Wikivoyage talk:Autopatrollers

Granting template editor rights to active autopatrollers

Swept in from the pub

Given the AndreCarrotflower has permanently protected the Chongqing article, I think it is time to revisit this. I have adjusted the protection level to allow template editors to edit it, and I don't see a reason why we should prevent trusted editors from editing articles if the goal is to prevent vandalism. If people find this a little too liberal, perhaps for a start, we can give template editor rights to all patrollers, and maybe grant them to our autopatrollers on request. What do people think? The dog2 (talk) 22:23, 28 July 2020 (UTC)

The permanent protection for Chongqing was done in error. As with Hong Kong, whose mistaken permanent full protection was noted on my talk page earlier, this stems from the Fuerdai vandalism spree of June 20-21 in which the bad-faith edits were coming so fast and furious that, in reverting the vandalism and preventing its recurrence, I didn't have enough time to pay attention to slips of the mouse hand. Sorry about this. Going forward, any other instances in which I've instituted permanent full protection on mainspace pages, specifically destination articles related to China where the protected status began in June 2020, should similarly be regarded as mistakes and the protection level should be reset to autoconfirmed and the duration to 3 months. -- AndreCarrotflower (talk) 22:31, 28 July 2020 (UTC)

Sometimes "indef" is the best setting for a persistent vandal. If we specify a certain date, the vandal can just mark his calendar and come back as soon as protection expires. "Indefinite" doesn't have to mean "permanent".

The English Wikipedia has invented a level called "Extended Confirmed", which is automatically granted to any account with 500+ edits that was created at least 30 days ago. It is popular for hot-button articles (e.g., Israel–Palestine, high-profile politics, etc.). The general notion is that it's hard to reach on purpose, so few people would go to that much trouble just to get access to a protected article. If we want that intermediate level of protection, then we could ask the WMF to set it up for us. WhatamIdoing (talk) 19:19, 29 July 2020 (UTC)

We could do that, or we could ask if the WMF can make it possible for us to grant additional editing privileges to autopatrollers, or at the very least patrollers. Some of the articles that I think we might need permanent protection for are the United States of America article and the China article. If we could have something along the lines of "allow autopatrollers only", that would work out. After all, autopatrollers are people we have deemed trustworthy, and AC is the only autopatroller I can remember who has gone rogue. The dog2 (talk) 21:19, 29 July 2020 (UTC)

Are there any articles where semi-protection hasn't been enough to bring vandalism down to a manageable level? —Granger (talk · contribs) 00:43, 30 July 2020 (UTC)

Other admins have been quicker than me in catching the vandalism, but the Fuerdai vandal is certainly a serious one for China articles. AndreCarrotflower, Ikan Kekek, Ibaman and SelfieCity, what do you guys think? You guys seem to be the ones to have caught most of the vandalism. The dog2 (talk) 01:18, 30 July 2020 (UTC)

I don't tend to pay close attention to whether a vandal is autoconfirmed, but I think that's rare, and if so, semi-protection should be sufficient in the great majority of cases, but I'd like to hear from anyone else on this. Ikan Kekek (talk) 09:25, 30 July 2020 (UTC)

It's not by any means unheard of for a vandal to become autoconfirmed. I think ExtendedConfirmed is worth looking into, especially for the highest-profile articles on the site as The dog2 suggested. It may be hard for a bad-faith editor to "reach 500 edits on purpose", but it's not that difficult for those who have a genuine interest in contributing to the travel guide. -- AndreCarrotflower (talk) 15:22, 30 July 2020 (UTC)

When I checked at the English Wikipedia, less than 1% of all editors ever reached extended-confirmed. There are more than 30 auto-confirmed accounts there for every extended-confirmed account.

If you want to see exactly who would be affected, then this RecentChanges list (with the 'new' 2017 filters enabled) will show you all the recent editors who have made more than 10 edits but fewer than 500, and so would be excluded by extended-confirmed. (This is also a good search system for finding promising contributors to encourage, so warm up your 'Thanks' button.) WhatamIdoing (talk) 16:40, 30 July 2020 (UTC)

┌────────────────────────────────────────────────────────────────────────────────────────────────────┘ Yeah. The Fuerdai vandal is very persistent, and if we don't permanently semi-protect some articles, all we will be doing is playing cat-and-mouse with this guy. So as of this post, I am formally proposing that we introduce Extended Confirmed status to Wikivoyage along the lines of w:Wikipedia:User_access_levels#Extendedconfirmed. The dog2 (talk) 16:32, 30 July 2020 (UTC)

Extended confirmed protection would also prevent helpful edits from new or casual contributors, so the benefits must be weighed against the costs. Can you or anyone come up with examples of articles where this is needed (in other words, articles where vandalism remains unmanageably severe despite semi-protection)? If not, I don't think this is something we need to worry about right now. We can always come back to this idea if it becomes necessary someday in the future. —Granger (talk · contribs) 16:37, 30 July 2020 (UTC)

@The dog2: Rereading your last comment, I'm confused. First you said "if we don't permanently semi-protect some articles, all we will be doing is playing cat-and-mouse with this guy", but then you said "I am formally proposing that we introduce Extended Confirmed status". Are you trying to argue for indefinite semi-protection or for extended confirmed protection? Indefinite semi-protection is already possible and would not be changed by introducing extended confirmed status. —Granger (talk · contribs) 16:45, 30 July 2020 (UTC)

Granger - "Helpful edits from new or casual contributors" probably aren't needed on the super prominent articles like United States of America that ExtendedConfirmed would cover. If a change takes place that's of such import that it would need to be noted on an article that high up the breadcrumb hierarchy, one of our regulars will almost certainly take care of it. Where edits from new or casual contributors are most helpful is mainly on bottom-level destinations that, by and large, tend not to be magnets for vandalism. -- AndreCarrotflower (talk) 16:53, 30 July 2020 (UTC)

@Mx. Granger: Yes, we can permanently semi-protect by allowing only autoconfirmed users, but as AndreCarrotflower mentioned, it is not unheard of for a persistent vandal to reach autoconfirmed status. Extended confirm will give us an additional tool in the arsenal to deal with persistent vandals like the Fuerdai guy. Of course, an extremely committed vandal could possibly reach extended confirmed status, but it will take a really serious amount to commitment to vandalism to do so. And if anyone is trying to use Wikivoyage to push a political agenda, we could probably catch them before they hit that status. The dog2 (talk) 16:57, 30 July 2020 (UTC)

(edit conflict) @AndreCarrotflower: Thank you for giving a concrete example. As far as I can tell, the United States of America has not gotten any vandalism while semi-protected, let alone an unmanageable amount, so extended confirmed protection doesn't seem like it would be helpful. Moreover, that article has gotten constructive edits from users who would not be extended-confirmed, such as this one. —Granger (talk · contribs) 16:58, 30 July 2020 (UTC)

@The dog2: I don't think extended confirmed protection would be particularly helpful in preventing Fuerdai vandalism, as that user tends to switch accounts frequently and edit articles that aren't semi-protected. With respect to the issue of pushing a political agenda, I again ask you to provide examples of articles where extended confirmed protection would be helpful. —Granger (talk · contribs) 17:02, 30 July 2020 (UTC)

@WhatamIdoing: Just a question here since you work for the WMF. If we do not introduce extended confirmed status, would it possible introduce a new protection level to only allow autopatrollers to edit specific articles? Of course, the autopatrollers group would presumably cover patrollers, administrators and bureaucrats as well. The dog2 (talk) 17:06, 30 July 2020 (UTC)

@Mx. Granger: I guess one that comes into mind is the China article, and also the Japan, Taiwan and Hong Kong articles. This is only my guess but given the articles the Fuerdai vandal targets, it is certainly possible that there is a political motivation behind it. The dog2 (talk) 17:06, 30 July 2020 (UTC)

Thanks for giving specific examples. It seems that extended-confirmed protection would not be appropriate for those articles, as they have not gotten any vandalism while semi-protected (as far as I can tell from looking at the past year or two of edits—please correct me if I missed anything), while they have gotten constructive contributions from users who would not be extended-confirmed. [1] [2] [3] [4] [5] It seems that in all four of these articles, extended-confirmed protection would prevent good edits but would not stop any more vandalism than semi-protection. —Granger (talk · contribs) 17:18, 30 July 2020 (UTC)

Granger - Your continued objections to stronger measures against vandalism are duly noted, but it does not follow that semiprotection will necessarily continue to prevent vandalism on prominent articles just because it has managed to do so thus far. For instance, the Fuerdai vandalism spree of last month just so happened to occur at a time when there were admins on Recent Changes patrol. As the person who most often handles the Deny Recognition procedures related to the Fuerdai vandal, there are many times when I have seen his sockpuppets amass as many as three or four dozen edits before being caught and blocked, apparently because no one was watching Recent Changes at the time they emerged. These are sockpuppets that would have attained autoconfirmed status and, if they had been part of a coordinated attack like last month's instead of one-offs, would have done great damage even to semiprotected articles if there hadn't been any admins around to notice. That that didn't happen was purely by chance. It is important that we stop burying our heads in the sand, pretending there aren't any limits on our policy of open access, and obstructing paths toward viable solutions to the problem of malicious edits and begin taking this issue with the seriousness it deserves. -- AndreCarrotflower (talk) 20:50, 30 July 2020 (UTC)

Citing facts, which Granger has done, is not "burying our heads in the sand". I take your point that it is possible that a vandal could become autoconfirmed and continue to vandalize before we can block them, but right now, these proposals are solutions in search of a problem, and therefore, I oppose them for the time being. And I think the solution going forward is probably something less drastic than anyone has mentioned so far: Increasing the number of edits for auto-confirmation to something like 50. Ikan Kekek (talk) 20:59, 30 July 2020 (UTC)

┌────────────────────────────────────────────────────────────────────────────────────────────────────┘ I support stronger measures against vandalism when they seem like they would be effective and would not cause excessive collateral damage. You may remember that I helped add to our policies in order to reduce Fuerdai vandalism not long ago.

Extended-confirmed protection is useful on Wikipedia, where vandalism and disruptive edits are a much bigger problem than on Wikivoyage and take different forms. But I don't think it would be useful here, and the examples AndreCarrotflower and The dog2 gave seem to illustrate that point.

In terms of ideas that might be effective here, what if we try copying Wikipedia's abuse filter 918 and implementing it here? User:Ground Zero might be able to help with that, as he's an admin on both sites. —Granger (talk · contribs) 21:09, 30 July 2020 (UTC)

The dog2, I'm not a dev (usual disclaimers apply, etc.), but what we call 'admin' or 'template editor' or whatever is usually a bundle of separable rights (e.g., viewdeleted protect editprotected, etc.; if you look at the middle of mw:Manual:$wgGrantPermissionGroups#Default_value, you'll see a typical list of the separate rights that we call "admin"). Which rights appear under which names is configurable. So with that in mind, whatever we're calling 'autopatroller' could likely have other rights assigned to it.

We could also change the autoconfirmed status, to make it harder for newcomers to edit semi-protected articles. Once upon a time, it was zero edits and zero days, and I don't know what it is now. That wouldn't do much good for pages that aren't semi'd already, but it's another option to consider if people feel like there are widespread problems.

On the theoretical side, some vandals might be pleased when articles get locked down. Making you restrict the entire world just to slow them down gives some people a sense of power. Sometimes, the best defense is making it boring. Having your edits immediately reverted by a tireless bot would probably have that effect. WhatamIdoing (talk) 01:58, 31 July 2020 (UTC)

Do you have any candidates for the role of "Tireless Bot"? Ikan Kekek (talk) 06:20, 31 July 2020 (UTC)

I hope she doesn't mean one of us :-)

Our page about autoconfirmed users doesn't mention a necessary number of edits, just a time period. Is this not accurate?

What does the Wikipedia filter 918 do, for those of us with insufficient rights to see?

As to the broader proposal, I am convinced by Granger's arguments that this is a step too far right now.--ThunderingTyphoons! (talk) 06:57, 31 July 2020 (UTC)

I don't know precisely what Wikipedia abuse filter 918 does, but it is the filter they use for Fuerdai vandalism according to this page.

I suppose one takeaway from the point about a tireless bot is that it's best to solve the problem (e.g., by blocking and reverting) and move on, without any fanfare or displays of irritation. —Granger (talk · contribs) 11:09, 31 July 2020 (UTC)

Yes, as opposed to having a police-style case file on them, like the page you linked. Good way to impart a false sense of notoriety.--ThunderingTyphoons! (talk) 11:19, 31 July 2020 (UTC)

Right, I'm glad we don't have those... I'll also spend more time patrolling recent changes (in as botlike a manner as I can) to try to reduce the load on other admins. —Granger (talk · contribs) 11:26, 31 July 2020 (UTC)

TT, I think we might see whether Rich Smith could set up w:User:ClueBot NG for us. An actual bot, not a human. WhatamIdoing (talk) 18:00, 31 July 2020 (UTC)

@WhatamIdoing: In theory yes, but it will use the ENWP dataset which means it may false positive on a few things on Voyage. I can look in to setting it up if this is really something WikiVoyagers are interested in and on a side-note, this Pub could do with some ClueBot III treatment :) Rich Smith (talk) 20:45, 31 July 2020 (UTC)

I have been impressed by how effective Cluebot is on WP, and think it could be useful here. I realise it may be a big job to adjust the dataset. AlasdairW (talk) 21:44, 31 July 2020 (UTC)

I think having a bot to do it will be useful. This will free up our time and energy to actually contribute constructive content. And perhaps the bot can also be configured to handle the Telstra vandal too. The dog2 (talk) 21:55, 31 July 2020 (UTC)

Rich Smith, we have an unusual archiving system that requires manual intervention. We put relatively little in central talk pages. If we talk about an article here, then the discussion gets archived to that article's talk page.

Is there a way to make a list of some sample diffs that ClueBot would revert, so we can check them? WhatamIdoing (talk) 17:24, 1 August 2020 (UTC)

Might be able to run it in "dry run" mode so it will log the reverts, but not do them - Rich^T|C|E-Mail 21:43, 1 August 2020 (UTC)

┌────────────────────────────────────────────────────────────────────────────────────────────────────┘Here's another suggestion in addition to having a bit. How about granting rollback rights to our patrollers? Wikipedia has a separate group of users called rollbackers. I'm not sure we need that here on Wikivoyage, but I think giving our patrollers rollback rights would give us a few extra pair of hands to fight vandalism. After all, the role of a patroller is to patrol recent changes, and giving them rollback rights will allow them to respond to vandalism much more effectively. The dog2 (talk) 19:03, 1 August 2020 (UTC)

I believe patrollers already have the rollback right, but if I'm wrong about that, I support granting it. —Granger (talk · contribs) 19:22, 1 August 2020 (UTC)

Yes, I remember using the rollback tool when I was (only) a patroller on WV. --Comment by Selfie City (talk | contributions) 19:35, 1 August 2020 (UTC)

@Rich Smith: That sounds like a good idea. Why don't we try that out first, and if it works well, we can then implement it. The dog2 (talk) 01:42, 2 August 2020 (UTC)

Speaking of which, would it also be possible for a bot to go after the Telstra vandal. It's not as pressing as the Fuerdai vandal because (s)he is not as destructive, but I think having a bot to do that will be helpful too. The dog2 (talk) 02:44, 3 August 2020 (UTC)