Wikipedia talk:Rollback for non-administrators proposal

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.

---

Disagree

This allows just about any editor to get the rollback tools. Once they get the tools, there is no easy way to get them removed. --Rschen7754 (T C) 21:09, 9 December 2007 (UTC)

It's easy for them to get them removed - any admin can remove them, all that's needed is a consensus on WP:AN. Also, not everyone would get them, there's minimum standards that are needed before you can get them. an edit warring block would almost certainly disqualify anyone from getting them. Ryan Postlethwaite 21:26, 9 December 2007 (UTC)

It's much safer to just allow the user scripts to continue - any Wikipedia users who are intelligent enough to find and operate the scripts should be intelligent enough to use them properly. —Vanderdecken∴ ∫ξφ 12:22, 16 December 2007 (UTC)

Installing a script takes much less effort than making thousands of contributions over several months. So any user able to find this page, request rollback and have it accepted is intelligent enough to use rollback properly too? – Gurch 16:30, 16 December 2007 (UTC)

Nice but...

Will it be easy for the bit about limitations (non-minor, usage limit) to be actually implemented? Do we really want these marked as major edits? Are these restrictions necessary with the fairly high entrance requirements? GDonato (talk) 00:13, 10 December 2007 (UTC)

As far as I'm aware, the non-minor edits and usage limits bit is already coded out and just needs to be given the green light, but I do agree that maybe we don't need the limitations with the fairly high entrance requirements - they could most probably be removed. Ryan Postlethwaite 00:16, 10 December 2007 (UTC)

Usage limits I would support, but since the primary use for this is vandalism reversions, I would suggest removing the non-minor edit part. Mr.Z-man 00:19, 10 December 2007 (UTC)

Comments

While I agree with the general idea, I think it is too strict. I would definitely ease up on the requirements. 2000 edits and 6 months of experience is on the low threshold for passing RFA. Since this would grant the most minor admin tool and the only one that can be replicated with scripts, I feel the requirements should be easier. Perhaps more like the requirements for Vandalproof. I would suggest:

1. Editing for at least 2 months
2. 250+ mainspace edits
3. No edit wars in the past 2 months
4. No recent blocks

OR Be an approved (or approved for trial) bot that needs rollback

The primary use for this is recent-changes patrol - I really don't think it takes 6 months to get used to that. Removal of the tool needs to be a bit easier as well. I would suggest that it be removed after discussion an AN or ANI or if it is being used in a content dispute that results in the disputed page being protected or the user being blocked. Mr.Z-man 00:16, 10 December 2007 (UTC)

Yeah, those limitations look a little better to be honest - the ones I put in were just off the top of my head, and with the possibility to remove the feature if there's abuse, we shouldn't set them too high. It's important that the users understand the use of rollback in mainspace, so I don't think we need to look at totals for other editing areas - all in all, I like those thresholds. Ryan Postlethwaite 00:20, 10 December 2007 (UTC)

So one could make 250 AWB edits, then wait 2 months, then get rollback? --Rschen7754 (T C) 00:23, 10 December 2007 (UTC)

It's better than autoconfirmed users only and it allows admins to take a thorough look at the users contribs before giving them it. As I've already stated, it can very easily be removed if there's any misuse (bear in mind that users wouldn't have AWB if they'd made less than 250 edits. Ryan Postlethwaite 00:27, 10 December 2007 (UTC)

The only requirement for Twinkle is a browser that supports it and even if it is removed from a user's JS file, it will still work until their cache clears. Mr.Z-man 00:45, 10 December 2007 (UTC)

That doesn't mean we should make the problem worse, however. So somebody makes 250 typo edits, maybe even to the same page, then wait 2 months and then ask for rollback. Do you see the problem with this? --Rschen7754 (T C) 00:51, 10 December 2007 (UTC)

I think if admins exercise a little discretion and don't just act as process bots, that won't be a problem. Note that they need to state a reason for the request. If they only fix typos, I'm not sure what that reason would be. Mr.Z-man 00:54, 10 December 2007 (UTC)

The problem is, in real life, they won't. And the problem with this proposal is that once rollback rights are given, they are hard to revoke. --Rschen7754 (T C) 00:56, 10 December 2007 (UTC)

How are they hard to revoke? No harder than stripping someone of twinkle (which I've done a couple of times). Ryan Postlethwaite 00:57, 10 December 2007 (UTC)

Not with the required ANI consensus... there are administrators who will let anything go (and I've been through SRNC and seen it happen...) --Rschen7754 (T C) 00:59, 10 December 2007 (UTC)

What about the suggestion I made above? After discussion an AN or ANI or if it is being used in a content dispute that results in the disputed page being protected or the user being blocked. Mr.Z-man 01:12, 10 December 2007 (UTC)

If any administrator can summarily revoke it, I suppose. But if there's a lot of damage done... --Rschen7754 (T C) 01:17, 10 December 2007 (UTC)

• I find this an awful lot of red-tape for functionality that Twinkle can provide me with as well with a lot less work. I'm perfectly happy with popups and twinkle, they suit my current needs just fine. Therefore i'm seriously starting to doubt the usefulness of this whole operation if this is the way it will come to be. One of the reasons this was developed, if I understand correctly, was that it should be lighter on the servers. But if we are only gonna give it to the few people who will go trough the red-tape, than that advantage is neglect-able and this an exercise in pointlessness. If the right is revokable, I think it should be automatically granted (within the limits of course [perhaps even more stringent than the ones proposed]). I'm not saying that I don't understand the worry some people have, I'm saying that if people worry so much, why would we enable the whole functionality in the first place, instead of letting people use twinkle like they do now??? --TheDJ (talk • contribs) 02:22, 10 December 2007 (UTC)
  • Correct. In order for this proposal to be acceptable and not damage the encyclopedia, these requirements would be necessary. Therefore, the idea of non-admin rollback is pointless. --Rschen7754 (T C) 02:24, 10 December 2007 (UTC)
    • Just to get this clear, i'm pro non-admin rollback, i just think the current proposal is totally stupid. --TheDJ (talk • contribs) 02:29, 10 December 2007 (UTC)
      • Red tape? All they have to do is put one line on the bottom of the page and wait a little while. Mr.Z-man 02:35, 10 December 2007 (UTC)
        • That is how RfA started. But before you know it you will see cases where someone once pissed off an admin over something little, and the admin doesn't grant the user these rights (or pulls him off the requestlist or something). Then we have an ANI thread, and then a couple of cases later there will be an arbcom case. People will start screaming CABAL etc etc etc. I think the current form is just asking and waiting for drama. --TheDJ (talk • contribs) 02:46, 10 December 2007 (UTC)
  • Admin rollback skips the editing screen, so therefore makes rollback quicker and therefore vandal fighting faster - it's quicker and better than any available script (and current scripts could be modified to use admin rollback) so people used to using twinkle would have it invaluable. The problem with giving it to everyone is that it can easily be misused - we need some way of trusting users who to give it to, and if they abuse that trust, an easy way to take it away. Ryan Postlethwaite 02:31, 10 December 2007 (UTC)
    • The proposal at the top of this section does not address those conditions. --Rschen7754 (T C) 02:33, 10 December 2007 (UTC)
      • Yes it does! Users that misuse the tools will have them revoked - I'm not sure why you're questioning that. Ryan Postlethwaite 02:35, 10 December 2007 (UTC)
        • In theory, yes. In practice, they won't be. --Rschen7754 (T C) 02:49, 10 December 2007 (UTC)
          • I thought for about 15 mins about your reply to Ryan's question, and the only thing I can come up with is: "You seem bitter...." --TheDJ (talk • contribs) 03:09, 10 December 2007 (UTC)
            • I'm being matter-of-fact. I've seen admins overlook things that were blockable many times. --Rschen7754 (T C) 03:41, 10 December 2007 (UTC)
              • I find the restrictions to sufficiently address that problem. Most trolls won't put in the effort to get something like this and those that do can be dealt with by an admin (there are enough of them to have at least one look at the problem seriously). If nothing else, there can always be a list compiled in the user's edits page, grouping together their rollbacks. The rollbacks could even be rolled forward by an admin or perhaps another user with the rollback function. Basically user checks user. --Alexnye (talk) 09:43, 10 December 2007 (UTC)
                • I've been thinking about this recently, and was pleasantly surprised that someone else had been as well :) It appears to me there is relatively little damage that could be caused by this, as it can be so easily revoked and reverted compared to something like deleting of article. Combine this with the fact anyone can use twinkle I don't think the requirements should be anywhere near those required for adminship. I would have thought about 1k edits over about 2 months or more and of course no edit wars or vandalism edits in around 6 months but admins should be allowed to use discretion in extreme cases. Tiddly-Tom 19:31, 10 December 2007 (UTC)
                  • I say that we AGF for most people with Mr. Z-man's minimal requirements, then deal with problems as they come up. To do the cliché thing and paraphrase Jimbo: we don't cage people while in restaurants because we're going to give them steak knives. Making the tool hard to get defeats the purpose of this proposal. The idea is - some good edits, no bad history: here you go! - simplicity is the goal here, please remember that. If there are problems, we remove the rollback - undo is nearly as good anyway. Nihiltres^{t.l} 12:59, 11 December 2007 (UTC)

                    Undo is not "nearly as good", not by a long way. It requires the edit form to be recieved and submitted, which is slow, it requires the user to modify the summary if they don't want to get yelled at, and most importantly it is impossible to revert multiple revisions at once using undo, whereas rollback reverts all revisions by the same user and so does what is needed automatically in 99% of cases 144.32.177.176 (talk) 13:06, 11 December 2007 (UTC)

                    Fair enough - what I mean to point out is that it is possible to revert edits without rollback, though it may be less convenient. I find undo to be more convenient for me sometimes even as an admin because it allows me to enter a custom edit summary for cases that may be borderline, or bad edits made in good faith (such as mistakes). The issue here is that we can restrict abuse if (and when) it happens. I happen to be arguing for minimal restrictions to the tool, ;) Nihiltres^{t.l} 14:58, 11 December 2007 (UTC)

                    Rollback also allows the summary to be changed, just pass the appropriate parameter in the URL. Rollback saves an expensive backend request and, from experience, cuts the time a reversion takes by almost three-quarters -- more on a bad day/slow connection. If you're talking about a thousand or so reversions a day, every day (several thousand in the case of bots), that adds up 144.32.177.175 (talk) 16:11, 11 December 2007 (UTC)

• I tried to change the proposal for those requirements, then got reverted partially. FunPika 01:43, 12 December 2007 (UTC)

Motivation

This proposal does not have a motivation. Why do (or should) we want this functionality? It seems as if some inherent need is assumed, but this assumption does not help bring clarity to the discussion, it does the opposite. I think the proposal and this discussion will benefit if the motivation is put in words. It will guide the process because the proposal naturally follows from it. And it will guide the proposal because its intent (and not only its mechanics) will be clear. -- Pepve (talk) 02:10, 10 December 2007 (UTC)

Rollback aides with vandal fighting, it's faster than any reversion script as it skips the editing screen. There's plenty of non admins here that are really good and trusted vandal fighters who could really benefit from having this tool, so I guess you could say there's the motivation - it gives regular users an admin tool to aide their work, without having to 'have them all'. Ryan Postlethwaite 02:23, 10 December 2007 (UTC)

You can convert Twinkle to not have the extra screen... --Rschen7754 (T C) 04:44, 11 December 2007 (UTC)

No, you can't. Whether it's visible to the user or not, it still has to make a (slow) request in the background – Gurch 09:41, 11 December 2007 (UTC)

Ryan, I used your words to write a new section, Motivation and intention. Hope you don't mind.

I think this is important to make explicit, the rest should follow from it. Although it will obviously lead to different conclusions. Thanks for your attention. -- Pepve (talk) 04:57, 12 December 2007 (UTC)

Two Points

(From original proposal) If we were to go ahead, we wouldn't want to see editors start utilizing it for edit wars ala Wikipedia:Requests for arbitration/Guanaco, MarkSweep, et al, and should have strong disincentives for those who use it for such a purpose. The way rollback is given must not give out a lot more heat than light as well, and heat is generated through unilateralism and arbitrariness. - Mailer Diablo 08:26, 10 December 2007 (UTC)

Bureaucrats, not administrators

As noted at the previous discussion, I oppose administrators granting access to admin tools. I think the giving of the rollback tool should be done by bureaucrats. We require it for the bot flag, and for promoting admins (among other things), we should require it for this. - jc37 13:29, 10 December 2007 (UTC)

But why bureaucrats? It will put a burden on an alreayd limited number of people when administrators a fully able to complete this task. Ryan Postlethwaite 13:31, 10 December 2007 (UTC)

Honestly, Administrators could be considered "fully able" to do a lot of things on Wikipedia. But that's not how things are broken up here. ROllback is part of the "admin package" of userrights, which requires at least a bureaucrat to assign. I don't think we should suddenly say that administrators can give out the separate tools, but not the whole package. And noting the various bureaucracy that's being put in place for this (seems rather similar to bot approval, to me), I don't see that this is in anyway not just an admin tool that's just being scaled down for non-admin use. - jc37 14:02, 10 December 2007 (UTC)

But rollback is the only admin tool that can be replicated with scripts, its hardly an admin tool at all. If it weren't for the risk of sleeper socks and the unwillingness to raise the autoconfirm threshold, I would support giving this out automatically. IIRC, the only reason it wasn't given more liberally before was the absence of rate limiting, which is completed now. It's basically just a faster way to do a certain type of edit, and the edit can be reversed by anyone - deletion, protection, and blocking need an admin to reverse. Mr.Z-man 18:25, 10 December 2007 (UTC)

I think that there more concerns than that. For example, marking the use as "minor", which means it won't show up on recent changes. And until this is done, it will require an admin to rollback a rollback. So no, I still don't see a difference. I see your perspective on this, I just also see the concerns, and I think going "slow" (as suggested in WP:SOP), is probably a good thing in this case. - jc37 12:13, 12 December 2007 (UTC)

I'm sorry, I really don't understand your argument. Special:Recentchanges shows minor edits by default. Nobody actually reads it, however, because there is too much there. Anyone can revert an edit whether it shows up there or not, because all edits are listed in the page history. They do not have to be an administrator, or even logged in. They are also all listed on the user's contributions page. All logged-in in users can and do mark edits as minor, especially reversions, without any issues. People do not watch established users on Recent Changes ready to criticise their edits if they do anything wrong. They watch for vandalism – Gurch 12:34, 12 December 2007 (UTC)

Well, I was repeating (from memory, which may be foggy) some concerns that I saw that others had. Personally, my main two concerns are that admins shouldn't be granting/removing; and that there should be a process for granting. My comparison is bots, and adminship (and from what I understand, tools, as well). - jc37 01:28, 13 December 2007 (UTC)

• Actually, I think giving bureaucrats the ability to give/remove rollback would be a better idea. Perhaps "requests for rollback" could be a similar process to WP:CHU. Also, I am concerned that giving administrators the ability to give/remove rollback would send people's RfA standards up, resulting in more good candidates not passing (the risk of RfA standards rising impossibly was one of the reasons the suggestion to make administrators and bureaucrats one user rights group didn't pass). Acalamari 18:06, 10 December 2007 (UTC)
  • I'm not so sure, I think giving/revoking rollback is a lot less of a big deal than blocking/unblocking. Mr.Z-man 18:25, 10 December 2007 (UTC)
    • I like the idea about user requesting to be granted the tools, like how AWB works. OhanaUnited^{Talk page} 22:32, 10 December 2007 (UTC)
      • Struck after more thought. Acalamari 18:30, 12 December 2007 (UTC)

5 rollbacks per minute? WTF?

"The user will be limited to five rollbacks a minute."

Why? While patrolling recent changes, especially during busy periods, it is frequently necessary to revert more than five edits per minute; indeed, I've done twice that many times – Gurch 14:44, 10 December 2007 (UTC)

Well...... that's what would happen if all autoconfirmed users got rollback, but I certainly think it's a good idea to remove that limitation if users have to request rollback like with this proposal. Ryan Postlethwaite 14:48, 10 December 2007 (UTC)

Then I will do so. – Gurch 15:10, 10 December 2007 (UTC)

Userrights

As a completely irrelevant aside, do we now have

$wgAddGroups['bureaucrat'] = array( 'sysop', 'bot' );
$wgRemoveGroups['bureaucrat'] = array( 'bot' );

rather than having pages like makesysop, makebot, on an ad-hoc basis? Special:Userrights still says it's limited to Stewards and Jimbo. Has this feature ever been tested? (maybe someone should code up a "makerollback") —Random832 19:46, 10 December 2007 (UTC)

Originally the page had links to a giverollback extension and even a screenshot of said extension. Then I replaced that with the information about $wgAddGroups and $wgRemoveGroups. [1] [2]

Also, the line that you had typed up there would have resulted in Steward assistance being needed to make bcrats. If you want to test that feature, just open up a test wiki running the latest stable release of MediaWiki (1.11.0 I believe). FunPika 11:17, 11 December 2007 (UTC)

I've tested Userrights on my own wiki which is the same version as what WMF uses (1.12.0 +/- a few revisions). The code should be something like:

$wgGroupPermissions['rollback']['rollback'] = true; 
$wgGroupPermissions['bureaucrat']['userrights'] = true; 
$wgGroupPermissions['sysop']['userrights'] = true; 
$wgAddGroups['bureaucrat'] = array( 'bureaucrat', 'sysop', 'bot' );
$wgRemoveGroups['bureaucrat'] = array( 'bot' );
$wgAddGroups['sysop'] = array( 'rollback' );
$wgRemoveGroups['sysop'] = array( 'rollback' );

This will give admins and bureaucrats access to Userrights, bureaucrats can add bureaucrat, sysop and bot, remove bot; admins can add and remove rollback; and people in the rollback group can use rollback. Mr.Z-man 00:44, 12 December 2007 (UTC)

As I've noted above, I'd rather that Bureacrats (not admins) grant/remove rollback. And in looking at the above, I see that that would be a simple change to your proposal. - jc37 00:48, 12 December 2007 (UTC)

For the number of requests that this would probably have, there would be far too many for the 'crats to handle, so it really would have to be administrators who would be able to grant the permission - this really isn't a big deal, but it needs some thought when being granted. Ryan Postlethwaite 01:46, 12 December 2007 (UTC)

Any more than "adminship is no big deal" : ) - No, I still disagree. And the "workload" may be larger at first, but after the initial, it would lessen, I presume. Note that I also support that this should require some approval process, as well. - jc37 12:08, 12 December 2007 (UTC)

What about removal? Would we have to wait for a crat to come by ANI? By the logic of "its part of the admin package" we should need a steward. Mr.Z-man 00:38, 13 December 2007 (UTC)

Rollback can already be taken away from admins via Special:Blockip. —Random832 01:22, 13 December 2007 (UTC)

(ec) - What's done for a bot? Block the bot? How about with awb? If necessary, block the user? I don't see a problem. And I don't have a problem with bureaucrats gaining an extra userright (removing rollback). But why should we skip from stewards to admins? Doesn't sound like a good idea at all. - jc37 01:28, 13 December 2007 (UTC)

Why? Which administrators do you not trust to give people the ability to do something they have done thousands of times before more efficiently? Feel free to name names, I'm sure they won't mind – Gurch 15:42, 13 December 2007 (UTC)

There are several presumptions in that statement. But rather than get distracted in a discussion of semantics, I'd rather discuss the topic at hand. I requested this below, and I'll request it again: let's create a new thread (or even a new sub-page) and list and discuss all the actual concerns, and see if there's a way to find consensus. - jc37 23:41, 13 December 2007 (UTC)

I don't think anyone will object if you start a new thread. But I'm against a new sub-page. There are quite some useful opinions and suggestions here, a clear sheet will only inspire to restate them. -- Pepve (talk) 23:54, 13 December 2007 (UTC)

You keep mentioning "concerns" and yet I see none other than the assertion that bureacurats are necessary to approve something, without any explanation as to why the potential for abuse is so great that this is necessary – Gurch 00:19, 14 December 2007 (UTC)

There is a major difference between rollback and other admin tools. Unlike deletion, protection, and blocking, which require someone with the same permissions to undo, rollbacks can be undone by anyone. Many of the "concerns" raised here amount to attempts to predict the future. The only things we have to compare this proposal to are rollback scripts and automated editing tools, many of which have no requirements or requirements similar to what is suggested here, none require a bureaucrat, but have yet to see the massive amount of abuse predicted with this. Mr.Z-man 01:57, 14 December 2007 (UTC)

Rollback is no big deal

This tool is simply a convenience. It doesn't enable one to do anything that even an anonymous editor can't do. There's no need to create a new tier of users and the bureaucracy and elitism that goes with it. Let's just give it to all autoconfirmed users, like we do with pagemove. If someone's abusing the privilege, it can be dealt with through the same channels as any abusive editing. Toohool (talk) 20:05, 10 December 2007 (UTC)

Have you ever used rollback? You'll understand why it's not just a normal tool - it's far far quicker than any other reversion tool. Ryan Postlethwaite 20:09, 10 December 2007 (UTC)

No I haven't, as I'm not an admin. Someone who wants to be disruptive is going to be disruptive, this would just make it a bit quicker for them. I don't think that justifies the departure from the Wiki spirit that this proposal represents. Toohool (talk) 20:21, 10 December 2007 (UTC)

Can't people get rollback from Twinkle and VandalProof? Marlith ^T/C 05:22, 11 December 2007 (UTC)

No. – Gurch 09:40, 11 December 2007 (UTC)

Rollback is "no big deal", the way that adminship is "no big deal". With all the benefits and liabilities (as well as trust and responsibilities) that go with it. - jc37 01:28, 13 December 2007 (UTC)

Huh? What "trust and responsibilities" do you have if you have a button that says "rollback", that you don't have if you have a button that says "revert" that does exactly the same thing but more slowly and less efficiently? – Gurch 10:32, 13 December 2007 (UTC)

Read my response to the thread below this one. (But yes, from what I understand, there's a difference.) - jc37 12:32, 13 December 2007 (UTC)

previous proposal

Note Wikipedia:Requests for rollback privileges, which was rejected in September 2006. I would actually be in favor of automatic granting of all admin privileges in the absence of explicit opposition, but I might be part of the lunatic fringe on that. - BanyanTree 21:35, 10 December 2007 (UTC)

What's different now than a year ago? Is there any reason why the proposal should be accepted now, rather than rejected for the same reasons as last time? ~MDD4696 21:43, 12 December 2007 (UTC)

Because consensus can change. Mr.Z-man 01:51, 13 December 2007 (UTC)

Yes it can, but in looking over the recent RfC, it doesn't look as if it has. The only difference now is that developers have some concerns about scripting tool usage in regards to server load. If the current scripting tools are slowed down, or eliminated, then those who use them are hoping for an "easy" replacement. - jc37 02:01, 13 December 2007 (UTC)

So you're basically saying that from now on people should do everything manually, and thus take five times as long over it, just because despite making thousands of good edits you don't trust them to use a different, much faster interface? – Gurch 10:31, 13 December 2007 (UTC)

No, I don't recall saying that or that. The conflation here is that there are technical concerns as well. It's not entirely my concern, but from what I was reading in the RfC, it's the "how" of how rollback works that is concerning people. Else it wouldn't be an admin userright. Though I think that that would be a good topic for discussion. Let's start a new thread describing exactly what rollback is and isn't. (Maybe even as a sub-page of this?). Maybe if we can deal with the concerns one-by-one, that would help? - jc37 12:32, 13 December 2007 (UTC)

I don't see any concerns at all, other than paranoia that if rollback was given to someone they might somehow abuse it in a way that wouldn't be noticed, and that somehow being able to block anyone who does abuse it isn't enough to stop them – Gurch 15:40, 13 December 2007 (UTC)

Then perhaps you're missing some concerns. Are you interested in discussing them or no? - jc37 23:41, 13 December 2007 (UTC)

All of us, including Gurch, are interested in that discussion. Otherwise we wouldn't be here. -- Pepve (talk) 23:49, 13 December 2007 (UTC)

Sysop in other languages

Hello,

I am a sysop in the French language wikipedia. I think it would be interesting to give this function to sysops in other languages even if they don't have the minimum of 2000 edits. We have been around wikipedia for a long time (more than 2 years, 21,000 edits on fr, around 1,900 here) and know how to use the rollback function. It's annoying to revert vandalism on this wikipedia due to the lack of this function. Thanks in advance. Poppy (talk) 09:10, 11 December 2007 (UTC)

I agree that on other projects it's hard for me to revert vandalism - more so than if I never had rollback and this is certainly something we should consider, especially if the user can prove they are an admin on another project. Ryan Postlethwaite 13:04, 11 December 2007 (UTC)

I updated the requirements with ideas from one of the above discussions and this one. I worded so that a sysop on another English Wikimedia project (English any Wikinews, English any Wiktionary, etc.) would also be eligible. FunPika 20:06, 11 December 2007 (UTC)

Requirements

Ladies and gentlemen, let's discuss the various options in more depth. I think that's more fruitful than just being bold and arguing through edit summaries (in this case). We have a strict option: 500 mainspace edits and 6 months of good behaviour. There's a looser option: 200 mainspace edits and 2 months of good behaviour. And there's an administrator know-how option, without strict criteria and instead assuming know-how in evaluating users and their behaviour. Finally, there may be other options, please bring them up. -- Pepve (talk) 04:27, 12 December 2007 (UTC)

There's also the option where we stop waffling on talk pages and actually decide on something for once – Gurch 11:38, 12 December 2007 (UTC)

I think it should be mainly admin discretion, but a set of advisory guidelines would help users to know if they will likely pass and help keep uniformity in admin decisions. Mr.Z-man 16:03, 12 December 2007 (UTC)

I think setting up criteria for this is like setting up criteria for RfA. How about just set up a process, similar to RfA and/or bot approval? This shouldn't be automated. - jc37 01:28, 13 December 2007 (UTC)

Because unless we eliminate or slow down rollback scripts, no one would want to go through something like that. Mr.Z-man 01:49, 13 December 2007 (UTC)

So? I'm sure that, given the choice, no one want to go through RfA or BAG... But they do. This is no different. - jc37 02:01, 13 December 2007 (UTC)

It is different. RfA is a vote with fixed percentages. (No, don't say it isn't, you know perfectly well it is, and I'm not sticking exclamation marks in front of anything). This is not – Gurch 10:34, 13 December 2007 (UTC)

This is slightly off-topic, but I won't argue that RfA currently is a bastardised hybrid of voting and consensus, and these days seems to lean more to the "voting" side. (Personally, I suggest removing the "support" section, and re-affirming bureaucrats' discernment in closure.)

That aside, Not sure what you're saying in relation to this discussion. Could you clarify? - jc37 12:32, 13 December 2007 (UTC)

That this is different from RfA – Gurch 15:38, 13 December 2007 (UTC)

If there's a process we want to base this off of, it shouldn't be WP:RFA: it should be WP:USURP. There is no exact cutoff point for whether an account requesting usurpation is too "new", or too infrequently used: that's up to the bureaucrat to decide. Anyone may participate in the discussion with bureaucrats, bringing points in favor of and against renaming, but when discussion does occur, it is certainly a discussion and not a vote. (Using WP:RFA as a guide would result mainly in vacuous comments.) If an account is not renamed, no one's feelings are hurt. In the same way that one checks if the latently usurped account has deleted contributions, e-mail enabled, etc., one could check the block log, last 50 or so reverts, and talk page of an individual requesting the rollback permission for (in common-sense terms) red flags. Red flags, if not adequately explained, might result in a request being turned down. Then, of course, we'd want to have some more formal rules: i.e., getting blocked for 3RR means that rollback is taken away from non-admins, which may be re-requested at any time. Gracenotes^T § 17:35, 13 December 2007 (UTC)

I don't disagree. I used the comparisons to RfA and BAG to indicate that "some" process should be used in granting, and that process should require final discernment by at least a bureaucrat. - jc37 23:36, 13 December 2007 (UTC)

Why? The consequences of abuse of the rollback feature are no worse than the consequences of abuse of the editing feature. Are you suggesting we require approval by a bureaucrat to edit, as well? – Gurch 00:17, 14 December 2007 (UTC)

What if the bureaucrats approve of the process itself and any associated criteria, so that admins can implement this approval? –Pomte 00:35, 14 December 2007 (UTC)

New processes are not required to meet the approval of bureaucrats. It's not up to them alone to decide how things are run, it is up to the community. – Gurch 10:23, 14 December 2007 (UTC)

Doesn't this just amount to a "sub-admin" position?

And aren't we, in general against that? I mean, admin is fine by me. If a user wants rollback, go through RFA like everyone else. Either a tool is to be trusted to ALL editors or to admins only. Creating multiple levels of adminship between regular user and full admin (which is what this proposal amounts to) seems silly. You would patently NOT want all regular users to have access to this (you think editwars are bad NOW...), and so why do we need to create sub-admin positions to have access to it? --Jayron32|talk|contribs 06:47, 12 December 2007 (UTC)

Someone who mostly fights vandals here and would find the rollback button helpful while vandal fighting wouldn't be able to get the button (All sorts of "Didn't write a FA" and possibly "Doesn't participate in AFDs or project space discussions" opposes). FunPika 10:48, 12 December 2007 (UTC)

Imagine that several tools could be separately enabled for users. Rollback being one of them, and others possibly being deleting pages or viewing deleted pages. Would we not want users to specialize for a certain task, and only have the tools to perform that task. Wikipedia might become (or be) too big to master all tasks at once. A whole spectrum of admins may then better suit the wiki approach. (Note that I'm not utterly convinced of this, but it's worth considering.) -- Pepve (talk) 15:39, 12 December 2007 (UTC)

On a technical level, it would be possible (with current software) to create a new usergroup for every admin task. Rollback however, unlike every other admin action can be undone by anyone in a couple clicks. As far as edit wars go, that would be considered an abuse of the tool and grounds for removal. Mr.Z-man 16:01, 12 December 2007 (UTC)

That sounds great in theory. However the potential for WP:BITE, and just the sudden deluge of edits will mean that it will be likely that many of those edits won't be "checked" if they were actual vandalism. We'll also potentially have a new flood of complaints at WP:AN/I (whether justified or no).

I realise that this is being pushed in order to replace one of the "tools" because heavy use is slowing down the servers. But that's doesn't mean we should push forward into potential hazard. - jc37 01:28, 13 December 2007 (UTC)

Why a sudden deluge of edits? The primary use for this is vandalism reversion. The vandalism will be reverted one way or another (hopefully). The question is will people do it manually, with scripts, or with rollback. I don't think people who are used to reverting vandalism are suddenly going to get extra aggressive and lazy once they get rollback. And who is going to complain at ANI? The vandals? Mr.Z-man 01:44, 13 December 2007 (UTC)

Just because you intend something to be used in a certain way, doesn't mean it will be so. - jc37 02:01, 13 December 2007 (UTC)

What do you know, we already have RfA-like drama before there is an actual procedure. That's even quicker than I predicted higher up in the discussion. --TheDJ (talk • contribs) 02:47, 13 December 2007 (UTC)

(Looking around for the "drama") - Personally I see an open, ongoing discussion. Not sure what you're seeing. - jc37 02:57, 13 December 2007 (UTC)

Jc37, indeed, that's why we should form a process that sticks to the intention. (Which may fail, but we should of course try.) -- Pepve (talk) 20:57, 13 December 2007 (UTC)

One problem here is the conflation of user groups and associated user rights. User groups are merely cascading associations (grab bags, if you will) of user rights; the fact that the rollback permission belongs to the sysop group doesn't mean the permission is forbidden from use elsewhere. Jayron32's proposal is nice in spirit, but RFA is simply too political for it to be practical. The progression "vandal fighter → admin" has been disowned much by the community. Gracenotes^T § 05:43, 13 December 2007 (UTC)

I thought I read somewhere that the policy of Wikipedia was that most responsible editors would become administrators as a matter of course. Sure, many editors might not know how to use all the administrator tools. So what. If they are responsible editors, they can be trusted to learn how to use a tool before trying it out in a real situation. --Gerry Ashton (talk) 00:08, 19 December 2007 (UTC)

Yes, the original idea, back when Wikipedia was set up, was that responsible editors would become administrators as a matter of course. Unfortunately, candidate administrators are today held to much higher standards and denied adminship unless they can demonstrate both article writing experience and a "need for the tools" – and for reasons that continue to escape me, merely dealing with vandalism isn't considered to constitute a "need for the tools", even though two of the 'tools' – rollback and blocking – are primarily used against vandalism. Hence this proposal – Gurch 18:53, 19 December 2007 (UTC)

Newbie Question

Can't non-admin do most of the rollback features via scripts that are available such as Twinkle? Gtstricky^{Talk or C} 22:10, 13 December 2007 (UTC)

Yes, but at a performance cost. See [3]. When I did this reversion, I sent the server 72 KB of data, and the server sent me 178 KB of compressed data (uncompressed, that would be 315 KB). With the rollback permission, the server would send 35 KB (142 uncompressed), and I would sent it 2 KB. (This is without any script, although a rollback request can be made with a script.) With script rollback, the server sends approximately 500% more data, and I send it 3600% more. This is slower for both me and the server. Nearly all of Twinkle's requests are backend, as they are API. The rollback permission minimizes this, a lot. I understand that we shouldn't worry about performance, and that none of my analysis takes into account how scripts like Twinkle affect things server-side. However, giving users rollback is something the community can do, and I certainly wouldn't mind sending 3% of what I send now just to revert some vandalism. Gracenotes^T § 23:50, 13 December 2007 (UTC)

The amount of data sent is not the only issue. Bandwidth is not usually a problem, at least for those of us with fast connections. Speed, however, is – it may be slightly so when reverting vandalism and every second counts, but more importantly, it is very much so at the other end, where thousands of requests are being handled every second and unnecessary milliseconds really do add up. Page views are cached, and cached requests are relatively fast, because the page can be served directly or with minimal processing. To revert an edit, however, it is necessary to make two requests neither of which are cached: bring up an edit form, and submit the edit. To revert multiple edits, the page history has to be requested too, and that sometimes isn't cached either. To roll back an edit, only one request (the 'submit' one) needs to be made, even for most cases where multiple edits need reverting, since it reverts all edits by the last contributor, and while the request that does have to be made is the slowest (though as Gracenotes points out, still far smaller), the speed benefits of cutting the number of requests in half or by two thirds are very real – Gurch 23:57, 13 December 2007 (UTC)

For those of you who like looking at pictures, look at this one. This is a graph showing the number of requests per second to Wikimedia projects. This peaks at around 45,000 requests per second. The blue, green and orange areas of the graph represent requests made to the US, European and Asian clusters, respectively. Virtually all of these requests are anonymous users viewing pages – in such cases, the rendered page is cached; once one person has viewed it, it can be served very quickly the next time it is requested. When a logged-in user views the page, a small amount of processing is necessary to, for example, add their username to the user 'portlet'. However, the content area of the page – i.e. everything underneath the page title – is still cached and can be displayed very quickly. (To see roughly what is cached, view a page and then add "?action=render" to the URL. This is an oversimplification of the real situation and ignores things like user preferences, but is enough to give the general idea). A very small fraction of requests are for advanced things like requesting an edit page, saving an edit, viewing a special page, moving a page and so on. These requests are represented by the little black line at the bottom of the graph: as you can see, they account for only a tiny fraction of all requests. However, most of the Wikimedia servers' processing time is expended dealing with these requests, as they require the relatively slow PHP scripts that form MediaWiki to be executed, in order to render a page that nobody has asked for before, or which has gone out of date since it was last asked for. These scripts will assemble the HTML of the page and then serve it to you, usually retrieving information from the database and sometimes parsing some wiki-markup as well. All of these things are very slow compared to just pulling a pre-rendered page out of memory and sending it. So you can see how it is better do only do it once (rollback) than twice or even three times (anything else) – Gurch 00:15, 14 December 2007 (UTC)

Thanks Gurch and Gracenotes, that makes sense. So we (us non-admin type) can do it now with a handicap. However, currently you need to know HOW to do it (find the script and put it to use). I think those steps create some current safeguards that need to be carried over. WP:AWB has similar requirements. Humbly I submit that there could be a blanket user right that would grant you advanced editing tools, WP:AWB and the new Rollback feature. Again my thought is that the tools are available now anyway, I think it would give admins more control over users using the rollback features not less. Cheers. Gtstricky^{Talk or C} 00:06, 14 December 2007 (UTC)

I think you're being just a little paranoid there. To use rollback you would also need to know how to use it. You would furthermore need to know how to find the request page and make a request there, and you would also need to have sufficient experience that administrators would not reject your request. This is, to be honest, far more complex than the knowledge needed to use a script (all you have to do is see it in use, follow the link it advertises itself with in every edit summary, and follow the easily-understandable instructions on the page you are sent to which involve copying one line of code into a page that is linked for you and therefore basically reduces to knowing how to edit a page. – Gurch 00:15, 14 December 2007 (UTC)

I might not be communicating it well but I think we are saying the same thing. Gtstricky^{Talk or C} 01:46, 14 December 2007 (UTC)

In simple terms, when you use a script, you edit the version you are reverting to, and then the page is automatically saved. With admin rollback, you completely miss the editing screen, so you go via one screen instead of 2. It's only possible with the wikimedia rollback function that administrators currently have. In a nutshell, it's far quicker than any other method (even with the best script). Ryan Postlethwaite 00:10, 14 December 2007 (UTC)

Again, why?

I've read the other posts above, and am not quite satisfied with the answers. Undo is available to all, and the only argument that I see for giving non-admins (such as myself) rollback is that it's quicker. I can't stress enough that speed and accuracy are inversely correlated. I have seen many, many good patrollers that often revert or undo to a bad version of a page. I'm not innocent of this (though I think that I've self-corrected all of my errors), but many of the most fervent denizens of RecentChanges have a worse record than mine. Quicker and quicker tools lead too easily to less and less thought, and to a more flippant attitude. It is already all too easy to get into a "click-click-click" routine of patrolling, which leads to a lot of errors. This, in itself, might not be a great problem, except for the fact that the vast majority of the time that an established user reverts a page, it bypasses the usual scrutiny of RecentChanges, since most patrollers have Logged-in users turned off. In short, why make something that often requires thought so much quicker, while not changing the fact that it requires thought? It seems an invitation for a huge number of much more difficult vandalism fixes, since we will no longer be able to say with relative confidence that the last non-ip user has left a good page. I hope this was relatively clear. If not, I'll try to reiterate in clearer language. --Storkk (talk) 15:53, 14 December 2007 (UTC)

With ClueBot around, you can't say that anyway. I don't see why you are so confident administrators get it right now but other (often just as experienced) users cannot; I'm sure there's some good faith to be assumed around here somewhere. Anyway, it's not just the increased speed for the user, but also the other end that matters. Reversion scripts make several requests in quick succession, in an attempt to emulate the much more efficient rollback from the point of view of the user. Compared to rollback, they still appear relatively slow to the user, and as has been explained above, they make two to three times as many slow backend requests and unnecessarily transfer more than 30 times as much data. If the developers thought that the reversion scripts were a good idea, they would have added the functionality to MediaWiki itself. Instead, if anything they are expressing concern about the resources that certain user scripts are consuming. They have already implemented the ability for rollback to be given out to anyone and it only requires the community to stop being paranoid in order for it to be available – Gurch 16:10, 14 December 2007 (UTC)

With ClueBot around, yes, I can still say that pages get left for long periods with bad versions because of careless reverting by over-zealous, non-thinking patrollers. That was my main point. I still stand by "time taken and accuracy are inversely correlated". The one saving grace of this idea, IMHO, is the server load. This is a major point, I just don't think that this is the best way to solve it. --Storkk (talk) 17:37, 14 December 2007 (UTC)

Note:, I'm not saying that if this proposal is "enacted", I wouldn't request the priviledge... I would, but I still don't think it's a good idea for us non-admins to have. Let's face it, 250 mainspace edits is very small to ensure familiarity with the policies/guidelines. --Storkk (talk) 15:59, 14 December 2007 (UTC)

I removed the 250 number several days ago. Whoever added it and the other figures that they plucked out of thin air seems, for reasons that baffle me, only to be able to think in terms of numbers. No doubt they are one of the people who complains if an AfD is closed as delete despite numerically more keep votes, or that an RfA should have been failed because the support percentage was 78.2%. If a user is capable of making normal edits, they can do no more damage with the rollback feature than without it, and can be blocked just as easily whether they have it or not. Suggestions to the contrary are sheer paranoia and I am sick of hearing them – Gurch 16:10, 14 December 2007 (UTC)

Also, thanks to RfA voters' inability to do anything other than act like sheep, Cobi is about to recieve adminship despite only having 300 mainspace edits, which kind of defeats your argument that such a number is too low for only rollback – Gurch 16:12, 14 December 2007 (UTC)

I agreed with you (and voiced my opinion) on that RfA before seeing this proposal. No, it doesn't defeat my argument. Your point seems akin to what we often see on AfD arguments, to which one replies WP:OTHERSTUFFEXISTS. Whether or not he gets adminship has nothing to do with this proposal. I might agree with this if there was an absolute zero tolerance policy on its use for anything other than RC-patrolling, and a very low tolerance for that as well: i.e. give it to almost anyone, but take it away very quickly when it's abused (it WILL be for edit wars), and almost as quickly when it's used repeatedly without proper investigation of where a page should be reverted to. Then again, this is all just my 2 cents: I very much doubt that this proposal will fail... why would anyone disagree with granting themselves more "!power" ? Anyway, if I had to put numbers on a "requirement", I'd say: min 3 months, at least 1000 edits to mainspace, and the last 1000 edits anywhere must contain no non-vandalism-related revert wars. --Storkk (talk) 17:37, 14 December 2007 (UTC)

How would it be abused for edit wars? Rollback allows edits to be reverted a few seconds faster than they otherwise would be. Edit wars usually take place at the rate of something on the order of one reversion every few hours, if that – and that would still be quite fast enough for a 3RR block to be issued before the day was out. If you're about to make your fourth or fifth revert of the same page within 24 hours, it doesn't matter if you make it at 21:08:19 (undo) or 21:08:16 (rollback). Unless you're dealing with vandalism, you're almost certainly going to be blocked anyway. Anyone foolish enough to repeatedly rollback another person's edits would be treated in the same way as people who currently do the same thing a few seconds slower, and blocked very quickly. If they find someone else with rollback stupid enough to engage in an edit war with them, then... well, assuming they were both going to edit war anyway, all that will happen if anything is that rollback will lead to them being blocked more quickly, as the faster rate of editing will draw attention. So... what is the potential for abuse exactly?

Three months and 1000 mainspace edits? I see editcountitis has infected you. We used to give out adminship to people with that not so long ago – Gurch 21:09, 14 December 2007 (UTC)

Please note WP:KETTLE [4]. Pragmatism and realism != editcountitis; I gave a number that seemed sensible to me: reversions of vandalism are a great way to rack up edits without gaining much familiarity with policies and guidelines. Anyway, this proposal will stand or fall as it will, and I've given my opinion. I don't think I have any more cogent points to make... failing something else I feel the need to reply to, I'm done here. --Storkk (talk) 01:11, 15 December 2007 (UTC)

Quote: "...can be blocked just as easily whether they have it or not. Suggestions to the contrary are sheer paranoia and I am sick of hearing them" : I suggested nothing to the contrary. I respectfully disagree with User:Rschen7754's argument, which I think you are referencing above. I agree, that it would be relatively easy to undo a granting of rollback. However, it is undeniable that there has to be some kind of review before granting, and perhaps some kind of review before revoking. Irrespective of whether accomplishing the removal of privileges is easy or not (I'm sure it's both easy and quick), it will be accompanied by a certain amount of bureaucracy. Increasing the amount of Bureaucracy here, many think, is a Bad Thing(tm). Whether the good outweighs the bad I'm not sure -- there is a big "invisible" (to most of us plebs) good to decreasing server load, but there is an evil inherent in increasing the number of !vote pages we have. There are already far too many, which inevitably prompts flippant or unconsidered !votes. Note that whether or not you are "sick" of an argument does not enhance or detract from its merit--but I agree that this particular argument has been refuted. Anyway, I've said what I needed to to give my opinion. Have a nice day! --Storkk (talk) 17:58, 14 December 2007 (UTC)

I certainly don't want too much bureaucracy, but as you said, we need some semblance of process. If a process is efficient at processing items, this might make it a bureaucracy, but it certainly doesn't make it bad. If the rollback-granting process becomes inefficient or unnecessary, it should be reformed or removed. However, concern that this process may become overly bureaucratic is not a particularly constructive reason to not do it, merely a good reason to plan it better. As I suggested above, this process need not be based off of WP:RFA, or any other system involving "votes" as primary units of discussion. Gracenotes^T § 23:04, 14 December 2007 (UTC)

I'd be more comfortable with this if revoking was substantially quicker than granting. How that would be done, I don't know. Just my final 2c. --Storkk (talk) 01:11, 15 December 2007 (UTC)

It will be, in practise; requests will probably sit on the page for several hours or even a few days before being processed, whereas anyone using rollback for edit warring will be blocked and have it taken away within minutes – Gurch 02:26, 15 December 2007 (UTC)

I agree with you there. Gracenotes^T § 17:25, 15 December 2007 (UTC)

And as I said in an above section, opposing because of what might happen is not really helpful; anyone can try to predict the future to suit their opinion. I suggest basing this off of the approval process for programs like AutoWikiBrowser and Vandalproof. Neither of those require a vote or or significant bureaucracy, and to date neither tool has suffered from widespread abuse and neither approval process has transformed itself into a bureaucracy. Mr.Z-man 00:29, 15 December 2007 (UTC)

Oh Gods, No.

I don't understand why this has even been made into a proposal. It looked to me like there was pretty clear consensus against this in the discussion here: Wikipedia:Rollback for non-administrators. At the very least I'd say there was no consensus to go ahead with it. - Kathryn NicDhàna ♫♦♫ 06:45, 16 December 2007 (UTC)

You are aware that this proposal is quite different from what was originally planned? Mr.Z-man 07:09, 16 December 2007 (UTC)

I'm seeing sections like this:

Due to the nature of the rollback button, access cannot simply be granted to everyone, but will instead be assigned by administrators, on request. Administrators should evaluate requests for rollback on individual merit and review the users contributors edit history before granting them the permission.

When my impression is that a significant number of editors in the discussion did not want admins to be handling this. I think the discussion there was a better way of handling this. - Kathryn NicDhàna ♫♦♫ 07:54, 16 December 2007 (UTC)

Wikipedia talk:Rollback for non-administrators#Random thought. --MZMcBride (talk) 08:01, 16 December 2007 (UTC)

Forget what you think others want for the moment: why do you not want administrators handling it? Do you trust them to block users, delete pages, protect pages, edit the Main Page and sitewide JavaScript and yet not give out what is merely a much more efficient version of something that can be done with scripts, and revoke it (along with a block) the moment something goes wrong? Or do you not trust administrators at all? Given that our ability to edit pages is, through blocking, already at the mercy of this group of people, adding what is merely a specific type of editing that wasn't previously available to the mix doesn't seem like a huge leap. As I have explained above at length, the benefit is real, the possibility for abuse is minimal (anyone seeking to use this for abuse would be better off saving themselves the months it would take to get approval and just reverting edits normally) and I really don't see why it would require bureaurat approval, RfA-esque voting nonsense or anything else like that. It's not going to be given out to all autoconfirmed users, in fact the way things are going the standard will (unfortunately) be so high you could have got adminship with less a couple of years ago – so just what is your concern? – Gurch 11:14, 16 December 2007 (UTC)

Perhaps I'm not totally clear on how the approval process will go. Will the requests just go to a board like AIV or similar, with whatever admins who are willing to handle it giving approval? I'm mostly concerned about adding to the backlog of tasks we have to handle as admins. I don't think the world will end if we have to add approving rollback users to the backlog, and I do think it's preferable to having mini-RfAs. I'm just surprised this has become a proposal at all because I think there was no consensus to give non-admins the rollback. - Kathryn NicDhàna ♫♦♫ 21:40, 16 December 2007 (UTC)

There is obviously no consensus for all logged-in users to immediately be able to access rollback without any approval process at all whatsoever. Or is the same applying to an approval process filled with editcountitis and timeitis? FunPika 21:43, 16 December 2007 (UTC)

I find it extremely unlikely that there will be a backlog of more than a day or so. After an initial rush, the number of users requesting this is probably going to be not that much greater than the number of users requesting adminship, if that – Gurch 23:48, 16 December 2007 (UTC)

Also, if you're worried about the number of tasks administrators have to handle, persuade people to drop the "vandal fighters don't need to be administrators" attitude at RfA – then you'll have more administrators and this proposal will become unnecessary – Gurch 23:51, 16 December 2007 (UTC)

By contrast, though, won't enabling this increase standards at RFA? Gracenotes^T § 22:50, 17 December 2007 (UTC)

I thought the same in this section, but after more thought, as Mr.Z-man said, giving/revoking rollback would be less of a big deal than performing blocks. Some people may increase their standards, but I don't think this will be too problematic. Acalamari 23:11, 17 December 2007 (UTC)

I'm not sure why it would increase standards at RfA, unless people decide that a certain candidate can be trusted to take away people's editing ability completely but not to merely modify it slightly. Though of course given the sort of RfA stupidity that made this proposal necessary it may well happen anyway – Gurch 23:51, 17 December 2007 (UTC)

My original concern was that some people may see it as "another admin ability", and increase their standards accordingly; but yes, I do agree strongly that blocking is a bigger deal than this. Acalamari 00:25, 18 December 2007 (UTC)

Good points, but that's not quite what I mean. Let me rephrase the question: would people be less willing to give a vandal-fighter sysoppery because he or she already has the rollback permission? Gracenotes^T § 00:44, 18 December 2007 (UTC)

That's a possibility, but don't forget, anyone arguing "Oppose - user already has admin rollback rights" can be countered with the following responses: (1): full-admin rollback would be unlimited in the amount of reverts, unlike non-admins with the rollback permission, who would be limited to a few reverts a minute. (2): being an admin would mean being able to block users as well as have unlimited reverts, therefore making a more efficient vandal-fighter. Acalamari 01:02, 18 December 2007 (UTC)

I'm hoping argument (1) won't be implemented because it would make this feature pointless and, yes, (2) is logically true but that doesn't stop people ignoring that face and persisting with the "RC patrollers don't need adminship" attitude. And Gracenotes, what you're basically saying is that RfA will continue doing exactly what it is doing now - refusing to give RC patrollers adminship unless they've done significant other work elsewhere. This proposal is only here to work around that – Gurch 11:38, 18 December 2007 (UTC)

In some ways it may reduce standards as it will have been proven that users are competent with one, however small, part of adminship. Tiddly-Tom 13:20, 18 December 2007 (UTC)

Yes!

I thought giving easy access to the undo function on article history pages was a bad idea, but this? It's suicidal! Here are some reasons why I am against this.

• The undo function gets abused often enough, without an easy access revert tool making things even worse,
• We have three highly effective vandal bots (although it's fair to say one is much more effective than the other two) and a whole host of admins and other decent Wikipedians who patrol the recent changes with glorious results. Making everyone an admin in terms of vandal fighting would just (at least) treble the amount of vandal fighters, and there'd be so much competition to actually get something done...
• In the hands of unestablished users, a free for all revert function is a great opportunity for vandals. For a start, it doesn't look like vandalism on the recent changes display list.

No, an altogether awful idea. I'm sorry if I seem harsh, but I'm just extremely surprised that such a proposal has been suggested. Lradrama 11:13, 17 December 2007 (UTC)

Apparently you haven't read the proposal. Nobody is proposing to give it to anyone. In fact it is precisely the "other decent Wikipedians" you list above to which this is intended to be given. Nobody will recieve it without the approval of an administrator, and it will be taken away at the first sign of problems. Your use of phrases like "unestablished users" conveys the impression that you think everyone will have access to it, which is not the case at all. It's actually very likely that nobody will get it unless they're sufficiently experienced they could have passed an RfA two years ago; this proposal is only here to compensate for the community's attitude that "vandal fighters don't need adminship" when in fact this feature is designed specifically for vandal fighting – Gurch 11:34, 17 December 2007 (UTC)

Undo is, in fact one of the most correctly used functions. So many IPs I have seen using it reverting vandalism has been astounding. Your second point - bots will remove maybe less than a 3rd of vandalism. Does it matter that people want to RC patrol? Your last point: Vandals can easily add a revert script to their monobook. What's the difference here, compared to rollback? Redrocketboy 11:31, 17 December 2007 (UTC)

Oo dear. I have misread it. I'm sorry everybody but this actually sounds like a very good idea. Many thanks to Gurch for pointing this out to me.

I remember very well, before becoming an admin, how frustrating it was patrolling the recent changes without the aid of a vandal tool. It was really frustrating. I had tried to install both Twinkle and an admin-like revert tool but both had failed, for various reasons. And I had to revert the old-school way. Now I have the tools, it is so much better and so much easier. There are lots of great users out there with the potantial to become good vandal fighters but, who like me, may have had trouble with other vandal tools that want to become a little more competant at it. Therefore, I am wholly in support of this proposal. :-)

Once again, thankyou Gurch for informing me of my huge mistake, and for not letting me look a complete idiot for too long! Sorry! Lradrama 11:44, 17 December 2007 (UTC)

Let's give this a go

I think we should really consider giving this a go. The major concerns here so far are misconceptions about this proposal - some people look like they haven't even read it. We're not giving every user the tool, just people that have shown to have a need for it and can be trusted. I don't suspect there to be huge backlogs, and it's not the end of the world if someone has to wait a couple of days to get it. I would suggest we ask the developers to implement it for a two week trial, don't advertise it and see who comes to request it. If there's huge problems, we can take it from there - but I really can't see the potential for harm given that proposed users will be vetted first. Ryan Postlethwaite 11:42, 17 December 2007 (UTC)

Yes, I agree with you. I'm one of those people who completely misunderstood it. I've read it carefully now, and, after reflection on the times I spent RC Patrolling before becoming an admin, I am in big support of the proposal. Lradrama 11:47, 17 December 2007 (UTC)

Agree. Redrocketboy 12:17, 17 December 2007 (UTC)

Agree, what is the worst that can happen? If there are still doubts you could limit it to a few [hundred?] users for a few weeks and see how it goes. I doubt you even need to do that because if it really does not work, the developers can remove it for everyone to whom it was granted in one sweep. Tiddly-Tom 14:28, 17 December 2007 (UTC)

I don't think more than a few hundred users are going to ask in the first few weeks, or possibly even months. As I mentioned above, I expect the request rate to quickly settle down to roughly the same rate as adminship requests – Gurch 14:34, 17 December 2007 (UTC)

I agree with a trial to test this out. Acalamari 17:35, 17 December 2007 (UTC)

One thing that isn't clear at the moment: is this going to be rate-limited? I took out the sentence in the proposal that said it would be because, as I have mentioned above, the rate that was given is somewhat below that which RC patrollers generally have to maintain during busy periods, and would thus be rather annoying to the very group of people for whom it is intended while providing little benefit. If someone went crazy with the rollback tool, it would only be a matter of minutes before they were noticed, reverted, blocked and permanently stripped of the ability to use it, which means that it's unlikely anyone will even try; the sort of people who currently vandalize with throwaway sockpuppet accounts are unlikely to suddenly start spending months making useful contributions with each account to get it rollback-approved only to then start reverting everything. The only potential abuse would seem to come from its use in edit wars, which as I have also explained above would happen at a slow speed and would still result in a block just like any other sort of edit warring, so I don't see how a rate limiter would help. It might also be abused if administrators start giving it out to just anyone, but my experience of this project suggests that generosity is not a concept most administrators are familiar with (if anything, I feel they're likely to go too far the other way, and block/de-rollback RC patrollers for making a single mistake) – Gurch 14:34, 17 December 2007 (UTC)

By default, it would still be rate limited, but there is a configuration variable that can be changed to exclude people in the rollback group from the limit. Mr.Z-man 20:49, 17 December 2007 (UTC)

Well that's no good, unless we want another bureaucratic process to decide who to exclude. Can the limit not just be removed, or at least raised to a more tolerable number like 10/minute? That would still prevent RC patrollers from using rollback for one of its intended purposes -- rolling back the edits of a high-speed vandalbot by going down its contribs pagfe clicking rollback links -- but would at least mean they wouldn't keep getting error messages during busy periods and having to go and revert manually. That would seriously begin to annoy RC patrollers after a few minutes – Gurch 11:40, 18 December 2007 (UTC)

I would say I would submit a Bugzilla bug if no objections were raised to a trial within 24-48 hours, but there is one thing I want to figure out. Are we going to use Userrights or Giverollback? If we are going to use Userrights the bcrats should be aware before their Makebot/Makesysop pages are suddenly replaced with a limited version of Userrights. Because it appears no one has a problem with switching to Userrights in this discussion, I will post a message on the bcrat noticeboard about this. FunPika 20:14, 17 December 2007 (UTC)

Yeah we would use userrights I think - it seems the simplest way to go. The coding that you've put on WP:BN is the idea we need (I think!). Ryan Postlethwaite 22:45, 17 December 2007 (UTC)

Realistically if the feature is added it wouldn't be a trial, it would be permanent: what would be the incentive of those who supported it ever to come back to discuss whether it was successful or not? Christopher Parham (talk) 01:47, 18 December 2007 (UTC)

None, unless it turned out not to work as well as they thought and they decided it would be better off removed. Similarly, no incentive for anyone else to come back here unless they thought there was a problem with it. So if people do come back, then there may be a problem; if not, there is no problem and it can be kept. Isn't that the whole point? – Gurch 11:34, 18 December 2007 (UTC)

Please read this; it will give valuable insight into the current situation from a technical perspective.

When MediaWiki was first developed, user rights were not nearly as customizable as they are today; there have always been issues with who's able to give and take which rights. When bureaucrats' ability to grant +sysop was revoked, Tim Starling created mw:Extension:Makesysop, which effectively created the "steward" group and allowed bureaucrats to have the ability to give, but not revoke, +sysop. The extension was created because at that time, Special:Userrights (which on a default configuration allows bureaucrats the highest powers) was not able to handle the ability to allow users to give certain rights but not take them away. Eventually, Rob Church created a similar extension that allowed for bureaucrats to have the ability to add +bot to an account (mw:Extension:MakeBot). This left Wikipedia and all related Wikimedia Foundation wikis with Special:Userrights, Special:Makesysop, and Special:Makebot, the latter two being the special pages associated with their extensions, respectively. A bug was filed about creating "more modular" user rights, bugzilla:6711. In June of this year, Simetrical resolved that bug with rev:23410, which would, at some point, allow the deprecation of the two extra extensions (Makebot and Makesysop). A bug was filed regarding this: bugzilla:11645. As of today, bugzilla:11645 is still open and unresolved.

All of that background was to explain this: currently Special:Userrights is limited on the English Wikipedia to anyone in the steward group (i.e., no one), or anyone in the founder group (i.e., User:Jimbo Wales). Special:Userrights is not used for +/-sysop, +/-bot, or anything else. Bureaucrats use Special:Makebot and Special:Makesysop, and Brion Vibber, the CTO for the Wikimedia Foundation and the lead developer of MediaWiki has stated that there is absolutely no intention of adding another extension as that would be silly. As it currently stands, the bug allowing for greater use of Special:Userrights seems entirely immobile. This is all to say that the Giverollback extension and or the switch to Special:Userrights will almost certainly not happen any time soon. These issues have been thoroughly discussed and reviewed by both developers and sysadmins; thus, the trial period cannot go forward at this time. While you're free to file a bug, it will almost undoubtedly be ignored; many, many other bugs face this fate.

There are other proposals that would not introduce as much bureaucracy and would not require a drastic shift in the way user rights are assigned and handled. I urge everyone to look at that discussion and please comment. Cheers. --MZMcBride (talk) 21:35, 17 December 2007 (UTC)

Are you sure the reason why the bug hasn't been resolved is that until now it would have added no new features and thus really only been a code change? If we announce that we need to be able to give out rollback separately the developers have the option of resolving the bug or adding another extension; they've already said they won't add another extension and since the code needed to resolve the bug has mostly already been written anyway resolving the bug would presumably be far less effort – Gurch 22:33, 17 December 2007 (UTC)

The proposal you link to suggests that all users be given rollback. Several people have raised objections to that here in the mistaken assumption that this proposal was suggesting the same thing, and consensus seems to be against it – Gurch 22:35, 17 December 2007 (UTC)

I doubt that is the reason. See bugzilla:10727. That is an attempt to get addgroups/removegroups enabled to create a "test-admin" group on the Wikimedia Incubator. FunPika 23:19, 17 December 2007 (UTC)

Well, until 7 days ago nobody seemed to have come up with a way for that to function correctly, but in the most recent comments it seems they have. Perhaps the developers haven't noticed yet? You should try giving them a nudge and they might fix it – Gurch 23:49, 17 December 2007 (UTC)

Bureaucrat noticeboard comments

• Wikipedia:Bureaucrats' noticeboard#Special:Userrights

User:Redux just left a comment which I think is worth investigating. (And, as you may surmise from my comments above, I agree with his comments concerning limitations in who can "grant".) - jc37 11:50, 18 December 2007 (UTC)

• He seems to be arguing that just because it involves user rights, it must be approved by the Foundation. As I've pointed out there, user rights is merely the most convenient way to implement this particular feature, and there are already many things that administrators can grant and revoke – including the most important one, the ability to edit – that just happen not to be implemented as user rights – Gurch 12:12, 18 December 2007 (UTC)

Page protection

I've read that it was once possible to rollback while blocked, and this has since been fixed. Since this indicates an underlying weirdness in the implementation of rollback, I'd just like to put this out there to make sure that the devs check to make sure that it's not possible for an ordinary user to rollback a protected page. —Random832 16:41, 20 December 2007 (UTC)

I think it was blocks that were dodgy. Admins could in effect do anything while blocked, except for edit. Gradually, all the abilities were removed except the ability to unblock themselves, and some other things, such as viewing deleted revisions and viewing Special:Unwatched page, the blocking interface etc - not too sure on those though. Redrocketboy 17:00, 20 December 2007 (UTC)

I just tested it on my wiki, non-admins get a permission error when trying to rollback on a full protected page. Mr.Z-man 01:01, 21 December 2007 (UTC)

Bots

I am not an expert on the topic so don't shoot me down ;) I don't know if they already do or not, but this may be a good way of reducing server load by anti-vandal bots. I can't myself think of any reason why they would not be allowed it. Comments? Tiddly-Tom 23:40, 21 December 2007 (UTC)

Whether bot accounts are granted access to the tool or not is for administrators to decide if the proposal is actually implemented. I don't see any reason why they wouldn't, provided the account had been in use for sufficient time that they were satisfied there would be no problems – Gurch 19:53, 22 December 2007 (UTC)

Too much bureaucracy

My response to reading this proposal is: This adds another layer of bureaucracy to Wikipedia, without sufficient return-on-investment. As everyone and their brother has already pointed out, tools like WP:TWINKLE and WP:POPUPS already enable the apparent functionality for the user. If we want to allow more people to rollback in the name of server efficiency, then just give it to all named users, or some other auto-determined status (by number of edits, age of account, phase of the moon, whatever). Again, everyone already has the same apparent functionality, just via methods which are inefficient for the servers. —DragonHawk (talk|hist) 08:13, 1 January 2008 (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.