Wikipedia talk:Reviewing pending changes/Historical/Archive 1

Archive 1

Archive 2

Archive 3

Promotion

• I think that manual would be best... have something like rollbacker's system. Basically, automatically giving users the ability to review pages seems kind of dangerous... although most users would probably do fine, there are going to be a number of people who aren't vandals but who shouldn't have it anyway (e.g., not enough experience), and just saying that "X edits in X days is enough experience" doesn't seem to work. Having a page where anyone can request the ability for the duration of the trial (and afterwords, if wanted), and allowing any admin to grant it, would probably be best. Admins can use their best judgment regarding whether or not someone should gain the ability, kind of like is done with rollback now. It should typically be granted, unless the user is too new or it looks like they may use the ability poorly. (also, is this supposed to be here or on the talk page? Please move it if I was incorrect). –Drilnoth (T • C) 02:13, 28 March 2009 (UTC)

• I view the issue like this: many users will want high requirements due to BLPs, and many users will want low requirements so that reasonably experienced good-faith users can freely edit flag-protected pages. An autopromotion is inevitable, even if high, but we can argue on the requirements. I think a way to compromise on this is to use a group intermediary to (auto)confirmed and reviewer, say 'established', granted automatically, with lower requirements than reviewer (that would also be granted automatically). Established users should be able to semi-review semi-flag-protected pages (but not intermediary flag-protected pages), but they cannot patrol, or access restricted reviewing-related special pages. So something like this, (auto)confirmed is given as reference and numbers as examples:

(auto)confirmed: 10 edits, 4 days

established: 200 edits (100 in MAIN, 15 articles, 2 recent edits in MAIN), 2 months

reviewer: 500 edits (250 in MAIN, 50 articles, 5 recent edits in MAIN), 6 months

Additionally, each group can be added and removed by admins. So users can request premature 'established' access at WP:PERM, or premature 'reviewer' access, admins may grant it or not, or only the established access, based on some guidelines. Rollbackers could also have established rights, but not reviewer rights by default. Cenarium (talk) 18:03, 28 March 2009 (UTC)

I agree with automatic promotion for reviewers. Doing it manually gives a very strong image of cabalism. Only those who know the right people can review. It is better to give it out easily and remove it from anyone who misuses it. Sure, bad edits will get flagged. So what? This isn't supposed to solve every problem. If it can stop most and the worst BLP issues, that is good enough for me.

For this trial, how about we give it to all rollbackers? Letting people try it out is a big part of the trial.--Apoc2400 (talk) 18:20, 28 March 2009 (UTC)

That is the whole point. We should avoid bad edits from getting reviewed as okay when they're not. The main goal is to improve accuracy. That is more important than a perceived image of people who don't know how things work around here. Allowing any admin to give out the ability when requested gets rid of the "have to know" because all requests will be seen whether they know someone in a supposed high spot or not. With a simple check like this we can avoid problems before they even present themselves which can only be good for WP content. - Mgm|^(talk) 14:01, 5 April 2009 (UTC)

There is a discussion with poll at Wikipedia:Village pump (proposals)#Poll on (auto)confirmed usergroup to allow premature granting and removal of the autoconfirmed status. This is relevant to this discussion, since autoconfirmed users are autoreviewed. Cenarium (talk) 22:03, 28 March 2009 (UTC)

Just to clarify things for me, how do "reviewers" and "established" compare with the "editors" and "reviewers" described at mw:Extension:FlaggedRevs? Kevin (talk) 21:58, 29 March 2009 (UTC)

Reviewer from mw:Extension:FlaggedRevs has no equivalent. Editor from mw:Extension:FlaggedRevs is vaguely equivalent to reviewer, while established is a little weaker than reviewer. But this is a heavily modified version, so we can't really compare the usergroups. Cenarium (talk) 15:02, 29 March 2009 (UTC)

(responding to a comment higher up): 500 edits and 6 months seems a bit much to me... maybe 100 edits and one month, but I think that most of the trustworthy Wikipedians who are going to be around for the long haul and would make good reviewers don't need 6 months to establish that. Anyone who would misuse the ability would probably be noticed before a 1 month/100 edits count. –Drilnoth (T • C) 18:43, 29 March 2009 (UTC)

Do you mean for manual granting or automatic granting ? For autopromotion, we should apply a higher threshold than for manual promotion, since we don't have a review of the user contributions by an admin. Manual granting should be more based on experience and contributions than on numbers imo. We need autopromotion otherwise it'll be too much of a burden for admins to grant the rights to trusted users and many will be left out because they won't be noticed or request the rights. The numbers were given as examples. Cenarium (talk) 19:41, 29 March 2009 (UTC)

Okay; I thought that it was going to be a limit like "500 edits and 6 months or you can't get it," but if an admin can grant it before then (and generally do, unless there's an obvious problem), I'm fine with it. I just didn't want reliable, 3-month contributors to be completely unable to become "reviewers". –Drilnoth (T • C) 19:56, 29 March 2009 (UTC)

100 edits and 1 month seems fine to me for a trial. WP:PERM would be appropriate for those who cannot wait, and WP:ANI for those who misuse it. Kevin (talk) 21:58, 29 March 2009 (UTC)

Why the different terminology to how the extension is documented? We're going to end up with 3 different groups: those who cannot review, those who can review semi-flagged, and those who can review fully flagged articles. These last 2 would seem to equate well with "editors" and "reviewers". Perhaps I'm misunderstanding something. I was thinking that sysops would be made reviewers, and editors would be autopromoted as you are discussing here already. Kevin (talk) 21:58, 29 March 2009 (UTC)

Because editor may imply that others are not editors... so we just call reviewers. There are no quality or in-depth content checking in this proposal, which was the main purpose of the reviewers in the flaggedrevs doc, so it's why it has no equivalent. We also have to consider who can patrol, and blp folk will probably want higher requirements for them, so it's why I proposed to distinguish established users with semi-review, and reviewers with semi-review and patrol. Yes, reviewers will be autopromoted in any case, and sysops can fully review. Cenarium (talk) 23:26, 29 March 2009 (UTC)

We don't have a project page for low-level usergroups, why should we have one for reviewers ? It should be merged in Wikipedia:Reviewing guideline. We can discuss the requirements at the /Trial subpage, where this discussion started. Cenarium (talk) 20:09, 3 April 2009 (UTC)

Okay, it'll allow to centralize discussion, and we can consider merging it somewhere (in WP:UAL, etc), later. Cenarium (talk) 21:29, 3 April 2009 (UTC)

Fine with that - just want to get the discussion going and focussed on this area. AndrewRT(Talk) 18:42, 4 April 2009 (UTC)

Established usergroup intermediary between (auto)confirmed and reviewer

I proposed to add an established usergroup intermediary to (auto)confirmed and reviewer because users will probably want higher requirements for users able to patrol, due to BLPs, but not too high for confirming on semi flag protected pages, so that minimally experienced users can edit more freely on them. Users with need for other reviewers tools would probably be among more experienced users, and both rights can also be requested at WP:PERM anyway. We could have criteria for autopromotion like those:

• 'Established': 100 edits (50 in mainspace, 2 recent) and 1 month

rights: confirm on semi flag protected pages

• 'Reviewer': 300 edits (200 in mainspace, 3 recent) and 3 months

rights: confirm on semi flag protected pages, patrol in mainspace, access to Special:OldReviewedPages and Special:UnreviewedPages.

Thoughts ? Cenarium (talk) 03:15, 4 April 2009 (UTC)

I'm not too keen on this - sounds like we're making it more complicated than it needs to be. Could you explain to me what is meant by "patrolling" - what user rights would a patroller have which a non-patroller would not have? AndrewRT(Talk) 18:44, 4 April 2009 (UTC)

It's explained here. Cenarium (talk) 23:14, 4 April 2009 (UTC)

So, in summary, all patrolling does is de-flag the article from the list used by new page patrollers. That has less impact than reviewing itself. It would not make sense to me to make patrolling rights more restrictive than reviewing rights - it may just encourage the creation of a patrolling sub-culture out of step with the rest of the community. I suggest we drop this idea and just have one new group - Reviewers. AndrewRT(Talk) 21:55, 9 April 2009 (UTC)

Any revision on any article can be patrolled, and it doesn't affect the version viewed by readers, only revisions to flag protected pages can be confirmed/validated, and there the version viewed by readers by default is the latest confirmed/validated one. Patrol links are given in watchlists, etc. Even though it's passive, it's actually a little more sensitive because a main goal is to use that to monitor BLPs. Adding the established usergroup would allow to reduce the requirements for confirming on semi flag protected pages (autoconfirmed users are already normally automatically confirmed when the previous revision is, but they can't confirm until becoming reviewer), while keeping the requirements high enough for reviewers to satisfy concerns over BLP monitoring. To keep it simple, we could start with only reviewer, but we'll probably have strong divergences due to the two needs. Cenarium (talk) 00:32, 12 April 2009 (UTC)

avoids image of cabalism?

This is not a good reason for autopromotion (though there may be others). We shouldn't worry about our image, but what the feature was designed for. Increasing the accuracy of Wikipedia. Some people shouldn't be given the ability, not because they're not part of the clique, but because they're known to make mistakes or are more likely to break rules. Wikipedia's well-being should come first, that of the editors and anyone else second. - Mgm|^(talk) 13:55, 5 April 2009 (UTC)

I've reworded it. It is important, however, that the policies and actions we take have the confidence of the wider community. AndrewRT(Talk) 21:58, 9 April 2009 (UTC)

It's not just an impression, it would _be_ cabalism. --Apoc2400 (talk) 17:23, 11 April 2009 (UTC)

Having this right handed out automatically and revoked if not deserving would be best IMO. We need a simple system. We need to keep bureaucracy to a minimum. If someone goes through the trouble of creating an account and makes 250 or 500 edits without being banned they deserve recognition. We should WP:Assume good faith until guilt is proven rather than the other way around.Doc James (talk · contribs · email) 05:34, 29 May 2010 (UTC)

Decision

How and when are we going to decide this? The "trial" is already nearly a sixth of the way through and it hasn't even started.

How about a poll on autopromotion Yes or No with a deadline of Sunday 19th April? AndrewRT(Talk) 22:00, 9 April 2009 (UTC)

The two month clock won't start ticking until the extension is activated, which could easily be two months from now :D. There are mountains of technical issues to iron out before activation, so there's plenty of time to iron out the implementation details. That said, I do agree that some level of expediency is probably not a bad idea. Happy‑melon 22:17, 9 April 2009 (UTC)

Yeah.. we'd need to decide this, but we desperately lack comments here and on the implementation in general. We'll probably have a massive involvement once the feature will be turned on, but before that, almost nothing. Added to WP:CENT, but something clear, like a poll is probably needed. Cenarium (talk) 21:16, 11 April 2009 (UTC)

The technical issues should have been ironed out (or at least surveyed) before there was even a poll on activation. And I'm lucky I even found this page. --Pixelface (talk) 06:30, 12 April 2009 (UTC)

Draft for a poll below, please help in the elaboration. Cenarium (talk) 00:01, 12 April 2009 (UTC)

Poll on autopromotion

See Wikipedia talk:Reviewing/Historical/Poll on autopromotion

Established usergroup with autopromotion

See Wikipedia talk:Reviewing/Historical/Established usergroup

Implementation request

See Wikipedia_talk:Flagged_protection_and_patrolled_revisions#Implementation_request. Cenarium (talk) 22:53, 29 April 2009 (UTC)

Promotion criteria

The previous poll ended with a majority view that the trial should be run using a "reviewers" usergroup that is manually promoted by admins at WP:Requests for permissions.

Promotion at RFPERM is normally decided based on presumptive criteria - i.e. users with a certain number of edits and/or history of editing would normally be promoted. We now need to decide what criterion/criteria, if any, should be applied to promotion of users to the reviewer group.

Of those that favoured autopromotion in the previous poll, the mix of edits and time was:

Edits\Time	Unspecified	1 day	1 week	3 months	6 months	1,000 years
Unspecified		Taku
25			Falcorian Nicolas1981
50			Admiral Norton
100				Apoc2400 Fahadsadah
200	Walkerma II			JoshuaZ
500	Cybercobra Darrenhusted			Chamal Roux Hans Adler Exit2DOS2000 utcursch Mav Unionhawk	lucasbfr MuZemike SyG
1,000	Mjquin_id
10,000,000						Lar

These results should be interpreted with caution as they were in answer to a different question and don't include those people who voted against autoconfirmation. Nonetheless, I think they're useful in pointing up popular mixes.

My suggestion is we start to convass views from Saturday 9th May, advertise on WP:CENT (anywhere else?) and run it for, say, two weeks. I suggest we just ask people to say what combination they support, e.g.:

• 200 edits, 3 months User:Low supporter) 23:09, 7 May 2009 (UTC)
• no edit limit, 6 months User:Time only) 23:09, 7 May 2009 (UTC)
• 1000 edits, 12 months User:High supporter) 23:09, 7 May 2009 (UTC)

etc. How does this sound? AndrewRT(Talk) 23:09, 7 May 2009 (UTC)

I think we'd need a list of users matching arbitrary time since registration/edit count criteria for analysis, and most importantly, for admins to patrol and grant on sight when they feel the user would make a good reviewer/satisfy the requirements if we have.

The problem, quasi-contradiction, is that we need to keep the backlog in old confirmed pages as short as possible to avoid delaying edits from view, but in the same time patrolling should only be done by reasonably trusted and experienced users. So one side calls for lax criteria, the other side for strong ones... In the long term, I'm afraid it can't be solved without splitting the usergroup. For the trial, I would support rather medium-high criteria to begin with, then downgrade if the backlog is too important. But it's absolutely essential to search for potential reviewers, not wait for them to request the rights, otherwise we'll inevitably fail the trial. Cenarium (talk) 00:10, 9 May 2009 (UTC)

One of the key things someone else mentioned as well is the need for us to recruit reviewer applicants. We can start doing this as soon as we have these criteria agreed! On your comment about analysing users - any idea how we could get this kind of analysis? AndrewRT(Talk) 00:28, 9 May 2009 (UTC)

I think it should simply be granted on sight when the admin feels the user is ok for the rights, and maybe leave a note on the user talk page for information (a template to subst would be handy for that). This kind of data can probably be generated through WP:DBR. It should skip admins, and when implemented, reviewers. Cenarium (talk) 00:42, 9 May 2009 (UTC)

Thanks - I've put a request in on WP:DBR so lets see what they come back with. in teh meantime, here are some random stats on edit counts:

From Special:Statistics: Users: 9,611,153 Active: 159,523 Rollbackers: 2,572 Admins: 1,650 From Wikipedia:List_of_Wikipedians_by_number_of_edits: Users>8,277 edits: 4,000 (2.5%) From http://stats.wikimedia.org/EN/TablesWikipediaEN.htm: In Oct-06 - 4,330/151934 (2.8%) had >100 edits

AndrewRT(Talk) 01:14, 9 May 2009 (UTC)

Can we autopromote the rollbackers at least? ViperSnake151  Talk  22:39, 10 May 2009 (UTC)

MZMcBride has compiled a list of users with more than 1,000 edits, their first edit more than a year ago, and their latest edit within the past month who are not bots or admins at Wikipedia:Database reports/Potential reviewer candidates. There are 11162 of them, which means it'll take a long time and many admins to parse through the list. An idea to handle this would be to enable the reviewer usergroup before the complete implementation and actual beginning of the trial, to give us time to grant those rights. Cenarium (talk) 02:15, 11 May 2009 (UTC)

Well I know it would expand the pool even more, but 3-6 months seemed like what most people were thinking of in the poll, and based on your report I would be excluded by your report criteria, since I have ~11 months since my first edit. Perhaps the time limit should be modified to 6 months. It might also be useful to run a report on rollbackers and find out how may are active, this would be a smaller pool to run a trial with, before expanding to everyone that meets whatever the criteria are.--Terrillja _talk 02:34, 11 May 2009 (UTC)

It would be more efficient to parse through a smaller list of users with a high probability to be granted the rights and then gradually expand the list to a broader criteria and remove those who have been granted the rights. A list of active rollbackers could be useful too, although most of them would probably be included in the list when the criteria are relaxed. But I wouldn't grant reviewer rights to them without review, or above others, as it's really a tool rights, sometimes given to users who were not even autoconfirmed, while many article writers don't have it. Cenarium (talk) 12:26, 11 May 2009 (UTC)

Autoreviewers

For those who missed this, the group Autoreviewer has been enabled, pages made by those users are automatically patrolled (admins were already). The name autoreviewer is used for compatibility with the FLP/PR implementation. If we're in sync, this means an autoreviewer will have each of his/her edits auto-patrolled (that's the second flag, the one available for all articles). In the configuration, edits by autoconfirmed users on semi flag protected pages are auto-confirmed, except if auto-confirmation of non-reviewers is specifically deactivated for a page by an admin; so for compatibility between flags, autoreviewers should also be auto-confirmed on those pages. This means that autoreviewers will have the same (passive) clearance than reviewers, except they won't be able to review other users' edits. It allows them to bypass most of our anti-vandalism/blp checks, including those introduced by T21107 for abuse filter tags, so we must be sure they are trustworthy, and simply adding the ability to review other users' edits doesn't sensibly change the sensitivity of the user right at all.

Thus, I don't believe another usergoup is necessary, as an option in the preferences to turn off the reviewing interface would be better fit for this small difference (some users may not want to have the reviewing interface activated) and turning it on or off as we want would be easier than changing of group, and allow easier maintenance. I agree that if we trust a user for creating pages, we trust him for editing, but the criteria for being granted the reviewers group are still undecided, and may be higher than those that have been used for jvbot whitelist, which has now, for the most part, been transferred to the usergroup. So imo we'll either have to move them from the autoreviewer group to the reviewer group when they 'fit', then delete the autoreviewer group, or remove those who doesn't 'fit', then rename the autoreviewer group to reviewer. Cenarium (talk) 02:14, 23 June 2009 (UTC)

I agree fully with you. Reading this, I originally believed that there was some sort of mistake and this was discussed with no foreknowledge of the autoreviewer group. I think that this group is really redundant. Kevin Rutherford (talk) 20:13, 23 June 2009 (UTC)

Which group is redundant? Ruslik_Zero 04:29, 24 June 2009 (UTC)

I'm not familiar with "autoreviewers" but I think this is a different concept anyway. This is about reviewing other peoples' edits, not autoflagging your own. AndrewRT(Talk) 16:44, 27 August 2009 (UTC)

New discussion and poll: reviewer criteria

The trial of flagged revisions is moving closer and it would be useful to get this group up and running as soon as we can so that sufficient reviewers are recruited in time for the start of the trial. This discussion is only intended to decide on the group for the trial, and this is open to change at the end of the trial if the community wants.

A previous poll with 54 participants showed a majority against autopromotion on the grounds that this may otherwise become easy to game, a back door for vandalism and reduce confidence in the system. However, some feel that the participation and majority was not enough to base a decision. Therefore, taking advantage the media publicity on this issue, I thought now would be a good time to start a new discussion and poll. I have copied over people's previous reponses from the previous poll - feel free to delete/change these if you object. I've taken the liberty of refactoring people's previous responses into this format - hope no one minds and apologies if I have made any mistakes.

The poll will close at midnight (23:59) on Sunday 13th September and I'll advertise at CENT and other centralised discussion places. If the participation is less than 100 I suggest we consider extending the poll unless the result is clear cut.

Note that admins will already be given the status as part of their admin rights.

The questions are:

1. Should rollbackers be autopromoted to the reviewer group?
2. Should other registered users be autopromoted to the reviewer group?
3. If no to autopromoting registered users, what criteria should be used as a presumption for promotions?
4. If yes to autopromoting registered users, what criteria should be used?

AndrewRT(Talk) 17:23, 27 August 2009 (UTC)

Autopromotion of rollbackers

Please state "oppose", "support" etc, with rationale and sign with ~~~~

• Support with No Requirements - Rollbackers have already earned the trust of an Administrator, whom have earned the trust of the community, to have the judgement to revert edits they find inappropriate. Therefore, they also have the judgement to determine which edits they find appropriate, and should therefore be automatically promoted with no requirements to do so (other than being a rollbacker)--Unionhawk ^{Talk E-mail} 18:59, 27 August 2009 (UTC)
• Support, as per Unionhawk. Stifle (talk) 19:05, 27 August 2009 (UTC)
• Oppose any autopromotion (since apparently this poll is going to be run again anyway... why the second bite at the apple???) ++Lar: t/c 20:21, 27 August 2009 (UTC)
• Oppose per discussion below. Rollback is rollback. No need to make it weightier or more politicized. Protonk (talk) 20:39, 27 August 2009 (UTC)
• Support - (a) It's simple - we don't need another request process. (b) It's fair - those who are rollbackers are already trustworthy. Bsimmons666 (talk) 23:15, 27 August 2009 (UTC)
• Support - rollbackers have already gone through the manual process to weed out the gamers, and this gives us a bank of people with the rights to start with. AndrewRT(Talk) 00:21, 28 August 2009 (UTC)
• Support - reduction of WikiBureaucracy; anyone with rollbacker parm has someone to vouch for them, or at least someone who can be whacked if they go on a frenzy. Katana0182 (talk) 01:45, 28 August 2009 (UTC)
• Oppose - This is a lot more dangerous than a quick rollback button. Wknight94 ^talk 02:31, 28 August 2009 (UTC)
• Oppose - Whilst having rollback may imply some level of trust, bundling this with it seems like a bad idea, not least because outside of the 'trust' element there is no similarity. It's not like rollback isn't being removed every day from people who abuse it, abuse of this could have much more severe consequences for the encyclopedia. — neuro^(talk) 03:31, 28 August 2009 (UTC)
• Support - Let's not get too paranoid. They've already earned the trust. If we can't trust our own rollbackers how can we trust any reviewers? -- œ^™ 04:55, 28 August 2009 (UTC)
• Oppose - I agree with those who say that 'rollback' and 'review' are different things, and I do not find simplicity to be a strong argument. Also, I do not find it to be solely an issue of trust, but also about knowledge of Wikipedia policy (Vandalism, BLP, and others). Most rollbackers may well become reviewers if they apply, but it should not be an automatic step. 2help (message me) 05:25, 28 August 2009 (UTC)
• Oppose – comparing rollbackers to reviewers is like comparing apples to oranges. Both are also given for separate reasons. MuZemike 16:32, 28 August 2009 (UTC)
• Support with Requirements - combine rollback and a time requirement (as I proposed below, suggesting 3 months plus 500 edits). The rollback bit should (theoretically) ensure that an Administrator has checked them and that they are not (obviously) maliciously-inclined, 3 months/500 edits ensures they are around a while, hopefully understand our policies, and makes it a lot harder (or at least, requires more effort) for somebody with malicious intent to game the system and become a reviewer. PGWG (talk) 17:29, 28 August 2009 (UTC)
• weak support I'm worried that people will response to this by drastically increasing under what circumstances they give rolback. I think that PGWG's suggestion above seems very reasonable. JoshuaZ (talk) 18:20, 28 August 2009 (UTC)
• Support, as per Unionhawk.Chhe (talk) 03:51, 29 August 2009 (UTC)
• Weak support Even though this privilege has been granted rather loosely, and on different criteria, it is still a reasonable starting point, as a matter of simplicity. this is going to be complicated enough without adding any complication we do not absolutely require. . DGG ( talk ) 07:44, 29 August 2009 (UTC)
• Support We should avoid splitting Wikipedians into unnecessary classes/ranks. Rollbackers is a good balance in that the right is easy to get for good users but a level of proper vetting is done. --Cybercobra (talk) 20:36, 29 August 2009 (UTC)
• Support in the context of flagged protection and patrolled revisions. While I agree with general concerns that rollbackers are not content reviewers, they have been trusted to protect the integrity of the encyclopedia. This is what "Flagged protection and patrolled revisions" is all about, surely! My understanding is that it provides minimal protection of the integrity of the encyclopedia, while safeguarding the principle that anyone can edit. It seems to me that several opposes are based on stronger implementations of flagged revisions (which are widely opposed), in which reviewers need broader judgement. In such implementations it would be inappropriate to autopromote rollbackers. In this implementation, rollbackers are exactly the kind of editors we need to help us. Geometry guy 21:52, 29 August 2009 (UTC)
• Oppose—the rollbacker permission is a clearly defined permission on its own, and we don't need some sort of "pre-admin" class, as it were. While it might be interesting at some point to create an explicit non-admin "established user" group, rollbacker is not that group, and the permissions inherent in review go far beyond the trivial ability to revert edits in a single click. {{Nihiltres|talk|edits}} 04:43, 30 August 2009 (UTC), careless summary mistake fixed 16:46, 31 August 2009 (UTC)
• Oppose. RB status is assigned arbitrarily and easily; content review requires different skills and experience. To continue Nihiltres' allegory above, it's not about "pre-adminship", it's about "uber-editing". If censorship is inevitable, it should not be taken lightly and issued to any volunteer. Simply checking new edits for f-words is not enough. NVO (talk) 20:06, 30 August 2009 (UTC)
• Support. Cla68 (talk) 00:54, 2 September 2009 (UTC)
• Support Walkerma (talk) 02:00, 2 September 2009 (UTC)
• Oppose rollback is rollback. Icewedge (talk) 03:04, 2 September 2009 (UTC)
• Support, as a minimum. - BanyanTree 07:46, 2 September 2009 (UTC)
• Support for simplicity and because the criteria used to give someone rollback include general requirements that the user in question be able to distinguish constructive edits from those that are not, i.e. "has clue". The precise guidelines for when to flag/confirm a revision for the public are still being worked out, but the current proposal sets the bar in the same range as that for vandalism, i.e. it does not mandate that the reviewer check the verifiability of statements in general, but only of those statements that are potentially damaging to living persons. Pcap ping 11:52, 2 September 2009 (UTC)
• Oppose. These are different skill sets, with one having an arguably greater risk. Kevin (talk) 11:57, 2 September 2009 (UTC)
• Support As a starting point to get a group able to use this, rollbackers seem to meet the requirements that manual promotion would have. They have shown some level of clue and been manually looked at by an admin. If people believe that rollbacker has been given too liberally for this, the manual promotion will suffer the same complaints anyway. Jim Miller ^{See me | Touch me} 13:29, 2 September 2009 (UTC)
• Support Dy yol (talk) 08:46, 3 September 2009 (UTC)
• Support - If we're stuck with this flagged abomination, we need to make the number of people who can approve as broad as humanly possible. The rollbackers have already proven a certain level of trust to the community and can be trusted with this as well. Nutiketaiel (talk) 14:16, 3 September 2009 (UTC)
• Oppose. The functions are separate. Combining the uses will add to the conceptual baggage that weighs down flagged revisions (which has too much now). No need to do it. (It's fine to immediately give all rollbackers the reviewer capability as an initial manual action, but it shouldn't be an ongoing activity.) -R. S. Shaw (talk) 19:45, 3 September 2009 (UTC)
• Oppose per Protonk and R.S. Shaw. Ironholds (talk) 06:32, 5 September 2009 (UTC)
• Support per Dy yol, Walkerma and Cla68 SyG (talk) 10:25, 5 September 2009 (UTC)
• Support per PGWG above. Ks0stm ^(T•C) 15:46, 5 September 2009 (UTC)
• Support Any manual promotion for reviewers would very likely not go beyond what currently exists to get rollback permission (except perhaps a min edit count or time since registration). Joshdboz (talk) 02:29, 6 September 2009 (UTC)
• Oppose - I concur with MuZemike and neuro. Some trust is involved with rollback, but the two functions are entirely different. We often grant rollback to users who have been on Wikipedia for a couple of weeks, and have 200 edits of primarily vandal-fighting-only experience; while that kind of editor is fine with rollback, I'd never want them to be a reviewer since they probably do not possess the proper knowledge of WP:BLP and experience in working with content to make good decisions. We need people who know how to keep a good standard of quality in the mainspace, and some of our rollbackers simply don't. JamieS93 00:11, 8 September 2009 (UTC)
• Oppose per Jamie593 and others. In addition, as 2help pointed out below, "Per Wikipedia:Reviewing guideline, reviewers should be concerned with vandalism, copyright, legal threats, and BLP policy." That seems to be quite a high standard, and one that has very little to do with trusting an editor with a "undo in one click" function. Rd232 ^talk 08:29, 8 September 2009 (UTC)
• ""Oppose"" as reviewers would need a better knowledge of policy than mere vandal-spotting. --Bfigura ^(talk) 00:48, 11 September 2009 (UTC)
• Support Wikipedia is getting dangerously class-segmented as it is, what with all the bizarre privilege types running around. The requirements for rollbacker are substantively similar to the desired properties here. Furthermore, restricting the reviewer group too heavily turns this into an effectively censored encyclopedia. Ray^Talk 12:01, 14 September 2009 (UTC)
• Oppose Any and all forms of Flagged Revisions, our current system isn't broke why do we need to fix it. Feinoha ^{Talk, My master} 01:16, 15 September 2009 (UTC)
• Oppose autopromotion of any kind.—S Marshall ^Talk/_Cont 13:01, 15 September 2009 (UTC)
• Oppose per JamieS93 et al. –Juliancolton | ^Talk 14:58, 15 September 2009 (UTC)
• Support, but should have Administrators monitoring for first few months and Administrators should be able to drop them from the group Reviewers, not have to drop from both Rollbackers and Reviewers, if Reviewer right is:
  

a) Not Wanted

OR

b) Misused

OR

c) User has little knowledge of Wikipedia Policy(s)

(Same reason as top two and others, Rollbackers should already be trusted)

Also, to avoid serious problems which have been mentioned above, the edits should be stored somewhere similar to the history, and only viewable to another group (Or perhaps Reviewers?) Similar to the tool used by Administrators to see Deleted Pages, just perhaps only temporarily
Limideen 18:36, 6 January 2010 (UTC)

• Oppose. Only certain users are interested in rollbacking. Anti-vandalism is a very different process from actually reading and thinking about an article. It shouldn't be required as a "stepping stone" for people who really only want to do the reviewing, nor does it seem like a good idea to distract those actually willing to do such scut-work from the useful task they're doing now. Wnt (talk) 00:01, 1 February 2010 (UTC)

Autopromotion of other registered users

Please state "oppose", "support" etc, with rationale and sign with ~~~~

• Oppose; why, I hope is obvious. Stifle (talk) 19:05, 27 August 2009 (UTC)
• Oppose any autopromotion (since apparently this poll is going to be run again anyway... why the second bite at the apple???) ++Lar: t/c 20:22, 27 August 2009 (UTC)

Erm, because some people thought there wasm't enough participation first time round, there's been lots of media publicity which should hopefully mean lots of participation and we need to agree the criteria anyway, so I though easiest thing would be to run it all together. I may be wrong of course! AndrewRT(Talk) 00:24, 28 August 2009 (UTC)

What people thought this? Where was this said? It almost seems like you created this new poll out of the blue, although perhaps I missed it. Unless you think there is going to be a different outcome, 50 people is plenty. This is a waste of time and effort, in my view. ++Lar: t/c 01:31, 28 August 2009 (UTC)

• Oppose See first poll for rationale AndrewRT(Talk) 00:24, 28 August 2009 (UTC)
• Support The flagged revisions proposal imposes a high level of prior restraint upon Wikipedia. That prior restraint should be automatically turned off once a user has enough history that we can WP:AGF about them (instead of the what I might believe is the 'new standard' with flagged revisions: WP:ABF). I would support requiring users who have been blocked more than once to formally apply for permission, but everyone else should be automatically promoted after a reasonable amount of time and a reasonable number of edits. Katana0182
• Oppose - No autopromotion. Wknight94 ^talk 02:35, 28 August 2009 (UTC)
• Oppose any autopromotion. ≈ Chamal ^talk ¤ 02:49, 28 August 2009 (UTC)
• Oppose - Oppose any autopromotion. This is a job for humans. — neuro^(talk) 03:33, 28 August 2009 (UTC)
• Oppose because (slightly) lesser rights like rollbacker are not granted automatically. A large number of edits and/or having an account for a long time could be easily be used to game the system if the edits are trivial. Granted, one can fool an administrator too and engage in malicious use of the revision rights after obtaining them, but it seems to me less likely. Pcap ping 12:46, 2 September 2009 (UTC)

• Support but only with very high standards, I recommend a minimum of 5000 edits and 1 years service, and a clean block log. -- œ^™ 04:58, 28 August 2009 (UTC)
• Strong Oppose - Reviewers should show that they understand Wikipedia policy regarding which edits are acceptable; not only a willingness to edit. 2help (message me) 05:27, 28 August 2009 (UTC)
• Oppose ---- Luk ^talk 12:58, 28 August 2009 (UTC)
• Support – For me, I'd rather not have things complicated further by setting up some miniature version of RFA or something to determine if a user should be able to have the reviewer privilege. MuZemike 16:35, 28 August 2009 (UTC)
• No. Glacier Wolf 00:24, 29 August 2009 (UTC)
• Support, as per Katana0182. With a required condition that they haven't ever been blocked.Chhe (talk) 04:10, 29 August 2009 (UTC)
• Support if the level is high enough. For example, 6 months and 5,000 edits & no blocks. That's enough to make a credible attempt at Admin., even.
• Neutral. There is potential to game the system here, however the criteria are defined. As with the assignment of rollback rights, human discretion probably works better, and when it goes wrong, accountability should be in place so that editors learn from their mistakes. Geometry guy 21:56, 29 August 2009 (UTC)
• Weak Support We just need to weed out the vandals and those unfamiliar with policy. Definitely needs to be higher than autoconfirmed though. --Cybercobra (talk) 01:02, 30 August 2009 (UTC)
• Weak oppose—while I like the idea of making simple the process of giving out this right, I don't think that autopromotion is a sound system. While we should be proactive in making sure that the permission is not overly hard to obtain, trust cannot be measured by edit count, block log, etc. Trust must be given by a human: manual promotion is quite desirable in that respect. That being said, a reasonable autopromotion threshold is fair if the right can also be granted and revoked manually. {{Nihiltres|talk|edits}} 04:43, 30 August 2009 (UTC)
• Oppose. Hazardous tools must be issued only to qualified volunteers, and never issued to those who don't want or don't need them. NVO (talk) 20:08, 30 August 2009 (UTC)
• Using database reports to detect potential candidates and letting admins grant those rights at their discretion should work fine. Cenarium (talk) 23:20, 31 August 2009 (UTC)
• Oppose. Human discretion is probably the way to go. Kaldari (talk) 18:34, 1 September 2009 (UTC)
• Oppose. Some discretion is necessary. Cla68 (talk) 00:55, 2 September 2009 (UTC)
• Support The autopromotion of basically everyone. Trust should be implicit, but that trust should be easily revoked.
  — V = I * R (talk to Ω) 02:14, 2 September 2009 (UTC)
• Support along the lines of Katana0182 above. - BanyanTree 07:46, 2 September 2009 (UTC)
• Oppose, unless the minimum criteria is along the lines of DGG's suggestion below. Kevin (talk) 12:09, 2 September 2009 (UTC)
• Oppose. Open for abuse. feydey (talk) 19:50, 2 September 2009 (UTC)
• Support - Again, if we're stuck with this flagged abomination, the list of people who can approve should be as broad as possible. Every registered user should have the power to approve edits untill this flagged test fails and we go back to everyone being able to edit. Nutiketaiel (talk) 14:18, 3 September 2009 (UTC)
• Support. The biggest reason to do this is to avoid adding to the impression that Wikipedia is controlled by a few elite editors, and that no normal person need bother trying to edit. Yes, the system could be gamed -- big deal, it won't be widespread and cleaning it up not a huge thing. Admins can remove the reviewer capability whether autopromoted or not. We can adjust the promotion criteria as experience indicates. -R. S. Shaw (talk) 19:57, 3 September 2009 (UTC)
• Support After a user has shown enough edits without block logs he/she should be auto promoted. The idea that I have to wait till someone reviews my edit then it can becomes trusted makes Wikipedia very restrictive to me with a User having to battle till he gets Reviewer status. This would certainly reduce moral and general edits to Wikipedia.--Diaa abdelmoneim (talk) 21:23, 3 September 2009 (UTC)
• Oppose Just to be a registered user is not enough to be trusted. There should be some kind of additional criteria like number of edits. SyG (talk) 10:28, 5 September 2009 (UTC)
• Support with a min edit count, min time since registration, no blocks, and a manual recourse for those who for one reason or another are not autopromoted but wish to become reviewers. Joshdboz (talk) 02:34, 6 September 2009 (UTC)
• Neutral. Not opposed to auto-promotion in principle, but it depends too much on the specific criteria, as well as the ease with which it can be removed. Rd232 ^talk 08:30, 8 September 2009 (UTC)
• Support. The group with approval rights should be as broad as possible. Exactly where to put the cutoff, I don't have a good suggestion. —Quasirandom (talk) 20:41, 9 September 2009 (UTC)
• Oppose Oppose auto-promotion for reviewers. Manual approval would be better here.--Bfigura ^(talk) 00:45, 11 September 2009 (UTC)
• Oppose autopromotion of any kind.—S Marshall ^Talk/_Cont 13:01, 15 September 2009 (UTC)

Presumed criteria for manual promotion

Promotion at WP:Requests for permissions is normally decided based on presumptive criteria - i.e. users with a certain number of edits and/or history of editing would normally be promoted. This is a presumption which may be ignored if, for instance, an editor has a history of vandalism.

Please state the criteria you would support, for instance, "200 edits, 3 months", with rationale and sign with ~~~~

• User has clue. Stifle (talk) 19:05, 27 August 2009 (UTC)
• Per Stifle. (although measuring clue with numbers is problematic, I know it when I see it) ++Lar: t/c 20:23, 27 August 2009 (UTC)
• Yep, no autopromote per above. Using WP:AWB to correct 10,000 spelling mistakes does not qualify someone to review content viability. I've done thousands of spelling fixes my own self, and those alone do not qualify me to verify a modified quantum mechanics formula. In fact, nothing qualifies me to verify a modified quantum mechanics formula and I hope by now I've gained enough trust that people know I wouldn't try to verify a modified quantum mechanics forumla. I just ain't that smart. Wknight94 ^talk 02:42, 28 August 2009 (UTC)
• I wouldn't like to set out any presumed criteria. I guess the 'presumed criteria' is that the person giving out this flag simply trusts the user with it. Making it more complicated than that seems like a silly thing to do. — neuro^(talk) 03:36, 28 August 2009 (UTC)
• Just like rollback is given to users who have shown that they fully understand what vandalism is, reviewing privileges should be given to users who understand which edits are acceptable to the encyclopedia (such as BLP policy). 2help (message me) 05:37, 28 August 2009 (UTC)

Comment: AutoWikiBrowser and NewPageWatcher both have presumed criteria of 500 edits (here). I thought I read somewhere that rollback had presumed criteria too, but I can't find that. AndrewRT(Talk) 00:33, 29 August 2009 (UTC)

• Admin discretion. Glacier Wolf 00:23, 29 August 2009 (UTC)

Comment: If we don't do autopromotion, admins are going to have to decide who they promote and who they decline. At the moment, do they have enough guidance on the required level to be able to decide? I say that becase when I applied I was knocked back saying wait for the criteria to be developed. AndrewRT(Talk) 00:32, 29 August 2009 (UTC)

What I mean is that we should have general guidelines. I hate things getting set in stone. Besides, some special cases do arise, and it is generally good for admins to have a little wiggle room like that. Glacier Wolf 00:44, 29 August 2009 (UTC)

• Make it up as we go along. This is, presumably, a test afterall. Nifboy (talk) 05:29, 29 August 2009 (UTC)
• Five Pillars. In my contributions here I have been amazed by how rarely other contributors refer to the five pillars. They are not that difficult to understand, yet they inform everything we do here. Any editor who understands what the five pillars are for and why they say what they say has enough clue to be both a rollbacker and a reviewer (in this implementation). Geometry guy 22:00, 29 August 2009 (UTC)
• While I don't have specific criteria in mind, I endorse 2help's comment above. Users essentially need clue. Admin discretion might work, and 500–1,000 edits without a block seems like it would be a reasonable bar that such users could be given the tool on a good-faith basis. {{Nihiltres|talk|edits}} 04:43, 30 August 2009 (UTC)
• Niftboy said it. To me, today, this is definitely not a single admin's call; I'm for a public vote. But see what time will tell us. NVO (talk) 20:10, 30 August 2009 (UTC)
• At admins' discretion, normally any good-faith user who has been around and active enough and who, based upon a review if the admin doesn't know the user before, should know what is vandalism, a BLP violation and other bad content. Admins can use database reports to detect candidates. Cenarium (talk) 23:33, 31 August 2009 (UTC)
• Many of the above comments are worrisome because they suggest misplaced emphasis and problematic underlying attitudes. First, the capability to be granted is minimal: just ability to make a checkmark that somebody who has been around awhile with no significant transgressions has glanced at the changes and seen no obvious crap (not whether a quantum mechanics formula is right or not). If a determined vandal has gotten the capability, there will not be major damage before he is unmasked. Second, having promotion left to admin discretion and making it up as we go gives the appearance of a ruling class trying to reserve to itself more powers, not because those powers are important but just to be able to feel good about being able to exercise them. Wikipedia needs to fight such appearances because it discourages participation. For this minor capability, there is no need burden Wikipedia's reputation with an additional bow to centralized authority. -R. S. Shaw (talk) 21:04, 3 September 2009 (UTC)
• I'm for "auto with discretion", based on 1000 edits, 6 months, no blocks as the auto threshold - identifying established editors with enough experience to do the job. Promotion could be applied manually for a variety of reasons, eg an editor having an edit warring block, or making lots of contributions and learning the ropes in 3 months. Promotion could also be suspended manually, eg if someone amasses 1000 edits with automated tools or otherwise looks like they're not ready. But the numeric threshold gives an idea of the experience level admins should be looking for in manual promotion. Rd232 ^talk 08:39, 8 September 2009 (UTC)

Criteria for autopromomotion

For information: WP:RFP requires 500 edits for someone to be given AutoWikiBrowser or NewPageWatcher functionality

Please state the criteria you support, for instance, "200 edits, 3 months", with rationale and sign with ~~~~

• 1 day Why delay? Taku
• 25 edits 1 week Seems fine, especially if we require a certain number of edits (25ish?). Vandals will either not make enough edits, or will be caught. I find the possibility of "sleeper" vandals unlikely, easily dealt with, and not able to be deterred by any reasonable wait period. 25 edits and 1 week seems to be what I've set in my mind. This should weed out any vandals, and let new users start editing in full quickly (and therefore hopefully continue to do so). Falcorian

I stick by this. I believe the energy the average vandal will invest is very low, below 25 good edits. However, I also worry that new editors (before they're hooked ;-) ) will have little patience for waiting for their edits to show up. Hopefully 25 isn't too onerous for the good contributors to overcome. --Falcorian ^(talk) 00:43, 28 August 2009 (UTC)

• 25 edits 1 week Knowing that what you write is seen by the whole world immediately is an important motivating factor for editors. It is what makes Wikipedia successful. Let's not erase that factor for a small benefit. Really, imagine a guy would make good edits during one week and then suddenly start vandalizing pages: He would just get banned and reverted, not a big loss, and very few people would take the pain to do this. With this "one week" setting we eliminate 99% of the vandals, and we keep the motivating factor for 99% of the editors. With a "3 months" setting we eliminate 99.5% of the vandals, but we also eliminate the motivating factor for 50% of the editors. (figures are just my guess). Knowing that what you write is seen by the whole world immediately is an important motivating factor for editors. It is what makes Wikipedia successful. Let's not erase that factor for a small benefit. Really, imagine a guy would make 25 good edits and then suddenly start vandalizing pages: He would just get banned and reverted, not a big loss, and very few people would take the pain to do this. With this "25 edits" setting we eliminate 99% of the vandals, and we keep the motivating factor for 99% of the editors. With a "200 edits" setting we eliminate 99.5% of the vandals, but we also eliminate the motivating factor for 50% of the editors. (figures are just my guess) Nicolas1981
• 50 edits 1 week 4 days is too little, but I think this is good enough to keep out most vandals and I suppose the flagging itself will be reviewed at some point, at least by the administrator removing the protection. That's the number I could rack up in one month and it's a lot for a new user. One can't expect new users to huggle to prove their worth. Admiral Norton
• 100 edits 3 months Long enough to gain some experience and too long for vandals to bother waiting. Just enough to prevent sleeper accounts. Apoc2400
• 100 edits 3 months Fahadsadah
• 200 edits Sounds about right. Walkerma
• 200 edits II
• 200 edits 3 months JoshuaZ
• 500 edits Darrenhusted
• 500 edits 3 months This plus the edit count of something like 500 (or close) would be good enough to find the people that can be trusted with it. Anyone can pile up a few hundred edits using something like huggle in a matter of days. But that wouldn't give him an understanding of the problem here. I think we should allow at least 500 edits to get enough experience (+3 months, as I have said there) so that the user has time to look around the place and understand the Policies and Guidelines properly. Chamal
• 500 edits 3 months Plus 500 editcount, or 200 editcount non-automated. Roux
• 500 edits 3 months Reasonable period. I agree with Chamal N. Hans Adler
• 500 edits 3 months Shows if they will be an Active Editor or an Active Vandal. 500 Article edits should be enough to show knowledge of Policies and Guidelines. Exit2DOS2000
• 500 edits 3 months Just right. utcursch
• 500 edits 3 months Seems reasonable to me. Sounds good if it is 500 non-reverted edits. -Mav
• 500 edits 3 months Reasonable enough. Long enough to be a good editor, yet not too long, as they might as well be admins by then. Sounds good to me. 500 edits is a good standard. Unionhawk
• 500 edits in 50+ articles, 3 months I can name axe-grinders who have made 5000+ edits to a single article. Good editors get around a little bit. - Manning
• 500 edits 6 months 500 edits over 6 months ought to be enough to prevent gaming the system and allow the older accounts to help ponctually without having to formally request the right (a manual check of the accounts newly meeting this criteria could also be done). A procedure should be created to grant it earlier, though. 500 edits over 6 months ought to be enough to prevent gaming the system and allow the older accounts to help ponctually without having to formally request the right (a manual check of the accounts newly meeting this criteria could also be done). A procedure should be created to grant it earlier, though. lucasbfr
• 500 edits 6 months Standards should be high (but still less than that of becoming a sysop, which is normally around 1 year). It should be longer than those who operate on a Tor network, which is currently 3 months. 500 edits sounds fair. Combine with my proposal for 6 months, it shows to those "slow editors" or those editors who emphasize on quality of edits over quantity of edits that they are more than capable of reviewing edits. MuZemike
• 500 edits 6 months Editors should have had enough time to get a strong understanding of Wikipedia before they upgrade to the "Reviewers" class. 500 edits is not that much if you are a newbie that needs to get accustomed to the inner behaviour of Wikipedia, and understanding its most basic guidelines. SyG
• 500 edits 6 months <= 1 block A reasonable amount of time and a reasonable number of edits to discourage griefers. I prefer time more than edits. I think a block criterion would be reasonable as well. 1 block is allowable for over-enthusiasm or exuberance, but when editors start racking them up, this might be a sign of problems that a manual process might be more appropriate for. Katana0182
• 500 edits Seems a reasonable amount of editor experience to know WP policy and to keep trolls out. Cybercobra
• 500 edits Will be enough to discourage vandals, but enough to encourage real editors to stay. Darrenhusted
• 1000 edits I believe it really took me a year to fully understand and appreciate the various duties, issues and obligations that being an editor entails. We need to give people time to trip over the various discussions, policies, standards and practices... Mjquin_id
• 5000 edits, 1 year, and a clean block log. Adminship criteria pretty much. -- œ^™ 05:04, 28 August 2009 (UTC)
• 10,000,000 edits 1,000 years As an alternative to "never". Lar
• Oppose all autopromotion, general bad idea. Stifle (talk) 19:05, 27 August 2009 (UTC)
• Oppose all autopromotion per my comment in the previous section. Wknight94 ^talk 02:42, 28 August 2009 (UTC)
• Oppose all autopromotion - See my comments above. — neuro^(talk) 03:37, 28 August 2009 (UTC)
• Oppose all autopromotion per the reasons I've given above; a willingness to edit is not sufficient to gain reviewer status. 2help (message me) 05:39, 28 August 2009 (UTC)
• 500 edits, 3 months and rollback - Edits plus time should mean that they understand our policies, having the rollback bit should mean that an administrator has already looked at them (at some point), and judged them to be less likely to be a threat to the project, and that they have experience in looking at recent changes (judging by the criteria that some administrators place on granting rollback). To my mind this is a possible compromise that ensures a time component (which some seem to want), plus a human component (which others want), without adding a new level of bureaucracy and more busy-work for the admins. PGWG (talk) 17:20, 28 August 2009 (UTC)
• 500 edits, 3 months, and rollback - once again, if a rollbacker can be trusted to decide which edits are inappropriate, they can also be trusted to know which edits are appropriate.--Unionhawk ^{Talk E-mail} 15:04, 29 August 2009 (UTC)
• 500 edits, 3 months, and rollback per PGWG and Unionhawk. Ks0stm ^(T•C) 15:50, 5 September 2009 (UTC)
• 5,000 edits, 6 months, no blocks, activity within the last two months. That's enough to make a credible bid for adminship. The suggestions of 500 edits are way too low--some people know the rules by then, but many people do not. DGG ( talk ) 16:36, 29 August 2009 (UTC)
• I'm generally against autopromotion, but an ungamed statistic of a few hundred edits over several months is the ballpark I would aim for here, because it gives an editor the opportunity to experience the wide range of our activity. Geometry guy 22:06, 29 August 2009 (UTC)
• Weak oppose, but… 500 edits with no blocks seems like a reasonable baseline if it has to go that way. Numbers can always be gamed. I don't want to rule out autopromotion, but I don't think it's a good idea. {{Nihiltres|talk|edits}} 04:43, 30 August 2009 (UTC)
• Strong oppose, but, perhaps, bureaucrats and higher-ranking deities may get it ex officio; but not the admins. Never. NVO (talk) 20:14, 30 August 2009 (UTC)
• 10,000 edits, 12 months, no blocks, activity within the last two months, and an official endorsement from the pope. No one expects the Spanish Inquisition! Kaldari (talk) 18:32, 1 September 2009 (UTC)

This will not work due to this. Sorry. :( 2help (message me) 19:06, 1 September 2009 (UTC)

• Support 200 edits, 3 months (JoshuaZ's idea above). Cla68 (talk) 00:57, 2 September 2009 (UTC)
• Autoconfirmed users who have not been blocked, subject to liberally-applied human review for rights removal. (Note to all those opposing autopromotion in this section: this is the section for pro-autopromotion users to outline their ideas. The section for those against all autopromotion to discuss their ideas for manual promotion is under Question 3. You're double voting if you're in both.) - BanyanTree 07:46, 2 September 2009 (UTC)
• Autoconfirmed users - Again, let's make this as broad as possible. Nutiketaiel (talk) 14:20, 3 September 2009 (UTC)
• 1000 edits, 6 months, no blocks. This may be too high or perhaps too low, but seems like a good starting point, and probably sounds pretty much like "experienced editor" to those not versed in wikilore. -R. S. Shaw (talk) 21:31, 3 September 2009 (UTC)
• 1000 edits, 6 months, no blocks - with caveats. This would be the auto threshold, identifying established editors with enough experience to do the job. Promotion could be applied manually for a variety of reasons, eg an editor having an edit warring block, or making lots of contributions and learning the ropes in 3 months. Promotion could also be suspended manually, eg if someone amasses 1000 edits with automated tools or otherwise looks like they're not ready. PS I could also live with PGWG's criteria, which several others have agreed with. (Theoretically there could even be multiple "packages" of criteria for autopromotion, eg if the editor has rollback, 500 edits and 3 months is enough for auto-promotion.)Rd232 ^talk 08:36, 8 September 2009 (UTC)

Other discussion

• I suggest rather loose criteria during the trial, and then have the real discussion at the end of it. --Apoc2400 (talk) 17:56, 27 August 2009 (UTC)
• I suggest this poll not be run again, the previous one was sufficient. ++Lar: t/c 18:04, 27 August 2009 (UTC)
• Agree with Lar. What purpose does this serve? → ROUX ₪ 18:26, 27 August 2009 (UTC)
• I agree w/ the above. Autopromotion defeats the "purpose" of flagged revisions. Autopromoting people w/ rollback is worse than nothing because it bundles another tool with a userright that is supposed to be the new "no big deal". Rollback is handed out if an admin sees an editor knows how to revert vandalism. That's not necessarily the same job as reviewing a revision. Protonk (talk) 18:42, 27 August 2009 (UTC)
  • I'm not sure another poll needs to be run given that the result will likely be the same as the previous. The primary thing I'm curious about is whether any form of compromise might be possible. I'm incidentally curious about what other projects with flagged revisions have done. On the English Wikinews for example one needs explicit permission but that's a much smaller project. Can someone maybe fill us in on what .de does in more detail? JoshuaZ (talk) 18:45, 27 August 2009 (UTC)
  • It depends on what one expects from the reviewers. Is it supposed to be absolutely certain that no crap gets through, or is it enough that the edit has been checked by an other person that probably isn't a total dofus? Lower requirements means more reviewers means we can flag protect more articles without getting a backlog. It's a tradeoff between checking fewer articles more carefully or more articles but with more crap getting through. That is why I prefer to have the trial first to see how many reviewers we need. I also concur with JoshuaZ about checking how it works on de.wp. --Apoc2400 (talk) 19:03, 27 August 2009 (UTC)
    • Is it correct to compare this with rollback? Rollback is given to people who are active vandal fighters. This thing is different. ≈ Chamal ^talk ¤ 02:46, 28 August 2009 (UTC)
      • Oh, yeah. Protonk mentioned that already :P ≈ Chamal ^talk ¤ 02:47, 28 August 2009 (UTC)
    • I'd also like to know about how the German Wikipedia handles this. Also, can someone direct me to the Flagged Revision page on the German Wikipedia so I can use Google Translate and read some of the details on it? Mahanga^Talk 21:05, 2 September 2009 (UTC)
    • Ok, found it. Here's the German flagged revisions page (translated). A user is auto-promoted to reviewer status after being a member for 60 days, 300 edits, empty block log. Poland's is 500 edits, 90 days, no blocks. Mahanga^Talk 21:19, 2 September 2009 (UTC)

My understanding was that Flagged Revision was only going to be used for vandalism, and not content, and so any revision that was clearly not vandalism and didn't violate BPL sourcing requirements was to be sighted. Is this still the idea? --Falcorian ^(talk) 11:45, 28 August 2009 (UTC)

Personally, I'd like to see it go further. Subtle misinformation is a frequent problem here (how many times do you see IPs changing 1943 to 1947 and such?) and is arguably more poisonous than blatant "poop" vandalism. Wknight94 ^talk 12:13, 28 August 2009 (UTC)

Subtle misinformation is also a problem that flagged revisions is less suitable for. --Apoc2400 (talk) 13:07, 28 August 2009 (UTC)

Unfortunately it is a problem that is handled poorly under our current system and was a big selling point for FR on this wiki. My strong suspicion is that FR will not look a lot like the huggle first line of defense. Rather it will look like what NPP is supposed to look like, editors handling a queue of revisions and dealing with them as needed. Since reviewers will have the benefit of time and sighting a revision will make it harder to detect down the road, FR will probably be pushed into reviewing changes for subtle vandalism or misinformation. Protonk (talk) 00:48, 29 August 2009 (UTC)

When a reviewer subscribes to police the article, he/she should subscribe to just this, watching every bit of content, including the invisible stuff and each external link. Anything less will fail BLP compliance. In theory, this should have worked before through ordinary editors' watching their fields of interest, but it did not. Time will tell. NVO (talk) 20:22, 30 August 2009 (UTC)

That shi t p has sailed, since it is clear that we will get flagged revisions, but the statement NVO just made encapsulates my fears about the limited upside FR will have on en.wiki. The missing ingredient remains constant diligence and elbow grease. Protonk (talk) 00:46, 31 August 2009 (UTC)

OHHH! Whoops. That ship has sailed. Wow. The p and the t are so far apart. How did I do that? Protonk (talk) 21:28, 8 September 2009 (UTC)

The first version was spot on... I'm more concerned about the verb. Has sailed? When? Sometimes it seems that FR is already on, but nobody really knows where and how... an undercover test. They're watching. Doctor, they're everywhere, duck and cover! NVO (talk) 02:13, 9 September 2009 (UTC)

Per Wikipedia:Reviewing guideline, reviewers should be concerned with vandalism, copyright, legal threats, and BLP policy. 2help (message me) 16:22, 28 August 2009 (UTC)

I know this is probably being taken care of, but I just want to remind people that User-Warnings will need to be created associated with Reviewers
Limideen 18:48, 6 January 2010 (UTC)

Also, surely, Rollbackers will have no job now, as every edit is checked, so it makes sense to give them another one automatically as a Reviewer!
Limideen 18:48, 6 January 2010 (UTC)

Eyeball request

(discussion merged from Help talk:Pending changes review process)

• You understand you should not use the tool where it might lead to concerns over bad faith by uninvolved administrators (for example, in a dispute, aim to check appropriate edits by users you do not agree with, not just those you do)

Are there any situations where the tool could be used in bad faith, given it's an affirmative tool only (not a restricting one), other users can also check, no obligation to check anything, etc? And does this bullet over-expose reviewers to wikilawyering and aggressive claims of bad faith, given a reviewer automatically checks their own edits and may simply not check others' all the time?

Apart from "serious or repeated poor use" (such as approving vandalism or junk) this is the only non-blatant bad-faith case I can immediately think of. Discussion of this bullet? Is it needed, can it be improved? FT2 ^{(Talk | email)} 08:55, 5 June 2010 (UTC)

I think it would be better to just say not to use it on articles where you're involved in a dispute at all. As for automatic reviews, these are specifically noted as automatic (see example). It is also possible to "unreview" an edit. I'm not sure exactly what the question here is, but in general, the only ways to misuse/abuse this would be to A) review an edit that you know is bad (this could include approving vandalism/spam or POV/BLP issues) or B) unreview an edit that you know is good (this would mostly apply to disputes). Mr.Z-man 19:04, 6 June 2010 (UTC)

Merge with Wikipedia:Reviewers

(discussion merged from Help talk:Pending changes review process)

This page (or sections of it) should probably be merged with Wikipedia:Reviewers. -- RobLa (talk) 01:31, 8 June 2010 (UTC)

Maybe in a page Wikipedia:Reviewing, where we can also merge Wikipedia:Reviewing guideline ? The technical info on the usergoup can also go to WP:UAL. Cenarium (talk) 14:06, 9 June 2010 (UTC)

Sold! I'll start consolidating. -- RobLa (talk) 18:29, 11 June 2010 (UTC)

Proposal to rename Autoreviewer to Autopatroller

Proposed here. Cenarium (talk) 16:27, 12 June 2010 (UTC)

Removal of rights

I have reverted Joopercooper's edit to this policy which purported to limit removal of reviewer rights to ArbCom and state that admins removing reviewer rights would be liable to be desysopped. I have been unable to find consensus for this major change to the page. If consensus has been reached somewhere, please kindly point out where and feel free to revert me. Stifle (talk) 19:57, 12 June 2010 (UTC)

There has been a bit of discussion here, at WP:AN, User talk:Risker and a few other pages, no new consensus formed. Cenarium (talk) 20:32, 12 June 2010 (UTC)

This guideline is still under development though, only the grand lines have been given in the initial proposal. Cenarium (talk) 21:34, 12 June 2010 (UTC)

Would you like to expand stifle on why you think that rights, which users currently enjoy, should now be able to be unilaterlly removed by administrators without the fullest of processes? --Joopercoopers (talk) 19:24, 13 June 2010 (UTC)

Unless you can explain why you have removed this, I think it needs to be reinstated, Stifle. Removing this is the editorial equivalent of desysopping, and it darn well needs more than just a random admin to decide whether or not a review decision was correct. Ideally, the editors who regularly maintain an article will be doing the reviewing, and they understand the content better than a drive-by admin. Because of the configuration of this extension, editors are forced to accept edits that could be wildly inappropriate, revert them, and then actually edit the article. This is precisely the opposite of what I would have hoped for because it makes it easier to make an error than to do things correctly. Removal of this permission is a major disenfranchisement of editors, and must be done at a higher level than just a single administrator. Risker (talk) 19:49, 13 June 2010 (UTC)

This is exactly what I said would happen, the likes of Stifle, overbearing and overpowering the content editors! - and to think , this is ony the start.  Giacomo  20:03, 13 June 2010 (UTC)

As strange as this may be, I agree with Giano. I don't think we need to say that only Arbcom can remove the right - as was discussed at Wikipedia:Requests_for_comment/Pending_changes_trial#Removing_reviewer_rights, there could be a legitimate reason to remove reviewer rights of someone who misuses it (eg, accidentally/incorrectly reviews vandalism, but is otherwise a good-faith editor), but inappropriate removal can and should result in summary desysopping. I can easily see, without the right protections being in place, this tool being used to enforce a POV on articles or some such thing. --B (talk) 20:29, 13 June 2010 (UTC)

I'm curious as to where the consensus for your preferred version was achieved, Risker. Blocking a user is a major disenfranchisement of an editor as well, and that is entrusted to Admins. Apparently it takes only a single admin to grant a right as well. If the reviewer ability is considered of greater importance than the ability to edit itself, then I would agree with the idea that more than the opinion of a single admin should be necessary. I am not, however, particularly enthusiastic about handing ArbCom more power. The community should rule, not an oligarchy. Resolute 20:30, 13 June 2010 (UTC)

• Why discussions are running in multiple places? Sole Soul (talk) 20:33, 13 June 2010 (UTC)

Whatever nascent consensus may existing, I'd suggest it should be born here - perhaps we might focus on arguments for and against rather than the circular versions of whether there is one at the moment or not - consensus can change, and it's not as currently displayed on this trial policy page for me. --Joopercoopers (talk) 20:33, 13 June 2010 (UTC)

<-- "Administrators should note that reviewing rights should not be removed without the approval of Arbcom and desysop is a likely consequence of vexatious removal." WHAT? I have to admit when I saw this get removed I thought it was quite obvious that stifle should remove it. This is supposed to be no big deal, given to trusted users to help fight vandalism (like rollback) even the description of "when to give it out" states that the level of trust is similar to rollback. Just like any admin can take away rollback if they think the person has abused it then any admin should be able to take away reviewer if they think someone deserves it. If not then why are admins able to GIVE reviewer? If that is going to be the case then we should be having some sort of community approval method for the flag (which is ridiculous) When in the world did this become equivalent to sysop? (or at least removing it is equivalent to desysop) this totally changes how I would ever give it out. Am I at least allowed to remove the ones I've given out so far since I now now I'm not allowed to remove it without going through a crazy process? We have to totally redo how we are giving out reviewer rights if this is how you have to remove them, the assumption (and how it is written) is reviewer is much closer to rollback. Currently you are saying it is closer to sysop. Is it just confusion that removing reviewer removed the fact that their edits get "auto patrolled/sighted"? The vast majority of pending changes get autoapproved if all you have is autoconfirmed. James (T C) 21:10, 13 June 2010 (UTC)

Removal exclusively by ArbCom is just absurd. Admins can remove rollback, autoreviewer, IP block exemptions and confirmed flags for a single misuse and can prevent someone from editing altogether via blocking, none of which requires any involvement from ArbCom, so why on Earth should this? Like controversial blocks and any other admin actions, it should be reviewed on ANI if need be.

I also agree with James above- if I'd known this before I started giving the flags out, I wouldn't have done so and I should be allowed to remove those I've granted so far if this to be how we handle removal. That said, though, I agree desysop is in order for malicious removal (as it would be for protecting an article you're edit warring on or a malicious block etc). HJ Mitchell | Penny for your thoughts? 21:26, 13 June 2010 (UTC)

"Admins ... can prevent someone from editing altogether via blocking" but they cannot topic ban a user, and there is a reason for that. Sole Soul (talk) 22:51, 13 June 2010 (UTC)

Umm, no. Then you clearly don't understand what this permission or protection level is about, Jamesofur and HJ Mitchell. It is about controlling *all* edits to the articles. Go and play at the testwiki, and you'll see how it is nearly impossible to approve some but not all pending changes, and how incredibly easy it will be to introduce the very vandalism we are seeking to keep out of articles, accidentally, by good editors. This is completely unlike rollback or account creator or edit filter editor, the other permissions that can be handed out by admins, because it is focused completely on content. A single administrator should not be capable of removing this permission from longstanding editors. Risker (talk) 21:31, 13 June 2010 (UTC)

Then establish the policy that for a one-time accidental mistake you don't remove it - only for repeated carelessness or intentional misuse. Micromanagement by arbcom is not the solution. --B (talk) 21:35, 13 June 2010 (UTC)

The problem, B, is that it will be the judgment of an admin as to whether or not the edit should have been accepted or not. And frankly, too few of our admins who are interested in handing out (and withdrawing) permissions are knowledgeable enough about content to make that decision. Realistically, we haven't created the rules that need to go with this. If someone rejects an edit that isn't obvious vandalism, is it automatically a bad rejection? If an anon rewords a sentence and the edit is rejected, is the rejecting editor going to be called on "misuse"? We're crossing a line here that nobody has given enough thought to or discussed adequately with respect to routine editing decisions. Risker (talk) 21:55, 13 June 2010 (UTC)

Risker, again, reviewers cannot reject edits, they can only accept edits or edit as usual. They are not 'forced' to accept bad edits, they can 'revert', by rollbacking or otherwise, bad edits. I've tried to clarify this here. The only possible misuse or abuse of this tool could be in accepting obvious vandalism, or un-accepting a previously accepted good edit which would make no sense because reviewers would see it again in the queue and re-accept it. Cenarium (talk) 22:45, 13 June 2010 (UTC)

Sounds like weak software design to me. Gwen Gale (talk) 21:39, 13 June 2010 (UTC)

God, no kidding. Risker just successfully argued that this right is poorly designed and its implementation is being rushed. Sounds like the problem of how we determine when to remove the right will be the least of our worries. Resolute 21:43, 13 June 2010 (UTC)

I think leaving removal to arbcom would be harmful. If an admin mistakenly or otherwise removes the right, it can always be quickly restored by any admin. Gwen Gale (talk) 21:36, 13 June 2010 (UTC)

Yeah, right. Malleus Fatuorum 21:39, 13 June 2010 (UTC)

There are sundry worries about this. Gwen Gale (talk) 21:41, 13 June 2010 (UTC)

Too true there are some worries, we have all seen some jerk pursuade admins to make bad blocks and sully clean block logs, haven't we Gwen? This wil be just one more opprtunity for Admin abuse.  Giacomo 

Not only admin abuse, Giano, as you know. Gwen Gale (talk) 22:04, 13 June 2010 (UTC)

@Risker: If admins can't remove it, why should we be able to grant it?

I'm not talking about removing it from someone who accidentally lets in some vandalism, but if someone consistently exercises poor judgement- for example, by not checking edits closely enough and letting in edits that could easily have been kept out- after having been spoken to, then the right should be revoked. If we're giving it out like candy floss, then why do we have to go through a process equivalent to that for removal of admin rights for its removal? HJ Mitchell | Penny for your thoughts? 21:49, 13 June 2010 (UTC)

Well, HJ Mitchell, I *don't* think admins should be involved in granting it; instead I think it should be automatically granted by the software once an editor has reached a specific, very low threshold. Remember that as we speak right now, any autoconfirmed editor can edit a semi-protected article directly, and contrary to popular belief it is extremely rare that we have autoconfirmed editors vandalising semi-protected articles; and that no editor can edit a fully protected article without obtaining consensus on the edit to be inserted. I think one month and 50 edits is more than sufficient for someone to be a reviewer. Once autoconfirmed, the only effective way to remove that "right" is to ban an editor. Frankly, I'm not seeing much reason for that threshold to be lowered. Risker (talk) 22:01, 13 June 2010 (UTC)

How is the reviewer right both so mundane that you believe anyone with autoconfirmed status should have it, but so critical that you also think only ArbCom should be able to remove it? Resolute 22:17, 13 June 2010 (UTC)

Try removing autoconfirmed status from editors, and you'll understand what I mean, Resolute. Risker (talk) 22:23, 13 June 2010 (UTC)

And what about blocking then ? And admins can set the abuse filter to remove autoconfirmed, by the way. Cenarium (talk) 22:45, 13 June 2010 (UTC)

Let me throw something out there. Cenarium wrote on the RfC thread: "if people see it's too difficult to remove the right, then they will want higher standards to grant it", which I think is a legitimate concern. However, I'm also very sympathetic to those that want to make it difficult to remove this right. So, perhaps a compromise:

1. The admin who originally granted the Reviewer right has 60 days (or some reasonable probationary period) in which they may unilaterally remove the right.
2. Beyond that, there's a more process-bound path to revoking access. I don't have a strong sense of what exactly this should be, but I can understand the desire to limit a single admin's ability to pull access to the tool.

Whatever the process is, I think it should be optimized to ensure a very large pool of reviewers. -- RobLa (talk) 22:02, 13 June 2010 (UTC)

RobLa, there is no reason that this "right" should be difficult to obtain. It's equivalent to autoconfirmed at present, and that is incredibly easy to obtain, but nigh-on impossible to remove. Risker (talk) 22:17, 13 June 2010 (UTC)

I think your arguments make great sense. Thank you. Dr.K. ^{λogosπraxis} 22:26, 13 June 2010 (UTC)

(Unindent) I've been discussing a similar issue to Riskers over at ANI. Some commentators seem very clear this right is for vandalism and rubbish only. Other's seem to think it may cover more - eg when an addition is uncited. I can see a huge problem here - particularly on controversial pages, involving what are essentially content, sourcing and NPOV disputes. Fainites ^barleyscribs 22:03, 13 June 2010 (UTC)

As with rollback, the bounds of what can be rejected must be very tight. If the software is lacking in how it handles many edits waiting for review at once, it's too soon to take it live. Gwen Gale (talk) 22:06, 13 June 2010 (UTC)

I agree! But this thing is being rolled out in 1 hr 51 minutes without editors being clear what those bounds are.Fainites ^barleyscribs 22:09, 13 June 2010 (UTC)

Please stop comparing this to rollback. Anyone can edit any article with or without rollback, a tool that is essentially useless and is just a click and an edit summary different than "undo", which every editor including unregistered ones has access to. This tool directly affects content in articles. Risker (talk) 22:17, 13 June 2010 (UTC)

I was only saying that the bounds of what can be spit out must ver very tight. The same is true of rollback. I agree that rollback is trivial and this is not, which is why I'm talking the time to post here to begin with. Gwen Gale (talk) 22:45, 13 June 2010 (UTC)

I don't think it was being compared to rollback other than "tight bounds", but I would predict that unless it is strictly restricted to vandalism/rubbish/BLP violations only it will quickly become another source of content disputes and disputes over the use and abuse of tools. POV pushing sockmasters can fool friendly admins.Fainites ^barleyscribs 22:31, 13 June 2010 (UTC)

• It's a disgrace that this is being foisted upon is. It is a half baked schemed, ill conceived and ill considered. It has been hatched in some recess of the encyclopedia without any full and frank discussion and presented as a "fait accompli" Had it publicised beyond the few interested in BLPs the debate would have been very different indeed.  Giacomo  22:13, 13 June 2010 (UTC)

Half baked at best. Moreover I still don't like the level 2 feature, which would put the edits of autoconfirmed users into the review queue (it's meant to handle socks). The worry is, lots of high traffic, core articles get socks all the time and level 2 could easily be used as a handy tool to screen for unwanted PoVs/editors. Gwen Gale (talk) 22:18, 13 June 2010 (UTC)

I've opened a discussion on reconsidering use of level 2 at Wikipedia_talk:Pending_changes#Level 2 protection. Cenarium (talk) 22:29, 13 June 2010 (UTC)

It's not being foisted on me, because I'll be ignoring any article to which it's applied. Malleus Fatuorum 22:22, 13 June 2010 (UTC)

• Yes you and I both, but I am very alarmed looking at my watchlist to see it being conferred like a knighthood on quite a lot of people tonight. The new elite?  Giacomo  22:27, 13 June 2010 (UTC)

Hardly. It is being cast about liberally. Fainites ^barleyscribs 23:04, 13 June 2010 (UTC)

Point of fact here: this is being rolled out June 15 U.S. west coast time. We're currently shooting for 4pm PDT on Tuesday to start the rollout of the actual feature, assuming that the software rollout (24 hours earlier) goes well. From an en.wikipedia.org end user perspective, June 15 is the only date that matters. I'd like to point out that this is a trial that has an upper bound (2000) on the number of articles. -- RobLa (talk) 22:40, 13 June 2010 (UTC)

In laymens terms folks, what's being proposed? GoodDay (talk) 22:25, 13 June 2010 (UTC)

A kind of protection (not default) whereby edits by IPs and sometimes, autoconfirmed users, must be reviewed for vandalism, BLP vios, copyright vios and legal/PA/libel woes before showing up in article versions seen by users not logged on. Gwen Gale (talk) 22:30, 13 June 2010 (UTC)

I've no problem with it where IPs are concerned. But, I'm not certain what auto-confirmed users are. GoodDay (talk) 22:32, 13 June 2010 (UTC)

That's the theory GoodDay. The practice is that any unregistered or newbie editor who tries to edit a protected page has his edit appearing in the History only until "accepted" or "unaccepted" by someone with reviewer "rights". There is currently a lot of misunderstanding about what the review criteria are, hence concerns that it could be used to further content disputes, or just with poor judgement as to content.Fainites ^barleyscribs 22:35, 13 June 2010 (UTC)

The implementation of the IP side (level 1) looks to me like it's sloppy as to both software and policy. The autoconfirmed side (level 2) is, IMHO, bound to be gamed by those little "teams" of editors that lurk at high traffic humanities articles (like anything at all linked with politics). I'm very wary of this feature being put to anything other than BLPs, for starters and then only when it's ready. Gwen Gale (talk) 22:37, 13 June 2010 (UTC)

What's an auto-confirmed user? GoodDay (talk) 22:38, 13 June 2010 (UTC)

Here. Four days and ten edits. Fainites ^barleyscribs 22:41, 13 June 2010 (UTC)

A newly created registered account. GoodDay (talk) 22:43, 13 June 2010 (UTC)

As Im in favour of mandatory registration, I'll accept any restrictions on IPs. However, we should tread softly on newbie registered accounts, as we don't won't to scare anybody off. GoodDay (talk) 22:52, 13 June 2010 (UTC)

It would be very disheartening for a sincere newbie to have his perfectly sensible addition not appear and be "unaccepted" because it was uncited.Fainites ^barleyscribs 22:55, 13 June 2010 (UTC)

Mandatory registration would prevent such an incident. GoodDay (talk) 22:57, 13 June 2010 (UTC)

Not with level 2 (and I don't think mandatory registration will help anything). Gwen Gale (talk) 23:01, 13 June 2010 (UTC)

Review at Level 1, deserves a trial run. GoodDay (talk) 23:03, 13 June 2010 (UTC)

The only way it could be 'unaccepted' is by being reverted, which is no different than what can happen now, except the edit wouldn't have publicly appeared, but if the article were instead semi-protected, it couldn't have been made in the first place. Cenarium (talk) 23:35, 13 June 2010 (UTC)

Faintes - that's the least of our worries - we were sold this programme on the prospectus it would be a way of limiting libellous additions to BLP's - something I'm largely in favour off - but it's already become a way of limiting vandalism in general, and the slip will be uncontestible as it's rolled out over your FA's/GA's and ultimately any bloody article. Now I might even be able to conscience that slippage, but what I can't accept is a unilateral admin say so, over my, good standing, content contributions. So if Arbom removal is not acceptable to admin turkeys voting for christmas - the discussion needs to move to what might be - but it's got to be something a damn site better than 1 admin and his view on it. --Joopercoopers (talk) 23:24, 13 June 2010 (UTC)

What's wrong with an admin exercising judgement (as with blocking, for example) and then we review it on ANI- if the consensus is that it was a bad call, the right can be re-granted. It's not difficult to add- it's 3 clicks. HJ Mitchell | Penny for your thoughts? 23:49, 13 June 2010 (UTC)

Joopers - I agree there is a range of worries - not least that if Level 1 replaces semi-protect, an autoconfirmed user (most of us) loses the ability to simply edit a Level 1 protected page without a reviewers or admins approval, if there is an edit awaiting review.Fainites ^barleyscribs 09:44, 14 June 2010 (UTC)

That would be level 2. Level 1 would only queue edits by IPs. Gwen Gale (talk) 10:10, 14 June 2010 (UTC)

You all need to figure out what this is going to be used for before handing out the rights and enabling the feature. But we aren't very good at planning, are we. Why does the reviewer right exist, why does Wikipedia:Requests for permissions/Reviewer exist, and why are requests there being fulfilled, if (as is the case based on the above) there isn't even agreement on how to use FR? Prodego ^talk 23:29, 13 June 2010 (UTC)

We know broadly which criteria to use for reviewers, so it's actually better to have some reviewers before it goes live. The disagreement above is about removing the right. FR is going to be used on articles otherwise semi-protected, limited to 2000, for the beginning articles at Wikipedia:Pending changes/Queue. Cenarium (talk) 23:39, 13 June 2010 (UTC)

Robla suggestion above is ok in principle for the initial phase. But as an editor I'll chose my admin thanks. --Joopercoopers (talk) 23:41, 13 June 2010 (UTC)

Practically - can we defer roll-out of this until we've answered some of the questions posed here? --Joopercoopers (talk) 23:44, 13 June 2010 (UTC)

I don;t see why not, and I think it would be a very sensible idea to do so. I don't oppose this, but if we're going to do it, let's get it right. HJ Mitchell | Penny for your thoughts? 23:49, 13 June 2010 (UTC)

The very fact that the powers that be chose to push for implementation in short order despite the number of unanswered questions and amount of confusion should answer the question, I think. I suspect the rollback plan is "deal with it". Resolute 23:55, 13 June 2010 (UTC)

Umm, the feature has been around and we've been talking about using it for THREE YEARS. We're not exactly implementing it in short order. The German Wikipedia has been using it since May 2008. Really, the Foundation needs to just say, "we're turning it on - this is a part of the service we're providing" and be done with it. There's too much upside in terms of protecting BLPs to not use it and for the three years of dragging our feet, we have let "perfect" be the enemy of "good enough". There are always going to be things that need to be worked out, but that's no reason not to take advantage of this valuable tool for preventing vandalism. --B (talk) 05:02, 14 June 2010 (UTC)

Umm, no the German Wikipedia has not been using this since May 2008; our little community stamped its feet and found the German extension to be completely unsatisfactory, and insisted on its own special little software. So what we have is some sort of hybrid between flagged revisions (what the Germans use) and page protection (which the Germans *still use*) that is completely untested. Aside from the software issues, the practical issues like what edits should and should not be accepted, and the fact that the pages related to this are contradictory on what its effects will be on non-admin editors, means we really haven't thought this one through. In fact, we haven't even bothered to state categorically that certain rules are strictly for the trial. Risker (talk) 13:43, 14 June 2010 (UTC)

You are confusing the technical aspect with the human and policy aspects, B. The technical function has existed for some time, albeit in a slightly different form than we are rolling out. But turning it on without any real clue as to what our policies surrounding its use means this obviously has not been discussed very well. We've already had some people running around in hysterics accusing admins of abusing this tool before it has even been turned on and we have arguments over who has the right to remove this tool from an editor. We have questions over whether this is akin to rollback or whether it is something far more significant. We have questions whether this is for straight vandalism only, or other things like NPOV problems. We have questions over whether this is intended for BLPs only or for everything in article space. This trial is going to be muddled in confusion while people argue over what this tool is for and how and when to use it. That part clearly was not thought out. Resolute 15:37, 14 June 2010 (UTC)

Arbitrary break

For what it's worth (if it hasn't been noted already), the major flaw in requiring ArbCom approval to remove the reviewer user group is that, in many cases, it will be patently obvious that a user shouldn't have it. It will be far rarer that it's an edge case. --MZMcBride (talk) 17:55, 14 June 2010 (UTC)

Proposal

The original proposal said that rights could be removed by admins, without more specifications; in subsequent discussions on the reviewer usergroup the issue of removal was not specifically discussed, it was not considered controversial at the time. It's important to put in place minimal due process and safeguards, but I'd venture to say that this is an issue which goes beyond the reviewer usergroup, as admins can block users so objectively it doesn't give them any more power. It is true, that, unlike rollback which just allows to revert faster, the reviewer usergroup grants a little (a tiny little) more editing capacity than does autoconfirmed. There are not many ways this right can be abused: either the reviewer reviews blatant vandalism, or un-accepts previously accepted good edits (which would make no sense since it would re-add the edit to the pending changes queue so any reviewer could re-accept it). Such abuse would be akin to vandalize or revert good edits for no reason. So I have difficulties to see this being done except by long term abusive users like Grawp et al. This is blockable abuse, as disruption, so admins have discretion to block, under the blocking policy, and to remove the right as well seems proper. What may happen though, is when an admin sees a reviewer accepting an edit (without further editing) which he/she would personally not have approved but was not against guidelines to approve, for example a non-obvious BLP violation, or a POV edit. In itself it's no ground for removal, and the admin should definitely not unilaterally remove the right. Mistakes in reviewing will probably happen, the reviewer should be advised but it's also not in itself ground for removal. There may be cases of repeated negligence in reviewing though, where advising and discussion failed to resolve the issue, or some unusual cases; a removal of a misused right (misuse which harms the encyclopedia) is preferable to blocking (it does happen that a user is only problematic in a specific area), so the access to this permission may be challenged.
Therefore, I think that (1) Admins shouldn't unilaterally remove reviewer rights except if the user has been blocked for vandalizing or knowingly accepting vandalism, or the user is indefinitely blocked. (2) It should be possible to request the removal of the reviewer right, the user seeking removal should open a discussion at WP:AN (or potentially some more appropriate place). An uninvolved admin should determine if there is consensus to remove the right, and if positive, may proceed to remove it.
Although my opinion could change with evidence in practice, I don't think that restricting the ability to remove this right to bureaucrats or functionaries would be beneficial, as it would make of its removal a bigger deal, hence of its granting too, and it is essential to grant this permission liberally. As for comparing with autoconfirmed, there has been a few proposals to allow it to be removed, but they didn't get much input and never reached consensus, the way autoconfirmed is implemented, you'd need to actually grant a right which removes the userright, which is kinda buggy, and we have the edit filter which handles various kind of potential abuse; it's much different from reviewer which adds a right we're completely unfamiliar with, and under trial, so we need to have some control over it. Cenarium (talk) 03:05, 14 June 2010 (UTC)

Also, I agree that admins should be allowed to remove a right that they have granted, for the duration of the trial or a fixed duration. Cenarium (talk) 03:25, 14 June 2010 (UTC)

I can live with that, but it seems odd that I have to jump through hoops to remove a mis-used permission when I can remove the ability to edit altogether at my own discretion. Just today, I blocked 4 established editors for edit warring and removed rollback from an editor, not a single one of which required consensus (or even a report) on AN and the blocks are far more significant than removal of a permission. HJ Mitchell | Penny for your thoughts? 03:29, 14 June 2010 (UTC)

Let's put it simply, Cenarium. If an administrator abuses this tool, the only recourse is to have him desysopped. That requires Arbcom input. I don't see why an editor should be subject to removal at a lower level of scrutiny than a sysop. Indeed, it is because of events over the last week or so, where admins have clearly abused tools that would have been stripped from non-admins who acted the same way, that has brought me (and a significant number of other editors) to this point of view. A decision not to run for RFA should not allow users to be more susceptible to removal of editing permissions. Risker (talk) 03:33, 14 June 2010 (UTC)

As I said, this issue goes beyond the reviewer usergroup and making more difficult to remove this right would not change that state of affairs in any significant way, users may still be blocked for behavior where it would be much less likely than an admin be blocked, and it's not possible to strip those rights off the admin package. So I appreciate this concern but this is not one which can be resolved or sensibly changed here. On the other hand, restricting removal more than proposed would have the adverse consequences enumerated above. Cenarium (talk) 03:55, 14 June 2010 (UTC)

Even if individual rights in the admin package can't be removed, the administrator can be temporarily blocked if they misuse a right. That doesn't require ArbCom intervention (unless they unblock themselves). HJ Mitchell | Penny for your thoughts? 04:02, 14 June 2010 (UTC)

It's a two-month trial, Cenarium. The tool, which has widely been promoted as an anti-vandalism tool, turns out to really be a content-control tool that requires that vandalism edits be accepted before they're reverted, and it is nearly impossible to address a mix of good edits, vandalism, and good-faith but problematic edits with it. Editors are going to make mistakes with it. There is going to be a huge amount of judgment involved. (Example: If an anon rewords a sentence to very subtly change its meaning, is it good faith--meaning it should be accepted--or vandalism--meaning it should be reverted?) These issues are only now coming up because people who work with content on a regular basis, who reasonably assumed that (a) it was only going to be used on BLPs and (b) the tool would work much differently than it actually does, are now finding out what it is. We can't afford to lose editors because of a two-month trial of a tool that is radically different than what was expected. Trials often have special rules attached to them, because it is understood that part of the process is identifying ease of making errors, finding out what the trialed tool really does and doesn't do, and those who participate almost always have special protections. And HJ Mitchell, just because admins could be blocked doesn't mean that they are; in fact, half the time they aren't even warned. Risker (talk) 04:11, 14 June 2010 (UTC)

Not if I have anything to say about it. If I do something blockable, I should be blocked and the fact that I'm admin shouldn't factor into the blocking admin's decision (though I appreciate that it inevitably does). Likewise, if another admin does something blockable, I'm not afraid to block them. But alas, I'm digressing from the matter at hand. HJ Mitchell | Penny for your thoughts? 04:25, 14 June 2010 (UTC)

Yes it's a trial, and we need finer control for a trial, and it's not like it was permanent. If it were so unreliable than you depict, de.wikipedia would have given it up long ago. It doesn't require that vandalism edits be accepted before they're reverted, I don't know where you got that, you rollback the edit and it automatically accepts the new revision, if you can't rollback you fix the vandalism then accept the new revision. For a "mix of good edits, vandalism, and good-faith but problematic edits with it", then you deal with the vandalism, fix what you can in it then accept what remains. The reviewer just has to make a good-faith effort to review, or leave to someone else. This is what people do all the time when reviewing new edits on their watchlist or recent changes, except there's no need to 'accept' at the end of the process (or beginning or middle, it doesn't really matter). The proposal was very clear on where this would be used. The tool works as the proposal said. The only technical difference with classic flaggedrevs is that it's not activated by default and autoconfirmed users are autoreviewed. What do you mean by losing editors ? And what about all the users we'll gain ? Cenarium (talk) 04:36, 14 June 2010 (UTC)

You're comparing apples with pomegranates here; the extension used at de.wikipedia is very different from this, and they have an extremely different ethos on that project. And yes, they use that tool for active content control, not just for vandalism reversion. This tool *can* do more or less what the proposal asked for, but it took me three minutes of puttering on the testwiki to realise it's the equivalent of handing the keys to a Peterbilt to someone who only needs (and expects) a motor scooter. You've not addressed why editors who undertake to participate in this trial should be held to higher standards than admins are. Risker (talk) 05:09, 14 June 2010 (UTC) Addendum: I don't know where this theory that we'll gain lots of editors is coming from. From what I have seen on OTRS, at least 85% of the people who are complaining about not being able to edit a protected or semi-protected article are the editors who have caused the article to need semi- or full-protection. Meanwhile, editors are indeed being disenfranchised by having to chose between having their edits reviewed by someone who probably knows nothing about the subject, or accepting permissions that will automatically get their edits accepted, but whose rules are so fuzzy and counterintuitive that they're at risk of being labeled either a vandal or an untrustworthy user. Risker (talk) 05:21, 14 June 2010 (UTC)

Assuming you consider this to be the case of any removable right assigned to non-admin users, then this is inherent, and there's no solution besides establishing new policy on removal of right, admin included, or new 'case law' by ArbCom.

There's no way those OTRS stats can reveal who would potentially edit an article, you'd need to make a proper scientific investigation with control groups and such to have any meaningful results. There's one thing that can be said though, since we put a big notice on the message displayed when viewing the source page of semi-protected pages, even with lots of instructions, we got much more edit requests, even though there's no edit section links and there's only by clicking on view source that users can get there, so there are definitely people trying to edit. The reviewing process, like editing Wikipedia, is based on crowdsourcing, it's not up to one user to do the work but to plenty of persons. That the user would have "their edits reviewed by someone who probably knows nothing about the subject" is true but is no different from how Wikipedia works right now, people review edits all the time, some exclusively for vandalism, some more in-depth, and they can alter or revert edits. The introduction of pending changes doesn't change that, accepting is a positive act, and it only takes one reviewer to accept the edit. If a reviewer sees an edit that is not obviously unacceptable, then he/she may accept it, or not, but then someone else could, or some other person would revert or alter it; that's the usual editing process. There's no new magic 'reject edit' button which reviewers could abuse on good faith edits, in fact it already exists: any user can revert or alter an edit. It's not possible for a reviewer to 'take control' of an article since all reviewers can see pending changes to articles so review and accept them. You know well that everyone can have their edits 'reviewed', then reverted or altered, by anyone; there's really only that the edits are hold from public view which matters here, and this is the price to pay for more editability than usual protection (which in turn allows more BLPs to have a minimal protection, as should be mentioned). The rules are really that simple: "there are certain things which you should make your best not to accept, if needed fix by usual editing before accepting, when unsure ask for feedback". Cenarium (talk) 06:36, 14 June 2010 (UTC)

Oh for pity's sake, Cenarium, I read the the OTRS requests themselves, and it takes only a few seconds to correlate "can't edit article, wanted to add XXX" with a page protection explicitly preventing "XXX" from being added. And nobody is threatening to take away a user's ability to edit an article for undoing edits anywhere else in the encyclopedia. I am an enormous supporter of taking extra care about information being added to BLPs, but this configuration isn't at all what I was expecting, and I am not alone in my thinking. Risker (talk) 13:43, 14 June 2010 (UTC)

But there are obvious reasons for this, there are some people who really wants something added to a WP article which is inappropriate, such as to advertise their website or whatever, or who have a really strong POV on a subject and absolutely wants to share it, or really have something against a living person. They are so motivated that when they see they can't edit, they go as far as contacting OTRS. But the majority of potential editors are not that motivated, and edit primarily to 'share their knowledge' or help WP, in general at the beginning they edit casually, there's no way they would be so motivated to go as far as contacting OTRS to get their edit through (if they even see that they can edit, since the 'edit' links are gone). There's no way a representative sample of potential editors go to OTRS.

Why would anyone threaten a reviewer from being removed reviewer status for undoing edits (assuming this is what you mean) ? And which configuration did you expect ? We have discussed this for years and there's never been consensus for anything, far from it, and then we've had flagged protection proposed, initially rejected, then this compromise proposal which got 80% support (and which you supported 'as a first step'). The proposal was clear on the configuration and how it would be used. It may not be what you wanted precisely but it is what had been proposed (minus sysop-level flagged protection). We can't have exactly what every person wants, this is a compromise. Cenarium (talk) 15:36, 14 June 2010 (UTC)

Abuse

I've been requested to chime in here, I suppose because I expressed a personal position some find important to voice on the subject.

Regardless of where the line might be drawn about what constitutes proper grounds for removal of the right, and regardless of the procedure we'll eventually settle upon for going about it, there is one thing that needs to be made very, very clear: removing this right is an admin action and, like blocking, is one where judgment needs to be good. Any administrator who would wield that ability as a weapon is abusing their sysop bit and risks loosing it. This would include anyone who threatens or removes the bit without good cause, as an act of retribution, or outside of whatever process eventually gets decided upon.

This tool is meant to eliminate (or at least greatly reduce) the amount of time vandalism is visible. I'd be very hard pressed to find a reason to remove the bit from someone who isn't misusing it to vandalize or let vandalism through within very narrow standards. There would definitely never be any valid reason for an admin to remove the bit from someone they are in a dispute with, or over review actions in an article in which they are editorially involved. — Coren ^(talk) 17:06, 14 June 2010 (UTC)

• Agreed, all admin actions are subject to review. Cenarium (talk) 17:46, 14 June 2010 (UTC)
  • I've added the proposed policy at Wikipedia:Pending changes#Reviewers. Cenarium (talk) 17:58, 14 June 2010 (UTC)

I would clarify that we're talking about review of the admins behavior, not simply an endorse/overturn.--Cube lurker (talk) 18:06, 14 June 2010 (UTC)

• There will be no need for any administrator to remove this right from any auto confirmed editor, if there is a claim that they will need to be able to do that then please tell me what it might be. Off2riorob (talk) 17:54, 14 June 2010 (UTC)
  • That's not yet clear one way or the other; though I admit that the number of reasonable scenarios is rather limited. What is clear is that doing so improperly will not be tolerated (which was my point). — Coren ^(talk) 19:07, 14 June 2010 (UTC)
    • Your comments are appreciated Coren. I can't see a senario but there may be one that reveals itself during the trial. As I see it, this tool is not an edit restriction for confirmed users and I can see no reason that an editor would need to have it removed, if they are adding violationg content they get blocked and if they do it again they get blocked for longer. This tool is not for that but is for the type of disruption from the type of accounts that have one edit to add a blp violation when something occurs and multiple new unconfirmed users come to add the uncited or violating content. Off2riorob (talk) 19:24, 14 June 2010 (UTC)
      • We need at least to be able to remove the right per consensus following a request at WP:AN, we don't know how this new tool could be used. And indefinitely blocked users have their rights removed. Cenarium (talk) 21:33, 14 June 2010 (UTC)
• I'm sorry, I started off this discussion and then never returned to it to defend my points.
  I don't think I said that I opposed limiting reviewer removal to ArbCom, only that there was no apparent consensus for it. I do think that there need to be careful checks and balances on this right, like WP:REVDEL, and would propose two options:
  • Removal of reviewer rights is treated as the same "level" of admin action as a block, so removing the rights from someone with whom you are an edit dispute would be treated as blocking them while involved for the purpose of sanctions.
  • Reviewer flag can be added by admins but only removed by bureaucrats.
• How does that sound? Stifle (talk) 08:24, 15 June 2010 (UTC)

That might make things worse, because the risk as I see it goes two ways: Some reviewers will use the tool for PoV and source screening (moreover at level2). The pith is, aside from BLP, legal and copyvio, that the tool not be used to stop any kind of good faith edit, but only straightforward vandalism, however awful or even editorially doomed any good faith edit may seem. If the tool is going to help do what one says it's meant to do, the bounds of its use must be kept way narrow and tight. Gwen Gale (talk) 08:51, 15 June 2010 (UTC)

That's fine in theory but what it means in practice is that one should accept a rubbish but good faith edit and then remove it in the normal way with an appropriate edit summary/message. How long before people just click "unaccept" because they can? Fainites ^barleyscribs 13:58, 15 June 2010 (UTC)

Quite. (and quibbling, isn't the opposite of 'accept', 'decline'? Unacceptable mangling of English if you ask me.) --Joopercoopers (talk) 14:36, 15 June 2010 (UTC)

My dear Joopers - if you go and practice using this shiny new tool via the links given you will see that "unaccept" is the word in the tempting little clicky box. Hence the inverted commas. Unfortunate but true. Maybe it's not too late to propose the change.Fainites ^barleyscribs 14:46, 15 June 2010 (UTC)

I have indeed, and it shall rankle with me every time I have to use the damn button; but in the grand scheme of things, there are better uses of my time than to encourage the software developers of an encyclopaedia to use better English. --Joopercoopers (talk) 16:13, 15 June 2010 (UTC)

But apparently not of mine (see below). Fainites ^barleyscribs 20:41, 15 June 2010 (UTC)

Oh you're on a hiding to nothing there Fainites - this page is stacked with those with a vested interest in the establishment of this scheme - fair enough, I'm sure they've all invested lots of their time. But as ever, getting them to change their plans at this stage is an uphill battle, and there are bigger things to lose sleep about on WP in my opinion. "Refer" sounds good btw. --Joopercoopers (talk) 20:57, 15 June 2010 (UTC)

Minimal change and consensus

I'd like to add a further comment to this discussion. I have been following the progress of flagged revisions off and on for the last year or two, and have been generally supportive of using the software to present more reliable content to readers, while opening up more articles for editors (rather than using (semi-)protection). The overriding impression I have had is that the introduction of flagged revision software will only be acceptable to the community if the changes it introduces are minimal.

The flagged protection (now pending changes) idea fits well with the minimal changes approach, but the same principle needs to apply to assignment and removal of the reviewer flag. My impression is that this has been misjudged. Introducing pending changes on some articles takes away the right of regular editors to have their edits to those articles stand without review. This is not a minimal change unless almost all established editors have the reviewer bit, and it cannot be removed unless willfully abused. I have read the above thread, and my view on this is close to that expressed by Risker.

The idea that one has to request the reviewer privilege turns it into a political football, rather than a key component of editing (like autoconfirmation). I tested this by setting the reviewer flag for a few editors I knew to be non-vandals, but who had not requested it. I received thanks on my talk page for being granted the "right" to use the "new tool". I have seen remarks elsewhere indicating that editors regard the reviewer bit as a status symbol, a vote of trust. This is inappropriate and is not minimal change.

There are some who see reviewing as an opportunity to control NPOV disputes, no doubt, and such influences will ultimately lead to the failure of any implementation of flagged revisions if they are allowed to affect who receives the reviewer bit. Minimal change is all the community will accept. Please take such considerations into account when forming this policy. Geometry guy 22:06, 14 June 2010 (UTC)

Unaccepting ?

When should a revision which has been manually or automatically accepted should be 'unaccepted' by a reviewer ? I can imagine the situation where a reviewer mistakenly accepts a revision, then self revert by unaccepting. There's no great need in unaccepting vandal/obviously inappropriate edits by autoconfirmed users which have been automatically accepted since it is sufficient to revert them to be done with them, but it seems acceptable. However, reviewers should be cautious in unaccepting edits accepted by reviewers, the edits would need to be blatantly inappropriate, and discussion with the accepting reviewer is preferable. Technically we could have review wars but this would be totally pointless. Cenarium (talk) 23:59, 12 June 2010 (UTC)

Help content

I suggest we move the help-oriented first three sections to Help:Pending changes and summarize them here, in a way that presents like a guideline. The guideline should cover in more details than WP:PEND the reviewing process, and the reviewer usergroup. Cenarium (talk) 03:50, 13 June 2010 (UTC)

Please don't merge this content into Help:Pending changes without being very thoughtful about keeping that article to the bare essentials. Help:Pending changes already has a lot of information in there, and is already at the threshold of what should be presented as introductory material. However, there is room for a new Help:Reviewing article, which might benefit from incorporation of the diagrams in Wikipedia:Pending changes/Terminology (or some improved version thereof) -- RobLa (talk) 22:27, 13 June 2010 (UTC)

Well, that's what I thought but then we shouldn't have merged Help:Pending changes review process. Cenarium (talk) 22:32, 13 June 2010 (UTC)

That's probably true (though I think there was at least some policy stuff on the original version). If you'd like to break that page back out, that'd be fine by me. -- RobLa (talk) 21:47, 14 June 2010 (UTC)

The help page is quite well done. I'm not that good at making help content :), so I'll just focus on the policy and guideline. Cenarium (talk) 23:43, 14 June 2010 (UTC)

Autoreviewer

Why is autoreviewer included in the new reviewer right? This marks all pages created by someone with the right "patrolled" on special:NewPages, but many of the people being given this right have never created a single article. Also, if they repeatedly create poor articles (which would see the autoreviewer permission revoked by an admin) what's to be done if removal of this permission by an admin results in a summary desysop? This whole thing has been very poorly thought through as half baked attempt to rush in flagged revs against consensus. The trial should be suspended until all the issues have been resolved (including this one and the one surrounding its removal). If we're going to have flagged revs (or "pending changes" or whatever euphemism we're using now), we might as well do it properly. HJ Mitchell | Penny for your thoughts? 22:55, 13 June 2010 (UTC)

There's certainly no need to rush. GoodDay (talk) 22:59, 13 June 2010 (UTC)

It had not been requested, I've asked if this can be removed. Cenarium (talk) 23:09, 13 June 2010 (UTC)

• Reviewers no longer have 'autopatrol', http://wikitech.wikimedia.org/index.php?title=Server_admin_log&diff=26992&oldid=26991. Cenarium (talk) 00:09, 14 June 2010 (UTC)

• FYI, it seems there's rough consensus to rename autoreviewer, see Wikipedia:Village_pump_(proposals)#Proposal_to_rename_Autoreviewer_to_Autopatroller. Cenarium (talk) 00:23, 14 June 2010 (UTC)

• Creepy! I made the exact same query exactly 1 minute before you here. –xeno^talk 14:33, 14 June 2010 (UTC)

This just defeats the point

"2 Edits by autoconfirmed users are automatically approved except in some rare cases."

Stupid. Why would we automatically grant users without any review of that user the right to have their edits go unchecked? An IP and a user with 15 edits are really not separated by that much. This defeats the purpose. If you have to be granted the right to review, you should have to be granted the right for your edits to be automatically approved. - ʄɭoʏɗiaɲ ^τ _¢ 23:08, 13 June 2010 (UTC)

No, because this is intended as an alternative to semi protection, and autoconfirmed users can edit semi-protected pages without restriction. Level 2 is equivalent to classic flaggedrevs, and can be used on pages subject to massive sock abuse. But overall, disruption by autoconfirmed users is quite rare. We don't want to protect more than necessary. Cenarium (talk) 23:14, 13 June 2010 (UTC)

(edit conflict)Well the vast majority (not all, but the vast majority) of vandalism-only accounts are blocked before they reach the autoconfirmed threshold, so vandalism by autoconfirmed editors is quite rare and their edits will be checked the same way they are currently- via recent changes patrolling, watchlists etc. HJ Mitchell | Penny for your thoughts? 23:15, 13 June 2010 (UTC)

I thought the point of this was to have a public version of articles, and that changes must be reviewed by any other editor (I think reviewing permission should go to autoconfirmed users, but not the automatic passing of their edits) unless a user is trusted enough. To be an Autoreviewer (or whatever the term is for users that can create new pages that don't go through New Page Patrol) requires 75 created articles. I've been here several years and have made over 10000 edits, but I still haven't created even 50 new articles. Why such a wide margin of trust between these two concepts? Now it just seems like something which users will have to jump over hurdles to be applied to an article (ala WP:RFPP), and that otherwise it'll be endlessly shot down as "not necessary, come back when there is more repeated vandalism". Do we even need this if it's essentially identical to the semi-protection status? Is this to be applied to all biographies? - ʄɭoʏɗiaɲ ^τ _¢ 23:52, 13 June 2010 (UTC)

No, the point is to screen out vandalism, but nobody knows what else it will be allowed to be used for screening out. As for autoreviewer, the 70 article guideline isn't rigidly stuck to at WP:PERM/A, but the right is purely technical, it's for people who are trusted enough not to write crap and who wrote new non-crap regularly enough that it's an inconvenience to new page patrollers to have to mark it patrolled so frequently. It's not really an advanced permission for the autoreviewer, just a way of saving time for new page patrollers. Finally, this isn't quite the same as semi prot- that prevents any unegistered or newly registered editor from editing the protected page, whereas this requires a "reviewer" to check the edit before it appears in the version that readers see by default. HJ Mitchell | Penny for your thoughts? 04:37, 14 June 2010 (UTC)

Level 2 protection

Please weigh in at Wikipedia_talk:Pending_changes#Level_2_protection. Cenarium (talk) 23:16, 13 June 2010 (UTC)

I wonder how many editors have even seen this function table. Gwen Gale (talk) 23:51, 13 June 2010 (UTC)

I hadn't. It's illuminating. Fainites ^barleyscribs 09:38, 14 June 2010 (UTC)

Proposal to restrict implementation solely to BLPs

Proposing that flagged revisions should be implemented solely on BLPs.

• Support --Joopercoopers (talk) 15:50, 14 June 2010 (UTC)
• Support. Then it can be easily ignored. I'm assuming that there will be some kind of way to easily indentify that an article's been flagged protected? Malleus Fatuorum 16:01, 14 June 2010 (UTC)
• Support, this is where it's truly needed and any glitches will seem easier to put up with. Gwen Gale (talk) 16:04, 14 June 2010 (UTC)
• Oppose Overtime I'm pretty sure we'd go beyond BLPs, for BLP concerns in non-BLP articles and others. Moreover there's no way to technically restrict this to BLPs and it would result in mass drama every time an admin uses it on non-BLPs. Plus, we'd lost plenty of editing opportunities for potential users. Cenarium (talk) 16:06, 14 June 2010 (UTC)

I don't understand how BLPs couldn't be flagged with a bit allowing FRs to show up in their protection menus but not elsewhere. Gwen Gale (talk) 16:12, 14 June 2010 (UTC)

It would be trivially easy to temporarily add Category:Living people to bypass this. And developers have almost always rejected to place arbitrary restrictions on possible actions, especially when trivially avoidable. Cenarium (talk) 16:19, 14 June 2010 (UTC)

Do you realise how ridiculous you arguing against "placing arbitrary restrictions on possible actions" looks in this context? – iridescent 16:20, 14 June 2010 (UTC)

It's not me arguing, it's what developers have done in the past. Cenarium (talk) 16:23, 14 June 2010 (UTC)

I don't understand. Why did you say "Moreover there's no way to technically restrict this to BLPs" only to follow it up with "It would be trivially easy to temporarily add Category:Living people to bypass this"? Why did you say it couldn't be done, then say it could easily be done? Gwen Gale (talk) 16:27, 14 June 2010 (UTC)

I meant there's no effective way to restrict it to BLPs. Even if developers would agree to only allow this to be activated on pages in Category:Living people, which I'm not sure is technically possible to do, it could be trivially bypassed. Cenarium (talk) 16:33, 14 June 2010 (UTC)

We don't need a technical restriction if we agree a policy restriction. --Joopercoopers (talk) 16:41, 14 June 2010 (UTC)

I didn't say the flag would be Category:Living people. Gwen Gale (talk) 17:04, 14 June 2010 (UTC)

What could it be then ? This is the only reliable way to detect BLPs (well, not as reliable as that actually, but we can't do better). Cenarium (talk) 17:13, 14 June 2010 (UTC)

It's easy to put a secure bit in a database bucket. The only pith of my carrying on about this is, I'm aware of no meaningful technical limitation in bounding FR to BLPs if the will is there to do it. I don't think the will is to be found, though. Rather, the will seems more or less targeted on rolling this out for the press beginning tomorrow. Hopefully some of the comments here will at least make folks more aware of the sundry worries foreseen. Gwen Gale (talk) 18:00, 14 June 2010 (UTC)

What is the problem exactly with using it on non-BLPs ? Why shouldn't we trial this on non-BLPs ? We're going to roll out progressively in any case. Cenarium (talk) 16:48, 14 June 2010 (UTC)

Because that's not what was advertised. I didn't !vote because the watchlist banner requested input on BLPs. I have no desire to spend already limited and overburdened time reviewing and approving changes to articles that have been reviewed. Truthkeeper88 (talk) 16:57, 14 June 2010 (UTC)

This is the poll and this is the proposal it approved, it doesn't single out BLPs and it has not been advertized as singling out BLPs. There has been many proposals prior to this one, some to use it on all BLPs, but they didn't reach consensus. Cenarium (talk) 17:09, 14 June 2010 (UTC)

• • Also oppose because this could be a slippery slope to use flaggedrevs on all BLPs, without regard for the protection policy, because of the sensible nature of BLPs, then nothing would stop from using it on all articles. While this proposal has the protection policy as safeguard. Cenarium (talk) 16:38, 14 June 2010 (UTC)
• Support. This is what was promised, not the vague "any article that might be edited by someone I disagree with" which seems to be the current plan. – iridescent 16:08, 14 June 2010 (UTC)
  • The proposal clearly stated that this could be used only as an alternative to semi-protection and in the same circumstances. This is what has been approved in the poll. You'd need to overturn that consensus of 80% of 300+ users. Cenarium (talk) 16:21, 14 June 2010 (UTC)
• Question: Is this proposed restriction for the trial? I'm of the mind that the trial period should be just that: try it out on a few different classes of articles and see how it fares. Restricting the trial won't allow any kind of data-gathering on how well it works on, for example, high-school science articles (which are indefinitely semi-protected for-the-most-part). –xeno^talk 16:15, 14 June 2010 (UTC)

• yes - I meant for the trial, for now. --Joopercoopers (talk) 16:41, 14 June 2010 (UTC)

• Thanks for clarifying. As intimated above, I think that would defeat the purpose of the trial. As far as I know, 'pending changes' is not just directed towards BLP concerns, it is also intended to re-enfranchise anonymous users on articles that are subject to lengthy or indefinite (indeed, permanent in some cases) protection (see WP:INDEFSEMI). –xeno^talk 17:50, 14 June 2010 (UTC)

• Support As per Iridescent. Also, 2000 articles is too many to test such a change. Suggest limiting to a much smaller pool of articles - even as low as 100 - to allow editors to understand what is involved here. Truthkeeper88 (talk) 16:30, 14 June 2010 (UTC)

  We'll progressively roll out anyway. Cenarium (talk) 16:33, 14 June 2010 (UTC)

• Oppose the consensus in the poll was quite clear that this was supposed to be applied as a replacement for protection only. In addition opening the feature up for use on all BLPs would be an enormous expansion of the scope of the proposal - there's only a few thousand semi-protected pages but well over 300,000 BLPs. Hut 8.5 16:35, 14 June 2010 (UTC)
  • Nobody has suggested that it should be used on all BLPs. The proposal is that it should be restricted to BLPs. Malleus Fatuorum 16:40, 14 June 2010 (UTC)
    • A trial on BLPs won't necessarily tell us how effective the tool is on other types of article. It's only reasonable to do a trial on BLPs if we're only going to use it on BLPs afterwards. Hut 8.5 17:55, 14 June 2010 (UTC)
• Support It's a great pity that "the poll" was not more widely advertised. There nees to be a guarantee that this will be confined to BLPs. Otherwise, it is eventually going to spread to all FAs, GAs and any other interesting page that takes an admin's fancy.  Giacomo  17:37, 14 June 2010 (UTC)
• Oppose. I believe we discussed this already and declined it - there are applications for this on various classes of article. Why change the rules of the trial at the last minute? Shimgray | talk | 17:06, 14 June 2010 (UTC)
• Oppose. I see no reason to restrict the trial in this way. And we do get libel of living people in other types of article by the way. -- zzuuzz ^(talk) 17:45, 14 June 2010 (UTC)
• Oppose If the supposed benefit is to open up protected articles, restricting the trial or the final implementation to just BLPs (which applies not just to biographies of course), is pretty pointless. MickMacNee (talk) 18:40, 14 June 2010 (UTC)
• Support. I don't believe community consensus has ever been gained for any other use. Espresso Addict (talk) 18:48, 14 June 2010 (UTC)
• Support, as that seems reasonable. GoodDay (talk) 19:29, 14 June 2010 (UTC)
• Support, I agree with GoodDay, it seems reasonable. —Preceding unsigned comment added by Cit helper (talk • contribs) 20:25, 14 June 2010 (UTC)
• Support this is what this tool was brought in for. We don't need software design to cover this, just a code of conduct. Just like admin actions in general - they get logged and discussed. Casliber (talk · contribs) 20:43, 14 June 2010 (UTC)

  But what is the rationale for not using this on non-BLPs ? This could allow more users to edit pages they would otherwise not be able to edit. This is a trial, we need to see if this can work before restricting, especially that no reason has been given to restrict it to BLPs. Why ? The tool may have been discussed in relation to BLPs but it was not brought in for BLPs exclusively. Cenarium (talk) 21:13, 14 June 2010 (UTC)

• Oppose no reason at all to limit it just to BLPs. While they are often protected more, that doesn't mean they are always the most vandalized, and I'd certainly hope this feature would never be limited to just BLPs so why limit the trial to them. -- AnmaFinotera (talk · contribs) 22:07, 14 June 2010 (UTC)
• Oppose. If this tool can be beneficial, I see no reason we should restrict its benefits to BLPs — although I concur that it should mainly be used there —. And as pointed out before, libel of living people can occur on non-BLP pages (such as an article about a political party). Salvio ( ^{Let's talk 'bout it!}) 22:17, 14 June 2010 (UTC)
• Oppose - Although helping to prevent BLP violations is certainly a major reason for implementing Flagged Revisions, in the feature now called Pending changes (PC), BLP is not the only reason. Vandalism prone and other vulnerable pages are also candidates for PC. In addition, there are potential BLP issues even in non biographical articles. So in order to get a better feel for how this feature works across a wide breath of articles, a large variety of articles should be used in a trial, not just BLP articles. — Becksguy (talk) 08:38, 15 June 2010 (UTC)
• Oppose. I've heard different reasons to introduce the novelty, from silly to sinister, and each of them makes sense only when all (or at least all high-traffic) articles are protected. And, anyway, it's a test and the test sample is minuscule. East of Borschov (talk) 08:44, 15 June 2010 (UTC)
• Oppose-All articles can have BLP content in them. Off2riorob (talk) 21:06, 15 June 2010 (UTC)

First phase of the trial

I post here because it's the most active talk page. Assuming the trial begins as scheduled, we have the following outstanding issues:

1. Obtaining (trial) policy status for Wikipedia:Pending changes
2. Obtaining (trial) guideline status for Wikipedia:Reviewing
3. Making sure the help page Help:Pending changes is ready, and consider creating a Help:Reviewing page
4. Decide whether we should use at all level 2 protection, consensus being required to remove it from the configuration
5. In the first phase of the trial only pages listed at Wikipedia:Pending changes/Queue should be PC-protected, in a controlled way, so we need to determine which pages should be listed and how to control deployment.

Cenarium (talk) 19:01, 14 June 2010 (UTC)

• Also deciding the details of the way in which success or otherwise of the trial will be measured, as I noted elsewhere previously. Espresso Addict (talk) 20:26, 14 June 2010 (UTC)

• Indeed, discussed at Wikipedia talk:Pending changes/Trial. Cenarium (talk) 21:26, 14 June 2010 (UTC)

Edit warring???

I do not believe that articles which are fully protected because of edit warring should *ever* be subject to pending changes, because any edit accepted is going to breach the expectation that there has been a consensus discussion and agreement on what should be inserted into the article.

In fact, I do not believe that articles which are *semi-protected* because of edit-warring should be included in this trial; the primary reason for semi-protection as a result of edit-warring is if unregistered/new users are repeatedly inserting vandalism, and it is the vandalism rather than the edit-warring that is leading to semi-protection.

I am removing references to edit warring from the body of this guideline. Risker (talk) 19:26, 14 June 2010 (UTC)

It is specifically mentioned at Wikipedia:Pending_changes#Scope that pending changes should not be used in case of dispute, it was also mentioned in the original proposal. The guideline should of course respect that. Cenarium (talk) 19:35, 14 June 2010 (UTC)

I totally agree with Riskers edits to the situations that may require pending protection. While I am here, if an article that has been pending protected and is showing only vandalism from unconfirmed users then semi protection would imo be preferable, and a lot less work and watching. If twenty edits from unconfirmed users are occurring and one from a good faith IP then imo semi protection would be better and the good faith IP can be encouraged to get an account and also can make an edit request on the talkpage as per usual. Off2riorob (talk) 19:49, 14 June 2010 (UTC)

• Unrelated: Are you sure that this change is not a result of misunderstanding of how pending changes work? Sole Soul (talk) 19:59, 14 June 2010 (UTC)

That content does seem a bit confluted are we as reviewers responsible for previously added content, no. We review the edit and as such add it and take responsibility for that addition. We are not as reviewers responsible for all the content in the article, just the addition we review. Off2riorob (talk) 20:08, 14 June 2010 (UTC)

No, not previously added content, rather "earlier pending revision". All pending changes will appear in one diff view, so approving the last pending revision separately is not an option. Sole Soul (talk) 20:15, 14 June 2010 (UTC)

Yes, that's one of the weaknesses of this process. If we have a bad edit then a good edit, and the good edit doesn't actually remove the bad edit (which is quite possible), then we have to revert the whole series of edits and start over. Risker (talk) 21:08, 14 June 2010 (UTC)

This also is inaccurate, we don't "have to revert the whole series of edits". The reverting part is still the same as what we currently have, we can revert each edit separately, but we cannot review the last revision separately. I think we should undo this change now. Sole Soul (talk) 21:21, 14 June 2010 (UTC)

Reviewers review the changes between the latest accepted revision and the latest revision, when you click on [pending changes], you are given the diff between the latest accepted revision and the latest revision, this is what you review, for vandalism, including removal of content. If there are consecutive edits, some good, some bad, then you fix the bad edits then accept, the same way you would do any time you 'review' new edits made to an article on your watchlist. There are no 'weaknesses', this is how we've been reviewing edits since the beginning of Wikipedia. Cenarium (talk) 21:18, 14 June 2010 (UTC)

I do and likely we all do revert vandalism and then go back and look again to see if there is some missed. Presently I have not used the new tool and am unsure and confused as to the actual actions of the new tool so for the time being support souls edit that imo removes some possible misunderstandings http://en.wikipedia.org/w/index.php?title=Wikipedia:Reviewing&diff=368057334&oldid=368049376 Off2riorob (talk) 22:38, 14 June 2010 (UTC)

Reviewing process

Wikipedia:Reviewing#Reviewing process

I've edited it so that reviewers are guided through the process. Comments ? Cenarium (talk) 02:39, 15 June 2010 (UTC)

Your editing is good and thank you for your efforts but the whole system sounds like a complicated mess. I think that just the sheer load of complicated edits the reviewers are asked to make is a great disincentive to editing any of the so protected articles. And that does not include the very probable edit conflicts which inevitably will arise from so many people doing so many reviewing edits at the same article. Dr.K. ^{λogosπraxis} 02:58, 15 June 2010 (UTC)

When users review new edits on their watchlists, don't they proceed in a similar way ? Cenarium (talk) 03:03, 15 June 2010 (UTC)

Of course they do. But they don't have multiple edits waiting in a queue for them. IMO edit pile ups are going to increase with this system. Dr.K. ^{λogosπraxis} 03:18, 15 June 2010 (UTC)

Actually, most don't, to my experience, Cenarium. And am I incorrect in understanding that any edit by a reviewer is automatically accepted? If so, then their only option is to revert/undo the whole mess and then re-add what might be worthwhile. Risker (talk) 04:26, 15 June 2010 (UTC)

I've added Wikipedia:Reviewing#Editing pages with pending edits. You could have asked a bit earlier, this would have avoided much pointless arguing. Try it on the testing wiki to see it live.

I don't think that in general there's going to be much piling up, except on articles heavily edited or if we have too large backlogs. Personally I (used to) check the recent edits to my watchlist in a similar way: see if this is not obvious vandalism, see if it should be reverted and in complex cases, 'good' and 'bad' edits piling up, then I undid some edits or fixed them manually. How do you proceed, Risker ? Cenarium (talk) 05:23, 15 June 2010 (UTC)

You're right, I probably should have brought this up earlier; unfortunately, this isn't the only iron I have in the fire today, and I've been a bit less focused than I intended. I'm going to be heading over to the testwiki to do some more playing about and make some more mock-ups for others to trial the tools. Cenarium, I want to take a moment to say thank you to you for all of the diligent, hard work you have put into this trial. We won't always agree, but please know that I fully recognise and appreciate the thoughtful effort you have brought here. Risker (talk) 05:43, 15 June 2010 (UTC)

Likewise, Cenarium and also, thanks for the hours of time you've put into steadfastly following these pages and answering questions. Gwen Gale (talk) 08:19, 15 June 2010 (UTC)

"Similar" to ... ? =

"If you have rollback or autoreviewer rights, you are a good candidate for reviewer rights as well – the level of trust is similar". Similar to what? Which of the two? the level of "trust" for rollback and autoreview is distinctly different (at least so it is declared). Perhaps "autoreview" should be struck out? Seeing an account promoted to reviewer with only 600 edits I'd say it's closer to rollback than to near-impossible autoreview. East of Borschov (talk) 07:40, 15 June 2010 (UTC)

I agree the level of trust should be looked at and talked about more. Finding out what that level of trust truly is may not be easy until the tool has been used live for awhile. Give the reviewer bit automatically at 300-1000 edits? Handle it like a rollback request? Ask for more trust than rollback? I'm tending to lean towards auto-granting, since indeed most vandal/LT-PA/copyvio/BLPvio accounts do get blocked before they hit 300-1000 edits. Gwen Gale (talk) 09:11, 15 June 2010 (UTC)

Proposal to change "unaccept" to "decline" in the Reviewer

A small but not unimportant change. That's what we'd all be doing. Declining. It never was accepted. Fainites ^barleyscribs 14:50, 15 June 2010 (UTC)

Support as proposer. Fainites ^barleyscribs 14:50, 15 June 2010 (UTC)

It's my understanding that "unaccepting" is the reversal of an acceptance, not an initial rejection. Risker (talk) 14:52, 15 June 2010 (UTC)

Where was the acceptance though? Aren't these edits applying for acceptance? Fainites ^barleyscribs 14:54, 15 June 2010 (UTC)

Looks to me like you're both saying the same thing. The fit word here would be decline... unaccept would indeed be a wonkish, bureaucratic-Latin construct meaning there was an acceptance, but then, it was reversed. On the other hand, someone who was declined might indeed feel unaccepted. Not accepted would be the same as decline... oh wait, no, a decline isn't needed for one to be "not accepted" :) Gwen Gale (talk) 15:28, 15 June 2010 (UTC)

There is a known usability problem with the feature that we plan to address, but unfortunately, simply relabeling "accept" as "decline" makes it worse.

The problem is that the mental model most people have for a review process involves three states: "accepted", "rejected", and "unreviewed". Unfortunately, the software really only supports two states ("accepted" and "unaccepted" in our configuration). The only way to truly reject a revision is to put an accepted revision after it (thus relegating the previous version to the history). Our proposed solution for this problem is to implement a "reject" button. Here's the current specification for the "reject" button.

Rather than pushing to relabel the "unaccept" button, I'd encourage everyone to think about how to make it a less prominent feature in the user interface, since it's not something people will typically care about doing. While "unaccept" is admittedly not the greatest word, whatever word that replaces it should mean something to the effect of "put this back into the queue for further review". -- RobLa (talk) 17:24, 15 June 2010 (UTC)

Support before the decline is terminal. --Joopercoopers (talk) 16:46, 15 June 2010 (UTC)

• Oppose Unaccepting an edit is reversing an edit that was accepted, not declining the initial change. Lots of people are confused about this.--Banana (talk) 17:06, 15 June 2010 (UTC)
• Oppose as per Banana. If someone unaccepts an edit, it goes back in the pending review queue. Salvio ( ^{Let's talk 'bout it!}) 17:20, 15 June 2010 (UTC)
• You can't unaccept a revision which has never been accepted. And see point 4 of the warning, you can't 'reject' besides usual editing, see Wikipedia:Reviewing#Reviewing_process. Cenarium (talk) 17:52, 15 June 2010 (UTC)

If it is just being put into a different queue (per Salvio), how about "refer"? Less brutal. Fainites ^barleyscribs 20:36, 15 June 2010 (UTC)

'Refer' sounds like and excellent compromise, 'defer' might work as well, as you are deferring acceptance (or not) to a later date. --Joopercoopers (talk) 20:58, 15 June 2010 (UTC)

Exercise extreme caution with the "reject" button.

The reject button on the review page was added to the software at the last minute. ... it was previously expected that everyone would use the history to undo bad edits just like they always have, but it was realized that a reject button would make things easier in the common case. Unfortunately the reject button has some consequences which were unintended when dealing with multiple edits: See Bugzill 23987 for my request to remove the reject button when there are pending changes by multiple authors and the rationale. --Gmaxwell (talk) 03:36, 16 June 2010 (UTC)

No "reject" button has been added yet. Do you mean "unaccept"? "accept"/"unaccept" just work on the right-sided (newest) revision on diff pages. Aaron Schulz 03:44, 16 June 2010 (UTC)

Okay then I must come from an alternative universe where there was a reject button in addition to the accept which rolled back all the pending revisions. Have you ever visited this alternative universe? --Gmaxwell (talk) 03:53, 16 June 2010 (UTC)

Howie wanted such a button, but it wasn't added yet. Aaron Schulz 03:54, 16 June 2010 (UTC)

It appears that I read the specification and then later ass_u_me_d the un-approve button did exactly that. Egg on my face. Fantastic. On the plus side, my misunderstanding seems to have resulted in a good reason why it shouldn't be implemented in a particular way, so it's not all bad. Thank you for the clarification. --Gmaxwell (talk) 03:57, 16 June 2010 (UTC)

I was opposed to the button before because it encourage "approve or REVERT IT ALL" which is often bad. I'll show your bug report to Rob and Howie. I didn't even think of that specific case before. Aaron Schulz 04:00, 16 June 2010 (UTC)

Hi Gregory and Aaron - I definitely want to implement this feature right or not at all. I've copied this conversation over to http://flaggedrevs.labs.wikimedia.org/wiki/Wikimedia_talk:Reject_Pending_Revision to make sure we consider these points as we develop the feature. I'm planning a more detailed reply there. -- RobLa (talk) 05:40, 16 June 2010 (UTC)

Userbox

(cross-posting to the userbox wikiproject too)

I just created {{User pending changes}}. Stick it on your userpage.

—ΩpenTheWindows™ 04:51, 16 June 2010 (UTC)

Good edits which should not be accepted

This is a different matter but one which is closely related to the above section.

Imagine an article with three revisions, A, B, and C. A is published, B and C are pending. You arrive and find that C is a great edit, but B is horrible vandalism that C didn't revert. In this case it is imperative that you do not approve _C_ (or _B_, obviously) because approving C implicitly approves the content of B inside C.

Instead you should hit undo on B. This will create an edit by you— revision D, which you should then approve (it will not be auto-reviewed because C was not reviewed, and the software will not assume that you have reviewed C). If the undo conflicts due to an overlap, you should manually undo the change made by B and (again) approve your resulting edit. In no case should C be approved.

Beyond being an important thing to keep in mind, this is an example of why PC isn't really a system for keeping track of the quality of single edits. --Gmaxwell (talk) 02:07, 16 June 2010 (UTC)

What you're describing, others have described as a software bug. Gwen Gale (talk) 02:19, 16 June 2010 (UTC)

It is not a bug, it is an unavoidable characteristic of the software which is a result of the single linear edit history. There is no way to improve in in this respect without resulting in absolutely awful results such as a requirement to manually merge changes when multiple edits are made between approvals. This kind of misunderstanding is why I ardently opposed the change of the name of the feature, because the new terminology creates a misleading impression that it is an edit review feature. I recommend studying the descriptive page I made about this feature back under the old name: [1]--Gmaxwell (talk) 02:24, 16 June 2010 (UTC)

Greg, I have just tried to do what you suggested over on the testwiki on this page, using *reviewer-level* permissions (i.e., not admin permissions), and I don't quite see how you are doing that. Any chance you could walk me through it, using this example? I'm on IRC, you can ping me. Risker (talk) 02:42, 16 June 2010 (UTC)

Just to keep current, Gmaxwell and I are working on the testwiki and communicating via IRC to figure out how exactly this will work, and how to document it. Risker (talk) 03:10, 16 June 2010 (UTC)

Risker might write more in summary but I wanted to get a quick update out. The primary source for the confusion seems to be the "pending review" review screen. It only allows you to accept or reject groups of edits and doesn't give you any access to the intermediate revisions. The way the system was originally designed you had to use the history page to "reject" changes, and as a result you had the full editing power to keep or revert individual edits. In this last cycle of usability improvements to the system we added the ability to "reject" from the review page. So if you use that then there is no easy way to make single revision changes and the above IRC quote is completely true. Risker's proposal to me was that if there is any vandalism you should go to the history and deal with it from there and that sounds completely reasonable to me. Also see the "extreme caution" section below for some of the more severe issues the reject button can cause. --Gmaxwell (talk) 03:44, 16 June 2010 (UTC)

I was told this was said on IRC about this subject, and I thought it needed a good public answer:

this seems like its going to make vandals even more effective because all they have to do is make one edit in a string of ten good ones, and then the entire set has to be thrown out

No. This isn't the case. If that happens you simply revert the vandalism exactly as you would always do then you or someone else approves your edit. The approval doesn't even need to be immediate, but if there is nothing else wrong after your edit then there is no reason you shouldn't approve your own edit. This all makes a lot more sense if you understand PC to be a system which controls publication rather than a system which reviews edits. Originally the language we used to describe this feature was centred around marking revisions for public display, but it was decided that it was too hard for the public to understand and so a press-relations driven decision was made to instead use "review" oriented language to describe the feature. --Gmaxwell (talk) 02:28, 16 June 2010 (UTC)

Example. I made a mockup of how this could happen at Wikipedia:Pending changes/Testing/CBM. I intentionally approved a bad edit at 3:06 and rolled back some vandalism in the following edit. Then the vandalism was re-inserted in the edit titled "sneaky". Later, I looked at this diff: [2] I didn't see anything wrong, so I used that diff to roll back the "new edit". But this automatically marked as reviewed, so the bad edit ended up in the reviewed version. — Carl (CBM · talk) 03:20, 16 June 2010 (UTC)

Step-by-step "how-to" for reviewing multiple edits

1. If you are reviewing multiple edits it is possible that there is a good edit that has been removed by a vandalistic one. Do not rely on what you see in the "pending review" page, but instead go to the page history, regardless of whether the version you see has vandalism or not.
2. Check the page history. If all edits are by a single editor, and the most recent pending edit is vandalism, it is reasonable to assume they are all vandalism. Return to the review page, undo the series, and you are finished your review. If the most recent version is good, you can review the last edit in the series from the page history, and accept all edits that way.
3. If there are multiple editors who have submitted pending revisions, review each edit individually from the page history, and undo any edit that is vandalistic, a BLP violation, or otherwise clearly unacceptable according to review criteria. Each undo will create a new edit under your username, but will not be automatically accepted. Leave good edits in place, unreviewed.
4. Once you are satisfied that all inappropriate edits have been undone, you will be left with good edits. Check the most recent pending edit to be certain you've removed all vandalism. Review that edit.
5. This will clear the backlog of pending changes.

Written by me, to be reviewed by Gmaxwell. Risker (talk) 03:41, 16 June 2010 (UTC)

We could simplify step 2 by granting rollback to all reviewers (you would simply rollback in that case). I personally think that such a change should be uncontroversial but I understand that my beliefs and reality do not always coincide. --Gmaxwell (talk) 03:59, 16 June 2010 (UTC)

I've transferred this over onto the project page. Risker (talk) 05:56, 16 June 2010 (UTC)

Reviewer is being granted with a far lower barrier to entry than rollback. –xeno^talk 05:58, 16 June 2010 (UTC)

• And you think that all you have given this "right" to are going to understand all this do you? 1,000s of good edits are going to be lost.  Giacomo  07:18, 16 June 2010 (UTC)

I agree with Giano. Gwen Gale (talk) 09:54, 16 June 2010 (UTC)

I know that, for pages on my watchlist, I look through most of the revisions by hand anyway. With the review system, there's no reason I have to change that. There's no need to use the "review" interface at all. If you simply edit the article like usual, reverting vandalism and keeping good edits, everything will be fine. The only change for a flagged page is that a reviewer needs to routinely mark vandalism-free versions as "reviewed". But these marks don't actually affect the page history, which still contains every edit, as usual. Nothing can be "lost" now that was not lost before, and the page can still be edited exactly like before to restore good content that is deleted by vandals. — Carl (CBM · talk) 10:45, 16 June 2010 (UTC)

• Please remove these complicated, unnecessary steps. It is not the responsibility of a reviewer, neither the purpose of reviewing to make sure that "good edit has not been removed". If the pending review page does not contain vandalism, then approve it, period. Any removed material can be reinserted and approved later. Sole Soul (talk) 12:29, 16 June 2010 (UTC)
• For example:

1. User A makes a good edit
2. User B makes a good edit
3. Vandal User C removes User B's edit
4. Reviewer X sees no vandalism and approves this version
5. User Y reverts User C
6. Reviewer X sees no vandalism and approves this version

Sole Soul (talk) 12:37, 16 June 2010 (UTC)

There's no reason for step 4 there, and without it you have exactly the instructions above. — Carl (CBM · talk) 13:02, 16 June 2010 (UTC)

Ironically, step 4 means that a reviewer does not have an extra step, checking the history. Re-inserting a good edit should be done by users who regularly edit the page and know enough about it, this is not the responsibility of reviewers. Sole Soul (talk) 13:24, 16 June 2010 (UTC)

That's a very different picture of the process that I have. The goal is not for a reviewer to go through a dozen articles in a minute. I would say that if the reviewer does not feel they know enough to comfortably edit the article, they should not review edits for it. There are many reasons for protection, such as long-term POV pushing, that would not be obvious to someone who is not familiar with the page. — Carl (CBM · talk) 13:30, 16 June 2010 (UTC)

If reviewing requires more than spotting an obvious vandalism or something similar, then reviewing itself can become a tool for POV pushing. Sole Soul (talk) 13:39, 16 June 2010 (UTC)

PC removes the _urgency_ in reverting. Readers aren't harmed by a pending edit— so there is no reason to act with excessive haste. This doesn't mean that it requires more careful review than spot checking, but it does mean that blinding reverting things is somewhat less excusable than in regular RC patrol activities. --Gmaxwell (talk) 13:43, 16 June 2010 (UTC)

If the goal of the tool is to help with BLP and NPOV issues, as I believe it is, then reviewing will necessarily have to go beyond spotting obvious vandalism. The goal that I have seen is that virtually all protected articles will be converted to the PC system once it has been tested in practice. — Carl (CBM · talk) 13:46, 16 June 2010 (UTC)

NPOV is not a primarily IP users issue, as opposed to vandalism. Autoconfirmed users are not less of a concern when it comes to NPOV issues. So are you talking about level 2 PC protection? Sole Soul (talk) 13:54, 16 June 2010 (UTC)

Sea of Japan is an example of a page semiprotected because of NPOV, rather than vandalism. It is one of the pilot pages for PC listed at WP:PCQ. Liancourt Rocks is another example of NPOV-related semiprotection. — Carl (CBM · talk) 13:59, 16 June 2010 (UTC)

If an autoconfirmed user is creating so many NPOV problems that protection is justified, _block them_. Geesh. Protection shouldn't be used as a crutch to allow keeping trouble makers around. --Gmaxwell (talk) 14:02, 16 June 2010 (UTC)

@Carl: That does not address what I've said. I know that there is examples of non-autoconfirmed users with NPOV problems, but there is a reason why autoconfirmed users are largely not affected by PC, though NPOV problems are primarily an autoconfirmed users problem. Sole Soul (talk) 14:11, 16 June 2010 (UTC)

Review page

Am I reading this correctly, that for most situations, the pending review page is useless of limited value, or worse, compounding the complexity of the instructions? Would it not be sensible therefore to either remove it, or make the instructions explicit to ignore it? --Joopercoopers (talk) 14:13, 16 June 2010 (UTC)

accepted version isn't showing up

My edit was reverted[3], but this reversion (automatically accepted) doesn't show up here [4] or in the article itself. Is there some kind of caching issue? bug? ErikHaugen (talk) 21:18, 16 June 2010 (UTC)

Ok, it is showing up fine now. There is some lag of some kind here. ErikHaugen (talk) 22:23, 16 June 2010 (UTC)

That's strange, but I will point it out to the people who run the servers just in case. Thanks for reporting it. — Carl (CBM · talk) 22:26, 16 June 2010 (UTC)

reviewing intermediate edits?

Something that is not clear to me yet from what I've read: when an article has some approved changes intermixed with some unreviewed changes, is the intent that all changes should be reviewed, including changes that fall in between sets of reviewed changes? I was originally under the impression that reviewing intermediate changes is not as important as checking the most recent unreviewed changes, but now I'm not sure. Tim Pierce (talk) 00:10, 17 June 2010 (UTC)

Why are you unsure? There is no need to review the intermediate edits directly. When you review it will show you a diff from the last approved version the the most current pending version. Providing that the aggregate change is acceptable, you approve it and your work is done. The main version of the article will then display the most current version of the article and there will be no pending changes. --Gmaxwell (talk) 00:22, 17 June 2010 (UTC)

The instructions say to check all intermediate changes in order to catch a vandal reverting someone else's good change that is not yet accepted:Wikipedia:Reviewing#Step-by-step_.22how-to.22_for_reviewing_multiple_edits ErikHaugen (talk) 00:25, 17 June 2010 (UTC)

You're right. And it's fine to go back and double check. But if there are revisions A, B, C, and D, and D is approved, there's no need to go back and mark the previous three as approved. If there was content lost between them and the current version, the only way to restore it is by editing the article as usual, just like you would if the page was not in the pending changes system. — Carl (CBM · talk) 00:36, 17 June 2010 (UTC)

That's what I was trying to get at, thanks. If the most recent revisions have been approved, is there much value in going back and reviewing older revisions that are not marked as approved? It sounds like the answer is "no." Thanks for clarifying. Tim Pierce (talk) 00:40, 17 June 2010 (UTC)

Show pending edit tag and reviewing screen examples

Reviewing for dummies, please. I just joined up and did not understand the instructions for reviewing because I had no idea how I would be getting to the "review screen" or what the review screen looked like. Only after I went to testing page and logged out, made some edits, and logged back in did I see what was being described. It would help a lot if an example of the pending changes tag and the reviewing screen were included with the instructions. Picture's worth a bunch of words. Jojalozzo 02:41, 17 June 2010 (UTC)

Proposing a delay to trial implementation

The following discussion 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.

---

Resolved

 – Since the trial has actually already started, debate on delaying the start seems moot. — Becksguy (talk) 10:30, 17 June 2010 (UTC)

I propose that implementation of the trial scheme be delayed until (1) the many issues discussed on this page have been ironed out; and (2) the criteria for success or failure of the trial have been discussed in detail by the community & formally set. Otherwise it's unclear what precisely is being trialled, and how we measure whether the trial has succeeded or failed. Espresso Addict (talk) 01:52, 14 June 2010 (UTC)

• Support. If we're going to do this, we may as well do it properly and implement it when everybody knows what they're supposed to be doing and how it works. This hasn't been widely enough publicised for my liking, either- the vast majority of editors who aren't involved in the "behind the scenes" running of the project likely have no idea about this. HJ Mitchell | Penny for your thoughts? 04:30, 14 June 2010 (UTC)
• Support. I agree with HJ Mitchell. This has not been very well publisized - I was unaware of the Reviewer right's existence until earlier today. -FASTILY ^(TALK) 04:45, 14 June 2010 (UTC)
• We have been a bit rushed by the development team which fixed a launch date only 10 days ago. Before the date was fixed, people didn't get involved, since the proposal was approved more than a year ago and the implementation never profiled. A few days of delay, no more than a week, would allow to better prepare us. The remaining issues and criteria for success or failure can be discussed, and refined during the trial. We could advertize through the watchlist notice. Cenarium (talk) 04:53, 14 June 2010 (UTC)
  • The watchlist notice is a good idea. Is that a mediawiki page (if so, which one)? I agree with the rest as well- I'm not talking about a major setback, just a few days so that everyone knows what's going on. HJ Mitchell | Penny for your thoughts? 06:18, 14 June 2010 (UTC)
    • It's mediawiki:watchlist-details. Though I'm not sure what we could say now. Cenarium (talk) 08:15, 14 June 2010 (UTC)
    • Oppose delay, we can restrict the use in mainspace at the beginning until policy issues are ironed out. Cenarium (talk) 15:44, 14 June 2010 (UTC)
    • Furthermore the team said they couldn't delay from a fixed number of days, they would have to cancel and find a new launch date. Cenarium (talk) 15:46, 14 June 2010 (UTC)
• It's already been delayed for three years. --B (talk) 05:05, 14 June 2010 (UTC)
  • So another few days won't hurt! :) HJ Mitchell | Penny for your thoughts? 06:18, 14 June 2010 (UTC)
    • ...except that it makes it more likely it'll become another never-implemented feature. Which would be problematic! Shimgray | talk | 17:07, 14 June 2010 (UTC)
• Oppose - It's my understanding that the review changes code does not apply until articles are set to "pending changes" protection. It will be easy to scale the number of pages at this protection level up or down based on how the reviewing process is working. Thus I see little risk to enabling the reviewing right on June 15th as planned. --Marc Kupper|talk 05:58, 14 June 2010 (UTC)
• Oppose Many issues will not become apparent until we have tested it on a limited number of articles.--Banana (talk) 07:27, 14 June 2010 (UTC)

• There's also the possibility, in case this is implemented as scheduled, not to PC-protect anything in mainspace, just allowing testing in Wikipedia namespace, until we're ready for use on articles. This would allow testing in conditions very close to mainspace before the rollout on articles. In any case we should deploy progressively in mainspace. Cenarium (talk) 08:15, 14 June 2010 (UTC)

• Support --Joopercoopers (talk) 08:38, 14 June 2010 (UTC)
• Support I am very unhapy about this, it seems to be being rushed through, without proper consultation or explanation. This needs to be heavily debated by the wider community, not just the few heavily involved in BLP affairs.  Giacomo  08:56, 14 June 2010 (UTC)
• Support Even if this is fated to happen, more time is needed to first make the software side robust as can be in off-wiki testing, to make more editors aware of this new kind of protection and to smooth out policy wrinkles. This (or something akin to it) is likely needed as to BLPs. However, I'm still wary as to how level 2, which would more than likely wind up being put to many high traffic, sleeper-sock-ridden core articles and then queue edits made by all editors other than "reviewers" and admins, could be easily gamed as a means of further swaying content which is already heavily skewed (full protection has been helpful because it stops everyone from editing until they've been nudged into finding ways to get along editorially at least somewhat and seldom lasts very long). Meanwhile, many helpful registered editors will either not be aware they can get the "reviewer" bit, or will not go through the bother of learning about, asking for and then dealing with it, hence, over time, level 2 would likely stir up a "caste," an online software ghetto wherein thousands of registered editors could not straightforwardly edit core articles, the very content which draws so many here to begin with. Gwen Gale (talk) 09:28, 14 June 2010 (UTC)
• Support We seem to have gone from an attempt to solve the BLP problem to a proposal with huge implications for newbies, established content editors and "rights" granting admins in a matter of hours. Clearly requires more thought. Is there any merit in bringing it in just for BLP's? Fainites ^barleyscribs 09:34, 14 June 2010 (UTC)
• Oppose The point of a trial is to try stuff. If you wait till you have the 'perfect' thing to try, you will never get there. We don't want to get in more analysis paralysis about whether the colour of the UI should be blue or whether one small set of people or another small set of people should be allowed to remove this right or that right. No process thing here is going to destroy the Wikipedia or any BLP article if it is a bit suboptimal, and we can make ongoing changes as we proceed. - Wolfkeeper 14:24, 14 June 2010 (UTC)
• Oppose - It's been a year and a half now and we're still not getting anywhere. Please, please, please do not delay this any further. Juliancolton (talk) 15:27, 14 June 2010 (UTC)
• Oppose Just launch it already! "Let's wait for this, let's wait for that." Oh come on, are we going to trial this or are we just here to talk about it? KTC (talk) 15:28, 14 June 2010 (UTC)
• Oppose, per William Pietri. Cbrown1023 talk 16:13, 14 June 2010 (UTC)

I'm sure Cbrown1023 didn't mean to suggest that I was advocating either way, but just to be crystal clear, while wearing my Foundation employee hat, I am studiously neutral on this decision, which I think is purely up to the community. I just wrote the linked message to make the community aware of the costs of delay. William Pietri (talk) 16:51, 14 June 2010 (UTC)

• I have seen many comments made from a deficient understanding of the pending changes facility - I'll even admit I'm still a little hazy on most details - delaying the trial won't remedy this. A trial is a trial, and if it completely sucks, we can petition to turn it off. But let's get the ball rolling, so people can at least make an informed decision. –xeno^talk 16:17, 14 June 2010 (UTC)
• Oppose delaying - there's been plenty of time to discuss all of this; let's get on with the trial and see how things work in reality. Mike Peel (talk) 16:48, 14 June 2010 (UTC)
• Oppose a delay - please, let's just bite the bullet and get on with this. I don't think we'll ever be perfectly prepared for this, but we've been waiting literally years, and the worst enemy of a new feature is the fear and uncertainty that builds (indefinitely) when it's not yet implemented. Shimgray | talk | 17:09, 14 June 2010 (UTC)
• Oppose Launch it already! Cit helper (talk) —Preceding undated comment added 20:26, 14 June 2010 (UTC).
• Oppose - we've been asking/waiting for this for how many years, now that it's ready we should launch it. If people are complaining that they weren't consulted they should get out from underneath the rock they've been hiding under. Don't say "I didn't know" after three years of planning and the last 6 months of very public trials/updates/announcements/fixes. Witty Lama 21:14, 14 June 2010 (UTC)
• Oppose Be bold, for once... --Magnus Manske (talk) 22:57, 14 June 2010 (UTC)
• Support as there's no need to hurry. We've got all the time we need. GoodDay (talk) 23:30, 14 June 2010 (UTC)
• Oppose. Throw the baby into the water and let it drown. Throw more (some said it will be 2000, and now it's a dozen a day). Kill them before they grow. East of Borschov (talk) 07:10, 15 June 2010 (UTC)
• Oppose per Xeno, Juliancolton, Witty Lama, Wolfkeeper, among others. Yes, there are those that didn't know/get involved in the endless debates about Flagged Revisions. I'm a major over analyzer, but at some point one has to launch, or give up. This is a modest trial, with very few articles and relatively few experienced and trusted editors, in addition to the admins, that will trying this process out. It's not the end of the world as we know it, and if an article or editor isn't working out for this trial, any admin can remove either one from the trial process. — Becksguy (talk) 07:40, 15 June 2010 (UTC)
• Oppose delay - Like Becksguy, I know I can belabour points. But it is time to trial this, acknowledging that it's a trial, and one of the major purposes of a trial is to learn what does and doesn't work, and whether or not there's something worth improving. It's two months. As long as nobody seriously pushes the ground-rule boundaries (by adding articles that don't qualify for protection, or by whacking people for learner/trialer mistakes, for example), we can find out whether or not this is worth the investment in time and human resources. Let us keep an open mind as we walk down this path. Risker (talk) 07:53, 15 June 2010 (UTC)
• Oppose delay - Lets just run with it. it'a time and there are plenty people around interested and keeping an eye out. As magnus says: "be bold, for once...." —TheDJ (talk • contribs) 11:43, 15 June 2010 (UTC)
• Oppose - although I think it is too late to oppose as it is revving up to start. Off2riorob (talk) 21:07, 15 June 2010 (UTC)

Unaccepting

See section above. In which case should an accepted revision be unaccepted ? It's ok when you made a mistake and revert yourself, but otherwise I can't think of cases where editing wouldn't be more appropriate. And unaccepting automatically accepted bad revisions seems largely superfluous. Cenarium (talk) 22:15, 17 June 2010 (UTC)

Unaccept can be useful where editing is going to take some time and where switching the default view to a prior approved version would be better thing for the readers in the interim. With the way PC is used these shouldn't be common cases, but the underlying software is more general. It's still useful with PC in any case: For example, say the article was at a good state, then there was a 50 edit long run of mixed of vandalism, subtle vandalism and good edits. ... and some of these versions got approved. It's going to take a while to sort it out, so it would be good to unaccept back to the last good state. You could revert instead, but you run the risk that people will continue to edit from the older version now at the front, thus creating a merge workload in addition to the cleanup workload. --Gmaxwell (talk) 05:59, 18 June 2010 (UTC)

Request to mention in diffs if all intermediary edits are by a single user

Requested in T26007. This way there would be no need to check the history to see if all intermediary edits are by a single user. Cenarium (talk) 23:49, 17 June 2010 (UTC)

Don't approve things that you know to be wrong.

I thought this would be common sense: If you know or strongly believe an edit to be wrong, you shouldn't approve it just because it was made in good faith.

Instead you should fix the edit and approve your fixed version. You are, of course, not _obligated_ to check for correctness as a part of reviewing and there is no guarantee that anyone has done so... but if you accidentally happen to spot in error in this process you should fix it. To do otherwise would make pending revisions into a tool for lowering the quality of articles, which is certainly not a goal.

I'm bringing this up because I noticed another reviewer approving an obviously (to me) incorrect but probably good faith change. I pointed it out just because I thought he'd be interested, and not claiming that he'd done anything wrong. On IRC he argued with me that it was acceptable to approve and leave standing changes which you _knew_ to be wrong in addition to ones that you didn't know (I agree that you're not obligated to know). I think this is pretty outrageous. --Gmaxwell (talk) 01:32, 16 June 2010 (UTC)

If it was made in good faith, you can accept it, then revert it, if it was clearly in bad faith, it probably qualifies as vandalism and shouldn't be accepted. Just my thoughts- not to be taken as "The Law^TM". HJ Mitchell | Penny for your thoughts? 01:36, 16 June 2010 (UTC)

There is no real difference between accepting and then reverting and just reverting, except that if you accept and then revert and sometime later the most recent version of the article gets de-accepted then the edit you revered may end up being displayed to the public because it is the latest remaining approved version. This is an undesirable outcome which could be avoided by skipping the approval and simply reverting. Still, would you agree that if you _know_ it to be bad, you ought to do _something_ and not just approve it and go on with life.--Gmaxwell (talk) 01:42, 16 June 2010 (UTC)

An edit made in good faith should always be displayed in the edit history, so as to give proper attribution and credit to the editor. -Quinxorin (talk) 17:50, 18 June 2010 (UTC)

All edits remain visible in the edit history, unless they're deleted. Not approving an edit will not cause it to disappear. Reach Out to the Truth 19:42, 18 June 2010 (UTC)

PC is a protection tool against V, CV, LT/PAs/libel and BLP, it is not meant to screen for what an editor may take as an editorial error of fact. Editorial quality is carried forward by open editing and discussion, as ever.Gwen Gale (talk) 01:37, 16 June 2010 (UTC)

There is a enormous difference between a screening tool, where you only approve things you know to be OKAY ... and blindly and willfully sitting by while incorrect changes are made to an article. As an editor you ought not let that happen regardless of the circumstances and PC should not be used as an excuse to cause Wikipedia to publish material which you know to be incorrect. I'm not arguing that you have any obligation to check for accuracy, only that if you already know something to be wrong you shouldn't contribute to the wrongness by approving. --Gmaxwell (talk) 01:42, 16 June 2010 (UTC)

Common sense is that if you know an edit is wrong, you should fix it. Reverting or fixing the edit does accept it, in the sense that it moves the "latest accepted" revid past the revid of the erroneous edit. There's no requirement that we have to accept the edit before editing it, as far as I have heard. — Carl (CBM · talk) 01:44, 16 June 2010 (UTC)

Correct. Although if the prior version to the one in question was not approved your revert will not be automatically approved. The software only assumes that you added no problems to a previously unproblematic version, it doesn't assume that you review the article every time you edit. In the common case where there is only one non-approved version, however, it will do exactly as you say. --Gmaxwell (talk) 01:46, 16 June 2010 (UTC)

Thanks, that clarified some things I had been wondering about. — Carl (CBM · talk) 01:51, 16 June 2010 (UTC)

PC is a protection tool. Good faith edits which might be taken as flawed are accepted and then undone if need be, as they've always been. Gwen Gale (talk) 01:49, 16 June 2010 (UTC)

I think it's fine to just pick up the article at the top, fix any problems you see in the overall diff to the last accepted version, and then accept the whole mess at once. — Carl (CBM · talk) 01:51, 16 June 2010 (UTC)

CBM and I are thinking along the same lines. Why would you knowingly approve an incorrect edit? If you _know_ it's wrong, simply fix it and avoid adding misleading data to the edit history as well as saving yourself a step. Gwen, can you spell out exactly why someone should approve something they know is wrong? I am completely confused by your position. --Gmaxwell (talk) 01:54, 16 June 2010 (UTC)

Because if it's not vandalism, a copyvio, legal threat/PA/libel or a BLP vio, it shouldn't be (and is not meant to be) "fixed" with PC. Gwen Gale (talk) 01:56, 16 June 2010 (UTC)

I'm not talking about using PC at all. I'm talking about just editing the article like usual without clicking "accept" or "reject". — Carl (CBM · talk) 02:00, 16 June 2010 (UTC)

Nor am I, what I'm talking about is _not_ using PC: Take no PC action when you know the edit is wrong. Fix it like a normal editor, or leave it to someone else. --Gmaxwell (talk) 02:02, 16 June 2010 (UTC)

Yes, outside of PC, fix it. Make your good faith edit as to what you see as an editorial mistake. Or leave it. Keep in mind, a volunteer editor on en.WP isn't required to edit anything at all. Gwen Gale (talk) 02:04, 16 June 2010 (UTC)

Right but that is also why they shouldn't approve it. No one is required to edit, but when you do make a change of any sort on the site you're required to not knowingly make the site worse. --Gmaxwell (talk) 02:08, 16 June 2010 (UTC)

The thinking is, you're not making the site worse, you're helping by using a protection tool (not an editorial tool) called PC to screen only for edits in four tightly bounded areas of worry. If you approve a good faith edit that still carries an editorial error, you've helped, by using the tool as it's meant to be used: Keep in mind, PC is only another kind of semi-protection and full-protection, but meant to allow editing, rather than stop edits, it's not even meant to be used on most articles. Gwen Gale (talk) 02:15, 16 June 2010 (UTC)

I think you're splitting hairs. If you approve a version containing information you know to be incorrect the public will see that incorrect information when they wouldn't otherwise. Full stop. Let my throw out a hypothetical: If, in good faith, I stuck you with an edit protected request for Johann Sebastian Bach asking you to add a date of birth of 31 March 1684— would you go ahead and make the edit if you knew or suspected it to be wrong? If not, why would you approve exactly the same edit once the article was changed from SEMI to PC? --Gmaxwell (talk) 02:37, 16 June 2010 (UTC)

I see, so if an admin does not agree with an edit, it does not get published. Are admins and reviewers now to be experts on all subjects? Are edits to wait days/months now in queus while every obscure fact is checked? This is a recipe for disaster and trouble.  Giacomo  13:10, 16 June 2010 (UTC)

Anyone else can approve any revision (assuming they have reviewer rights). It isn't really a queue of edits that have to be approved individually; it's more about how long it has been between the last reviewed revision of an article and the current revision. Indeed, once a later revision is approved, there's no reason to go back and mark earlier revisions as approved. On the other hand, there's no way to mark a revision as "unapprovable" or "rejected", and all edits are visible in the same edit history. — Carl (CBM · talk) 13:21, 16 June 2010 (UTC)

Sounds far too complicated, glad I am giving it a miss.  Giacomo  13:36, 16 June 2010 (UTC)

Again, I agree with Giano on this, this tool as designed could wind up being a very slippery slope to heavy PoV and source screening (with reviewers either hiding what they mean to do behind the tool, or good faith reviewers quite mistakenly carrying forward their own too-narrow or flawed outlooks on sourcing). I don't like how this has been designed. If this kind of protection, as it is now, winds up in say, 6 months, being put to hundreds or thousands of high traffic, core articles, moreover at level 2, it will harm the project.

As for swapping a birth date in the JSB article, if such an edit went unaccepted as vandalism and was tweaked/fixed as such, that would more than likely be an ok use of the protection tool, because it would likely be vandalism. However, even b-dates get disputed sometimes, so much care is needed. Either way, PC isn't meant for (and by policy cannot be used for) screening out what a reviewer may take as "mistakes." Those must be undone or otherwise handled with an edit summary outside of the protection tool, as they've always been. Gwen Gale (talk) 13:23, 16 June 2010 (UTC)

The PC tool cannot remove or screen out mistakes, it simply does not have that feature. The mistake is still in the article and still has to be removed from the article by normal editing. Otherwise it will be in every subsequent revision until it is removed. There is no way to mark an edit as "unaccepted" or "rejected"; the PC tool cannot be used to hide, suppress, or remove edits.

The underlying goal is that the latest reviewed revision of the page should not be significantly behind the latest revision. But this is often best achieved by simply editing the article and then accepting the new revision you create. — Carl (CBM · talk) 13:44, 16 June 2010 (UTC)

Let me elaborate. Suppose there is a pending change by Bob and then I, as a frequent editor of the article, make some edits. I want my edits to go live, obviously, so I mark the page reviewed when I'm done with it. At this point, if I haven't gone out of my way to remove Bob's edit, it will also be part of the live version that is shown to everyone – regardless whether Bob's edit was explicitly reviewed or not. The only way that I can remove Bob's edit is to actually undo it with an edit of my own. This is true with or without PC, and is actually completely independent of the PC system. — Carl (CBM · talk) 13:51, 16 June 2010 (UTC)

Awful design, dreadful. Gwen Gale (talk) 13:58, 16 June 2010 (UTC)

At least it has the property that nobody can remove anyone else's edit without explicitly undoing it with another edit. That means any sort of censorship that might become possible under PC was equally possible before. I had no part in the development of the software at all, BTW. — Carl (CBM · talk) 14:34, 16 June 2010 (UTC)

So a PoV editor clicks into the PC revision, "fixes" the unwanted PoV and the unwanted sources, then accepts the revision. If someone had asked me to come up with a screening tool for spotting and skiving off encyclopedic but otherwise unwanted PoV and sources in humanities articles, I couldn't have come up with a handier way to do it than this. I already see posts by editors happily chatting about how it will help them carry forward the "integrity" of PoV disputed articles, which has nothing to do with vandalism, copyvios, legal/attack/libel or BLP woes at all. Gwen Gale (talk) 14:50, 16 June 2010 (UTC)

To start, don't give bad editors the review tool. _Continued_ PoV editing against the communities wishes is no less a form of vandalism than any other. But more importantly, how is this any different than the case without PC? You make an edit, I don't like it so I "remove your unwanted pov, unwanted sources". Perhaps after that you revert again and maybe we get into an edit war. PC doesn't change the behaviour at all, because PC is not a tool to control _edits_, it's a tool that reduces the harm of bad edits so we can exercise fewer restrictions over editing. --Gmaxwell (talk) 14:58, 16 June 2010 (UTC)

Why do you saw awful, Gwen? The behaviour follows directly from the requirement that there must never be multiple separate forks of history which must be merged, from the requirement that no one be prohibited from editing just because no one has yet reviewed the prior version, and from the requirement that if multiple edits taken together have a good final state that you can get the final state published up on the site quickly without reviewing each and every diff. I don't believe that it is possible to get all of those properties without also resulting that CBM is describing. But perhaps you see something I don't. --Gmaxwell (talk) 14:52, 16 June 2010 (UTC)

Good faith edits which are not vandalism, copyvios, legal/PA/libel or BLP woes can be held up in the queue (from going live for readers not logged in) until they've been "fixed." The hapless IP/non-reviewer editor gets the hint quick and is driven off even faster than before. Gwen Gale (talk) 15:04, 16 June 2010 (UTC)

If your edit is ultimately reverted by some WP:OWNer does it really change how you feel about the situation if it was displayed to the public for a whole two minutes before you were reverted? --Gmaxwell (talk) 17:20, 16 June 2010 (UTC)

I tried to explain in my post in the "Sea of Japan" section below why I feel that list of concerns is incomplete. — Carl (CBM · talk) 15:12, 16 June 2010 (UTC)

Yes, I saw that, my understanding is that you hope to use PC as a tool to screen good faith but unwanted editorial content which isn't vandalism, copyvio, legal/PA/libel or BLP woes. This is not what the community consensus asked for and it is not within the policy. However, it seems to be spot on what I said would happen. Gwen Gale (talk) 15:22, 16 June 2010 (UTC)

I agree with you Gwen and I add that this looks more like censorship than anything else. Dr.K. ^{λogosπraxis} 15:31, 16 June 2010 (UTC)

I don't like the word censorship outside the bounds of a government halting speech through police powers. This aside, yes, I believe I know what you mean, it's a means to skew PoV and sourcing before readers who aren't logged on can ever see it and hence drive off unwanted editors. Level 2 can squeeze out all editors who aren't reviewers or admins, all one need do is claim the article is disrupted by sleeper socks. Gwen Gale (talk) 15:43, 16 June 2010 (UTC)

The real purpose of this ill-considered new feature is becoming more evident by the day. Malleus Fatuorum 15:38, 16 June 2010 (UTC)

Yes. I've been worrying for a few days now that this is true. Gwen Gale (talk) 15:43, 16 June 2010 (UTC)

A minor point of correction— Non-logged in Readers can _always_ see the pending changes, they just have to click over to the pending changes tab. I guess you know this, but some people may have gotten another impression from the above so I think that it would be good to make it clear. I don't think anyone here is advocating that PC be used to impose particular viewpoints on articles— I know I believe that it isn't particularly useful for that, at least not compared to the already existing "undo" button— but I think we'd all be interested in seeing any examples of it being used that way. I will _gladly_ file an RFC against anyone doing that. Rather than worrying about what might happen, lets just agree that if it were to happen we wouldn't tolerate it. --Gmaxwell (talk) 17:26, 16 June 2010 (UTC)

This seems pretty straight forward to me. If there is a pending change that says something obviously wrong, I think people should revert that change, noting in the edit summary why they did it. This has nothing to do with PC, and is what people should always do. If the person stumbled on the article with the intention of reviewing a pending change that is irrelevant. The PC system isn't being subverted into editorial control. The human who was previously doing PC is now going to do editorial. Likewise if I am disambiguating, and notice a typo, I might fix that typo, but that is not subverting the disambiguation policies. I've simply changed what I was doing for a second. - cohesion 21:26, 16 June 2010 (UTC)

I agree with Gmaxwell and others; this seems like common sense. If you know something is wrong, you shouldn't approve it. You should fix it. The behavior should not be substantially different to how it would work without PC. If you saw an edit that you knew was wrong, would you really just completely ignore it? Because that's basically what you're doing if you approve a change that you know is wrong. Its actually worse with PC, because without it, ignoring it is a passive process – you just don't do anything. But with PC, you're actively approving of it by clicking the review button. Changes that are just POV or that you aren't sure are wrong is a different matter; that's a judgment call, but content that is clearly factually incorrect should not remain in the article, and should definitely not be approved. Mr.Z-man 03:03, 17 June 2010 (UTC)

Exactly. In addition to the 'normal editing' bit, I'm concerned that if someone approves something they _know_ is bad because they don't have the time or patience to fix that they will deny all of other other eager reviewers who would have the time to fix it a chance to even look at the edit. Thanks. --Gmaxwell (talk) 04:39, 17 June 2010 (UTC)

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.