Wikipedia

Wikipedia:Edit filter/Requested/Archive 4

< Wikipedia:Edit filter‎ | Requested


Completed requests

Black Baboon

  • Task to flag up the addition of the phrase "Black Baboon" (with and without caps) to any article. IP vandal on multiple IP addresses.
  • Reason racist vandalism to various Formula 1 related articles, replacing Lewis Hamilton with "Black Baboon". Please see WP:AIN#Racist vandalism. Mjroots (talk) 16:18, 27 March 2010 (UTC)
    I've never created a filter before, but I'd like to take a shot at this :). I'll let you know how it goes. - Kingpin13 (talk) 16:23, 27 March 2010 (UTC)
    P.S. I think it would be better to make sure that "Lewis Hamilton" was replaced by "Black Baboon", rather than just preventing "Black Baboon". I'm also not sure this is happening often enough to justify a filter, but we'll see what an admin more experienced in this area says...? - Kingpin13 (talk) 16:25, 27 March 2010 (UTC)
I wasn't asking for the phrase to be banned, only flagged up. There may be a legitimate use when discussing baboons (not my area of expertise). Mjroots2 (talk) 18:07, 27 March 2010 (UTC)
Well, I've created Special:AbuseFilter/311 as log only for now, to see if this is happening often enough (hopefully I've done this right). I've left in the check for Lewis Hamilton, since baboons are real animals, and can be black, - Kingpin13 (talk) 19:39, 27 March 2010 (UTC)
It looked right to me, but I made a few changes to optimize it. --Shirik (Questions or Comments?) 23:56, 27 March 2010 (UTC)
Thanks for the creation of the filter. In this case, it's not a question of how often it occurs. The vandalism is sporadic, but it is important that it is flagged up when it happens so that the IPs can then be blocked, and the vandalism reverted. Mjroots2 (talk) 09:54, 28 March 2010 (UTC)
Shouldn't this, this and this have tripped the filter? Mjroots (talk) 12:50, 28 March 2010 (UTC)
  Fixed - Kingpin13 (talk) 13:35, 28 March 2010 (UTC)
I've just looked at the filter code, and I'm not sure about two of the tests. These are:
count("Hamilton",removed_lines) > 1 & count("Black Baboon",added_lines) > 1
which I think should be 
count("Hamilton",removed_lines) >= 1 & count("Black Baboon",added_lines) >= 1
because otherwise it'll only pick up on two or more changes, rather than one or more, which I believe to be the intention. --Redrose64 (talk) 19:52, 28 March 2010 (UTC)
  Fixed - Kingpin13 (talk) 07:46, 30 March 2010 (UTC)
The edit filter log indicates that the filter picked up this edit at 16:00, March 30, 2010 (and earlier ones at 12:33 and 12:17) but didn't pick up this one at 16:35, March 30, 2010. Any idea why? DH85868993 (talk) 07:33, 30 March 2010 (UTC)
Hmm.. No idea why that doesn't show up in the log, when testing the filter it matches that edit, - Kingpin13 (talk) 07:46, 30 March 2010 (UTC)
Also, is 4 edits in just over 4 hours sufficient justification to change the filter to Disallow the edits rather than just Log them? DH85868993 (talk) 07:50, 30 March 2010 (UTC)
I don't think we need to disallow the edits. It's not like there are so many that it is a huge problem. I'm quite content with them being flagged so that the IPs can then be reverted and blocked. Of course, if a bot could handle this task it would be better, but I don't know if such a thing would be allowable. Mjroots (talk) 11:53, 31 March 2010 (UTC)
Can we please have the edit filter changed to disallow the edits? We've had to revert 8 edits to the same article in the past 3 hours. Every time we get his IP address blocked, he just changes to another one. Thanks. DH85868993 (talk) 13:19, 7 April 2010 (UTC)

  Done I don't see any reason to not set this to disallow. - Kingpin13 (talk) 13:29, 7 April 2010 (UTC)

Can the filter be tweaked to pick up edits like this one and this one as well? Thanks. DH85868993 (talk) 03:29, 10 April 2010 (UTC)
This edit didn't trip the filter either. Maybe we will need a blanket ban of the phrase "Black Baboon" being added to any article by anyone other than an admin or higher. Mjroots (talk) 18:24, 11 April 2010 (UTC)
If it's possible, a ban on the use of this phrase by unregistered editors would probably suffice. This abuse is solely by an indefinitely blocked editor using multiple IPs. Bretonbanquet (talk) 19:19, 11 April 2010 (UTC)
We could try that, but the vandals are now registering accounts. See this edit for which I indeffed the vandal. Mjroots (talk) 14:37, 12 April 2010 (UTC)

your mom

  • Task: Search for the insertion of the phrase "your mom" into articles. Applies to all pages.
  • Reason: It's a very common form of vandalism but no bot seems to pick it up.

- Nunquam Dormio (talk) 12:30, 20 April 2010 (UTC)

  • I think that this has been asked for in the past, and possibly even run for a while, but that it was disabled due to false positives (as "your mom" can have many legitimate uses). —Soap— 13:07, 20 April 2010 (UTC)
Quotes, probably. I'm going based off of this thread, though, not any actual experience (that thread was before I was involved with edit filters in any way). —Soap— 15:55, 20 April 2010 (UTC)
If a famous person has said something quotable, they'd almost always be saying "my mom". There might be one or two instances where they'd say "your mom", but these would be outnumbered 10.000:1 by this childlike vandalism. Nunquam Dormio (talk) 19:58, 20 April 2010 (UTC)
Well a mainspace search shows about 200 pages that have legitimate uses of "your mom" and about 2000 that have "your mother". Still I could see a good use for a filter for this, I just don't think it should be set to disallow. —Soap— 21:46, 20 April 2010 (UTC)
I really couldn't help but laugh while following those links. Just what is it about "your mom" that seems intrinsically funny? --jpgordon::==( o ) 23:24, 20 April 2010 (UTC)

No 19 in the search list was Food distribution, which had the text "i did your mom ha ha ha ha" until I changed it. It seems to me that most additions of "your mom" will be vandalism and it should be flagged up in some way. Nunquam Dormio (talk) 06:21, 21 April 2010 (UTC)

Special:AbuseFilter/320 is now running in log-only mode. If it works, I'll add it to your mom Special:AbuseFilter/260. Someguy1221 (talk) 06:33, 21 April 2010 (UTC)
I withdraw my objections since this is picking up on average about one hit every 20 minutes and there are no false positives so far. It can always be changed later if the false positives do become a problem. —Soap— 16:34, 21 April 2010 (UTC)
Actually, this is a false positive, though one that would have been prevented if the immediately preceding edit to the article it is a talk page of had been blocked. —Soap— 17:24, 21 April 2010 (UTC)

your mom no longer has any objections. Just saying. --jpgordon::==( o ) 06:09, 23 April 2010 (UTC)

  Done Confirmed that this has been merged into 260 so marking this as done. --Shirik (Questions or Comments?) 04:30, 26 April 2010 (UTC)


Images and unblocks

  • Task Filter and disallow the adding of an unblock template to a talk page which includes an image.
  • Reason A troll creating unblock requests including graphic sexual images (User talk:Teaboy5, User talk:RabAllanReturns, User talk:JamesAllansHole are the latest few), plus I've never seen an unblock request with a legitimately used image, its always trolling somehow--Jac16888Talk 11:39, 8 March 2010 (UTC)
    Nawlin Wiki has already created a filter for this purpose, though at present it is specific to going after "Blu Aardvark", and may need to be expanded in the future if other users start doing this. Special:AbuseFilter/301 —Soap— 12:31, 8 March 2010 (UTC)
      Already done per Soap --Shirik (Questions or Comments?) 06:17, 15 March 2010 (UTC)

Template spam from 88.105.0.0/16


Remove Skype Formatting

  • Task: Edit filter should look for known Skype formatting text (such as begin_of_the_skype-highlighting and end_of_the_skype-highlighting) and warn/prevent the user from saving the edit. modified text to archive because filter is in use. - Hydroxonium (T•C•V) 03:26, 22 April 2011 (UTC)
  • Reason: Per this discussion it appears that Skype, when installed with certain browsers, is incorporating Skype formatting commands into the edit window which subsequently get saved. About 100 known occurrences since mid-Feb. -  7  02:24, 28 March 2010 (UTC)
    • Propose the warning text: "You are a user of a Skype extension for FireFox. This extension has a bug that at times inserts begin_of_the_skype-highlighting and end_of_the_skype-highlighting into the editfield and this problem was detected in this edit. Please remove these texts and you might want to disable the Skype extension." —TheDJ (talk • contribs) 10:30, 28 March 2010 (UTC)modified text to archive because filter is in use. - Hydroxonium (T•C•V) 03:26, 22 April 2011 (UTC)
      • Something like that works for me. Maybe my adaptation of your version is too verbose, but for less technical users this may be easier to understand: "It appears that you are a user of a Skype extension for FireFox. This extension has a known bug that at times inserts extra text into editable fields on webpages, such as the edit text box here on Wikipedia. These text strings, such as begin_of_the_skype-highlighting and end_of_the_skype-highlighting have been detected in this edit. Please remove these text strings manually before trying to save your changes. To address the root cause of the problem you can follow the instructions disable the Skype extension here." Feel free to copyedit further.  7  03:26, 29 March 2010 (UTC)modified text to archive because filter is in use. - Hydroxonium (T•C•V) 03:26, 22 April 2011 (UTC)
  • +1 for me on the need for this. It makes a mess of isbn and other identifier-numbers (and their resulting template/autolinkifications) too. DMacks (talk) 21:20, 2 April 2010 (UTC)
  Done, log only for now at Special:AbuseFilter/313. Nakon 00:19, 3 April 2010 (UTC)
8 hits since creation, adding the warning message. Nakon 16:41, 3 April 2010 (UTC)


Redirects becoming undirected

  • Task: Provide a tag that flags an edit as something like "redirect becoming article".
  • Reason: These types of edits are essentially new-page edits, but they go undetected. Recently someone changed Emancipator to be a band page (from a redirect to Emancipation), but this wasn't caught. A flag like this could be helpful.

— Timneu22 · talk 13:21, 17 June 2010 (UTC)

  Done in log-only mode for now. See filter 342. עוד מישהו Od Mishehu 13:12, 13 July 2010 (UTC)
I have a question about the specific intent of the filter at WT:EF#Filter 342. PleaseStand (talk) 04:52, 15 July 2010 (UTC)

Appending to the end

  • Task: Disallow appending plain text to the very end of an article or category, after categories/interwikis. Check IPs and new editors. Here's the example edit (taken from the section "very large text" below).
  • Reason: Prevent such incorrect edits. The warning text should instruct an editor to move the text above, to the appropriate place in wikitext; if edit is made on category page then explain that the correct way to categorize an article is to add [[category:xx]] to the article itself (example edit 1, example edit 2).

I have this filter in a another project and it triggers 20+ times a day. I can provide the code if someone is willing to create a testing filter and see how it works. — AlexSm 23:57, 26 June 2010 (UTC)

Looks reasonable, although I don't see wh you can't publish the code publicly. Which other wiki project is this from which filter is it there? עוד מישהו Od Mishehu 13:03, 13 July 2010 (UTC)
I did not want to spend time explaining if nobody was interested. The code can be found at ru:Special:AbuseFilter/74, the suggested code for enwiki is this:
action == 'edit' &! 'autoconfirmed' in user_groups &
(article_namespace == 0 | article_namespace == 14) &
removed_lines rlike '^\[\[([a-z]{2,3}|Category):.*\]\] *$' &
strpos(added_lines, removed_lines) == 0 &
(
add := substr(added_lines, length(removed_lines));
substr(new_wikitext, length(new_wikitext)+1-length(add)) + '\n' == add
&! contains_any(add,'{{','[[')
)
The code has to be this complicated because of the way the filter works when text is appended to the end, see mediazilla:20310, my comment #4. The last unchanged line is included in both removed_lines and added_lines, for example see this edit and these variables in the filter. — AlexSm 14:59, 13 July 2010 (UTC)
Very good filter, we should have it. Sole Soul (talk) 19:11, 13 July 2010 (UTC)
Filter 351. עוד מישהו Od Mishehu 12:11, 11 August 2010 (UTC)

Template spam from 88.105.0.0/16

No English language letters

  • Task: Tag edits by new users which either 1) create new articles that do not contain English letters, or 2) add large text that do not contain English letters
  • Reason: These edits are either vandalism or should be added to another Wikipedia language project.

A suggested code: 

article_namespace == 0
&! 'autoconfirmed' in user_groups
&(old_size == 0|edit_delta > 500)
&! lcase(added_lines) rlike "a-z"

- Sole Soul (talk) 21:01, 13 July 2010 (UTC)

Not necessarily. However, I have encountered cases in which a new article has been submitted in Arabic to this Wikipedia and it turns out to be advertising. PleaseStand (talk) 04:59, 15 July 2010 (UTC)
Notice that this is not supposed to catch edits which contain non-English letters, rather it catches edits that do not contain a single English letter. Sole Soul (talk) 09:27, 15 July 2010 (UTC)
created in filter 346. In log-only mode for testing. Someguy1221 (talk) 09:46, 20 July 2010 (UTC)

Proxy Connection: keep-alive

  • Task: Check for addition of the exact text string "Proxy Connection: keep-alive", all pages, and on a match (if implemented) ask user to re-check their edit
  • Reason: Some unknown technical glitch appears to cause this text to be added, see this VPT archive.Removed dash between "proxy" and "connection" because filter is in use - Hydroxonium (T•C•V) 14:29, 3 May 2011 (UTC) It can cause users to be incorrectly accused of vandalism. Hard to say how frequent this problem is, as recentchange patrollers will catch most; if this does not add any significant overhead, I ask for a bit of a trial, for e.g. 2 weeks, just to see if it picks up enough to warrant the overhead. Open to alternatives too, of course. -  Chzz  ►  09:41, 18 July 2010 (UTC)
Sure, lets see how it works. Tim1357 talk 23:16, 18 July 2010 (UTC)
  Done See here. Is there any other conditions you can think of to exclude some edits from this filter. Perhaps admins? Tim1357 talk 00:00, 19 July 2010 (UTC)
313 exempts only bots. —Soap— 00:36, 19 July 2010 (UTC)
This certainly seems to be helping, in warning people. Some ignore the warning, but in those cases, it helps log it for fixing.  Chzz  ►  06:13, 29 August 2010 (UTC)

Removal of LGBT project banner with disruptive edit summary

- –xenotalk 17:06, 26 July 2010 (UTC)

I'm in the process of hanfdling this. Please don't add code for this, per WP:BEANS. עוד מישהו Od Mishehu 12:05, 25 August 2010 (UTC)
Filter 356, though there are no hits yet (nearly 4 days after creation). If there are any missed by the filter, please e-mail me the URLs of the diffs. עוד מישהו Od Mishehu 05:26, 30 August 2010 (UTC)
Disabling due to a complete lack of hits. עוד מישהו Od Mishehu 07:20, 7 September 2010 (UTC)

Unique blacking vandalism

  • Task: Revert blacking vandalism
  • Reason: Recently today, I discovered hundreds of IPs vandalising pages by adding <div style="position: fixed; width: 100%; height: 100%; top: 0; left: 0; background-color: black; opacity: 1; z-index: 99;"></div> in several article spaces, therefore blacking out the page. Not only the people around the world would see the information, but also, it makes the edit impossible to roll back without going to the history box unless you're using any separate AV tools like Huggle and Igloo. Here's an example of the one I reverted.

- Minimac (talk) 06:15, 19 August 2010 (UTC)

  Done Older filter modified. -- Gogo Dodo (talk) 06:32, 19 August 2010 (UTC)
Sigh. I thought this was a thing of the past. Yes, we had a pre-existing filter for this, which was loosened up a bit because it was producing false positives. But now I see we're going to have to go back to the stricter form unless someone can think of a genius solution that will stop the vandalism without stopping people from putting templates, etc. with fixed positioning in them. —Soap— 10:17, 19 August 2010 (UTC)

Stronger filter rules for high school articles

  • Task Prevent edits that replace phrases in the article with inappropriate language
  • Reason High school articles are a frequent target of vandalism, and there are a number of words and phrasings that would cause false positives elsewhere that would be almost always inappropriate in a high school article.

- Triona (talk) 04:46, 2 September 2010 (UTC)

Not quite specific enough, but one shouldn't be too specific here per WP:BEANS. Please email me with:
  1. which words should be excluded here?
  2. What is the criterion for recognizing that an article is about a high school?
עוד מישהו Od Mishehu 07:18, 2 September 2010 (UTC)
See filter 39. This filter picks up two different things: Words which should (almost) never appear in school articles (like 'cunt\b'), and words which can appear but should be flagged for review (like 'paedophile'). It's quite possible, with some thought, to split out the former category into a new warning (or blocking) filter, but the other type should also remain to allow the 'edgy' edits to continue and be flagged for review. I've always thought however that it would make more sense to place the pure vandalism terms in more generic vandalism filters. -- zzuuzz (talk) 07:30, 2 September 2010 (UTC)

Tampering with subject name in lead sentence or infobox of a BLP

  • Task: Stop anonymous or unconfirmed users from altering, replacing, or deleting the subject name in the lead sentence or infobox of a BLP article.
  • Reason: A lot of such edits by new or anonymous users turn out to be vandalism; here is a recent example.

- --SoCalSuperEagle (talk) 18:44, 8 September 2010 (UTC)

I believe that it's technically impossible to detect the change. We can test for such changes in the infobox, but not in the actual article. עוד מישהו Od Mishehu 12:42, 12 September 2010 (UTC)
If it's possible to detect and stop such changes only in the infoboxes, that's fine with me; it's certainly better than nothing. --SoCalSuperEagle (talk) 17:58, 5 October 2010 (UTC)
I wrote a first version of it, and have this version in my test filter. Please let me know if it misses any (filter 358, anything in October). Please let me know if you know of any the filter missed. I did check that it catches the diff you suplied above. עוד מישהו Od Mishehu 20:20, 5 October 2010 (UTC)
Filter has gone live with its own filtr, although it's still set to log onlt - see filter 364. I'm not sure it should be set to disallow, as there are possibly too many false positives - see here. עוד מישהו Od Mishehu 12:38, 11 October 2010 (UTC)

QuickiWiki Look Up

I picked the examples with extraneous text to show it is hard to revert these as often they have baked in, with intervening edits, and it is not always clear what is cruft and what is not. - Colfer2 (talk) 17:31, 14 September 2010 (UTC)

Is it always followed by the words "Look Up"? Adding a link by the way, since it's harder than I expected to find it on Google since there is other software by that name, and Google autocorrects it to QuickWiki if you don't use a + sign before it. It could perhaps be added to the Skype filter since the problem is very similar. —Soap— 17:46, 14 September 2010 (UTC)
Yes, the bad edits always say "QuickiWiki Look Up". Thanks... -Colfer2 (talk) 18:06, 14 September 2010 (UTC)
It seems to be the Firefox extension WikiTweak, according to this comment: http://en.wikipedia.org/w/index.php?title=User_talk:Leonard^Bloom&oldid=383912819#X-15_spam_.26_edit -Colfer2 (talk) 18:13, 14 September 2010 (UTC)
  DoneI've added it to the "codespill" filter, which is warn-only, in the hopes that people using this extension will see the problem and not save the edit. It will also help us track the edits. However if this turns out to not be enough (especially if the extension becomes more popular), it can be moved to the Skype filter, which is set to "disallow". —Soap— 18:29, 14 September 2010 (UTC)
Thank you Soap. -Colfer2 (talk) 18:34, 14 September 2010 (UTC)
So far only one hit, but it's still worth keeping. —Soap— 12:13, 19 September 2010 (UTC)

The filter missed two new edits, one in article space:

  1. diff
  2. diff

-Colfer2 (talk) 00:08, 26 December 2010 (UTC)

Filter 331

This filter shouldn't be visible to the public (i.e. set it to private). Minimac (talk) 09:32, 7 October 2010 (UTC)

  Done Thanks for pointing it out. - EdoDodo talk 16:45, 7 October 2010 (UTC)

Game prevention

  • Task: Undo/prevent/whatever is best redirects to The Game (mind game).
  • Reason: In the last two days, there's been a spate of redirects on Main Page featured articles (see Shield nickel and Convoy GP55, and it appears that it may be spreading [1]. GedUK  14:33, 8 October 2010 (UTC)

- GedUK  14:33, 8 October 2010 (UTC)

And another [2] by the same IP, blocked for now. Acroterion (talk) 14:53, 8 October 2010 (UTC)
Created as filter 363. Log only for the moment (hits) but will change to prevent if it works fine and there are no false positives. - EdoDodo talk 14:59, 8 October 2010 (UTC)
Couple of hits, no false positives. Changing to disallow. - EdoDodo talk 20:21, 8 October 2010 (UTC)
Thanks for this, it seems to be working well. GedUK  11:09, 11 October 2010 (UTC)

Random characters

  • Task: Detect addition of seven consecutive consonants (excluding y), e.g. sdfsdfs. This is a very common type of vandalism.
  • Reason: Seven or more consecutive consonants are very uncommon. They are almost certainly caused by randomly typing characters on the keyboard.
  • Code:
!("autoconfirmed" in user_groups) &
(article_namespace == 0) &
(added_lines rlike "[bcdfghjklmnpqrstvwxz]{7,}") &
!(removed_lines rlike "[bcdfghjklmnpqrstvwxz]{7,}") &
!(added_lines rlike "https?://[^ ]*?[bcdfghjklmnpqrstvwxz]{7,}")

- Ideal gas equation (talk) 13:56, 12 October 2010 (UTC)

Created as filter 366. Log only for the moment (hits) but might consider changing it to tag if there are few false positives. - EdoDodo talk 15:35, 12 October 2010 (UTC)
The false positives I can see this catching are nucleotide and peptide sequences in biology-related articles. I'll try to keep in eye on it to see if that happens. Someguy1221 (talk) 20:10, 12 October 2010 (UTC)
The filter seems to be doing good, but I think it should be added as a line within the Repeating characters filter rather than a separate filter all its own, unless you are planning to bring it to a level above "tag". Merging it would also allow us to re-use the code that exempts URL's, such as URL's containing long strings of base-64, or even short strings such as you often see on YouTube, without having to run the code again. —Soap— 15:59, 13 October 2010 (UTC)
I'm not sure if this is technially possible, but if all the false positives are coming from peptide and nucleotide articles, could you say something like "If article is member of Category:Peptides, ignore" and If article is member of Category:Nucleotides, ignore". That seems like an easy solution to the problem. Sven Manguard Talk 16:23, 15 October 2010 (UTC)
Unfortunately, I don't think that would be technically possible. - EdoDodo talk 18:43, 27 October 2010 (UTC)
If there are no objections I will change this to tag the edit, since there is a fairly low number of false positives. - EdoDodo talk 18:43, 27 October 2010 (UTC)
Found a false positive earlier here; apparently seven consonants isn't that uncommon in German (Rückwartsschnitt). No real way to code for that, I just wanted to point it out since if this filter goes to "tag" mode it may lead to people posting on the false positives page wondering why they're being tagged. Perhaps we just need to look for more than seven consonants. Also I think it would be good to consider merging this into 135 to save time. —Soap— 10:47, 28 October 2010 (UTC)
I've merged it into into filter 135 and marked it as deleted. - EdoDodo talk 11:14, 28 October 2010 (UTC)

Jcarleo

  • Task: Prevent IP sockpuppets of User:Jcarleo from adding {{sockpuppet|jcarleo}} to User:C.Fred's (and potentially other users') user and user talk pages.
  • Reason: Prevent sockpuppets of Jcarleo from adding false sock templates to userpages.

- Eagles 24/7 (C) 23:37, 15 October 2010 (UTC)

I think if you want to do this it would be best to put it in a filter such as 58, not a separate filter, to save time. That way people that aren't affected by 58 will not have to run through another filter afterwards. I'm not sure how to type out the | symbol though and couldn't seem to work it out in my sandbox. (Don't just use |, the system will see it as an "or" operand and therefore treat "sockpuppet" as the string to look for. "1" doesn't seem to work either, despite the system saying it should.) —Soap— 00:17, 16 October 2010 (UTC) 
&#124;
Access Denied [FATAL ERROR] 00:19, 16 October 2010 (UTC)
I don't think an HTML code would work either, it would just be parsed as a literal, including the ampersand. I'll try it though. —Soap— 00:21, 16 October 2010 (UTC)
  Done added to 58 with "1" as the connector. I don't know why it wasn't working before, I must have made some kind of mistake. If anyone thinks that 58 is not a good place for this feel free to move it around. —Soap— 00:26, 16 October 2010 (UTC)

Nowiki tag abuse

  • Task: Prevent the use of randomly placed nowiki tags to evade other filters.
  • Reason: Becoming popular among 4chan vandals.

- Access Denied [FATAL ERROR] 00:04, 16 October 2010 (UTC)

What exactly do you mean by "randomly placed"? /ƒETCHCOMMS/ 03:41, 16 October 2010 (UTC)
Example code: FU<nowiki />CK YO<nowiki />U AND YO<nowiki />R MO<nowiki />M HAG<nowiki />GER??? (often in the middle of curse words) Access Denied [FATAL ERROR] 03:45, 16 October 2010 (UTC)
I edit conflicted typing out a much less interesting example. Anyhow, the best approach is probably just to block edits by non-autoconfirmed users where the added text contains more than (say) three nowiki tags. — Gavia immer (talk) 03:50, 16 October 2010 (UTC)
It would need to cover both nowiki varietes: {nowiki}{/nowiki} and {nowiki /}. (Of course with < and > instead of braces) Access Denied [FATAL ERROR] 03:52, 16 October 2010 (UTC)
Also make it cover non-autoconfirmed users adding nowiki tags around user or user talk pages. Access Denied [FATAL ERROR] 15:03, 16 October 2010 (UTC)

Created, log only for now, at Special:AbuseFilter/367. Someguy1221 (talk) 06:59, 17 October 2010 (UTC)

New user editing Wikipedia:Articles for deletion page

  • Task: Tag edits of users with less than four edits elsewhere.
  • Reason: To help keep track of possible sockpuppetry.

As a suggestion: 

!("autoconfirmed" in user_groups)
& (wikipedia_namespace == 0)

:| TelCoNaSpVe :| 09:04, 22 October 2010 (UTC)

Created as filter 369 (hits). Log only for the moment, but once I'm sure that it's working well and there are no false positives I might change it to tag the edit. - EdoDodo talk 06:34, 26 October 2010 (UTC)
I would oppose that as it would likely just make people who get tagged more angry than they would be already. Personally I think the filter itself is a bad idea, as there's genuinely nothing wrong with a new user editing an AfD, and in fact, if the new editor is also the author of the article being discussed, it's pretty much inevitable. Is the incidence of sockpuppetry at AfD's so extreme that we have to screen every edit made by every editor to see if it might be a new user editing an AfD? —Soap— 11:52, 27 October 2010 (UTC)
Yeah, looking at the hits, I have to agree that it doesn't look like there's much cause for concern. Most hits are just new users replying to people's concerns about the article they created. - EdoDodo talk 12:06, 27 October 2010 (UTC)
What about the same concept for RfA/RfB's? Tim1357 talk 01:15, 17 November 2010 (UTC)

Would it be possible to restrict the filter to tagging only those who have not previously created any pages before !voting in the AfD? :| TelCoNaSpVe :| 00:41, 21 November 2010 (UTC)

titaniumwf.com spam

  • Task: Block additions of the text "titaniumwf"
  • Reason: There has been a months-long effort by this website to spam Wikipedia (including other language project). The website has been blacklisted on meta, but various users and IPs continue to add links without formatting. See the contributions of Managermerrill (talk · contribs · deleted contribs · nuke contribs · logs · filter log · block user · block log), for example. A filter that would block these attempts would be useful. Deli nk (talk) 21:07, 8 November 2010 (UTC)
  Done as 374. —Soap— 22:58, 20 November 2010 (UTC)

InfoCheck

  • Task: Flag edits with edit summaries that include the strings "www.InfoCheck.com", "InfoCheck.com", "InfoCheck", and "InfoCheck(r)" for strong possibility of vandalism, often on BLPs. Possibly disallow these edits if it turns out there are no constructive edits being made with these in their summaries.
  • Reason: All edit summaries I have seen with these strings in them have been vandalism of the most obvious kind, but the edits have come through multiple users over time. Most recent IP blocked for such edits, as an example: [3].

- keɪɑtɪk flʌfi (talk) 14:45, 24 November 2010 (UTC)

Created as filter 376 (log). Log only for the moment, will change to tag or disallow later if it works fine and there are no false positives. - EdoDodo talk 17:57, 24 November 2010 (UTC)
  • I'm already dealing with this vandal in filter 52, which covers edit summary vandalism in general. I've added this item to that filter and am disabling #376. NawlinWiki (talk) 20:05, 26 November 2010 (UTC)

Denied requests

Link-spam www.fibroidsuterus.com

  Resolved
 – 21:10, 12 May 2010 (UTC)
thanks, I was not aware of the spam-blacklist filter, seems more appropriate. I will watch it for a while and request if I see more activity. Richiez (talk) 11:11, 12 May 2010 (UTC)

  • Task: For the range 208.54.14.0/8: disallow page creation with "*onzale*" in title, disallow editing of "Due Date" and flag any edits that include "*onzale*".
  • Reason: There is a very persistent paid editor on this dynamic IP range who has been spamming on behalf of his client Joel William Gonzales for a long time now. See the SPI archive and one of over 16 related AfDs. Unfortunately there are also good faith editors on this IP range, so a netblock would have collateral damage. Smocking (talk) 17:32, 3 March 2010 (UTC)
Created 302 and 303. Log only for now to see how well they do. If the hit rate is insignificant, these filters will probably be disabled. --Shirik (Questions or Comments?) 18:18, 9 March 2010 (UTC)
I have seen no hits on this filter yet. Is the user currently blocked? Is he being missed by the filter? If this filter doesn't get that many hits, it may not be a good candidate for a filter for performance reasons. We have a limited number of conditions to work with, and the fact that this would require 2 separate filters is a problem in and of itself. --Shirik (Questions or Comments?) 02:26, 12 March 2010 (UTC)
I dont believe the filter can detect the true IP address of a user who's editing with a named account, so these filters will only detect anonymous users, and anonymous users can't create pages. Also, /8 is a very large range; is it really necessary? —Soap— 02:30, 12 March 2010 (UTC)
Good point. I wasn't thinking and created 303 as an IP user creating a new article. I deserve a trout for that one.... That would explain the lack of hits. 302 could still match, but only if it's anonymous. --Shirik (Questions or Comments?) 02:50, 12 March 2010 (UTC)
Just wanted to mention that the user's (EZW) edits may be in rather confused, but good-faith, edits. I've dealt with the user multiple times on the IRC channel #wikipedia-en-afc, and each time, I've concluded that although the user has a major conflict of interest, the issue is a lack of policy comprehension. Frankly, after speaking with the user over 5 times on IRC, I think that the edits are in good faith, as the user quite obviously believes that their client is notable. I've attempted to explain our policies which conflict with their edits. However, I do not argue the fact that a filter should be in place.  IShadowed  ✰  03:06, 12 March 2010 (UTC)
EZW and his sockpuppet gallery have recreated the Joel W Gonzales page at least seven times now (not to mention the other pages and adding Gonzales to Due Date even though he has a tiny role). He hasn't made a single edit that wasn't reverted, created a page that wasn't deleted through AfD or shown any intention of stopping the cycle anytime soon. This has been going on for over half a year now and sorry, with all the best intentions in the world, I can no longer assume good faith in this case. Unfortunately he seems to be using other IP ranges as well now, so this filter will probably not be very effective anymore. Smocking (talk) 18:07, 12 March 2010 (UTC)
Not asking for one's sympathy, simply mentioning that I've dealt with this user, and the problem seems to be ignorance rather than bad-faith. Either way, the user and his many socks are blocked and apparently we're working on a filter.  IShadowed  ✰  21:28, 12 March 2010 (UTC)
  Not done Deleted the filters. 302 is functionally impossible, and 303 does not get enough hits to justify the performance hit and condition consumption. I would advise simply watchlisting the affected pages and keeping an eye on them. --Shirik (Questions or Comments?) 03:04, 17 March 2010 (UTC)

Extra trigger for possible cut and paste moves

  • Task: Would it be possible to add the {{Hangon}} tag to the list of templates that trigger this filter?
  • Reason: Cut and paste page recreation immediately after speedy deletion with the {{Hangon}} tag left in place is a fairly common occurence at NewPages. Jeffrey Mall (talk • contribs) - 10:29, 8 April 2010 (UTC)
    Or any deletion template, really. There was once a tag called "possible cut and paste move or recreation", but it's inactive now. It looks like filters 164 and 237 at various times in the past did track creating a new page with a hangon tag, but it was removed because any such pages would be automatically placed in the CSD category anyway. Still, it shouldnt be much trouble to bring it back if people agree that it's a good idea. —Soap— 11:26, 8 April 2010 (UTC)
    Nevermind, just viewed the tag's source and wasn't aware {{Hangon}} added pages to both Category:Contested candidates for speedy deletion and Category:Candidates for speedy deletion. Thanks anyway, Jeffrey Mall (talk • contribs) - 12:24, 8 April 2010 (UTC)
  •   Withdrawn request (withdrawn). Stifle (talk) 11:12, 12 April 2010 (UTC)

Stopping Mermaid Melody Pichi Pichi Pitch respellings

  • Task: Stop the respelling of the names "Michal" and "Michel" to "Mikaru" and "Mikeru" respectively on Mermaid Melody Pichi Pichi Pitch Pure and its related articles. The only instance the latter pair should appear is within the parameters of the "nihongo" template for pronunciation of the kana associated (みかる and ミケル respectively).
  • Reason: The former pair is the official spelling used in the primary media of the title, which in this case is the English manga. However, certain anonymous editors kept changing to the latter pair without any good/plausible reason to do. It is also due to this very mentality that Gryffinclaw (talk · contribs) got into a pseudo-edit war with several other editors. - 上村七美 (Nanami-chan) | talkback | contribs 09:42, 28 March 2010 (UTC)
    • Update: These two most recent edits on Mermaid Melody Pichi Pichi Pitch Pure should justify to make this edit filter request a priority: [4] [5] - 上村七美 (Nanami-chan) | talkback | contribs 16:49, 31 March 2010 (UTC)
      • Edit filters aren't really designed for issues with a single article, and there's no way for the edit filter to know what is a "related article". Nor will we lock out someone who's in a content dispute.   Not done. Stifle (talk) 11:15, 12 April 2010 (UTC)

Unregistered user creating a talk page for a nonexistent article

  • Task: Warns an unregistered user if he/she attempts to create a talk page for an article that does not exist. If the user insists on creating said talk page after being warned, then the filter should tag it to assist new talk page patrollers.
  • Reason: All such talk pages automatically qualify for speedy deletion under the G8 criterion. Many anonymous users have been creating these talk pages for experimentation purposes. Unfortunately, some of those pages would even meet the G3 or G10 criterion for speedy deletion. In addition, some anonymous users have apparently been using the talkspace to request article creation instead of using the AFC process. --SoCalSuperEagle (talk) 19:14, 27 March 2010 (UTC)
    Unfortunately I dont believe it's possible for the filter to tell whether a page exists or not. This is a limitation that would do a lot of good for us if it was fixed. (Someone please correct me if Im wrong.) —Soap— 19:31, 27 March 2010 (UTC)
      Not done Correct. I believe this has been asked for in the past but unfortunately does not appear to be possible. An adminbot that handles this would be interesting, though. --Shirik (Questions or Comments?) 23:57, 27 March 2010 (UTC)

Mediation Cabal cases

  • Task: Applies to all pages beginning with Wikipedia:Mediation Cabal/Cases. Please warn when user attempts to create a new page that starts with that prefix and ends with a "/".
  • Reason: Needed to prevent people from using our system to create cases without a real name, such as Wikipedia:Mediation Cabal/Cases/2010-3-17/ which later have to be moved (by myself or another coordinator) to a valid location.

- The WordsmithCommunicate 17:41, 17 March 2010 (UTC)

Created 308, log only while unit tests are being performed. --Shirik (Questions or Comments?) 18:24, 18 March 2010 (UTC)
  Not done No hits in that time period. Additionally, filter was disabled by User:Prodego with the comment "The abusefilter is intended to stop abusive edits. This is neither problematic, nor abusive." I might recommend a bot to help detect this instead. --Shirik (Questions or Comments?) 12:38, 23 March 2010 (UTC)
Just noting that this filter has been revived in a different form, Filter 314, whose log (it's log-only) can be seen here. I dont think it should ever be set to Disallow, because there are legitimate pages like boot that need to end in a slash, but it seems to have caught a malfunctioning archive bot of some sort which created a page titled User talk:RashersTierney/Archives//Archives//Archives//Archives//Archives//Archives//Archives//Archives//Archives//Archives/. —Soap— 11:34, 12 April 2010 (UTC)
Oddly, the link above isnt working, even though it works in article space. It seems to be a bug relating to articles that begin with a slash, as the software assumes they're a subpage of the current page. Well, in case it wasnt obvious, the page Im linking to is /boot/. —Soap— 11:42, 12 April 2010 (UTC)
The link isn't working here because pretty much every namespace except articles interprets a link formatted like [[/blahblah]] as a link to NameOfPage/blahblah, but it doesn't do this in article space because subpages are disabled in that namespace :) (probably a totally unnecessary explanation, but I couldn't resist) — Preceding signed comment added by Cymru.lass (talk • contribs) 07:42, 13 April 2011 (UTC)

Rayman Unlimited/Advent/RX

  • Task: Prevent any edits containing the text "Rayman Advent", "Rayman Unlimited", and "Rayman RX" from being made to any articles.
  • Reason: For several months now, since July, a vandal has been adding false entries to various pages, primarily those of voice actors, claiming that they are somehow involved with a series of nonexistent projects related to the Rayman franchise (i.e. Rayman Unlimited, Rayman Advent, Rayman RX). The vandal's actions have been going on for several months now, and since he's hopping between IPs (with 70.188.151.127 and68.0.233.157 being the latest examples), it makes him very hard to track without regularly having to search for the above keywords. Nary a week goes by without him making another attempt under a different IP, and quite honestly, it's getting irritating. Therefore, due to the constant hopping and the fact that no one would make any entries for these without the sole intention of vandalism, I propose a filter be made to prevent the problem in the future. — Cyberlink420 (talk) 00:24, 10 March 2010 (UTC)
Created 304, currently log only, to evaluate performance and accuracy. --Shirik(Questions or Comments?) 02:29, 12 March 2010 (UTC)
  Not done Filter deleted. Filter has had no hits during the trial period, so the benefit does not seem to outweigh the performance hit. I would advise simply watchlisting or perhaps coding a bot to track the edits. --Shirik(Questions or Comments?) 18:53, 20 March 2010 (UTC)

Zalgo

  • Task A filter to trap edits like those of the history of this page and this edit and elsewhere, inserting huge chunks of unreadable and often vertical repeated text which are difficult to revert for those without rollback since the screen itself is blanked (similar to the "fixed position" vandalism) and because the amount of text is so large that any non-automated reversion will take a long time to load. NOTE: loading these diffs will be slow on most PC's; I use an add-on for Firefox to help.
  • Reason Vandalism. Could track strings such as "ͯ̋̇ͯ̎ͮ̕" which are repeated many times throughout the text and unlikely to ever appear in legitimate edits, or go for very large additions of any type, or adapt the fixed position filter (if it still is on). Note: I don't believe user:Zalgo himself is responsible for this vandalism, as he has not edited in over a year, though he created a page called Zalgo that an admin might want to see (unless it too is unreadable). —Soap— 00:59, 17 May 2010 (UTC)
Note, the "Zalgo" text is likely to be the output of a script on this page, in case that helps to filter this stuff. — Gavia immer (talk) 02:16, 17 May 2010 (UTC)
Comment: I think this doesnt really need a whole filter of its own; it could just be a single string or condition within one of the existing vandalism filters. —Soap— 12:42, 18 May 2010 (UTC)
 N Withdrawing my own request for now; I think it would be better to just block outrageously huge changes in page size (e.g. a million bytes or above) and hold off on dedicated "zalgo" code for now since it seems to be rare. —Soap— 21:45, 29 May 2010 (UTC)

Targeted vandalism by users with disruptive usernames

  • Task: If a new user registers with a username that clearly appears to attack some subject, such as usernames of the form "(insert subject name) sux" or "Ihate(insert subject name)", and if said user attempts to edit an article about that particular subject, disallow the edit attempt. Examples of what this filter should stop can be found here.
  • Reason: Out of all the edits that I've seen which met both of the above conditions, 100% of them were personal attacks, BLP violations of a defamatory nature, or simply pure vandalism.

- --SoCalSuperEagle (talk) 00:13, 21 May 2010 (UTC)

How is this better than using the WP:UAA bots, which flag usernames based on a regex blacklist. What additional benefits would using an edit filter have? Tim1357 talk 01:28, 21 May 2010 (UTC)
Also, note that sucks and sux are already blacklisted words, if they are in new user names. It is set so that the bot waits until the user edits to report it. Tim1357 talk 01:33, 21 May 2010 (UTC)
Actually, we do have a filter, Special:AbuseFilter/102, that blocks usernames, though I can't see it anymore because it's a private filter. But Im pretty sure it hasnt changed greatly since I resigned. In theory we could fulfill the request, in practice though it would be pretty difficult, since there'd be three steps to it ... trapping the username in such a way that it is allowed to edit (because probably at least a few legitimate names will have words like 'suck' in them), looking at the edits it does make to see if they qualify, and then disallowing those edits. In fact, I'm not actually sure it's possible to do that since as far as i know one edit filter cannot be used to pass information to another edit filter. We would need a separate filter unrelated to the existing username filter. —Soap— 01:45, 21 May 2010 (UTC)
  •   Declined, already served by the UAA bots. Stifle (talk) 20:50, 5 June 2010 (UTC)

Removal of good & featured page templates

  • Task: Notify non-autoconfirmed users about removing any good or featured page templates, including those in articles, portals and lists etcetra.
  • Reason: Removing this template also removes the category related to it; for example: Removing the featured article template also takes that article out of Category:Featured articles, and so on.

- Minimac (talk) 13:23, 31 May 2010 (UTC)

Is something like this terribly common? Someguy1221 (talk) 05:56, 8 June 2010 (UTC)

Could be done like this: 

!"autoconfirmed" in user_groups
& contains_any(removed_lines, "{{Featured article}}", "{{Featuredarticle}}", "{{FeaturedSmall}}", "{{Good article}}", "{{GA article}}", "{{Good Article}}", "{{Featured portal}}", "{{Featured list}}")
& !contains_any(added_lines, "{{Featured article}}", "{{Featuredarticle}}", "{{FeaturedSmall}}", "{{Good article}}", "{{GA article}}", "{{Good Article}}", "{{Featured portal}}", "{{Featured list}}")

Not sure it's something very common though, but it might be worth testing to see if it does happen sometimes. - EdoDodo talk 13:30, 22 August 2010 (UTC)

Tested; so far, all such cases were caught by otherfilters (page blanking, removing all categories, etc.). Updated the test filter; if there are no more results, then I'll close this as not done due to lack of hits., עוד מישהו Od Mishehu 08:46, 12 September 2010 (UTC)
Only 3 edits in 4 days; 1 of these edits was caught by yet an other filter (New user removing references), and one of the others had an other aspect of vandalism in it. I think there;s no need for this filter. עוד מישהו Od Mishehu 09:00, 16 September 2010 (UTC)

Deleted page recreation

  • Task: Tag newly-created articles that were once deleted.
  • Reason: Again, helps with vandalism patrolling

— Timneu22 · talk 17:28, 2 June 2010 (UTC)

I don't believe this is technologically possible right now, because the edit filter is unaware of whether a page has been deleted or not, and it can't search the deletion log either. I agree though that it would be a good idea, so if I'm wrong about it being impossible I would like to see it happen. —Soap— 18:17, 2 June 2010 (UTC)
  •   Technically impossible. Stifle (talk) 20:48, 5 June 2010 (UTC)

Targeted editor death threats

  • Task: Block and log edits by IP editors which match a significant number of the following words:

<snip>

  • Reason: Block death threats made against regular contributor by high volume IP hopping vandal. They have used a large number of IPs to make similar threats on AN and ANI at least in last 24 hrs.
I have looked into creating filters before but not done so and don't want to screw it up.

- Georgewilliamherbert (talk) 23:59, 7 June 2010 (UTC)

Added to filter 58. Someguy1221 (talk) 05:55, 8 June 2010 (UTC)
Followup: functionaries have requested that this be removed from the filter, as all hits are for material needing oversight. Someguy1221 (talk) 18:24, 10 June 2010 (UTC)
Removed the text per DENY, as should be done in all circumstances where 4chan or JarlaxleArtemis is discussed. —Jeremy (v^_^v Hyper Combo K.O.!) 20:59, 15 December 2010 (UTC)

Unsourced article creation

  • Task: Warn a new user who attempts to create a new article that does not cite any references whatsoever; if the user insists on creating such a page anyway, tag the creation. If possible, this filter should also apply to AFC submissions. Note that pages created as redirects should be exempted.
  • Reason: The absence of references may be a red flag for notability issues. Thus, unsourced articles have a high risk of being marked for deletion (possibly of the speedy type). Also, the absence of sourcing is generally considered grounds for automatic rejection of AFC submissions.

- --SoCalSuperEagle (talk) 19:24, 14 June 2010 (UTC)

I don't think this is possible. The edit filter can only catch purely mechanical things like an absence of <ref> tags, rather than semantic problems like a lack of sourcing. An article without formatted references might still have references, and an article with <ref> tags might not have valid references. If an editor makes a good-faith effort to add sourcing, but they happen not to understand our grossly overcomplicated referencing system, then we should thank them and fix the references - not scream an edit filter warning at them because the filter couldn't detect their efforts. — Gavia immer (talk) 00:37, 15 June 2010 (UTC)

Controversial BLP categorization

  • Task: If a new user attempts to add a possibly controversial or negative category tag such as [[Category:LGBT sportspeople]] or [[Category:Criminals]] to an article that contains the [[Category:Living people]] tag, prompt user to confirm if such category addition is appropriate. If user chooses to go ahead, tag the edit.
  • Reason: Inappropriate addition of such a category tag to a BLP page constitutes a form of BLP violation. Depending on the particular category tag that is abused, such an edit could also be tantamount to libel.

- --SoCalSuperEagle (talk) 18:28, 17 June 2010 (UTC)

Unfortunately, I thinkj the list of relevant categories here would be too big to handle. There are other relevant categories (such as Category:Arsonists and Category:LGBT rugby union players, both parts of the cateogry trees you mentioned, which would need separate entries). עוד מישהו Od Mishehu 13:08, 13 July 2010 (UTC)

Chinese Wooden...

  • Task: Flag a new page named "Chinese Wooden..." something.
  • Reason: A page named Chinese Wooden Pony Torture was created and deleted via discussion. A new page was created today, called Chinese Wooden Donkey. These pages were almost identical, and both vandalism. I realize this is a one-off request, but if I hadn't remembered the first one, the second one may not have been caught. Could a tag be added that finds pages with this word pattern? — Timneu22 · talk 19:17, 29 June 2010 (UTC)
I don't thin k so - the title black list is probably a better corse to take for this problem. And the second page is Chinese wooden donkey - the capitalization of all but the forst letter is important. עוד מישהו Od Mishehu 13:00, 13 July 2010 (UTC)
What's the title blacklist process? I haven't heard of it. — Timneu22 · talk 14:37, 13 July 2010 (UTC)
MediaWiki:Titleblacklist; information is on the talk page. PleaseStand (talk) 04:50, 15 July 2010 (UTC)

IP talk page edits adding large amounts of text

  • Task: Stop IP editors adding large amounts of text to talk pages.
  • Reason: Several editors and myself have been fighting a long, slow vandalism war at Talk:Disney's House of Mouse, against suspected Bambifan101 (talk · contribs) IP socks. These IPs add a huge amount of irrelevant information about every Disney hero and villain ever to this talk page. The text added takes the page from 5,640 bytes to 192,497 bytes. The article itself is longterm protected to stop the same abuses. I'm not at all comfortable with protecting a talk page.

It's been suggested on my talk page that an edit filter may be a good way to protect, as a rangeblock is not possible.

IPs involved this year alone are:

I'm open to discussion around the size of edits added that would trigger this, and whether it would be best on this page only (if that's possible) or wider. I'm not aware of this occuring on other pages, but if it is Bambifan101, then I'd be surprised if it was just this one. GedUK  12:17, 6 July 2010 (UTC)

Hello, everyone. My thanks to Ged UK for starting this discussion. I'm one of the several editors who've been fighting this particular vandal alongside Ged UK, although I've only just recently joined the fray. I suggest that the proposed edit filter be set initially at a fairly conservative level to minimize the number of false positives. Perhaps the following conditions, ANDed together, might be sufficient:
  • user_age == 0 ;<-- anonymous IPs
  • edit_count <= 20 ;<-- "new" IPs
  • action == "edit" ;<-- edits to existing pages
  • article_namespace == 1 ;<-- talk pages
  • (new_size / old_size) >= 30 & (new_size - old_size) >= 180000 ;<-- really BIG size jumps
My apologies if I've mangled the edit filter syntax, but I hope this is close enough to convey the basic gist of what I'm trying to describe. The intent is to catch only relatively new anonymous IP editors adding large amounts of text to existing talk pages. I realize that the term "large" is relative, and admit that my use of 180,000 bytes is rather arbitrary (30 x 6,000, roughly reflective of the current pattern of vandalism); maybe some analysis can be done throughout the namespace to find a more objective value. Cheers! -- Bgpaulus (talk) 14:22, 6 July 2010 (UTC)
A few comments: It's unfortunate, but a filter that is designed for only one page will still run (at least the first line of code) on every page. Even though it will do nothing because it won't meet the initial testing condition, just the fact that it's there still adds to the processing time for every edit on every page. Were it not so, I would have no objections to this filter in its original proposed form. However you've also proposed that it could be set up to run on talk pages in general. If so, I note that blanking of talk pages is not currently disallowed by any filter (although it is tagged), and that there is no way for the edit filter to know whether a given page has been recently blanked, so it's possible that if this filter is set up to run on talk pages in general that a page could be blanked and then an IP trying to revert it would be unable to do so. Lastly, the second line is unnecessary because the edit filter considers all IP's to have an edit count of zero, even if they have actually made thousands of edits from that IP address. I don't mean to sound negative, I just want to point out a few possible drawbacks of this filter. —Soap— 10:16, 15 July 2010 (UTC)
Also, it seems that even when creating a new page, the action is classified as "edit" rather than something such as "create". This is not necessarily bad, as the creation of huge talk pages where there previously was nothing is suspicious in itself. —Soap— 10:19, 15 July 2010 (UTC)
Thanks Soap. I feared it might be more effort than it's worth, and I think we're pretty well on top of it, but it's something that we might have to revisit. GedUK  11:42, 28 July 2010 (UTC)

Mangled image names

  • Task: Detect changes that turn images into redlinks, or break gallery displays, during spelling fixes or other mass editing. Examples are [6] (fixed here), [7] (fixed here), and [8] (fixed here).
  • Reason: Because some editors seem to not realise that they have broken an image link and this is not always spotted until much later. Either alert them to this, or output a file for others to check. See also Wikipedia:Database reports/Articles containing red-linked files. It might be better to deal with this as a database report, but am hoping people here will have some ideas about this as well.

- Carcharoth (talk) 09:53, 17 July 2010 (UTC)

  • This is a pretty tall order, and I don't really think it can be done. We do have Special:AbuseFilter/220, which detects when someone tries to hotlink an external image directly into an article (which is allowed on some non-WMF wikis), but there's no way for the filter to be able to know whether a given link is a redlink or not (it can often detect whether a page has redlinks, but not tell you which ones they are). It would definitely be a good filter if it's possible though. Since these examples are all from automated tools, it might be possible for the makers of the tools to be able to sieve out image names from their find/replace functions. —Soap— 11:09, 17 July 2010 (UTC)
  • Apparently AWB already does that. Can you tell what the automated tools were that the other editors were using? And if the edit filter can detect whether a page has redlinks or not, can it not check before and after an edit, and if the page now has redlinks to ask the editor if they intended to add redlinks? Trouble is, article redlinks are legitimate. Carcharoth (talk) 11:44, 17 July 2010 (UTC)
  • I read something about use of the "old_html" variable being discouraged, because for some reason it apparently consumes more processing power than new_html. The reason I bring that up is because the only way to detect the existence of a redlink is to look at the HTML source of the page, rather than the MediaWiki code, as all links appear the same in raw MediaWiki form. Special:AbuseFilter/163 is the only filter I know of that uses this. As for what tools those editors were using, I dont know. Rogriv seems to be using something automated since I doubt someone would type out summaries like that; the other person may just have been using simple find/replace and doing it by hand. —Soap— 21:32, 17 July 2010 (UTC)
  • Looks like it can't be done, but thanks anyway. The explanation taught me a bit more about this sort of thing! Carcharoth (talk) 23:42, 17 July 2010 (UTC)

Page blanker

  • Task: Disallow blanking of pages or massive content removal from the range 112.203.0.0/16
  • Reason: This is a persistent vandal that keeps blanking pages, going on for about two months now. Single blocks are not effective and there are too many pages to protect. Rangeblock is also not feasible as this is a highly active range.
Some IPs used are:

There's many others, but it's too long to list. Elockid (Talk) 20:18, 30 June 2010 (UTC)

  • There's already a filter against page blanking. Stifle (talk) 10:31, 11 August 2010 (UTC)

Top pages vandalizing bot

  • Task: Detect and filter edits similar to those of Jarjarbinks10 (talk · contribs), Jetlagorange (talk · contribs), and Irrypride (talk · contribs) that target Wikipedia's top rated pages (or some such list) inserting large text directing readers to an external website. The website varies, but so far the text has been attributed as large (5em or 10em) and "font-weight:bolder". Might also key off of marginally auto-confirmed editor status (less than 20 edits, less than 10 days) if that is possible. Likely only needs to be active for a few weeks or months.
  • Reason: Repeated recent attacks. -- Tom N (tcncv) talk/contrib 02:56, 20 July 2010 (UTC)
  • We don't have a way of identifying "top rated pages" for an edit filter. Stifle (talk) 11:34, 28 July 2010 (UTC)
    Wikipedia:Popular pages is inactive. Hazard-SJ Talk 13:39, 21 August 2010 (UTC)
    Wikipedia:Most vandalized pages? Hazard-SJ Talk 13:41, 21 August 2010 (UTC)
    The edit filter can't parse a random page to see if some other page is listed on it. Stifle (talk) 15:06, 29 December 2010 (UTC)

Laughing

  • Task: Prevent and warn editors who introduce 'lol' or 'LOL' into articles.
  • Reason: Many bad-faith vandalism edits contain the phrase 'lol' or 'LOL'. If this filter could be limited to the Article namespace, as sometimes editors may use 'lol' in discussions, but there are almost zero situations when it should be included in an article.Acather96 (talk) 16:37, 3 August 2010 (UTC)
    I think there would be too many false positives, actually: search results in article space for 'lol'. We do block longer strings such as "lololol" though. If this does get created, I think it should not be set to disallow, because answering the false positive reports would be quite a headache ... similar to the current "your mom" filter but probably more extensive. —Soap— 10:10, 5 August 2010 (UTC)

Personal information of specific individual

  • Task: Prevent the repeated addition of a specific individual's personal information including phone number
  • Reason: An anonymous editor using a variety of IPs has been repeatedly posting someone's personal phone number along with sexually suggestive text. Because of the multiple IPs in a broad range involved, blocking has been ineffective and range blocking is not practical. Here is one example of the offending text. (Only administrators will be able to view this edit because I have hidden it using Revision Deletion, but if a non-admin would like to help out here, I'll email the text.) At this point it seems an edit filter would be the best way to handle this situation. There is a bit of additional info at my talk page here. -- Ed (Edgar181) 19:00, 5 August 2010 (UTC)
This has been done before, on 293 and one other one I can't find right now. I believe it was turned off because we decided it was better off going straight to the oversighters, and edit filter logs can't be oversighted yet (though a fix is coming soon). However it's possible it was turned off because the vandal was stopped by an enormous rangeblock we laid down at about the same time; basically, if the oversight team is OK with restoring this filter then we can do it and maintain it very easily; if not I would recommend waiting until the fix that will allow us to oversight edit filter logs is in place. —Soap— 22:36, 5 August 2010 (UTC)
I'm not sure I understand the reason that it might have been turned off. Would the edit filter unavoidably create some kind of public log entry that contains the information that shouldn't be seen? If so, I agree that it would make sense to go directly to oversight. -- Ed (Edgar181) 23:16, 5 August 2010 (UTC)
Yes, in Special:AbuseLog, which can be searched by filter ID and other things, and will show the text of the hits even if the filter is private. —Soap— 23:50, 5 August 2010 (UTC)
OK, it definitely looks like an edit filter is not appropriate this case then. Thanks for the explanation. -- Ed (Edgar181) 00:37, 6 August 2010 (UTC)

Cross-namespace redirect

  • Task: Detect creation of cross-namespace redirects.
  • Reason: This is a common issue here at Wikipedia, as CNRs often dominate the lists of WP:RFD - ANDROS1337 03:23, 23 August 2010 (UTC)
  •   Technically possible to handle all current namespaces and update in case of any change, but comlpicated. עוד מישהו Od Mishehu 08:42, 23 August 2010 (UTC)
  • Would be a small bit easier if the abuse filter supports lookaheads in the regular expressions... does it? I couldn't find whether it does anywhere. - EdoDodo talk 12:15, 23 August 2010 (UTC)
Be careful not to make policy with this filter. The types of CNRs found in CSD R2 are all subject to speedy deletion, and hence can be regarded as "prohibited". Other CNRs are not subject to speedy deletion, even though they are usually deleted if presented at RfD, so using the filter to block them would go beyond current practice. — Gavia immer (talk) 15:34, 23 August 2010 (UTC)
I agree with Gavia immer here. Detect might be ok, but it probably should not warn or prevent. Might be best handled by a WP:DBR (which already exists)... –xenotalk 15:36, 23 August 2010 (UTC)

Multiple unexplained section blankings at one article

  • Task: Disallow a single anonymous/unconfirmed user from making 5+ edits in the form of unexplained section removals at a single article within a 30-minute period.
  • Reason: Some vandals have wrecked pages by systematically removing one section at a time to avoid attracting the attention of antivandal bots or tripping the filter that stops large deletions of content.

- --SoCalSuperEagle (talk) 08:55, 24 August 2010 (UTC)

There's already a filter in place for this that just tags it at the moment, Special:AbuseFilter/172. This is actually one of the most-often-hit filters on Wikipedia. I don't know why it doesn't actually do anything, maybe you could ask some of the people who have worked on it, or at WT:EF. Someguy1221 (talk) 09:05, 24 August 2010 (UTC)

Example file tagging

  • Task: Identify edits where example files are included (example.jpg, example.png, etc)
  • Reason: To make edits where example files are included easier to find.

- JV Smithy (talk) 02:38, 6 September 2010 (UTC)

Are you talking about any file which begins with the word "Example"? If not, then what are the files/criteria for identifying them? עוד מישהו Od Mishehu 05:46, 6 September 2010 (UTC)
For example: [[File:Example.jpg]] included in articles is what I meant. JV Smithy (talk) 01:22, 8 September 2010 (UTC)
This was once tagged by Filter 18. I don't know why it was deleted. Someguy1221 (talk) 09:07, 19 September 2010 (UTC)
Too many people were making positive edits with test-type formatting included by accident. I don't know whether that was the only reason for turning it off or if it was also during a time when we needed to cut back in general, though. —Soap— 10:05, 19 September 2010 (UTC)

Filter 231

Next to "article namespace", perhaps == 0 should be changed to >= 50 (or was it => 50 ?). Smaller ones can presumably be tagged as blanking. mechamind90 15:28, 16 September 2010 (UTC)

No, this condition means that the page is an article. עוד מישהו Od Mishehu 08:52, 19 September 2010 (UTC)

Social network catcher

  • Task: It would catch anything with a Social network link, and tag it.
  • Reason: I see lots of pages with things like Facebook, and Myspace, and almost all of the time it's not needed.

Oh yeah, I could also help write the code if anyone wants. Endofskull (talk) 02:44, 17 September 2010 (UTC)

Please e-mail me the recommended code. Do not publish it publicly, per WP:BEANS. עוד מישהו Od Mishehu 11:29, 5 October 2010 (UTC)
But it shouldn't be blocked outright, nor do I think that it should be tagged. A lot of time, a facebook or myspace link may be included in the External Links section within reason (in fact, there are more than 1300 uses of myspace and facebook in the article space right now). Even if it isn't really a good place for the link, it's not the worst thing in the world, and not something serious enough to deserve a tag, warning, or block. Tim1357 talk 01:24, 17 November 2010 (UTC)
I think this is already covered by XLinkBot. Stifle (talk) 15:03, 29 December 2010 (UTC)

N-word Filter

  • Task: Disallow and warn any nonconfirmed users who insert the "n-word" in either the 5 of 6 letter forms.
  • Reason: This edit and about 50 more I each day demonstrate to me that non-autoconfirmed users cannot be trusted with the freedom to use this word "constuctively." Yes, there are legitimate uses of the word, such as pages on historical discrimination, but the vast number of uses are vandalism. The legitimate uses can be put it by autoconfirmed users.
BTW, if there already is a filter that supposedly does this, it wasn't working in the past few days.

- Sven Manguard Talk 02:56, 17 October 2010 (UTC)

We have filter 9, which tracks things like this. But it's not set to disallow because it would have a lot of false positives. —Soap— 12:04, 17 October 2010 (UTC)

Rickrolling, Wikipedia style

  • Task: Edit filter to note (with exceptions for pages where such a link does legitimately belong) edits which add the text "Rick Astley" to an aritcle.
  • Reason: I don't know if this is a one-time gag or an emerging epidemic that Wikipedia's going to need to start watching out for long-term, but take a look at this edit. I don't think we should blacklist the phrase entirely, since there will always be some articles where a link to Astley's article is warranted and legitimate — but to prevent rickrolling from becoming a Wikipedia phenomenon as well, I think there should be a filter in place to at least flag such edits for double-checking. An alternative might be to explicitly blacklist the phrase "Rick Astley" if and when it's followed by unrelated pipetext. Bearcat (talk) 19:48, 31 October 2010 (UTC)
  • I think WP:RBI will suffice for this. Stifle (talk) 15:08, 29 December 2010 (UTC)

Default Text Editor Filter

  • Task: Edits in which the diff includes additions like '''Bold text''' and ''Italic text'' and other defaults from the wiki editor.
  • Reason: Identify vandalism.

No expert with edit filters, but throwing the idea out there. - Allmightyduck  What did I do wrong? 20:33, 1 November 2010 (UTC)

Already exists as Filter 18. We disabled it about a year ago because a lot of times people will make edits that have helpful information and "toolbar spam" in the same edit, and in some cases may not even be aware of it. —Soap— 20:36, 1 November 2010 (UTC)
Gotcha. Thanks. Allmightyduck  What did I do wrong? 21:11, 1 November 2010 (UTC)

Filter 50

  • Task: Instead of using [A-Z], wouldn't be better to use unicode property \p{Lu}?
  • Reason: This way it would match other uppercase letters which are not in [A-Z], like those with have diacritics.

I'm also asking because at Portuguese wikiprojects we also need to match characters like [ÁÀÂÃÇÉÊẼÍÓÒÔÕQ̃ÚŰÜŨ] and I though of adding \p{Lu} instead of specify every possible character. Would it be efficient enough or is it better to add the characters explicitly in [...]?

Thanks for your help, - Helder 23:56, 6 November 2010 (UTC)

Perhaps using ccnorm to get rid of accents and other things like that would be a good idea. I won't make any changes though, because I'm not familiar with this filter. - EdoDodo talk 09:09, 7 November 2010 (UTC)
Unfortunately, currently ccnorm doesn't work for most of those characters, as pointed out in bugzilla:25619. Helder 12:21, 7 November 2010 (UTC)

User creating talkpage of nonexistent mainspace page

  • Task: Tag, warn, and possibly even disallow on the second try editors who attempt to create talkpages of nonexistent articles; note that this does not apply to other namespaces, such as File talk and Wikipedia talk namespaces, which may be important for later use. If possible, and for the purpose of helping the people at WP:CP please restrict this filter to talkpages that are not of subpages, such as Talk:'Article'/temp, which are meant to store versions of articles with copyright removed for later improvement. If it is not possible to exempt subpages from the filter, then it may be best to restrict the talkpage creation filter to new editors.
  • Reason: To clean up after useless talkpages created in error or obvious vandalism, because they waste time and resources.

- :| TelCoNaSpVe :| 00:50, 21 November 2010 (UTC)

  Not done As far as I know this is not possible, because the edit filter does not have access to information about other pages, only the page that was edited. There is a database report at Wikipedia:Database reports/Orphaned talk pages that lists talk pages without a corresponding article though, it's updated every few days by a bot. - EdoDodo talk 14:42, 24 November 2010 (UTC)
Retrieved from "https://en.wikipedia.org/w/index.php?title=Wikipedia:Edit_filter/Requested/Archive_4&oldid=1214632685"