Wikipedia:Edit filter noticeboard

"WP:EFN" redirects here. For information on the {{efn}} template, see WP:EXPLNOTE.

Welcome to the edit filter noticeboard

WATCH Recent filter changes (): Filter 1165 — Flags: disabled Last changed at 22:04, 18 October 2021 (UTC) Filter 970 — Pattern modified Last changed at 21:46, 17 October 2021 (UTC) Filter 869 — Pattern modified Last changed at 23:37, 16 October 2021 (UTC) Filter 1125 — Flags: disabled Last changed at 10:58, 15 October 2021 (UTC) Filter 1159 — Actions: tag; Pattern modified Last changed at 11:01, 15 October 2021 (UTC) Filter 1170 (new) — Actions: none; Flags: enabled,public; Pattern modified Last changed at 13:33, 14 October 2021 (UTC) Filter 1169 (new) — Actions: none; Flags: enabled,public; Pattern modified Last changed at 00:23, 14 October 2021 (UTC) Filter 1171 (new) — Actions: none; Flags: enabled,private; Pattern modified Last changed at 01:25, 14 October 2021 (UTC) Filter 1014 — Flags: enabled; Pattern modified Last changed at 00:13, 14 October 2021 (UTC) This is the edit filter noticeboard, for coordination and discussion of edit filter use and management. If you wish to request an edit filter, please post at Wikipedia:Edit filter/Requested. If you would like to report a false positive, please post at Wikipedia:Edit filter/False positives. Private filters should not be discussed in detail here; please email an edit filter manager if you have specific concerns or questions about the content of hidden filters. There are currently 293 enabled filters and 26 stale filters with no hits in the past 30 days (). See also the profiling data and edit filter graphs. Click here to start a new discussion thread

Edit Filter Helper for Taking Out The Trash

Taking Out The Trash (t · c · del · cross-wiki · SUL · edit counter · pages created (xtools • sigma) · non-automated edits · BLP edits · undos · rollbacks · reviews · logs ^{(blocks • rights • moves)} · rfar · spi) (assign permissions)^{(acc · ap · fm · mms · npr · pm · pcr · rb · te)}

Okay, I recognize off the bat that this request is a bit unusual, but please hear me out and don't automatically oppose due to the (seemingly) young age of my account, as that is incredibly misleading. This is my second account on Wikipedia, and is a clean start. My previous account has been inactive since 2017. Under my previous account I attempted to do some content work, but just wasn't any good at it. As such, under this account I've decided to dedicate the vast majority of my efforts to countervandalism work. Most of my CV efforts are done in the form of patrolling the filter log looking for vandalism that slips through the filters, and reverting it accordingly. Additionally, I will also monitor accounts or IP addresses that are repeatedly tripping disallow filters, thereby flooding the log, and, assuming the flood is not the result of a false positive, I will issue {{uw-attempt}} warnings on their respective talk pages. Speaking of false positives, that's the other thing I do in addition to reverting vandalism - working WP:EFFP which appears to be extremely "short staffed" if you will. Many of the reports there are frivolous, but I've found a handful that are legitimate FPs (most commonly new users or IPs attempting to add titles of songs or song albums that happen to contain profanity, but it's not exclusively limited to that). However, the vast majority of the reports that get filed involve private filters. Sometimes, when the same edit trips multiple filters, and at least one of them is public, that's enough information for me to handle the report. Reports that involve exclusively private filters, however, especially ones where the filter title is ambiguous and therefore doesn't provide any hints whatsoever as to what it's looking for, I can't handle. Many of these reports simply go unactioned and will be automatically archived by the bot without a response, which isn't a good thing. The same problem exists with filter log flooding - if a particular user is flooding the log, but all of the filters they are tripping are private without an explanatory title, I have no idea whether it's a vandal being successfully stopped, or if it's a good-faith editor being caught in a string of false positives. That's not good either - even though a majority of the time it's the former, it only takes the latter to happen once for a potential new contributor to be driven away. After all, it's a bad first impression to have your edits blocked by some automated process and not get an answer as to why.

In summary, I am requesting Edit Filter Helper permissions for two reasons. One, to be able to assist better and more completely with false positive reports. As it stands now, it seems to only be myself and one other editor who routinely check that noticeboard, which is a problem in and of itself, but would be less of a problem if I could evaluate more of the reports. Secondly, when patrolling the edit filter log for vandalism that slips through the cracks, it would be helpful for me to know when an edit that tripped a private filter should be reverted (or even the user reported, in dealing with long-term abuse cases), or if something is a false positive that can be ignored assuming it wasn't disallowed. This access will greatly improve my ability to keep Wikipedia free of vandalism and other unwanted content, which, after all, is my primary purpose here at the present time. I will answer any questions to the best of my ability - please ping me upon asking a question.

Rundown of requirements for granting as listed at WP:EFH:

Must meet all of the following criteria:

Y Demonstrated need for access (e.g. SPI clerk, involvement with edit filters) – I think that being able to help evaluate false positive reports more completely and being able to evaluate filter log hits more completely both qualify as a "demonstrated need for access", especially considering that I seem to be one of only two active editors at the FP report page

---

Y No recent blocks or relevant sanctions – clean block log

---

Y At least basic understanding of account security – can confirm that I have a strong password

---

Yg At least basic understanding of regular expressions if the intent is to assist with authoring filters – Not Applicable as I do not plan on assisting with authoring filters at the present time

---

Y Sufficient ability with the English language to understand notes and explanations for edit filters – native speaker of American English

Must also meet at least one of the following criteria:

Y Currently-active extended confirmed editor on the English Wikipedia (i.e. has made edits or logged actions within the last 12 months)

---

N Current administrator on another WMF project – Not Applicable since the first criteria is met and only one is required

---

N Current edit filter manager on another WMF project – Not Applicable since the first criteria is met and only one is required

---

N Current WMF developer or staff member who needs access for WMF-related purposes – Not Applicable since the first criteria is met and only one is required

---

Taking Out The Trash (talk) 15:42, 17 September 2021 (UTC)

Discuss (EFH Taking Out The Trash)

• @Taking Out The Trash: you have a relatively new account and small number of contributions - are there any specific admins you have been working closely with if you have been resolving false positives? Having them speak on your behalf here could go a long way. — xaosflux ^Talk 23:28, 17 September 2021 (UTC)

@Xaosflux: My response to this question would essentially be "what ProcrastinatingReader said below". Taking Out The Trash (talk) 13:55, 18 September 2021 (UTC)

• From my experience, Taking Out The Trash has been competent, has a valid need, and would make good use of the permission. The account is newer, but I don't think the editor is a sock, which is the degree to which tenure matters as this is just a viewing permission. So honestly I'd probably lean support on this one. It's kinda hard to be working closely with an admin or EFM on filter-related things, in part because non-EFH editors don't have access to any of the interfaces so it's excessively burdensome to get much done. Generally I think this would only happen in niche cases (eg people from different areas working together, like a SPI regular working with a filter regular). ProcrastinatingReader (talk) 09:24, 18 September 2021 (UTC)
• I oppose granting the permission, at least for now. While I appreciate that you may have more experience because of a prior account, your current one (and the only one whose track record we can assess), was registered only 38 days ago, and only obtained (temporary) rollback rights today. Some of these filters contain fairly sensitive material, and I wouldn't be comfortable granting the right to someone where there is so little track record to be evaluated. --Blablubbs (talk) 16:05, 18 September 2021 (UTC)
• I cannot support at this time, regretfully. Taking Out The Trash seems competent enough, but I just can't support granting EFH - a sensitive perm whose abuse is nigh-impossible to detect - to an account that's barely a month old. GeneralNotability (talk) 16:08, 18 September 2021 (UTC)
• (edit conflict × 2) The point of a clean start is that it's clean. I don't think it's appropriate to rely on your status of "Someone who had another account previously" without disclosing what that account is. Otherwise, you should be held to the same standards as any other month-old account. I myself picked up perms quite quickly as a new user (rollback after two weeks, among other things), but even if EFH had existed back then I would have never dreamed of requesting it. I don't think there's any amount of good work someone could do in a month (or really less than six months) that would convince me they can be trusted with EFH. -- Tamzin^{[cetacean needed]} (she/they) 16:12, 18 September 2021 (UTC)
• Requestor Comment If I may, I honestly find the comments above regarding tenure to be a little insulting, especially when the comments amount to "they know what they're doing but they need to wait longer for some unstated reason". Tenure is not a measure of competence - someone could have years of editing experience and not be competent to work with filters (there are probably existing admins to whom this would apply). Likewise, someone can be competent to work with filters with less total tenure, which is the situation that I'm in. As it stands now, there is no tenure requirement listed at WP:EFH other than being extended confirmed, though it sounds like there might be a desire to change that (?). I would oppose such a change though - saying that someone needs X number of months before even being considered for Y permission regardless of their edits and productivity is simply not fair. We shouldn't be setting arbitrary tenure requirements for permissions - each and every application should be evaluated on its merits and based on whether or not the user in question is believed to have enough experience to use the permission correctly and appropriately. If there is reasonable confidence that a user will not misuse a permission they have requested, that should be adequate to approve the request, regardless of how long they've been around. Taking Out The Trash (talk) 18:01, 18 September 2021 (UTC)

  P.S. As I noted in the section above, this request wouldn't even be neccesary at this time if we didn't have so many private filters (too many IMO), but it seems there's no consensus to change that. Taking Out The Trash (talk) 18:01, 18 September 2021 (UTC)

  Taking Out The Trash, to me EFH is about trust more than anything else. Whether or not you have the competence to have EFH is not in question here. The question is whether or not I am willing to trust you with the details of our private filters. To be blunt: no, I do not currently trust you that much, and that is heavily influenced by account age. Am I being excessively paranoid? Quite possibly. Perhaps you will earn that trust at some point in the future, at which point I will happily support you for EFH. GeneralNotability (talk) 18:08, 18 September 2021 (UTC)

  I still fail to understand what the (supposed?) connection is between account age and the risk of, oh, I don't know, me going and leaking all of the private filters on Wikipediocracy. As far as I can fathom, account age has nothing to do with mitigating that risk. Taking Out The Trash (talk) 18:55, 18 September 2021 (UTC)

  Account age does not per se mitigate the risk. What mitigates (or lessens) the risk is people having earned the community's trust that they won't abuse the perm. It's not that we would automatically grant EFH to anyone who's been around 6 months or a year or whatever; it's that, after that much time, we would (hopefully) have developed senses of whether we can trust them. Not just whether we can trust them to not intentionally abuse the permission, but also whether we can trust them to make good decisions, and, particularly for EFH, to know which things should and shouldn't be said in public. It's entirely possible that you have amazing judgment and are a master of OpSec. But I have no way to know that yet. There's just not enough data. -- Tamzin^{[cetacean needed]} (she/they) 19:22, 18 September 2021 (UTC)

•  Request withdrawn The responses here are, to put it politely, disappointing. Additionally, I'm concerned that some sort of yet-to-be-explicitly-defined tenure requirement is being applied here, but no such requirement is listed at WP:EFH beyond being extended confirmed. I'd suggest a discussion to consider changing that if this is the norm/expectation among most editors. Finally, I'd encourage existing EFH/EFM/admins to please, please, help with the false positives board, specifically the reports where all of the filters that were tripped are private (i.e. there was no public filter tripped simultaneously that allows others to see the blocked edit in question) and/or where private filters with ambiguous titles are involved (i.e. cases where you can't really deduce what the filter is checking for without seeing the filter itself). Taking Out The Trash (talk) 19:24, 18 September 2021 (UTC)

  The Wikipedia:Edit filter helper policy has not lined up with practice at this noticeboard for a long time, possibly since the day the user right was first created. ProcrastinatingReader (talk) 23:59, 18 September 2021 (UTC)

  And that is a serious problem. Either this noticeboard needs to get in line with what the policy says, or the policy needs to be changed to reflect what this noticeboard expects/desires. I'm not sure which one is more prudent in this circumstance, but something needs to be done. Taking Out The Trash (talk) 14:28, 19 September 2021 (UTC)

  I made a slight modification. ProcrastinatingReader (talk) 15:14, 19 September 2021 (UTC)

  I think that's a good way to put it, yeah. I would object to any change to the policy that focuses specifically on account age, because that's just one factor of many, but specifically noting the importance of trust, and of the discretion of individual editors responding here, seems like a good approach. -- Tamzin^{[cetacean needed]} (she/they) 15:29, 19 September 2021 (UTC)

Template:Z33 and Filter 602

AbuseFilter 602 was created in 2014, with Template:Z33 being set as the trigger when Template:Ds/alert was placed on a userpage. Following a recent TFD it was decided that the z-number templates should be deleted as being no longer necessary, mainly because the insource search functionality makes them redundant. Since Z33 triggers an edit filter, I thought I would check here to make sure the filter is properly updated to no longer look for this string before it is removed from the ds/alert and from current use. (please ping on reply) Primefac (talk) 13:19, 24 September 2021 (UTC)

At a glance I'm not sure why that check is required. Why is a check for -- derived from template:ds/alert -- (a variant of the line above the Z33 one) not sufficient? Ping @AGK: who I think wrote that functionality. ProcrastinatingReader (talk) 13:41, 24 September 2021 (UTC)

That change produces zero false negatives in the last 300 hits. Not sure about FPs, but that seems unlikely. Suffusion of Yellow (talk) 21:13, 26 September 2021 (UTC)

So the Z33 check can be removed without issue? Primefac (talk) 21:19, 28 September 2021 (UTC)

@Primefac:  Done. It should be fine to remove the template from ds/alert now. Suffusion of Yellow (talk) 22:02, 2 October 2021 (UTC)

Excellent, thanks. Primefac (talk) 12:34, 3 October 2021 (UTC)

Racist labeling edit filter possibility

This is an issue that has been increasing over the last couple of years, due to incidents in the news, involving both BLP and historical bios. It involves the words "white racist", "racist" "racist white supremacist" or variations - always in the lead sentence of an article, "(person) is/was a white racist supremacist". That seems to be a going trend to edit Wikipedia to insert racism labels in the lead sentence. Is there a way this can be filtered out, or do we editors just keep on manually doing the reverts? We can't watch list every bio out there. — Maile (talk) 17:37, 30 September 2021 (UTC)

@Maile66: if a set phrase is constantly, and inappropriately, being added - it can likely be handled by a filter. Please gather a few recent diff's, the collection of phrases - and drop a request over at WP:EF/REQ. A filter can first be built in logging mode to validate it, then it can either "tag" or actually block edits if warranted. — xaosflux ^Talk 19:00, 30 September 2021 (UTC)

Thanks. I'm going to copy this thread to my own talk page, and take care of it at the right time. Today's edits were on historical figures. But I'd like to act on this when I see more serious ones, as in the recent months, I've seen these labels added to BLP office holders. Thanks for the advice. — Maile (talk) 19:24, 30 September 2021 (UTC)

Moving 1169 to warn

• 1169 (hist · log)
• EFR Request
• VPT discussion

The filter log for 1169 shows a reasonable volume in bad telephone links being added, apparently unintentionally and seemingly due to a bug in Minerva (mobile web). Would it be fair to move 1169 to warn? ProcrastinatingReader (talk) 10:29, 10 October 2021 (UTC)

A couple of thoughts: We should probably at least exclude bots. I'm thinking mainly of revert-bots, although reverts (i.e. content restoration) may generally be an issue as long as this text already exists (I expect most humans could probably just deal with it, although maybe not ideal). We'd also need a meaningful error message, since it's usually (but not always) an unintended action, and it's a bit difficult to respond in the affirmative without seeing that. -- zzuuzz ^(talk) 11:08, 10 October 2021 (UTC)

This can definitely be improved, but it's the best I can do right now. Invalid^OS_talk 11:50, 12 October 2021 (UTC)

Well they wouldn't need to revert it, they would abandon it. And couldn't they just review and fix it? — xaosflux ^Talk 13:17, 12 October 2021 (UTC)

I'm not sure if the problem is in the browser or mobile VE. I think it's the latter, in which case they could only fix it by source editing or on a desktop. ProcrastinatingReader (talk) 14:05, 12 October 2021 (UTC)

@ProcrastinatingReader: from my earlier tests, I'm 90% sure this is a client-side browser issue. If it was VE while in minerva it should be trivial to reproduce, and we would want a high priority phab ticket opened on it. — xaosflux ^Talk 15:30, 12 October 2021 (UTC)

@Xaosflux: Here's a revised version of the warning:

Invalid^OS_talk 11:51, 13 October 2021 (UTC)

• I excluded bots. — xaosflux ^Talk 13:15, 12 October 2021 (UTC)

If we can limit this with user_mobile == true, either with an a/b filter setup or by looking through the logs, I think it would be preferable to do that - it would save hassling most of the reverters. -- zzuuzz ^(talk) 16:25, 12 October 2021 (UTC)

On looking further, I see that Special:AbuseLog/31062535 was not on a mobile. -- zzuuzz ^(talk) 16:38, 12 October 2021 (UTC)

I think the combination that causes this is Mobile Safari + VisualEditor. At least, I couldn't reproduce it on Chrome or Firefox on Android. I don't think it matters if they're using the mobile or desktop version of Wikipedia; I see <meta name="format-detection" content="telephone=no"/> on both the desktop and mobile site; it's just Safari being brain-damaged here. Suffusion of Yellow (talk) 22:08, 12 October 2021 (UTC)

"LTA username / impersonation creations"

Why is this filter tripping on the action autocreateaccount, which I assume means an already-existing global account being automatically created locally upon the user's first visit to this wiki? This is generating a ton of false positives, way too many to report, of users with preexisting global accounts (in good standing) who are blocked from merging their account locally on enwiki, despite their username not closely resembling any high-profile user here (and I think AntiSpoof would block the creation of a global account that has a username extremely similar to a preexisting account, if I'm not mistaken). Chances are, if someone created their account initially on a different project, they are not attempting to impersonate anyone here, since someone attempting to impersonate someone from enwiki would attempt to create the impersonation account on enwiki. Therefore, should this filter be changed to only trigger on createaccount, meaning manual account creation on this wiki? I'd also wonder why global AntiSpoof isn't sufficient for this purpose, since it would catch most would-be-impersonations before they can even be created elsewhere. Taking Out The Trash (talk) 18:18, 12 October 2021 (UTC)

@Taking Out The Trash: which filter ID# are you referring to? — xaosflux ^Talk 18:24, 12 October 2021 (UTC)

@Xaosflux: I don't know, because it's private. All I see is the filter being tripped in the logs, with tons of false positives where the global account attempting to be merged doesn't impersonate anyone or anything (at least not obviously) and the global account is otherwise in good standing. Taking Out The Trash (talk) 18:25, 12 October 2021 (UTC)

My understanding is that Global Antispoof...isn't great. And regarding if someone created their account initially on a different project, they are not attempting to impersonate anyone here, since someone attempting to impersonate someone from enwiki would attempt to create the impersonation account on enwiki - doubtful, I have absolutely seen harassment accounts created on other wikis to evade local filters. GeneralNotability (talk) 18:27, 12 October 2021 (UTC)

• OK, seems that this is private filter Special:AbuseFilter/874. — xaosflux ^Talk 18:27, 12 October 2021 (UTC)
  • I'm not seeing any overwhelming number of FP's on this. — xaosflux ^Talk 18:29, 12 October 2021 (UTC)

Well, the impetus for me filing this thread was this lovely flooding of the filter log (granted, the first set of attempts were a few days ago).. but I don't see anything particularly wrong with that username, and the global account appears to be in good standing. This also isn't the first time I've noticed these types of situations. Taking Out The Trash (talk) 18:37, 12 October 2021 (UTC)

@Taking Out The Trash: that specific ones seems to be a FP, and you can report it at WP:EF/FP; that being said it seems to be a suspect account based on the contributions at commonswiki so far so I don't think we would attach it locally without a specific request from that account holder (such as via VRT). — xaosflux ^Talk 18:51, 12 October 2021 (UTC)

@Xaosflux: Firstly, what the heck is "VRT"? (and if I have no clue what that is, can we expect some new user caught in an FP to know what it is?) Secondly, I guess I am of the opinion that we shouldn't be disallowing the automatic attachments of global accounts with the exception of usernames that would be immediately blocked at UAA without waiting for edits. Especially because, now that I think about it, chances are they won't even see the disallow message, since auto account creation/merging usually doesn't require a specific action from the user. Their account just wouldn't get created here, and they'd have no reason to know why. Taking Out The Trash (talk) 18:56, 12 October 2021 (UTC)

@Taking Out The Trash: WP:VRT - and again, for this specific case feel free to report at FP. — xaosflux ^Talk 18:59, 12 October 2021 (UTC)

Anti-spoof has limits - it's basically a character substitution test, and not a particularly good one, whereas we can use regex and substitution and some other stuff. I think this filter does tend to have too many false positives, but as for the rest, the filter is mainly for LTAs and not just impersonations. A lot of LTAs create their account elsewhere. For example, we're glad that Special:CentralAuth/V4Nd4L15M_45_4_53rV1c3 couldn't register here. -- zzuuzz ^(talk) 18:35, 12 October 2021 (UTC)