraphimblade ^{Talk to me} 17:02, 11 September 2010 (UTC)

I don't see any way to support that interpretation; that's a highly POV rendering of the facts. A position is supported by 65% of the community - I won't defy them and shut this thing off. I think the thing to get busy on now is a more precise specification of what we expect from the Foundation in terms of version 2, what deadline we will have for them to get that out to us, and what precise parameters we will use for the next poll. In my view, following the desires of the *entire* community (rather than the vocal minority trying to block this) will involve acknowledging that there can be consensus to keep, consensus to get rid of it, and a range of options in the middle which tell us "keep working on it". What I recommend at this point is prioritizing for the Foundation just what they need to fix to meet the most important and fixable objections.--Jimbo Wales (talk) 18:27, 11 September 2010 (UTC)

What about those whose "keep" !votes were conditional or for one position only, huh? If it doesn't go according to their wishes, are you not defying them? —Jeremy ^{(v^_^v PC/SP is a show-trial!)} 20:26, 11 September 2010 (UTC)

Removing pending

User:Dabomb87 is removing pending protection from what looks like a random position, without consideration to whether or not it was beneficial at the article or not, just with the summary, trial ended.. Is this necessary or beneficial, he appears to be reverting back to the level of protection previous to pending being applied. Off2riorob (talk) 18:35, 11 September 2010 (UTC)

I've stopped for now, since it seems to be more controversial than it was originally (and there seemed to be reasonable support for it here). As for reverting to to the previous level of protection, I'm just returning to the status quo, and am not necessarily endorsing it. Dabomb87 (talk) 18:41, 11 September 2010 (UTC)

Thank you for stopping. What I propose we do now is have a discussion of where Pending changes is useful, even in the current configuration, and where it is not as useful, and make some gentle adjustments based on that. I do not want to see sudden fast moves in any direction, as I think that would not be helpful for general peace and harmony. There is no reason to rule out removing it from some articles, but there is also no reason to rule out adding it to some articles that don't have it. (When should we add it? When it improves the encyclopedia. When should we remove it? When it improves the encyclopedia.)--Jimbo Wales (talk) 19:56, 11 September 2010 (UTC)

Yes, there is no reason why it can't continue to be requested when in some situations it is clearly the better method of protection and allows more unconfirmed users to contribute than semi protection there are clearly occasions where I have been wanting to add it and have seen articles semi protected for a year when against the attacking additions of a single user pending protection would have been plenty. I also find it a poor idea to be removing pending when it was working fine and reverting back to indefinite semi protection, that is clearly a restrictive retrograde step. Off2riorob (talk) 20:25, 11 September 2010 (UTC)