The following discussion is closed. Please do not modify it. Subsequent comments should be made in a new section.

The scheduled two-month trial has ended. The community should now decide if the implementation is to be continued, and it should discuss possible adaptations, in terms of policy. Developers have indicated it would be too complex to turn off the feature, then turn it back on in case the decision is in favor of continuing the implementation, so they will wait for the community decision — unless it takes more than a month, in which case they will turn off the feature. In the meantime, it is possible to ask administrators to stop adding new pages under pending changes.

The trial was approved in this poll. A community consensus is required to continue the implementation. A discussion phase of roughly two weeks analysed the trial, and a decision phase with a classic support/oppose continuation poll of two weeks began on 22 August 2010 (00:00 UTC). It will end on 5 September 2010 (00:00 UTC). The organization of the closure is still under discussion.

For information, you may consult the feedback given during the trial. As of 20 August 2010, there were 1,409 articles under pending changes, and the average lag for pages with unreviewed edits pending was 4 minutes, 30 seconds — see Special:ValidationStatistics for updated statistics, and Wikipedia:Pending changes/Metrics for metrics on usage of pending changes. The Preliminary Analysis page has some cuts of data that may be useful for discussion. There are also Village Pump proposals relating to the discussion (FA Pending Revisions proposal).

Working Summary (unofficial)[edit]

This section should briefly list issues. Transfer ideas to this informal summary. Neutral phrasing appreciated. May get edited for clarity. Please keep discussion in other sections. It is not necessary for everyone to agree that every point is valid, just to get a consensus on which issues have been identified.

Pros (what worked)[edit]

Cons (what didn't work)[edit]

User interface / Usability

Vandalism / Workload

Community

Feature Requests (what might make it better)[edit]

User interface / Usability

Policy

Expansion or contraction (beyond policy and pure numbers of articles)

Open questions (what was unclear)[edit]

Evaluation

Implementation

  • Check for obvious deficiencies?
  • Check for vandalism only?
  • Enforce reliable sources?
    • All of the time?
    • When a change looks questionable?
  • Disregard for reviewer guidelines?
  • Disregard for WP:COI
  • Editors with a recent history of vandalism, blocks, or other sanctions that would indicate poor judgment?
  • Editors who have previously (recently?) had reviewer status taken away for poor judgment?
  • Should be relatively easy to get, but not auto-confirmed - a defense against "sleeper" accounts is required to prevent vandals from getting the reviewer permission, especially if we let up on RC patrol on "reviewed" articles.

Expansion

Notes[edit]

  1. ^ see point 1: Pending Changes feedback discussion, also
  2. ^ For example, see comments by 84user (in response to Tryptofish) below
  3. ^ see here
  4. ^ see here
  5. ^ see here
  6. ^ see here, here, here, and here
  7. ^ see here, here, and here
  8. ^ see here
  9. ^ see here and here
  10. ^ See Wikipedia:Pending changes/Feedback#Unapprove button and Wikipedia:Pending changes/Feedback#Accept/Unaccept
  11. ^ See Wikipedia:Pending changes/Feedback#Unapprove button and mediawiki specifications
  12. ^ See Wikipedia:Pending changes/Feedback#Usability: Rename Accept/Unaccept actions to state names and Wikipedia:Pending changes/Feedback#Succinct yet descriptive list of issues and suggestions point 7 and Wikipedia talk:Pending changes/Closure#PC Terminology
  13. ^ See Wikipedia:Pending changes/Straw poll#Keep: options 2, 3, or 4 !vote 406 by User:Hullaballoo Wolfowitz
  14. ^ see comments by Scott Mac below
  15. ^ See comments by Fayenatic below
  16. ^ See also Wikipedia:Pending changes/Feedback#Level 3 PC
  17. ^ See Wikipedia:Pending changes/Straw poll#Close: option 1 vote 171 by UncleDouggie and User talk:UncleDouggie#Re: Edit Filter
  18. ^ see comments by Cptnono
  19. ^ see comments by Jeremy

Preliminary discussion[edit]

Note that pending changes is still active. It will be shut off in 1 month unless there is consensus to continue

  • Note my comments about rapid attacks and sockpuppets above. PendingChanges is useless against either. —Jeremy (v^_^v Carl Johnson) 20:18, 21 August 2010 (UTC)[reply]
  • It only works well because nobody's sussed out a way to bypass it. As I stated below, Murphy's Law in regards to our antivandalism measures has never been proven fallacious. —Jeremy (v^_^v Carl Johnson) 20:18, 21 August 2010 (UTC)[reply]
  • As a form of protection it's useless. PC just clogs the history with drek and will only increase the workload of admins, nonadmins, and Oversighters alike. The downsides are far too serious and outweigh the upsides. —Jeremy (v^_^v Carl Johnson) 20:18, 21 August 2010 (UTC)[reply]
The not so good: editors bitching at other editors accepting not vandalism but maybe not correct edits. In other words, for following the WP:RVW policy as written. Further work needs to be done in establishing a consensus for what reviewing means. I see that as a separate discussion that whether to accept pending revisions. I assume that discussion would take place at Wikipedia_talk:Reviewing pending changes. Gerardw (talk) 15:35, 14 August 2010 (UTC)[reply]
I know this is aimed at comments another editor and I made on your talk page regarding infactual edits accepted. It was an attempt to be polite and I felt you replied to us very standoffishly by pointing at a policy with no further comments yourself. My complaint about the policy is something I'll address if pending changes DOES stay, and my comments were not meant to be "bitching" as you say, just constructive criticism. For my part, I apologize if you thought otherwise. I agree that "further work needs to be done in establishing a consensus for what reviewing means". CycloneGU (talk) 16:51, 14 August 2010 (UTC)[reply]

Keep in mind that not only is it an anti-vandalism tool, it can be used as an alternative to page protection in some cases (see my reasoning above). Tyrol5 [Talk] 16:13, 14 August 2010 (UTC)[reply]
From the ones that I've seen, I'm having a hard time thinking of page protections situations where this would be applicable / useful
One other note, right now you are polling the converted. Persons interested in this, and used to dealing with the complexities to the point where you can't understand that they are complexities. If you would liked a sampling of the rest, I'd cast the RFC net wider, and provide a simple 1 paragraph summary of what this change would mean. North8000 (talk) 16:23, 14 August 2010 (UTC)[reply]
As an alternative to page protection, I think it would be useful mainly in the area of constant WP:BLP violations, rather than consistent vandalism (where an RFPP may be necessitated). Tyrol5 [Talk] 16:33, 14 August 2010 (UTC)[reply]
I think that aa lot of the problem folks on those are auto-confirmed accounts, and and nothing "higher". E.G. that's what I am. It's unclear whether or not their edits would need to go through the gauntlet. North8000 (talk) 18:31, 14 August 2010 (UTC)[reply]
And that's certainly a valid concern, which is why different levels of PC protection were proposed originally (see this). Tyrol5 [Talk] 18:49, 14 August 2010 (UTC)[reply]
Just to respond to that point about vandalism getting quickly reverted. Yes the vast majority is, but we have no way of knowing when recent changes is unattended or swamped. Today I found an article that was vandalised 18 months ago, pending changes is a better system and will make that sort of thing rarer. ϢereSpielChequers 19:40, 14 August 2010 (UTC)[reply]
It's definitely a mistake to think that vandalism can always be reverted quickly. Among the articles I've applied pending changes to are: the president of a major international organisation called a penis for nearly a day, a school article containing outrageous libel for five months, another school prominently described simply as "full of slags" for ten days, the leader of the opposition in Israel with a vagina for a photo for six hours, and the White House Chief of Staff labelled dead for a while (a similar edit to the President's top advisor lasted 3 hours). It's stuff like that that slips through the net and does nobody any good. With pending changes we don't have to lock them up to all new users. -- zzuuzz (talk) 18:49, 16 August 2010 (UTC)[reply]
I see what you are saying (I couldn't agree more), but keep in mind that this was only a preliminary trial. Tyrol5 [Talk] 04:06, 15 August 2010 (UTC)[reply]

  • As an aside, I also vehemently oppose, should this end up being implemented, compulsory reviewer rights based off of account age or edits. Not only does this guarantee that someone who is unwilling or unworthy of using the userright is given it, but it will only make some of the issues brought up (such as accept/unaccept ambiguous edits that aren't blatant vandalism) worse. —Jeremy (v^_^v Carl Johnson) 21:25, 16 August 2010 (UTC)[reply]

Procedural issues
  • The 'poll' forming here on the Closure page is not a representative sample of the community; most people here will be reviewers. Acceptance needs broader discussion in the community; I don't know if RfC is planned, or what.
  • The above 'in favour' do not make it clear what they are in favour of; some people object to 'across the board', or specifying ideas of limits, but there are no specific implementation plans that have been put forward. As PC can mean over 9000 different things, we need ultra-clear proposals that we can support or oppose.
Observations
  • I'm concerned that it is only too tempting to see this as a quick solution to too many page problems; if a page is getting a bit of vandalism, it's tempting to just apply this, but we don't seem to know what impact that has on productive editors; we know that it has at least some impact, if only for the numbers of Wikipedians expressing their extreme disdain for the whole idea.
  • A somewhat oxymoronic fact I found was, that it can work better when PCs are not reviewed so quickly. When there were many users vying to 'approve' a change as soon as it happened, there was a tendency for the editors to just briefly check if it was vandalism, and approve or reject it. That's fine, in as far as it goes - however, when things slowed down a bit, the reviewers tended to check the edits more thoroughly - for example, validating references or improving the format of the edit.
  • I agree with CycloneGU above, that we need to clarify the expectation of reviewers - whether they just check for pure blatant vandalism, or if they should check further
  • The terminology and interface is extremely confusing, 'unaccept' and the lack of a 'yes/no' type interface; this has been discussed elsewhere, so I won't elaborate here
  • Speed concerns esp. on large articles
  • If it is only used for vandalism, then not only is there the danger of things slipping through, but also people perceive that, because an edit has been 'accepted', it must be OK, therefore may not check it a second time
Conclusions
  • I feel that the way forward is with improvements to the system to address concerns raised, followed by a further trial. I find it hard to judge if I would support or oppose other suggestions, without very specific proposals. Chzz  ►  05:21, 16 August 2010 (UTC)[reply]
Note that the pending changes is still active. CycloneGU (talk) 05:26, 16 August 2010 (UTC)[reply]
Also, this was listed at WP:Central discussion, so it should attract a wider audience. Ocaasi (talk) 15:43, 16 August 2010 (UTC)[reply]
  • While it may look like that is the case, the reality is far grimmer. FraggedRevs cannot handle mass attacks on an article (such as from 4chan threads) as the amount of edits made outstrips the capacity of reviewers to address them in a timely manner, and it is far too easily circumvented by autocon-buster sockpuppets who can outright bypass FraggedRevs altogether. In short, while it does reduce IP vandalism, it only does so if they aren't editing the article en masse (as those who attempted to use PC on, well, 4chan found out to their chagrin). —Jeremy (v^_^v Carl Johnson) 20:05, 21 August 2010 (UTC)[reply]
  • Comment While I don't support the PC feature mainly because of the 'cracks' in its system(users falsely accepting changes, technical issues, etc.), an amount of vandalism does slip through recent changes patrol. Pending changes might be able to keep that in check in a certain measure, but the outstanding issues has got to be resolved first. Bejinhan talks 05:37, 19 August 2010 (UTC)[reply]

Arbitrary Break[edit]

Reviews are quite visible, e.g. [[1]] Gerardw (talk) 17:05, 16 August 2010 (UTC)[reply]
Not as visible as edit history. You have to cross-ref the edit history and the review log to get the entire picture. OhanaUnitedTalk page 04:26, 17 August 2010 (UTC)[reply]

  1. First and foremost, this trial hasn't captured and can't capture certain aspects of a real deployment. In particular, without the 2000 page limit, we are certain to see backlogs and blind acceptance (or blind reverting) of edits without checking the immediate prior history. Another point is that we haven't particularly tested what happens when vandals know how PC works. Remember, when vandalism is accepted, it becomes harder to get it off of the article, since not everyone can revert it anymore.
  2. The "unaccept" option and the option to leave a note when accepting are both fairly useless. If you want to revoke acceptance, then in practice you surely want to revert, and if you revert you ought to leave an edit summary. If you accept an edit, the reason had better be plain: "no problems here". Both of these should be removed.
  3. My major concern has always been that PC will be used to own articles or lock particular editors or opinions out of them. I was pleased to find that I didn't encounter this on any article I was watching. However, I have seen a great number of naive proposals to use Pending Changes as a quality control, or to revoke reviewer permission for reasons unrelated to reviewing or vandalism. Any such practices could become soft censorship over time, so any serious proposal to continue pending changes needs to stomp hard on these urges.
  4. I haven't seen any case being made for the usefulness of Pending Changes level 2, and it runs a much greater risk of enabling article ownership. Unless there is a very good reason out there that hasn't been articulated, I think Level 2 ought not to be approved.
  5. Of course, I also oppose deploying Level 1 protection, though that doesn't stop me from having the opinions above. Simply put, there is a case to be made for using Pending Changes on unwatched BLPs, though nobody is seriously pursuing that option. In all other cases, it offers only illusory advantages over semiprotection. Semiprotection prevents vandalism and constructive edits equally, and administrators understand this and avoid using it where it is not needed. Pending Changes does not prevent vandalism; it only makes it somewhat unlikely to be seen by the general readership. In exchange for that, it makes reverting (accepted) vandalism a much more arduous and failure-prone exercise for the very same general readership.
More disjointed ramblings later if I happen to think of them. — Gavia immer (talk) 05:47, 17 August 2010 (UTC)[reply]
  • To some extent the trial was too short, since there are clearly still a number of issues that need to be resolved like the speed, UI etc; and editors and admins alike are still just getting used to the system like how to review (when to accept etc) and when to request or grant/use flagged protection. In particularly we need a way to deal with seemingly good faith but poor edits, perhaps some sort of reject with comments option.
  • Of course it's also something that vandals are getting used to, and we don't really know how they will respond (for example it's clearly possible one of the reasons we still get a lot of vandalism is because vandals are unfamiliar with PC so see they can edit an article and vandalise not really understanding few are going to see it if it isn't approved and while there's always likely to be a level of this, it may drop as people get more familiar with PC).
  • I do think it's clear that it isn't always a good alternative to semi-protection and I think Jimbo Wales' comment a while back that we could open the main page up to editing is never going to happen.
  • So all in all I wouldn't mind if the trial is continued, perhaps in an expand form. In particular, I would like to see more wide use of flagged protection in BLPs if not used on all unwatched BLPs then with a very low threshold for acceptance for BLPs (e.g. only one problematic edit which lasts for say at least 1 hour is enough to use flagged protection rather then requiring ongoing problems). As I've already said, I'm not suggesting semi-protection should be abandoned and this applies to BLPs as well. Some may considered widespread use of semi-protection on BLPs better then PC, but from my experience with discussions on this, it seems unlikely that will happen so if that's more likely with PC, why not?
  • Note that I'm also not opposed to the implemention of PC on a non trial basis since IMHO we've seen enough to know it is a useful tool that isn't going to destroy wikipedia.
  • Also if PC is going to be turned off in a month, either we have to make a decision before then or it should be continued until a decision is reached. It IMHO would be a silly waste of effort to turn off PC only to turn it on a few weeks later although at the same time it would be inappropriate for PC to continue just because supporters continue to hold up any failure to achieve consensus to either continue the trial or accept PC.
  • (BTW, for all those comments on the German wikipedia and citizendium, the vast majority of users and contributors to the English wikipedia are almost definitely not familiar with either and therefore have no opinions on us following them. In fact I only actually see one comment on the German wikipedia and one on citizendium here anyway.)
  • P.S. I should mention despite some initial interest I actually barely used PC myself including reviewing etc so I don't think I should be considered the 'converted'
  • P.P.S. While the timing of the trial may not have been the best, it's IMHO offensive to suggest the developers somehow timed this specifically so it was the end of the school year in the US or whatever. Anyone following the protracted development that went into it would know clearly that was just an accident. Also in terms of why the trial was so short, AFAIK this was primarily because of detractors who were afraid that wikipedia wasn't going to survive the trial so a short trial was chosen to try an allay their concerns.
Nil Einne (talk) 18:57, 17 August 2010 (UTC)[reply]
A small aside on the subject of German Wikipedia:
I am not sure what the intended criticism of German WP is in this case.
German WP does insist that a comment be provided with every edit, but so do a few other projects, like Polish and Georgian. Regardless, a "comment" can be as simple as a single blank.
Varlaam (talk) 02:20, 19 August 2010 (UTC)[reply]
I'm not sure whether you were addressing me or speaking general. But for clarification, I have no views or for that matter little knowledge of how the German wikipedia works other then an understanding they have had something similar to pending changes for a while now and therefore are often an example, either of criticism or support for such a system. I was partially address this comment "It also drives off many would-be contributers who think we're going the way of the german wikipedia" and another one on citizendium. My point is that whatever you think of the German system or citizendium, I'm quite sure few people here on en.w really care that much about what they do and most people particularly non German speaking non regular contributors are almost definitely don't think we're trying to follow them or whatever. They may dislike what we're doing but they're not going to think, 'oh great, they're going the way of the German wikipedia/citizendium/whatever'. SO while they may be useful for comparison purposes, I'm quite sceptical of any claims that many people actually think we're following them. Nil Einne (talk) 14:33, 19 August 2010 (UTC)[reply]
  • There were also a lot of reviewers making, frankly, stupid decisions on accepting vandalism. If this is continued, we should make the reviewer right more strict and less of a giveaway. Solid guidelines on what is OK to accept would also help. Still, I'm against it, as I noted way above. —fetch·comms 22:08, 17 August 2010 (UTC)[reply]

Ryan Norton 09:21, 18 August 2010 (UTC)[reply]
Could you (or others) provide some diffs of outright vandalism that was accepted? Ocaasi (talk) 04:10, 18 August 2010 (UTC)[reply]
IP-users often correct spelling in the more esoteric pages (e.g.) and are not all vandals or idiots (which is common). Has anyone ever done a thorough study though on it? --Squidonius (talk) 18:58, 18 August 2010 (UTC)[reply]
Then your field of play is very narrow. I've seen countless registered vandals. Also, IP editing is a Foundation issue, not one en.wiki can take up. —Jeremy (v^_^v Carl Johnson) 19:02, 18 August 2010 (UTC)[reply]
I'm aware this IP-only issue is very much one of the Basic Commandments of Wikipedia, and indeed my 'niche' is just the small specialisation on mountaineering and the higher ranges, and central Asia and Tibetan subjects. Having said that, I wanted to coin my 2 cents in this discussion because from what I experienced it's more of a hassle and a burden than a solution to an intrinsic problem we're facing. Qwrk (talk) 21:08, 18 August 2010 (UTC)[reply]
  1. If you use it to cover large numbers of articles (or small number of heavily edited ones) you'll lots of edits to review and the quality of the review will decrease. That's fine if you want to as a tool to decrease the visibility of simple vandalism. But it is worth the effort for that? IS simple vandalism such a big problem? Isn't it reverted pretty quickly anyway? Conclusion: small gain, lots of effort, perhaps many new editors discouraged.
  2. The other possible use is to target a small number of articles (or a large number that get fewer edits) and ask people to do quality reviews. Target it on the articles where bad edits might not get spotted, and there visibility for a short time is a real problem. That is use it ONLY for underwatched BLPs. Less edits to review, more care gets taken, less new editors are affected, and we help what we know to be a real problem.
My point is that you can't simultaneously use this for 1 and 2. They are contradictory strategies. I'd strongly suggest #2 is more worthwhile. I proposed something similar before at Wikipedia:Targeted Flagging - something like that might be worth people looking at again now.--Scott Mac 18:48, 18 August 2010 (UTC)[reply]
Or on popular pages you just let the people with the page on watch to review the edits - they will still get reviewed pretty quickly. -- Eraserhead1 <talk> 18:59, 18 August 2010 (UTC)[reply]
No, you can do one or the other. If it is used on popular pages, then you'll have lots of reviewers, and people won't distinguish between "popular" and "sensitive" pages. You can either use this as a mass took, for screening out obvious stuff, or you can use it for protecting sensitive (low traffic) pages where people are told NEVER to approve an edit without actually reading the article, and checking the sourcing. Try to have it both ways, you fail.--Scott Mac 19:20, 18 August 2010 (UTC)[reply]
As someone who regularly edits articles within various scientific and international areas of interest, I find the premise of purposely singling out low traffic articles objectionable. The idea that someone out there might be discouraged from anonymously contributing specialized knowledge to articles like MON 863, Cry3Bb1, Hailong Market, Haidian Christian Church, Solar cycle 24, Hexatriacontanoic acid, Aluminium acetotartrate, Camp Rustamiyah, Vimbayi Kajese, mu Opioid receptor, Yangonin, Flavokavain B, Dihydromethysticin, Salvadora persica, and FKBP1B disturbs me. Moreover, if reviewers are encouraged to evaluate content on any basis of quality, then we can say goodbye to many of our anonymous contributors that speak English as a second language, and consequently, many of our international articles that require the knowledge of a local. Similarly, if reviewers are encouraged to evaluate content on the basis of whether or not it is factual, then we are going to see gross POV-based decisions become commonplace. The fewer people that view an article, the easier it would be for a reviewer to veto or extinguish an unpopular viewpoint.   — C M B J   07:54, 20 August 2010 (UTC)[reply]
Personally, I like the second idea. The first is most certainly not worth it; the seocnd is the better, although I still don't think that pending changes should be kept at all... ~~ Hi878 (Come shout at me!) 20:05, 18 August 2010 (UTC)[reply]
  • The problem is not the workload, but the southward effect that this would have on all IP contributions, beneficial and baneful. It's safe to say that Wikipedia's success has mainly been tied to instant gratification (i.e. edit an article, see your edit immediately). It is also for this reason that projects embracing the opposite have stagnated (i.e. Citizendium). If a beneficial IP finds that every edit he makes has to go through a bozo-filter before it shows up in the article, the likelihood of them editing is going to be decreased. Also, as has been pointed out FraggedRevisions can be a severe hindrance if the wrong people staff it - those who have preconceived notions, those who view it as "quality control", and those forced to take the tools and use them. Lastly, unlike most of the other antivandalism measures the Foundation has used FraggedRevs is the least effective due primarily to the press and the hype. A vandal will either attack nonrev'd articles or, worse, attain the userright and cause mayhem. —Jeremy (v^_^v Carl Johnson) 20:37, 19 August 2010 (UTC)[reply]
  • Correct me if I am mistaken, but don't editors editing PC articles get to see their changes on their screen immediately? Ideally, the review is completed within a few minutes, and then everyone else sees it, too. This of course assumes beneficial edits, not vandalism. CycloneGU (talk) 13:28, 20 August 2010 (UTC)[reply]
  • Instant gratification doesn't work if it's based on a fundamental lie. Just because it shows up for them does not mean that it will stay in the article, and for IPs making obviously-beneficial edits with a few stylistic errors in them, this can be a major issue. When - not if - an IP's edit gets rejected because he (in the process of expanding an article) misspelled "benign" or a similar highschool or college word, then the façade of instant gratification vanishes and the IP gets jaded by the experience. —Jeremy (v^_^v Carl Johnson) 19:31, 20 August 2010 (UTC)[reply]

Arbitrary Break 2[edit]

For some vandals, wasting volunteer time gives them their jollies. —Jeremy (v^_^v Carl Johnson) 21:20, 18 August 2010 (UTC)[reply]
As a reviewer, our time is not wasted to such efforts but rather well spent. My76Strat 02:56, 19 August 2010 (UTC)[reply]
Then you are the kind of person who gives the vandals I mentioned immediately above their jollies. —Jeremy (v^_^v Carl Johnson) 19:21, 20 August 2010 (UTC)[reply]
Reverting any vandalism is a waste of volunteer time, IMO. Sometiems I spend half my edits a day reverting vandalism, but that isn't going to change, with or without PC. - BilCat (talk) 03:42, 19 August 2010 (UTC)[reply]
That doesn't make any sense. It's not as if an accepted edit has to be treated differently than any other edit. It can be reverted as usual, and anyone 'monitoring the page' can see it and change it as needed. ♫ Melodia Chaconne ♫ (talk) 13:58, 19 August 2010 (UTC)[reply]
I think his point is that having a notice saying "THIS IS THE *ACCEPTED* VERSION" along the top will tend to lead the casual reader to think that it's been peer-reviewed, or something like that, so their likely to give anything they read in it more credence, thus potentially increasing the damage done by sneaky inaccuracies.☻☻☻Sithman VIII !!☻☻☻ 15:05, 19 August 2010 (UTC)[reply]
  • Those "other specific cases" were attempted and had to be rejected as test subjects because the vandalism overloaded the FR. —Jeremy (v^_^v Carl Johnson) 21:38, 19 August 2010 (UTC)[reply]
  • Using it on particularly-liked or -disliked politicians would end up overloading the FR, as those articles are subject to heavy amounts of editing. You think Barack Obama and Bush weren't already tried? —Jeremy (v^_^v Carl Johnson) 19:36, 20 August 2010 (UTC)[reply]
  • In the pursuit of a vandalism-free encyclopedia where anyone can edit we cannot sacrifice article veracity. If anything, the comment above only serves as an argument against using FraggedRevs on articles about complex subjects. —Jeremy (v^_^v Carl Johnson) 19:43, 20 August 2010 (UTC)[reply]
That is a fair point. If the page is protected because of errors about complex subjects, then PC is probably the wrong tool for fighting that. In general, though, there is a balance between veracity and the ideal of anyone-can-edit; I think PC can be useful if the alternative is (semi-)protection, etc. ErikHaugen (talk) 20:33, 20 August 2010 (UTC)[reply]
  • Murphy's Law indicates that they'll create sneakier vandalism or otherwise defeat antivandalism measures with or without FR. Thus far, it hasn't been proven wrong. —Jeremy (v^_^v Carl Johnson) 19:49, 20 August 2010 (UTC)[reply]
I think most ips don't know this poll is happening. Many i.p.s are 'drive-by- editors, and though they may not care about this discussion, that doesn't mean they don't care about the ability to edit articles without having to go through an oversight procedure. It's up to us as regular editors to think about the effect of policy on users who might not be a part of the discussion. Ocaasi (talk) 11:52, 20 August 2010 (UTC)[reply]
  • What you said above begs the question: What about edit-warring on such articles? Is it a violation of CRASH policy to accept edits furthering an edit war, and would doing so make you as guilty of edit-warring as the actual edit-warriors? There's a rather major rub here, because both edit-warring and abetting the edit war would fall under disruption, I'd think. —Jeremy (v^_^v Carl Johnson) 20:43, 20 August 2010 (UTC)[reply]
We need more graphs, and then need to work out which ones are useful. I started with these, and believe that the magnitudes are concerning. We are adding complexity to a lot of apparently low-traffic articles -- why? Its possible I have made a mistake (I am currently quite tired), but perhaps not. User A1 (talk) 23:55, 20 August 2010 (UTC)[reply]
Also it appears that the local vandalism ratio is quite constant, except in the very end, suggesting that vandalistic edits are not actually more problematic than any other article, except in the last 200-300 articles in the plot; which is about 20% of what we have currently marked for pending change protection. This seems like we are overusing the tool User A1 (talk)
  • I think its liberal application was part of the test. Any page could have been requested for the trial period, and it was granted much of the time. Heck, Bible and Barack Obama (the latter which I reviewed an edit on once but was beaten to reverting it) both got added to the queue, but those were reversed (quickly) as they are heavy targets evidently. CycloneGU (talk) 13:08, 21 August 2010 (UTC)[reply]
  • I am afraid I do not follow your reasoning. Patrolling recent changes also takes the manpower of experienced editors. Also, while vandals CAN register, they have to have a good activity before they will get the autoconfirmed flag. They can't just register and start vandalizing; we'll catch it immediately, PC or not. Further, what is the roadblock you speak of? Editors who previously could not edit some articles can now make good-faith edits that can be accepted by any member of the reviewing community; if the change is a good one, then it SAVES manpower on the talk page having to add it manually ourselves, and it's NOT an obstacle to the user who was able to add their bit to the article. CycloneGU (talk) 13:08, 21 August 2010 (UTC)[reply]
  • Speaking from experience, Cyclone, no they don't. I see autocon-busters all the time; they spend their first ten edits tweaking their talk page or wikiGnoming on an article before they start spreading havoc. —Jeremy (v^_^v Carl Johnson) 19:30, 21 August 2010 (UTC)[reply]
  • Point taken. However, the current guideline I remember reading is you have to have 75 edits - typically mainspace edits - before being granted the userright (if am I mistaken, correct me please). So a vandal spending ten edits on their page will contribute nothing to the edits required to qualify as a reviewer. Maybe we do need stricter standards for the role, I'll be first to admit that. The pending changes itself, however, is not an obstacle to good editors; only to bad. CycloneGU (talk) 23:01, 21 August 2010 (UTC)[reply]
  • You said nothing about the reviewer userright - "Also, while vandals CAN register, they have to have a good activity before they will get the autoconfirmed flag." In any case, autocon-busters still render PC:L1 useless because they can bypass PC altogether, making it no better than semi-protection with regards to defeating determined sockpuppeteers (whom commonly create autocon-busters). —Jeremy (v^_^v PC/SP is a show-trial!) 22:09, 28 August 2010 (UTC)[reply]
I don't think I follow your points:
  • "With rarely edited pages that very few or no editors have watchlisted, edits by IP's or registered but not autoconfirmed users are likely to stay unreviewed for a long time... They can also easily get staggered and nested which will make it even harder to eventually untangle them." Isn't that why PC would be useful, because it prevents unreviewed edits from just sitting there and getting tangled?
  • "PC feature could only work reasonably well for a relatively small number of pages - where a sufficiently large number of editors are watching them but the editing traffic is not so high as to make dealing with reviewing new edits difficult." The balance only has to be in favor of edits/reviewers. Reviewing few high traffic pages or many low traffic pages should be about the same in terms of difficulty.
  • "Most of non-vandalism IP edits, on any articles, consist of adding unverified information. Without the PC feature, when I see such an edit... I sort of let it slide based on AGF. However, with the PC feature, we are required to accept an edit which looks like a higher degree of endorsement." The badge of 'acceptance' has some weight, but unsourced BLP edits should be looked at with scrutiny to begin with, and I'm not sure assuming good faith is the driving motivation when working with BLP edits, in fact, it might be part of the problem. Ocaasi (talk) 09:05, 21 August 2010 (UTC)[reply]
  • I don't see the situation where IP edits on a low traffic article sit unreviewed for weeks and months as beneficial. The IP editors would only get confused and frustrated by such a situation, and, after a bunch of unreviewed edits pile up, experienced editors will be discouraged from editing the article further and trying to untangle a pile of unreviewed edits. It is simpler to use regular semi-protection on a wider range of articles. This way people will not be confused and regular editors will not get saddled with a huge new area of backlog. Nsk92 (talk) 14:22, 21 August 2010 (UTC)[reply]
  • Whatever configuration people settle on, the assumption is it would only work if the reviewer backlog clears quickly. In that case, edits won't pile up. So I agree what you described is a bad situation, but don't think anyone is proposing it. Ocaasi (talk) 14:56, 21 August 2010 (UTC)[reply]

Arbitrary Break 3[edit]

"Verified?" Where does it introduce this word? A reviewed edit is merely "accepted". This tool works very well for Telephone and others which get a steady though not heavy stream of IP graffiti and rarely a substantial improvement. Yes, it will confuse those who haven't logged on. It's worth the cost. Option 3, a modest expansion from the current thousand to a few thousand articles while continuing to tinker with the workings, is appropriate. Jim.henderson (talk) 18:14, 21 August 2010 (UTC)[reply]
He's saying that accepting the edit gives the impression that it has been vetted and verified. Reread his comment. —Jeremy (v^_^v Carl Johnson) 19:28, 21 August 2010 (UTC)[reply]
  • "Three-quarters" prot isn't likely to see use. "Quarter" prot is too ineffective to even be a "quarter" because it jams the history full of garbage and/or more serious material, such as BLP violations. —Jeremy (v^_^v Carl Johnson) 19:28, 21 August 2010 (UTC)[reply]
  • Any content you add or accept into an article you are responsible for, legally. You might get mitigation because you didn't actually present it for acceptance, but you are the one accepting it so you have a duty of care to yourself to review it well. Of course it rarely, almost never comes to such extreme situations.Off2riorob (talk) 19:44, 21 August 2010 (UTC)[reply]
  • That's not what I'm referring to. Even if the edit is reverted instead of accepted, the bum edit remains in the history (since we're obligated by the licenses to maintain a list of all authors), as will any other edit that a reviewer rejects. As a result, the history will be more difficult to browse through because of all the vandalism. —Jeremy (v^_^v Carl Johnson) 19:57, 21 August 2010 (UTC)[reply]
The author in the case of an accepted edit would be the unconfirmed user supported by the accepting reviewer. I haven't seen this you are referring to be an issue at articles I have worked that are pending protected. Off2riorob (talk) 20:08, 21 August 2010 (UTC)[reply]
Alright, then, you want Cliffsnotes: PENDING CHANGES CLOGS THE HISTORY WITH VANDALISM AND BLP VIOLATION EDITS. (I concede the authors point, but not the revision pollution one.) —Jeremy (v^_^v Carl Johnson) 20:21, 21 August 2010 (UTC)[reply]
Pending changes doesn't clog the history with anything of the sort. If there's that much vandalism you should be using semi-protection instead. That much is obvious from this trial. What PC is good for is preventing vandalism showing in articles which aren't subject to heavy editing and which wouldn't normally be semi-protected, where the vandalism would otherwise be completely unattended - in Google and visible to visitors. -- zzuuzz (talk) 20:51, 21 August 2010 (UTC)[reply]
Actually, it does, since all edits remain in the history. Even if the article sees only one bum edit every few days, those edits add up. —Jeremy (v^_^v Carl Johnson) 20:58, 21 August 2010 (UTC)[reply]
If they're not protected at all then the vandalism stays visible in the article instead. That's what adds up. -- zzuuzz (talk) 21:02, 21 August 2010 (UTC)[reply]
I couldn't have put it better myself. CycloneGU (talk) 23:03, 21 August 2010 (UTC)[reply]
Believe me when I say, chummer, that the objects in the history can be just as much concern as on the actual page. —Jeremy (v^_^v Carl Johnson) 03:14, 22 August 2010 (UTC)[reply]
As I have also pointed to, I would like to see Pending Changes be implemented more as a pathway for a longterm semi-protected article to become unprotected and hence allow more people to edit. However, I don't think the rest of the community may be on the same wavelength as me, nor do I see much motivation for users to try and make any efforts to unprotect some longterm semi-protected articles. –MuZemike 00:40, 22 August 2010 (UTC)[reply]
  • If it were used on low-traffic pages, it would be likely that the people reviewing a page's edits would not really know much about the subject anyway, and so would not be able to make a good judgement as to whether the edit (barring very obvious vandalism) was reasonable or not.
  • If it were used on high-traffic pages, incorrect edits would most likely be caught soon anyway, and be able to be corrected almost as easily, if not more easily.
Thus, I oppose the Pending Changes process. Layona1 (Talk) 04:09, 22 August 2010 (UTC)[reply]
  • Non-existent, or at best totally inadequate, protections against censorship, instead of ensuring that it is used for nothing more than vandalism protection
  • Inadequate guidelines/process for choosing editors as reviewers, in order to prevent picking people who will abuse tool
  • Waste of time and resources, with no clear benefit -- yet another layer of bureaucracy that will suck away community time
  • Increased complexity
  • Discourages new and anonymous users from editing, while doing almost nothing to prevent vandals
  • Directly contradicts Wikipedia's mission of being an open encyclopedia that anyone can edit
--Jrtayloriv (talk) 17:28, 22 August 2010 (UTC)[reply]
  • It actually encourages new and anonymous users by letting them edit articles that would otherwise be protected, so this works towards our mission. I do agree that censorship is a potential, but we just have to agree to have no tolerance for abuse of the reviewer power. —Arctic Gnome (talk • contribs) 21:05, 22 August 2010 (UTC)[reply]
  • No, chummer, just like DRM it screws the beneficial and empowers the baneful. You need to think less "beneficial" anons and more "all" anons - even the trolls, even the drive-by vandals, even the wethers of Grawp, even the IP sockpuppets. Thinking like yours is the very same logical fallacy FraggedRevs of any implementation runs on, and it's so severe as to be insurmountable. —Jeremy (v^_^v PC/SP is a show-trial!) 18:42, 9 September 2010 (UTC)[reply]
  • I actually found an Australia television show under Pending Changes as well and quetioned its inclusion. I agree that your sports scores example is another good example. My policy has been that if it's not sourced, it's not valid and should be reversed. If a source is provided, check the source; if it agrees, approve it. Sometimes I did my own research, too, and actually found the information; I can think of once this occurred and I approved the edit, then applied the source. Just my stance, though. CycloneGU (talk) 17:58, 23 August 2010 (UTC)[reply]
Unfortunately, that's the problem; everyone is acting based on their own stance instead of clear guidelines for approving. —Ost (talk) 17:39, 24 August 2010 (UTC)[reply]
In particular, the people who are using it to reject all unsourced edits are using it to apply an imagined policy. We do not normally reject unsourced edits unless they are challenged,or are unsourced negative BLP information, and a decision to change this to include all unsourced edits would require a very wide consensus indeed--even if only on BLPs, as most edits are entirely non-controversial. My understanding was that it would be used only to prevent clear vandalism or unsourced negative BLP edits. DGG ( talk ) 23:32, 24 August 2010 (UTC)[reply]
On that note, I did accept an edit that looked all right once and moments later had a talk page posting pointing out that the edit WASN'T all right. It was a mistake on my part, it seems, but what I accepted was not clear vandalism or negative BLP to me upon reading it. CycloneGU (talk) 00:30, 25 August 2010 (UTC)[reply]
Hmm, not a bad summary.. but maybe it would be more helpful if someone could conclude by summarizing both sides of the debate?? -- Ϫ 11:42, 25 August 2010 (UTC)[reply]
Larry Sanger has done that for us. —Jeremy (v^_^v PC/SP is a show-trial!) 21:03, 26 August 2010 (UTC)[reply]
Statistics seem largely irrelevant on such a small sample, the wheels did not drop off so perhaps it is just a simple question of, shall we now test it on what it was designed for, applying to all BLP articles. Off2riorob (talk) 14:33, 25 August 2010 (UTC)[reply]
I don't think there's support for a major expansion until at least the interface and policy issues are worked out. The 10k expansion seemed useful towards the end of better stats, even if it's still relatively small. Ocaasi (talk) 15:18, 25 August 2010 (UTC)[reply]
The more I think about it and see the discussion the more I start to support a full trial of all BLP articles. The trial was basically to see if the wheels dropped of and they didn't as so we should logically take the next step. Off2riorob (talk) 15:30, 25 August 2010 (UTC)[reply]

Continue but scale back[edit]

While the straw poll outcome was a majority in favor of continuing pending changes, a significant minority (~33%) want to end the trial. As a compromise, I propose the following:

  1. Roll back the current trial by removing pending changes protection from the pages that have it now, to be completed within 30 days.
  2. After pending changes protection has been removed from all articles on which it is currently used, continue the trial in 3 month increments at a smaller scale to allow the developers time to make improvements, adjusting the scale of the trial at each increment, based on consensus. (100 articles to start with?)
  3. Strictly limit the time a page is under pending changed protection to 1 week increments so that we can gain better experience with where the feature works and does not work, and allow the feature to be tried in other ways.
  4. Ask the opposers to elaborate on what doesn't work, so that we may make improvements.

Discuss? Triona (talk) 06:48, 7 September 2010 (UTC)[reply]

The discussion has been had. I sincerely doubt you will get any points that are not listed in the "working summary" section above, which pretty much says it all. There are some issues that improvements (technical changes) can address (feature request section), but there are questions that no amount of technical trickery is going to help you with (mostly the community and workload section). These are all well thrashed out above. Statistical analyses might answer some of the open questions above, but who will do it, and how will we know that they have answered the questions correctly? These things can take months to analyse in any meaningful manner. My personal opinion is that implementing this will generate a significant number of disgruntled editors, not implementing it propagates the status quo. User A1 (talk) 15:18, 7 September 2010 (UTC)[reply]

The first step is to honor the original trial agreement, which limited the trial to two months. That's over. That trial needs to stop. Once that's done, it's time to figure out what to do next. If PC proponents won't honor the original agreement, what faith can be placed in subsequent agreements?—Kww(talk) 15:20, 7 September 2010 (UTC)[reply]

I wish we could come to a consensus on this. I see two obvious things: 1, a lot of issues were raised (see above); 2, a lot of people want to remove PC. So, why not keep it enabled but end its use until the issues have been fixed, and then just start a new trial with a better technical system? Even if 1/3 oppose might be seen by some as not enough to be "no consensus", why keep a trial going on indefinitely with no apparent purpose, as the issues haven't all been addressed yet? —fetch·comms 00:02, 8 September 2010 (UTC)[reply]
I'm in support of PC but I'm more in support of the consensus operation of WP. Given the lack of clear consensus, the trial should end now. I'm starting to see backlogs on the PC editing list. Gerardw (talk) 02:39, 8 September 2010 (UTC)[reply]

I'm still confused how 2/3 support to go ahead with it is not consensus to move ahead when 3/5 was enough to turn it on originally. NW (Talk) 22:04, 7 September 2010 (UTC)[reply]

259-61 is 3/5 now? --Yair rand (talk) 22:20, 7 September 2010 (UTC)[reply]
I was talking about the original poll to turn it on, the one that Jimbo (not uncontroversially) closed. NW (Talk) 22:43, 7 September 2010 (UTC)[reply]
Notwithstanding Jimbo's close, the WMF didn't act on that previous proposal because it was not supported by at least 66% of editors, unlike the present trial which achieved 80%. Erik Moller had stated that for a FlaggedRevs implementation, 2/3 of editors in support were required [2]. Cenarium (talk) 23:48, 7 September 2010 (UTC)[reply]
Obvious vandalism decreases the opinion our readers hold of Wikipedia and thus potentially decreases people desire to join in editing.Doc James (talk · contribs · email) 23:33, 8 September 2010 (UTC)[reply]
So does subtle vandalism that is later proven false and becomes difficult to remove because someone gave it a green light, false promises, a very deeply entrenched troll community, and an already-prevalent mood that the unwashed anon peasants are diluting the glorious registered-editor master race's say. —Jeremy (v^_^v PC/SP is a show-trial!) 18:38, 9 September 2010 (UTC)[reply]
Do you realise that when you say "this needs to stop while we discuss what to do next" that it then no longer matters what you discuss and what the outcome of the discussion is? If it stops then it is not so likely to come back even if it is wanted. I rather like flagged revisions. I don't like calling it pending changes. Just typing that makes me feel dirty. Maybe it has something to do with my little corner of Wikipedia but i never came across any horrible issues with the feature. I read the poll. My thoughts on consensus are a poorly kept secret. In short it is 'good luck getting people to agree - we are not Cybermen'. To the person coming to an article to learn things are not always obvious. If it were obvious to them they would not necessarily be there reading the article or you would see a hell of a lot more anonymous correcting of obvious nonsense and errors that were snuck in for shits and giggles. I did that for about 7 years. With flagged revisions enabled i wouldn't have been able to correct the obvious errors i found however there would likely have been a lot less of them to be found. I voted in the keep section for the wildly crazy idea to actually not set some random restriction on the use of flagged revisions, forcing it there and disallowing it here but to use it as warranted on any given article be it about a person, an event, a place, a whatever. Technically that was not even an option in the voting, hence my calling it a wildly crazy idea. People read more than biographies. Every article makes mention of someone if it is at least a proper stub because someone had to write the reference used. Obvious nonsense gets added to any article not deleted. Proscribing an artificial limit to the scope of use or quantity of pages it is applied to is just setting up failure. Not every biography needs flagged revisions enabled; to claim so is just a whee bit exaggerated. To deny flagged revisions where it would be beneficial simply because an article is not a biography is likewise lacking some common sense. Scale back use? Maybe. Practical, common sense use? Yes, please!. delirious & lost ☯ ~hugs~ 06:23, 9 September 2010 (UTC)[reply]
To be perfectly honest 66% of the vote is a pretty resounding vote of support. If such a large percentage voted for one party in a US election or one coalition elsewhere it would be a very resounding victory. Obama won by 53% to 46% or something. -- Eraserhead1 <talk> 06:59, 9 September 2010 (UTC)[reply]

Announcement of result[edit]

Announcement about Pending Changes --Jimbo Wales (talk) 19:33, 10 September 2010 (UTC)[reply]

The discussion above is closed. Please do not modify it. Subsequent comments should be made on the appropriate discussion page. No further edits should be made to this discussion.
More