![]() | This is an archive of past discussions. Do not edit the contents of this page. If you wish to start a new discussion or revive an old one, please do so on the current main page. |
Archive 1 | Archive 2 | Archive 3 | Archive 4 | Archive 5 |
Filled requests
Signatures in article space
- Task: Detect the addition of signatures in the article space by anon editors.
- Reason: All too often I see vandals going and signing the article space. Most, if not all of the time, it is unconstructive, with anon's often putting things like "So and so was here! ~~~~", and other things. Since signatures do not belong in the article space, why not create a filter for it?
- UntilItSleeps PublicPC 18:08, 20 May 2009 (UTC)
- See Wikipedia:Abuse_filter/Requested#Wikilink_username and Filter 176. Rather than searching for Smallman12q (talk) 19:46, 31 May 2009 (UTC), we would search for a link to "user:" and "user talk:" Smallman12q (talk) 19:46, 31 May 2009 (UTC)
Section removal
- Task: Detect the deletion of a complete section
- Reason: I see often 'vandals' deleting random sections from articles (see e.g. diff). Often this is vandalism (someone clicking the edit button on a section, delete the content, press save. It is annoying, it happens often, but I don't see how to write a filter for this. Someone an idea? --Dirk Beetstra T C 17:57, 18 May 2009 (UTC)
- I see I was corrected on this removal, it was deemed unreliable. Nonetheless, can we at least try to find a filter and log it, and see how often it is vandalism. --Dirk Beetstra T C 18:40, 18 May 2009 (UTC)
- I actually created such a filter in another project and the code looks like this:
action == 'edit' & user_editcount < 100 & edit_delta < -500 & (!summary | summary like '/\**\*/' | summary contains 'Replaced content with ' ) & (!added_lines | (edit_delta < -1000 & !count('\n', added_lines))) & user_name != article_recent_contributors &! ((article_namespace==2 | article_namespace==3) & article_text rlike ('^'+user_name))
- It is triggered when a big piece of text is removed or replaced with some words and there is no edit summary. Filter has a warning only that roughly says this: "you are removing big piece of text without any explanation; please make sure your edit is constructive and explain the removal in edit summary field". P.S. This needs some more code so the filter is not triggered together with other "page blanking" filters. — AlexSm 19:58, 18 May 2009 (UTC)
I am now thinking along lines of :
old text contains (removed text + "==") AND new text does not contain (removed text + "==")
I'll play a bit. --Dirk Beetstra T C 08:59, 20 May 2009 (UTC)
- Tried in Special:Abusefilter/172. Seems to work, set to log/tag only. --Dirk Beetstra T C 11:27, 20 May 2009 (UTC)
- I often remove spam and copyvio in this manner, by removing an entire section. So do many other patrollers. I'd be pretty annoyed to get a message about it. DGG (talk) 06:14, 7 June 2009 (UTC)
Image:http://
- Task: Warn new users trying to add
[[image:http://...]]
to any page. - Reason: Help users to do the right thing and prevent some useless edits.
I noticed how often I see this happening and then I created this filter in Russian Wikipedia. It's been running for two weeks with approx. 21 hits per day. To be more user-friendly, I then split it into two filters like this:
- filter file-http (in ruwiki), should warn and tag; warning should say that Wikipedia doesn't support external images for illustration, suggest to find alternative existing image in Wikimedia projects and suggest to upload this image if it's free or can be used as fairuse.
action=='edit' & user_editcount < 100 & lcase(added_lines) rlike '\[\[:?(file|image|media):http:\/\/' &! lcase(added_lines) rlike '\[\[:?(file|image|media):http:\/\/(upload\.wikimedia|en\.wikipedia)\.org\/"
- filter file-http-wp (in ruwiki), should warn about incorrect syntax and suggest to correct it to e.g. [[Image:name|thumb]].
action=='edit' & user_editcount < 100 & lcase(added_lines) rlike '\[\[:?(file|image|media):(http:\/\/)?(upload\.wikimedia|en\.wikipedia)\.org\/'
— AlexSm 17:42, 14 May 2009 (UTC)
- This seems like a good idea. Stifle (talk) 09:19, 18 May 2009 (UTC)
- Done - I actually had the same idea before I saw this. It's here: Special:AbuseFilter/220. King of ♥ ♦ ♣ ♠ 05:24, 7 August 2009 (UTC)
([^\s]+(?=\.(jpe?g|gif|png|bmp|svg))\.\2)
Some useful syntax from taken from a regex cheat sheet=D.
(indent) I don't think it was a good idea to squeeze that much into one filter, it might be confusing for editors. Also, after reading Abusefilter-warning-external-images, I can guess that you haven't seen {{External media}} yet... — AlexSm 18:53, 19 August 2009 (UTC)
- I exempted that template (and all its redirects) from the filter criteria. -- King of ♥ ♦ ♣ ♠ 19:01, 19 August 2009 (UTC)
Blanker
- Task: The filter should prevent the blanking of lengthy pages.
- Reason: There's no reason for an IP to blank a page.
- Smallman12q (talk) 20:10, 4 May 2009 (UTC)
- Filter 3 does this. Gurch (talk) 20:52, 4 May 2009 (UTC)
Catching new pages that are copies of existing pages
- Task: Logs if a new user creates a page that contains a maintenance tag (or at least one of the more common ones) or a page that contains the text [edit].
- Reason: Similar to the cut-and-paste filter below that was ruled out, this would catch the worst examples of copied articles. I have encountered a number of such pages over the years and would like to see a filter run in log only mode to find out how common it is.
- Someguy1221 (talk) 08:36, 24 April 2009 (UTC)
- Special:AbuseFilter/164. I used the ambox style as a target for "maintenance tag" but that may be too general. Dragons flight (talk) 19:24, 2 May 2009 (UTC)
- It is quite common for people to create articles with maintenance tags on them, and as far as I'm aware there is nothing wrong with doing so. Or are only administrators allowed to do that now? Gurch (talk) 20:04, 2 May 2009 (UTC)
- I have no desire (and I'm sure Dragons flight doesn't either) to prevent people from creating an article with a maintenance tag. For this I only want the filter to catch precisely what I stated it should catch. I'm merely trying to find a good set of conditions that indicates such a thing has happened. Someguy1221 (talk) 10:11, 5 May 2009 (UTC)
- I don't think I've ever done that, but looking at the log, you may be correct. Dragons flight (talk) 23:35, 2 May 2009 (UTC)
- It's been known to happen when someone recreates a page that was deleted. But I agree with Gurch, this can generate false positives. Stifle (talk) 09:01, 5 May 2009 (UTC)
- I looked through the log and they do seem to be mostly false positives. I think those could be eliminated by confining the search for maintenance tags to the more common ones you wouldn't expect anyone to tag their own creation with (unreferenced, advert, notability). Would I be correct that this would be done by making something like the following change:
- It is quite common for people to create articles with maintenance tags on them, and as far as I'm aware there is nothing wrong with doing so. Or are only administrators allowed to do that now? Gurch (talk) 20:04, 2 May 2009 (UTC)
(
("[edit]" in new_wikitext) | (new_html rlike 'class="[^"]*ambox[^"]*"')
)
to
(
("[edit]" in new_wikitext) | (rlike ([({{unrefeferenced)({{advert)({{advertisement)({{notability)]|?.*}})
)
- Didn't make the change myself since I'm unfamiliar with regex, and I've broken enough of Wikipedia today. Someguy1221 (talk) 09:34, 5 May 2009 (UTC)
- SmackBot likes to convert these to use uppercase letters (e.g. unreferenced becomes Unreferenced) when dating them. Could you use a case-insensitive match? --Damian Yerrick (talk | stalk) 19:36, 7 May 2009 (UTC)
- Done. Someguy1221 (talk) 19:41, 7 May 2009 (UTC)
- SmackBot likes to convert these to use uppercase letters (e.g. unreferenced becomes Unreferenced) when dating them. Could you use a case-insensitive match? --Damian Yerrick (talk | stalk) 19:36, 7 May 2009 (UTC)
- Didn't make the change myself since I'm unfamiliar with regex, and I've broken enough of Wikipedia today. Someguy1221 (talk) 09:34, 5 May 2009 (UTC)
User Name check
- Task: Detect when a new user creates a page that matches their username(or almost matches) and then display a WP:COI message.
- Reason: Prevent COI issues.I'm not sure if this filter currently exsists. It also may cause problems in that there are people who want to create nuetral articles regarding themselves/their work. But this is still a proposal I had in mind.
- Smallman12q (talk) 00:47, 17 April 2009 (UTC)
- User:COIBot's definition of "username almost matches page title" causes me to be listed as a suspicious contributor every time I edit an article with "church" or "gulch" in the title. If you do implement something similar, please make it more strict. Gurch (talk) 17:18, 17 April 2009 (UTC)
- I've just attempted to make one at Special:AbuseFilter/148. Xclamation point 02:57, 21 April 2009 (UTC)
Another problem here is, while COIBot understands whitelisting etc. (so now it is ignoring Gurch and many others), we do have some editors with a confirmed and discussed conflict of interest, who still may turn out positive on this filter. By the way, Special:AbuseFilter/148 seems to be doing something different. --Dirk Beetstra T C 08:55, 1 May 2009 (UTC)Strikethrough, did not read properly. Sorry. --Dirk Beetstra T C 08:56, 1 May 2009 (UTC)- The filter seems to be working well. Perhaps it can be modified to also prevent a user from creating a redirect to their user/talk page.Smallman12q (talk) 16:53, 10 May 2009 (UTC)
Kamen Rider Blade/Card suit vandal
- Task: Block edits to the card suit articles where an editor on Japanese ISP(s) adds content relating solely to Kamen Rider Blade by adding the names of the cards from Blade to the suits. Example edits are as follows to each of the suit pages (followed by a link to the relevant Kamen Rider article where the content is normally found):
- Reason: Blocking these ISPs from editing may be detrimental, even though the vandalism is taking place on both en.wiki and ja.wiki (I am trying to work with the Japanese Wikipedia to help them prevent the same vandalism on their project). With the fact that this vandal is now IP hopping, and the vandalism is easy to catch, it'd be easier to prevent it entirely with the edit filter.
- —Ryūlóng (竜龙) 21:09, 8 August 2009 (UTC)
- Done - Special:AbuseFilter/222. Currently log mode only. King of ♥ ♦ ♣ ♠ 16:39, 10 August 2009 (UTC)
Japanese police vandal
- Task: Prevent IP editors originating from the 220.111. and 221.191. ranges, from adding Japanese language text to talk pages of articles related to police around the world. A filter would need to apply to Talk:Police (edit | article | history | links | watch | logs), Talk:United States Capitol Police (edit | article | history | links | watch | logs), Talk:Scotland Yard (edit | article | history | links | watch | logs), Talk:Law enforcement in Italy (edit | article | history | links | watch | logs), Talk:Chief police officer (edit | article | history | links | watch | logs), Talk:Chief of police (edit | article | history | links | watch | logs), and maybe others. Typical text to look for is: "一美のママ一美をたすけて" or "一美一美のママをたすけて".
- Reason: The admin who blocked one IP in the range as a result of my AIV report, recommended I raise the issue here and request a filter. I'll see if I can come up with more examples to refine the filter's effectiveness.
- Astronaut (talk) 14:04, 8 August 2009 (UTC)
- Done - Special:AbuseFilter/221. I don't know if it's possible to check for Japanese text though. King of ♥ ♦ ♣ ♠ 01:03, 10 August 2009 (UTC)
Non-admin reviewing unblock request
- Task: Prevent non-admins from reviewing unblock requests
- Reason: It's a common thing for "new" users to go around the current requests and declining them with a personal attack, or for blocked users to review their own requests
- Kingpin13 (talk) 21:04, 17 July 2009 (UTC)
- Done - Special:AbuseFilter/207. As written, it's probably going to have plenty of false positives, so I'm leaving it in log-only mode for a while. King of ♥ ♦ ♣ ♠ 00:02, 18 July 2009 (UTC)
Removal of WebCite URLS
- Task: Warn users if they attempt to remove a link to webcitation.org
- Reason: Webcitation.org (linkedover 20K times) is currently experiencing technical problems which may make archive links appear dead even though they aren't and make people want to remove the broken links. Additionally, even without this bug inexperienced editors may remove or replace these links with "current" ones if they don't understand why they are used to begin with.
- ThaddeusB (talk) 05:07, 14 July 2009 (UTC)
- Done - Special:AbuseFilter/208. Currently in log-only mode. King of ♥ ♦ ♣ ♠ 00:19, 18 July 2009 (UTC)
Tag removals of PROD templates
- Task: Tag revisions where a {{dated prod}} template is removed.
- Reason: I was thinking of a bot that would notify users that the prod they placed on an article has been contested, so they can decide whether they want to pursue deletion via AFD. But in order to not make this bot to have to check every new edit, it needs a way to identify such changes and I think the AF can do that quite well.
- Regards SoWhy 06:23, 8 July 2009 (UTC)
- Done - Special:AbuseFilter/200. How does it look? King of ♥ ♦ ♣ ♠ 21:01, 10 July 2009 (UTC)
- Looks fine as far as I can tell but I am not an expert. I tested it against a test-edit I made and it seems to work fine, so I added the tag and enabled it. Thanks for the code :-) Regards SoWhy 21:14, 10 July 2009 (UTC)
- Jeff friedl - First catch! King of ♥ ♦ ♣ ♠ 21:19, 10 July 2009 (UTC)
- Looks fine as far as I can tell but I am not an expert. I tested it against a test-edit I made and it seems to work fine, so I added the tag and enabled it. Thanks for the code :-) Regards SoWhy 21:14, 10 July 2009 (UTC)
plwha.org
- Task: Prevent the spamming of www.plwha.org and related domains.
- Reason: One long-term spammer, hopping across a range of IPs too broad to effectively rangeblock, has been repeatedly adding links to www.plwha.org for months. The site is blacklisted, but the spammer is now adding it without the "http://" so that it's not formatted as an external link. Reverting, blocking, small rangeblocks and blacklisting have yet to be effective. Due to this spammer's persistence another solution is needed. See 86.121.194.209 (talk · contribs) for an example of recent activity. - -- Ed (Edgar181) 14:17, 6 July 2009 (UTC)
- Done - Special:AbuseFilter/201. King of ♥ ♦ ♣ ♠ 21:14, 10 July 2009 (UTC)
Saban troll prevention
- Task: Prevent addition of libelous edits relating to the individual Haim Saban
- Reason: See User:Ryulong/Sandbox#Saban troll for extreme detail.—Ryūlóng (竜龙) 08:03, 5 July 2009 (UTC)
- This user seems to make a lot of different types of edits; it is impossible to create a filter without knowing what specifically we are going after. Is there any particular type of edit that you are interested in? King of ♥ ♦ ♣ ♠ 21:42, 10 July 2009 (UTC)
- There is one definite edit that the user performs which are the page blankings.—Ryūlóng (竜龙) 02:17, 16 July 2009 (UTC)
- Is it recent enough? A few hits per month do not justify an abuse filter. Have there been any such edits since July 5? King of ♥ ♦ ♣ ♠ 04:19, 18 July 2009 (UTC)
- The user is currently editing now as 67.70.153.70 (talk · contribs · WHOIS). He has not performed the page blanking, but he's doing the other edits which identify him.—Ryūlóng (竜龙) 01:05, 27 July 2009 (UTC)
- Done - Special:AbuseFilter/213. King of ♥ ♦ ♣ ♠ 16:01, 27 July 2009 (UTC)
- The user is currently editing now as 67.70.153.70 (talk · contribs · WHOIS). He has not performed the page blanking, but he's doing the other edits which identify him.—Ryūlóng (竜龙) 01:05, 27 July 2009 (UTC)
- Is it recent enough? A few hits per month do not justify an abuse filter. Have there been any such edits since July 5? King of ♥ ♦ ♣ ♠ 04:19, 18 July 2009 (UTC)
- There is one definite edit that the user performs which are the page blankings.—Ryūlóng (竜龙) 02:17, 16 July 2009 (UTC)
Masked Rider Blade Knight vandalism prevention
- Task: Prevent vandalism to main space, talk space, template space, user talk space, etc. addition of hoax content
- Reason: See User:Ryulong/Sandbox#Masked Rider Blade Knight for extreme detail.—Ryūlóng (竜龙) 08:03, 5 July 2009 (UTC)
- Done - Special:AbuseFilter/202. King of ♥ ♦ ♣ ♠ 21:42, 10 July 2009 (UTC)
Filter to catch template vandalism originating from 88.105.0.0/16
See Wikipedia_talk:Administrator_intervention_against_vandalism#template_spammer
- Task: Catch vandalism originating vandal using dynamic IP addresses in 88.105.0.0/16 range.
- Reason: Range is too wide for a rangeblock which would block 65536 IPs, but vandalism mostly consists of the same template additions. See this permalink for details[1]. Not sure if it a feasible thing to do so thought I'd pose the question here. Mfield (Oi!) 23:42, 4 July 2009 (UTC)
- Done - Special:AbuseFilter/203. King of ♥ ♦ ♣ ♠ 22:14, 10 July 2009 (UTC)
- I've disallowed the action, as it seems to work. -- King of ♥ ♦ ♣ ♠ 23:53, 10 July 2009 (UTC)
Duplicated section
- Task: Block creation of new section with exaclty the same content of a section created recently at the same page.
- Reason: Some times when the user clicks at [+] and create a new section in a page, he could acidently insert the same section again if he reload the page (maybe because the conection is slow, or because of clicking the submit button twice). Example: this edition.
- Helder (talk) 21:34, 25 June 2009 (UTC)
- Done -- Special:AbuseFilter/197. -- Cobi(t|c|b) 12:27, 26 June 2009 (UTC)
- Disabled due to performance issues. Stifle (talk) 15:56, 3 July 2009 (UTC)
4chan/crabcore
- Task: Detect widespread 4chan use of the term "crabcore"
- Reason: Apparently there is a current 4chan meme where they are trying to get the term crabcore adopted. The AfD for Crabcore shows lots of first-time IPs commenting on the AfD, and a lot of bands (such as Attack! Attack! and Attack Attack! have been subjected to drive-by IP vandals. It's never repeat IP addresses, which is what makes me think a filter would be a great way to handle this. My suggestion would be to reject edits from first-time editors (editcounts < 5) who add "crabcore" to a page in article space. Thoughts? tedder (talk) 02:42, 24 June 2009 (UTC)
- Special:AbuseFilter/190 -- running in log-only mode to see about false positives. Will put it in enable-mode later. -- Cobi(t|c|b) 07:28, 24 June 2009 (UTC)
- Done -- Cobi(t|c|b) 08:14, 25 June 2009 (UTC)
- Thanks, Cobi. tedder (talk) 11:04, 25 June 2009 (UTC)
- An update- this is working quite well- the filter log shows it's doing precisely what it should. Cheers! tedder (talk) 21:26, 26 June 2009 (UTC)
- Thanks, Cobi. tedder (talk) 11:04, 25 June 2009 (UTC)
- Done -- Cobi(t|c|b) 08:14, 25 June 2009 (UTC)
Gunfu vandalism
- Task: Block/flag edits made by IPs in the 69.124 range to add Category:Gun fu films or Category:Martial arts films to any article or category
- Reason: User hits hundreds of articles at a time before noticed, falsely adding the categories to irrelevant articles and categories; keeps changing IPs so blocking is ineffective.
- -- Collectonian (talk · contribs) 22:01, 15 June 2009 (UTC)
- Special:AbuseFilter/187 -- I'm running it in log-only mode for a few hours to see about false positives, but there should be none. I'll put in full effect later today or early tomorrow morning, assuming there are no issues. -- Cobi(t|c|b) 12:54, 23 June 2009 (UTC)
- Thanks. Just wanted to let you know, so far so good! No more cat spamming :) -- Collectonian (talk · contribs) 13:25, 8 July 2009 (UTC)
INFRINGEMENT vandal
- Task: This vandal abuses AOL's dial-up to add ==INFRINGEMENT== to totally random pages, usually with an edit summary of "INFRINGEMENT" as well. See Special:Contributions/172.130.240.199 for a typical example. These edits are always made as an anon, and all the offending IPs thus far have come from 172.128.0.0/10.
- Reason: A /10 is too large to rangeblock effectively, and due to the nature of AOL's proxy services, any rangeblocks not covering the entire /10 would cause massive collateral damage, and would be mostly ineffective.
- J.delanoygabsadds 17:57, 9 June 2009 (UTC)
- Only devs can rangeblock anything larger than a /16 anyway. I would support this. Stifle (talk) 15:06, 19 June 2009 (UTC)
- Done -- Special:AbuseFilter/186. -- Cobi(t|c|b) 12:24, 23 June 2009 (UTC)
No Wiki Markup at all
- Task: Detect when a page newly created has no wikimarkup at all and tag
- Reason: No Wikimarkup is a sign of a copyvio or other improper copy/paste move.
- Triplestop (talk) 20:44, 4 June 2009 (UTC)
- I have seen many such newpages created by inexperienced editors. In fact, I have seen advice given to someone new to the project to get some sort of text in, and worry about the markup later. Many people with slow connections are used to saving after each small edit and would naturally enter text this way. It's not necessarily a bad idea. I wouldn't even log, let alone give a warning. DGG (talk) 06:11, 7 June 2009 (UTC)
- Already covered by #180. Stifle (talk) 15:09, 19 June 2009 (UTC)
New users altering file copyright tags
- Task: A filter that logs all changes to copyright tags of images, sounds and other media files by new accounts.
- Reason: New users are often not aware of copyright rules and are more likely than others to break the rules by changing these tags without the proper permission being given. (I am not sufficiently well-versed in regular expressions to code this one).
- Mgm|(talk) 19:23, 18 March 2009 (UTC)
- One can get at changes to templates (sort of), but there is no uniform way to distinguish copyright tags from other templates. Dragons flight (talk) 19:53, 18 March 2009 (UTC)
- That is, not without setting up a massive regex, but that would probably encounter some performance issues - we've already had several today. Hersfold (t/a/c) 19:57, 18 March 2009 (UTC)
- If all you cared about was finding the edit and not warning the user or disallowing it, you could probably just look for any template change on a File page. I don't think there's that many templates used for files that aren't copyright-related. Mr.Z-man 04:38, 19 March 2009 (UTC)
- Testing as Special:AbuseFilter/59. Would grab all edits by new users that modify templates on file pages. Dragons flight (talk) 04:50, 19 March 2009 (UTC)
- I see possible valid reasons to change such templates, so I wouldn't add blocks or warnings to it any time soon, but I feel better now these are logged. - Mgm|(talk) 08:49, 19 March 2009 (UTC)
MascotGuy username detection filter
- Task: Filter should be programmed to detect the very predictable naming conventions and multiple sockpuppets of what may well be this site's most prolific and pesky vandal as outlined at WP:LTA/MG.
- Reason: This user has access to a seemingly endless number of IPs. September 2009 will mark this vandal's fifth anniversary of sockpuppets and misinformation.
- PMDrive1061 (talk) 22:34, 28 April 2009 (UTC)
- Note: In the month of April alone, this individual has created 12 "master socks" which in turn were responsible for 43 sub-accounts. Despite an overall sock count well into the thousands, there are only 23 basic variations of the usernames, all of which are listed at the LTA page. Create a bot or bots which recognize these variations and I think we'll finally be rid of this little scourge. --PMDrive1061 (talk) 03:04, 29 April 2009 (UTC)
- Created as log: Special:AbuseFilter/159 (log). Blocking all those variations may hit too many legit users, but the log should be a good start. Dragons flight (talk) 21:48, 29 April 2009 (UTC)
Yay! I plan to monitor that filter, believe me. Thank you! --PMDrive1061 (talk) 22:19, 29 April 2009 (UTC)
- So he uses accounts to create other accounts? Is there a way to catch when that happens? Wknight94 talk 14:51, 1 May 2009 (UTC)
- The abuse filter is not presently capable of detecting that, though I'd like to see that function added. Dragons flight (talk) 10:02, 2 May 2009 (UTC)
- Indeed, right now with
action=='createaccount'
when A creates B unfortunately we haveuser_name == accountname == 'B'
. Werdna on IRC told me that there is a bug open, but I could not find it. —AlexSm 15:08, 2 May 2009 (UTC)
- Indeed, right now with
- The abuse filter is not presently capable of detecting that, though I'd like to see that function added. Dragons flight (talk) 10:02, 2 May 2009 (UTC)
Addition of 'TBA' and '<artist's> X-th studio album' to discographies
- Task: A filter that prevents users from adding 'TBA', 'TBA (20xx)' and similar to the discography sections (example diff here) of musical artists and the 'Next album' field in {{Infobox Album}} and also prevents the addition of "<artist's> X-th studio album"-style entries in the same locations, unless an actual article exists on the forthcoming-but-unnamed album.
- Reason: Violation of WP:V and WP:CRYSTAL (see also Wikipedia:NMG#Albums) and a frequent mistake made by new users and anons. I have personally removed these kinds of entries more times than I can remember over the past couple of years. Disallowing such edits and providing the user with a simple explanation and some links would save a lot of people's time and effort.
--Kurt Shaped Box (talk) 03:03, 26 March 2009 (UTC)
- It is impossible to do article existence checks, so that is out. Also, it would be difficult to look for the <artist's> part unless that happens to be the page title it is happening on. One can go after people adding "TBA", "To be announced", "<some number>th studio album" and phrases like that. The infobox field is obvious enough but is the discography section always titled "Discography"? Dragons flight (talk) 03:36, 26 March 2009 (UTC)
- Always? No. High 90 percent range? Yep. Logging "TBA" and "nth studio album" in sections or articles named "discography" would be helpful. I'm not sure that you can go past warn stage on this one, though, because you can't do the existence check, and if the article exists, the addition is legitimate. Similarly, there are rare cases where TBA is OK.—Kww(talk) 03:47, 26 March 2009 (UTC)
- I've seen discography sections that were named "Albums" or "Studio albums" in the past but I'm struggling to think of any other titles. Under what circumstances would 'TBA' be a legitimate entry into an artist's discography? I certainly can think of circumstances where listing the *release date* as 'TBA' would be fine (e.g. the Chinese Democracy saga)... --Kurt Shaped Box (talk) 03:58, 26 March 2009 (UTC)
- I don't think there's a reasonable way for the filter to distinguish a title of "TBA" from a release date of "TBA". Release date was actually what I was thinking of as the occasional legitimate use of TBA in a discography.—Kww(talk) 04:03, 26 March 2009 (UTC)
- Trying out Special:AbuseFilter/105. Dragons flight (talk) 05:58, 26 March 2009 (UTC)
- This seems to be good, I recommend putting it on warn. Stifle (talk) 17:49, 25 April 2009 (UTC)
- MediaWiki:Abusefilter-warning-crystalball would be an appropriate warning. Stifle (talk) 17:57, 25 April 2009 (UTC)
- Changed to warn. Stifle (talk) 17:44, 26 April 2009 (UTC)
- Trying out Special:AbuseFilter/105. Dragons flight (talk) 05:58, 26 March 2009 (UTC)
- I don't think there's a reasonable way for the filter to distinguish a title of "TBA" from a release date of "TBA". Release date was actually what I was thinking of as the occasional legitimate use of TBA in a discography.—Kww(talk) 04:03, 26 March 2009 (UTC)
- I've seen discography sections that were named "Albums" or "Studio albums" in the past but I'm struggling to think of any other titles. Under what circumstances would 'TBA' be a legitimate entry into an artist's discography? I certainly can think of circumstances where listing the *release date* as 'TBA' would be fine (e.g. the Chinese Democracy saga)... --Kurt Shaped Box (talk) 03:58, 26 March 2009 (UTC)
Muhammad images
- Task: Disallow removal/change of depictions of Muhammad on the Muhammad article.
- Reason: On-going removal against consensus, see Depictions_of_muhammad#Wikipedia_article. The filter should disallow and warn any editor which removes or changes any of the 5 depictions:
- Image:Siyer-i_Nebi 151b.jpg
- Image:Siyer-i_Nebi 298a.jpg
- Image:Maome.jpg
- Image:Mohammed kaaba 1315.jpg
- Image:Gagarin PropovedMagometGRM.jpg
A note will also be added to the top of Talk:Muhammad on how to change the filter, after there is a consensus to do so. Or the filter should not apply to admins so users can request the change using Template:Editprotect.--Otterathome (talk) 17:11, 3 April 2009 (UTC) - Otterathome (talk) 17:09, 3 April 2009 (UTC)
- If we choose to make a filter that disallows removal of images that have been prone to censorship, then File:Mormon garments.jpg and File:Garment.jpg from Temple garment and Image:Mormon woman's temple clothing circa 1870s.jpeg from Endowment ceremony should get the same treatment. Not sure this should go over "warn", though.—Kww(talk) 18:21, 3 April 2009 (UTC)
- Testing Special:AbuseFilter/131. Dragons flight (talk) 21:35, 3 April 2009 (UTC)
- Do we really want to set the filter to disallow? Consensus can change, after all, and this might give the wrong impression. --Conti|✉ 22:25, 3 April 2009 (UTC)\
- I don't think consensus will ever change to allow censorship.— Dædαlus Contribs 21:59, 14 April 2009 (UTC)
- Consensus may change on which images to include. Can't you just have the article semi-protected instead? --Apoc2400 (talk) 18:55, 15 April 2009 (UTC)
- If consensus changes, it is trivial for an admin to make a change based on that consensus. The article is already semi-protected or full-protected. Semi-protection can easily be gotten around (I have seen several instances of new accounts making and reverting 10 edits to some random talk page to get "established" for the purpose of such deletion vandalism). Selectively protecting a small piece of the article so that the rest may be unprotected is desirable, because it allows more people to edit the content. ~Amatulić (talk) 00:36, 24 April 2009 (UTC)
- Consensus may change on which images to include. Can't you just have the article semi-protected instead? --Apoc2400 (talk) 18:55, 15 April 2009 (UTC)
- I don't think consensus will ever change to allow censorship.— Dædαlus Contribs 21:59, 14 April 2009 (UTC)
- Do we really want to set the filter to disallow? Consensus can change, after all, and this might give the wrong impression. --Conti|✉ 22:25, 3 April 2009 (UTC)\
- Testing Special:AbuseFilter/131. Dragons flight (talk) 21:35, 3 April 2009 (UTC)
Non-notable name addition filter
- Task: To block the addition of 2 names by a chronic IP vandal. For the article Name-dropping, to block multiple IP ranges from adding 2 non-notable names.
- Reason: Per This ANI discussion. These IP ranges have been doing this for about 15 months, everytime page protection ends, the IP's continue this vandalism. I think a filter should stop them in their tracks.
- Momusufan (talk) 14:32, 20 April 2009 (UTC)
Vandalism on Pioneer Courthouse Square article
- Task: To prevent edits such as this by sockpuppets.
- Reason: As an alternative to protection. All other articles with full protection due to vandalism have had the protection reduced or removed, and replaced by the abuse filter (specifically filters such as 47 and 54, and possibly 4 and 137).
- —Snigbrook 20:39, 25 April 2009 (UTC)
- What specific edits do you have in mind to add to the filter? Generally, abuse filters are used to detect vandalism across a broad class of articles, rather than just one or two. Stifle (talk) 08:51, 5 May 2009 (UTC)
- See Special:Contributions/Meanskeeps and Special:Contributions/Samoboow - also it wouldn't be the only filter for a specific article - there are a few I've mentioned for a similar purpose, but also filter 153 (where the alternative is semi-protection, which wouldn't prevent many constructive edits), and several that are probably to prevent spam. —Snigbrook 02:39, 10 May 2009 (UTC)
- Done Special:AbuseFilter/175. Just noticed this request and have independently filled it. —EncMstr (talk) 23:36, 27 May 2009 (UTC)
Denied requests
Sad Tragedy
- Task:Filter edits beginning with the phrase "My dear friends, it is with great regret that I must announce a sad tragedy", revert and block the account.
- Reason: The Runtshit vandal has made scores, if not hundreds, of abusive edits starting with this phrase. - RolandR 22:34, 1 June 2009 (UTC)
- Note: The abuse filter cannot revert or block. Stifle (talk) 10:36, 2 June 2009 (UTC)
- I thought that the abuse filter could prevent/block.65.51.38.194 (talk) 21:53, 9 July 2009 (UTC)
- It can prevent, but not revert.Smallman12q (talk) 01:56, 10 August 2009 (UTC)
- Not done Filter 17 covers this. -- zzuuzz (talk) 16:31, 1 September 2009 (UTC)
Jimmywales.org
- Task: To block the addition of this website in any form to pages especially to User:Jimbo Wales
- Reason: IP vandals have added this website lately. See here and here.
- Momo san Gespräch 03:44, 7 May 2009 (UTC)
- Declined, please use User:XLinkBot or the spam blacklist for this. Stifle (talk) 13:58, 7 May 2009 (UTC)
Willy on Wheels
- Task: I was able to add "on wheels!" to an article just now; surely I shouldn't be? Could the strings "on wheels!", "on Wheels" and "willy on wheels" be filtered?
- Reason: To stop anyone making "wheely" edits in the article space.
- The helpful vandal aka 86.138.73.150 (talk) 10:37, 6 May 2009 (UTC) :-)
- WOW seems to be gone at this stage and I can think of several valid reasons for adding "on wheels" to articles. Stifle (talk) 11:54, 6 May 2009 (UTC)
- My suggestions weren't just "on wheels" but slightly more specific. If nobody does WOW vandalism anymore though, I do apologise for wasting time here! 86.138.73.150 (talk) 12:24, 6 May 2009 (UTC)
- Not done Most of this is covered by existing filters and MediaWiki:Titleblacklist. -- zzuuzz (talk) 16:31, 1 September 2009 (UTC)
- My suggestions weren't just "on wheels" but slightly more specific. If nobody does WOW vandalism anymore though, I do apologise for wasting time here! 86.138.73.150 (talk) 12:24, 6 May 2009 (UTC)
Creating redirects to non-existent pages
- Task: To prevent users from creating redirects to non-existent pages. It will also have a positive side effect of preventing abuse such as this.
- Reason: Prevents both abuse and mistakes.
- -- IRP ☎ 22:11, 27 April 2009 (UTC)
- Oppose. I have occasionally done this due to the limitations of AWB. For example, to create redirects from SR x ([state]) to [state] State Route x, I might create new pages with #REDIRECT [[[state] State Route {{PAGENAME}}]] and then on a second pass remove the "SR" and "([state])". This is necesary because AWB doesn't pre-parse PAGENAME. (Yes, I could do any number of tricks to make sure the link is blue, but who's to say that those won't be classified as "abuse" too?) --NE2 04:09, 28 April 2009 (UTC)
- But you're autoconfirmed. We could at least start with non-autoconfirmed, eh? They aren't allowed to use WP:AWB anyway. Wknight94 talk 14:53, 1 May 2009 (UTC)
- Ugh. Spamming extra actions because an automated tool is too simplistic to do what you need it to do is not a good way of doing things. Gurch (talk) 20:49, 10 May 2009 (UTC)
- I don't think it's possible to detect in AF whether some page exists or not. —AlexSm 15:20, 2 May 2009 (UTC)
- It's not possible in the general case, but it is possible in this case. See Special:AbuseFilter/163. I think I have personally seen enough of this nonsense to justify filtering the not autoconfirmed editors, but for now running as a log. Dragons flight (talk) 18:18, 2 May 2009 (UTC)
- I wouldn't take this any further than warn as there are some constructive uses (e.g. a new user redirecting his user page to his talk page, which isn't specifically wrong). Stifle (talk) 08:50, 5 May 2009 (UTC)
- I'll sometimes do this deliberately, in adding a page with a group of alternative names, or in fixing a group of related pages. DGG (talk) 06:17, 7 June 2009 (UTC)
Abuse pattern on User:Jimbo Wales
- Task: Prevent abuse such as these edits to User:Jimbo Wales
- Reason: To prevent a continuous abuse pattern.
-- IRP ☎ 01:34, 21 April 2009 (UTC)
- Is it always the same kind of vandalism? --Conti|✉ 14:07, 21 April 2009 (UTC)
- The same "Jimmywales.org" and "Jimmywales.net" spam happening often. -- IRP ☎ 20:02, 21 April 2009 (UTC)
- It would seem more appropriate to semi-protect the page. Stifle (talk) 15:44, 27 April 2009 (UTC)
- The same "Jimmywales.org" and "Jimmywales.net" spam happening often. -- IRP ☎ 20:02, 21 April 2009 (UTC)
- Is it always the same kind of vandalism? --Conti|✉ 14:07, 21 April 2009 (UTC)
Jimbo refuses for his page to be protected.--Ipatrol (talk) 02:22, 30 May 2009 (UTC)
- ...because legions of editors and admins watchlist his userpage and revert everything quickly. Plus we don't need the abuse filter heavy artillery for a single page.--chaser (talk) 02:50, 2 June 2009 (UTC)
New User Imposed Restrictions
- Task: To impose certain restrictions on users who aren't autoconfirmed (or anonymous users who aren't in the ip_block exempt). These restrictions would be regarding tagging articles for afd, prod, and more. New users shouldn't be prodding/deprodding(to be fair) articles.
- Reason: Users who aren't autoconfirmed often are unaware of what the qualifications for a prod/afd.
Something like(this isn't complete):
(article_namespace == 0)
& !("autoconfirmed" in user_groups)
& !("ipblock-exempt" in user_groups)
& !("sysop" in user_groups)
& (action == "edit")
& ((lcase(removed_lines) rlike "{{(prod|prod-2|Huggle/prod)}}"
| (lcase(added_lines) rlike "{{(prod|prod-2|Huggle/prod)}}")
Also, is there one for preventing people from prodding pages they created(that's not really what prod was meant for)? - Smallman12q (talk) 01:44, 9 August 2009 (UTC)
- Not done I don't think we have a policy preventing IP's from PRODding articles. I also don't see enough examples of people PRODding their own pages to warrant a filter. We currently have in place filters for new users removing AfD and speedy tags. -- King of ♥ ♦ ♣ ♠ 19:15, 12 August 2009 (UTC)
Persistent re-adding of image that fails WP:NFCC to articles
- Task: Revert edits by IPs adding the file File:Beeny.jpg to articles.
- Reason: As soon as semi protection expires on articles, an IP on a very wide range (BT Broadband UK) adds them back in. This has resulted in articles having to be permanently semi-ed.
- Black Kite 19:01, 1 July 2009 (UTC)
- Sounds like a job for MediaWiki:Bad image list. Stifle (talk) 15:54, 3 July 2009 (UTC)
- Good point, I didn't realise that the bad image list was that flexible. Done. Black Kite 17:01, 3 July 2009 (UTC)
Page created with a deletion tag
- Task: Detect a page created with a deletion tag and warn the user
- Reason: This is an improper use of Delete tags and users should be warned as such. Pages created with delete tags are often reposts of previously deleted pages
- Triplestop (talk) 20:47, 4 June 2009 (UTC)
- No, it's a good faith mistake by new users, and we shouldn't "hold their feet to the fire" and scare them off even more with an abuse filter. MuZemike 15:57, 5 June 2009 (UTC) comment copied from ANI repost of this
- I'll disagree there. It's nearly universally a sign of abuse: someone copying an article immediately before deletion and pasting it to recreate it (or, cutting and pasting an unattributed copy from NationMaster or a similar Wikipedia mirror, because they tend to cache the last copy before deletion), ignoring the message that said that they were recreating a deleted article, and restoring a completely inappropriate copy. I will say that warning the user is probably useless, but logging and tagging it is probably worthwhile. Actually, warning the user is probably counterproductive: if the goal is to immediately catch and re-delete the inappropriate recreation, then warning the creator would do nothing but prompt him to remove the tag and help him to evade detection.—Kww(talk) 01:02, 6 June 2009 (UTC)
- Anything copied from a Wikipedia mirror is unlikely to have a deletion template as the source would probably be HTML not Wikitext. One likely explanation is that the user was editing the article in an attempt to improve it, but it was deleted; another is that it is a new user who doesn't know about the criteria for speedy deletion, or the deletion process, and disagrees with the deletion. I agree with Kww that an abuse filter warning could be counterproductive; the most appropriate way of dealing with these is probably through the deletion process, not the abuse filter. —Snigbrook 12:46, 6 June 2009 (UTC)
- Logging it is possibly useful, although I don't think it would be needed as the previous deletion would be in the deletion log, and the recreation would put it in the deletion category. —Snigbrook 12:53, 6 June 2009 (UTC)
- I would say definitely tag the article, so that an admin could easily delete on sight any inappropriate article that they just deleted but was recreated. Logging is possibly useful as well so admins can investigate such occurrences. Triplestop (talk) 15:38, 6 June 2009 (UTC)
- I agree with Snigbrook. If a new user is actually letting us know on CAT:CSD that they've recreated a deleted page, wins all around. All the abuse filter would accomplish (if it warned or prevented) is to let the user know how to keep the article less noticed. Someguy1221 (talk) 19:36, 9 June 2009 (UTC)
- I would say definitely tag the article, so that an admin could easily delete on sight any inappropriate article that they just deleted but was recreated. Logging is possibly useful as well so admins can investigate such occurrences. Triplestop (talk) 15:38, 6 June 2009 (UTC)
- I'll disagree there. It's nearly universally a sign of abuse: someone copying an article immediately before deletion and pasting it to recreate it (or, cutting and pasting an unattributed copy from NationMaster or a similar Wikipedia mirror, because they tend to cache the last copy before deletion), ignoring the message that said that they were recreating a deleted article, and restoring a completely inappropriate copy. I will say that warning the user is probably useless, but logging and tagging it is probably worthwhile. Actually, warning the user is probably counterproductive: if the goal is to immediately catch and re-delete the inappropriate recreation, then warning the creator would do nothing but prompt him to remove the tag and help him to evade detection.—Kww(talk) 01:02, 6 June 2009 (UTC)
- Not done per Snigbrook and Someguy1221. In any case, an article could reasonably be created with {{db-unpatrolled}}. Stifle (talk) 15:08, 19 June 2009 (UTC)
I will sue
- Task: Catch occurances of the phrase "i will sue" (for various appropriate permutaions of capitalisation). It should apply to user talk pages (and talk pages, maybe), and to IP editors.
- Reason: Per WP:ANI#Legal threat., various IPs seem (ahem!) unsure as to what extent free speech is guarenteed by the US constitution. This filter would help prevent legal threats in general, and this annoying free speech troll in particular.
- This flag once was redpropagandadeeds 19:43, 1 June 2009 (UTC)
- Wikipedia policy does not prohibit legal threats or legal action, it simply asks that one does not edit Wikipedia whilst a legal action is planned or ongoing. –xenotalk 19:51, 1 June 2009 (UTC)
- Could a filter maybe remind users of that? Or alert admins? I'm not necessarily looking for legitimate users to be prevented from posting with that phrase, just for a way to manage it. Cheers, This flag once was redpropagandadeeds 20:02, 1 June 2009 (UTC)
- Wikipedia policy does not prohibit legal threats or legal action, it simply asks that one does not edit Wikipedia whilst a legal action is planned or ongoing. –xenotalk 19:51, 1 June 2009 (UTC)
- Given that every filter adds processing time to every page safe (not much, but they add up), I think it'd be a misuse of this resource to filter every page for a single troll.--chaser (talk) 22:11, 1 June 2009 (UTC)
- Resolved– Overkill for one troll. This flag once was redpropagandadeeds 17:51, 5 June 2009 (UTC)
Invalid languages
- Task: To prevent the addition of nonexistant/invalid languages
- Reason: There is no reason for people to add invalid languages other than through vandalism such as in this diff.Smallman12q (talk) 15:37, 31 May 2009 (UTC)
- Is there anything to suggest this isn't just a random instance of vandalism? There's hundreds of possible valid languages and no way to tell the difference between an invalid language code and a valid page name that just happens to contain a colon somewhere. Mr.Z-man 17:03, 31 May 2009 (UTC)
- According to Wikipedia:Language_link, each language has a two letter code followed by a colon. I don't think there will be such a high false positive rate.Smallman12q (talk) 21:30, 31 May 2009 (UTC)
- That's not true though. Some languages have 3 letter codes, there's "simple" for the simple English WP, and others have longer, hyphenated ones like zh-classical:. In any case, there are more than 200 pages (and 57 non-redirects) in mainspace that consist of 2 letters, a colon, followed by anything. You couldn't just disallow anything that matched that, you'd have to have a list of every possible valid language code so people could still add real interwikis. Mr.Z-man 20:44, 1 June 2009 (UTC)
- I've looked it over again, and you are right. There are a number of languages that exceed the two letters, and this would indeed create a high false positive rate. I withdraw the request. Smallman12q (talk) 01:14, 2 June 2009 (UTC)
- That's not true though. Some languages have 3 letter codes, there's "simple" for the simple English WP, and others have longer, hyphenated ones like zh-classical:. In any case, there are more than 200 pages (and 57 non-redirects) in mainspace that consist of 2 letters, a colon, followed by anything. You couldn't just disallow anything that matched that, you'd have to have a list of every possible valid language code so people could still add real interwikis. Mr.Z-man 20:44, 1 June 2009 (UTC)
- According to Wikipedia:Language_link, each language has a two letter code followed by a colon. I don't think there will be such a high false positive rate.Smallman12q (talk) 21:30, 31 May 2009 (UTC)
Filter name
- Task: If 'thespoof.com' is added, give user a warning that it is a parody site and does not comply with with WP:EL/WP:RELIABLE/WP:BLP
- Reason: Sometimes gets added as a source/external link in good faith.
- Otterathome (talk) 15:23, 31 May 2009 (UTC)
- Seems like a job for User:XLinkBot. Stifle (talk) 17:23, 6 June 2009 (UTC)
Catherine Crier
- Task: Prevent recent autoconfirmed users from vandalizing Catherine Crier
- Reason: High profile, underwatched BLP with a specific vandal targeting with highly defamatory, false info.
- Hipocrite (talk) 08:36, 30 May 2009 (UTC)
An abuse filter is too big a hammer just to defend a single article. It would be far more preferable to fully protect the article and require use of {{editprotected}} from the talk page. —EncMstr (talk) 08:44, 30 May 2009 (UTC)
- Seems Rootology agrees, and has fully protected it. —EncMstr (talk) 08:45, 30 May 2009 (UTC)
- Wasn't sure how the technicals of the system worked. I agree that if it has to check every edit to see if it's to the specific page, that's too much of a performance hit vs the inconvenience of a fully protected forever article. Thanks! Hipocrite (talk) 08:54, 30 May 2009 (UTC)
- Not done, as per the above logic. Stifle (talk) 10:37, 2 June 2009 (UTC)
NIGGER vandalism
- Task: Prevent anon editors from inserting the word "NIGGER" into articles or talk pages.
- Reason: Although the word "nigger" can be used in constructive edits pertaining to the actual article and related ones, I've seen virtually no constructive IP edit that puts "NIGGER" in all caps. This is almost always used in a defaming fashion.
- Until It Sleeps 22:37, 23 May 2009 (UTC)
- Is there a filter to which we can simply add this?--chaser (talk) 02:29, 2 June 2009 (UTC)
- Not done, already a part of Filter 9. MacMedtalkstalk 04:32, 7 August 2009 (UTC)
Retired Tag Checker
- Task: Prevents regular users(eg, it doesn't apply to admins) from placing {{retired}} tags on userpages which are not their own.
- Reason: This is not something that a user should be doing, and in fact, many sockpuppets/stalkers/ and vandals do this sort of thing to users they don't like/etc.- — Dædαlus Contribs 20:48, 23 May 2009 (UTC)
- The following code may help.
(article_namespace == 3)
& !("sysop" in user_groups)
& (length(added_lines) > 7)
& (lcase(added_lines) rlike "{{retired}}")
Not sure if it works as I'm not an abuse filter sysop, but it may help=D.Smallman12q (talk) 01:00, 1 June 2009 (UTC)
- Is this happening enough to justify the extra overhead? Stifle (talk) 15:29, 19 June 2009 (UTC)
- We get spurs of activity in this area caused by prolific sockpuppeteers.— Dædαlus Contribs 19:05, 19 June 2009 (UTC)
- I don't think userpage vandalism, particularly oddities like this, is enough to justify an abuse filter. Plus, this would inevitably generate false-positives for non-admins properly adding the tags to userpages of de-facto retired editors.--chaser (talk) 19:50, 19 June 2009 (UTC)
Keep templatespace clean of user templates
- Task: To warn users when they try to create a page with the prefix "Template:User:", suggesting that they probably want to create it in their userspace rather than templatespace.
- Reason: To prevent us from having to userfy these erroneous creations. These are often created when create and click a link like {{User:Xeno/userbox}} (
{{tl|User:Xeno/userbox}}
)... It happens alot when people work backwards when creating userboxen (diffs on request ;>). –xeno talk 20:42, 12 May 2009 (UTC)
Related link: new templates. The code might also check for '{{userbox' in added_lines
— AlexSm 17:42, 14 May 2009 (UTC)
- How frequently is this happening? Each abuse filter adds a delay to every edit, so unnecessary ones are avoided whenever possible. Stifle (talk) 09:21, 18 May 2009 (UTC)
- It's not that frequent, so I'm going to self-reject this one. –xeno talk 04:48, 22 May 2009 (UTC)
Amazon.com affiliate links
- Task: Prevent addition of affiliate spamlinks into product URLs
- Reason: There are many valid examples of having product links in an article. My example, and the prompting of this request, is the Amazon Kindle page, which includes a link to the actual product on Amazon.com. Those links are in the format of http://www.amazon.com/gp/product/B0015TCML0. An addition of the keyword &tag indicates an affiliate program, which is essentially spam. Anyone who adds an afflink to these articles will reap monetary benefits of clicking through a (theoretically) non-profit website such as Wikipedia. See my revert, [2], for an example of removing an afflink. There should never be an afflink on any pages which link to amazon.com. Yngvarr (t) (c) 13:16, 7 May 2009 (UTC)
- Just to confirm: we're looking for a link to any amazon site with the text ?tag= or &tag= somewhere in it? Stifle (talk) 13:57, 7 May 2009 (UTC)
- That should pretty much cover it. Any link with a "tag=" is an affiliate. Thanks Yngvarr (t) (c) 15:50, 7 May 2009 (UTC)
- This would be more appropriate on the spam blacklist. I notice that "\bamazon\.com/.*&tag=" is already on there, it now just needs to catch the first parameter in the querystring by adding "?". -- zzuuzz (talk) 20:29, 9 May 2009 (UTC)
- This has now been added to the meta spam blacklist. -- zzuuzz (talk) 10:17, 11 May 2009 (UTC)
Anon edit summaries
- Task: Warn anons who don't provide an edit summary.
- Reason: Have a look at the history of a page like endocrine system. It doesn't appear to have many established users keeping an eye on it like plenty of other articles do. Many of the anon edits are constructive, but plenty are not. Sometimes these are spotted and reverted, but sometimes there are intervening edits and no one then notices for a long time, if at all. If nonsense is left in these article someone spots it eventually and takes it out, but no one may ever notice the absence of good material that has been lost. Very few anons leave edit summaries, and so it becomes difficult when glancing at the anon-filled history to tell who's who and what they might have done. I believe logged-in users can set their preferences to prompt them to enter summaries if they neglect to, so maybe this is not something for the abuse filter but can be turned on in the software elsewhere. I appreciate that edit summaries are not compulsory, but they certainly are good practice and to be strongly encouraged. The help page about them even says "It's a good idea to set your user preferences (under Editing) to 'Prompt me when entering a blank edit summary'." Why not essentially make this the default for anons? If it annoys them then it's another reason to sign up for an account, and memorable usernames make scanning page histories much easier. If they really don't want to enter a summary they don't have to. Some might enter fairly useless summaries like "editing article", but this still makes it easier to pick out those who don't bother or who are only interested in vandalism and type in gibberish because they've been told they need to say something. It's a lot easier to spot potentially problematic edits if there's something to differentiate them from everything else from a glance at the page history.
- 79.64.170.147 (talk) 01:51, 7 May 2009 (UTC)
- I could think of several non-anons who don't use edit summaries 99%+ of the time. Also, edit summaries are not a good way of determining if an edit is vandalism or not. x42bn6 Talk Mess 19:29, 7 May 2009 (UTC)
- This would make it harder to spot some vandalism, as the automatic summaries wouldn't be used. Mr.Z-man 05:43, 8 May 2009 (UTC)
- The abuse filter is designed to deal with abuse, not with behaviours that are mildly tiresome. Stifle (talk) 20:03, 9 May 2009 (UTC)
Overlinking related edits
- Task: Prevent against edits which consist of only wikilinking a substantial amount (i.e. upwards of eight words) of common words and terms seemingly at random.
- Reason: Some IP editors continuously overlink as the bulk of their contributions which usually leads to several warnings, this filter should reduce the amount of editor time taken up reverting overlink violations. I'm aware that it's likely to get a lot of false positives so a logging only period would come first before going live.
- treelo radda 22:45, 29 April 2009 (UTC)
- This would seem to require a definition of "common words". Or are you proposing to prevent all edits that link more than eight words? Gurch (talk) 13:35, 30 April 2009 (UTC)
- The latter. treelo radda 14:44, 30 April 2009 (UTC)
- The filter is here (log), but I presume it's going to be hard to weed out the false positives. —Admiral Norton (talk) 15:45, 30 April 2009 (UTC)
- That's what I figure especially regarding edits which are expansions of articles, I'm looking for this filter to pick out edits which wikilink several words only, not those that add content to lessen the chance of false positives. Those which do add content and also overlink in addition would need to be checked over as it starts getting a bit more grey by that point and probably beyond a filter to do effectively. treelo radda 16:14, 30 April 2009 (UTC)
- I've added a restriction to target only edits that increase the article size by greater than 0 and less than 80 bytes. If someone is just adding links, the edit size diff would seem very likely to fall in that range. Dragons flight (talk) 17:52, 30 April 2009 (UTC)
- That's what I figure especially regarding edits which are expansions of articles, I'm looking for this filter to pick out edits which wikilink several words only, not those that add content to lessen the chance of false positives. Those which do add content and also overlink in addition would need to be checked over as it starts getting a bit more grey by that point and probably beyond a filter to do effectively. treelo radda 16:14, 30 April 2009 (UTC)
- The filter is here (log), but I presume it's going to be hard to weed out the false positives. —Admiral Norton (talk) 15:45, 30 April 2009 (UTC)
- The latter. treelo radda 14:44, 30 April 2009 (UTC)
(undent) How long is the usual period for logging before letting a filter loose? From the logs I'm seeing a few false positives but mostly overlinking examples even if not for common words. treelo radda 19:22, 2 May 2009 (UTC)
- Wait, you're seriously treating any edit that adds more than eight links as abuse? WTF? Gurch (talk) 20:54, 4 May 2009 (UTC)
- Suggestions for improving it? treelo radda 21:18, 4 May 2009 (UTC)
- Oh, only under 80 bytes. Didn't see that at first. Still, it seems Dragons flight had to add that for you, so apparently you still think any edit that adds more than eight links is abusive. It's almost like you're trying to stop people writing good articles... Gurch (talk) 21:23, 4 May 2009 (UTC)
- Whatever you heard, it's all lies. Still, an oversight is an oversight there so if you wish to trout me for my nefarious ways, go ahead. treelo radda 21:49, 4 May 2009 (UTC)
- This works, but I don't see how it's especially productive. And as Gurch says, this does not necessarily qualify as abuse. We're not in the business of harassing editors with abuse filters, I hope. Stifle (talk) 08:48, 5 May 2009 (UTC)
- No, we're not and I know I'm not but what do you do to prevent those who specifically overlink nearly every word in one paragraph or whole sections without preventing those editors who do want to build and also need to make a lot of links? Yeah, we only give out notices to those who do overlink but there is an abuse issue with the edits I mean for this filter to target even though right now it catches fully legitimate edits as well. So, improve or dump? treelo radda 09:57, 5 May 2009 (UTC)
- I looked through the last twenty or so hits and none of them seem to be what the filter was meant to catch. A few of those twenty could be described as overlinking, but not the disruptive sort of linking every other word regardless of value. Someguy1221 (talk) 10:07, 5 May 2009 (UTC)
- Needs tweaking then. treelo radda 10:57, 5 May 2009 (UTC)
- Bear in mind that contributors who don't understand how to do links often write articles without them; someone then has to go and add them all in. Gurch (talk) 22:18, 5 May 2009 (UTC)
- Also, if someone vandalises an article by removing all the links or replacing the text with nonsense of roughly the same size, this filter will call reverting that vandalism 'overlinking', e.g. this. Gurch (talk) 22:20, 5 May 2009 (UTC)
- Needs tweaking then. treelo radda 10:57, 5 May 2009 (UTC)
- I looked through the last twenty or so hits and none of them seem to be what the filter was meant to catch. A few of those twenty could be described as overlinking, but not the disruptive sort of linking every other word regardless of value. Someguy1221 (talk) 10:07, 5 May 2009 (UTC)
- No, we're not and I know I'm not but what do you do to prevent those who specifically overlink nearly every word in one paragraph or whole sections without preventing those editors who do want to build and also need to make a lot of links? Yeah, we only give out notices to those who do overlink but there is an abuse issue with the edits I mean for this filter to target even though right now it catches fully legitimate edits as well. So, improve or dump? treelo radda 09:57, 5 May 2009 (UTC)
- Oh, only under 80 bytes. Didn't see that at first. Still, it seems Dragons flight had to add that for you, so apparently you still think any edit that adds more than eight links is abusive. It's almost like you're trying to stop people writing good articles... Gurch (talk) 21:23, 4 May 2009 (UTC)
- I checked about 10 examples from the log; all but one of them seemed a valid edit,k and that one was justa n error, not abuse. Just the opposite of what treelo seems to have found. Myself, when i wikify an article I do all of them I can find at a time, which is generally about a dozen, more if its an article with many names. Which ones are though to be vandalism or even improper? I would strongly oppose activating this filter in any form. DGG (talk) 06:56, 6 May 2009 (UTC)
- Suggestions for improving it? treelo radda 21:18, 4 May 2009 (UTC)
- I more or less concur with the above comments. The false positive rate appears to be very high. There are some cases of inappropriate overlinking in there, and a couple cases of outright vandalism, but a majority portion of the edits appear legitimate. Given that, I don't see it as reasonable to warn or prohibit using this filter. Dragons flight (talk) 07:26, 6 May 2009 (UTC)
- Yeah, sorry. So, consider it declined then? treelo radda 07:37, 6 May 2009 (UTC)
Talk Page Move
- Task: To detect when an active (and probably lengthy) talk page is wrongly moved into userspace.
- Reason: Prevent vandals from causing confusion when there is suddenly no talk page=P.
- Smallman12q (talk) 00:40, 17 April 2009 (UTC)
- How often does this happen? Aside from legit userfications, I don't think I've ever seen anything like this. Mr.Z-man 00:51, 17 April 2009 (UTC)
- I've only seen it happen once, but I don't see why it couldn't happen in the future.Smallman12q (talk) 21:55, 17 April 2009 (UTC)
- The problem with allowing "only seen it happen once, it might happen again" filters is that you end up with hundreds of them, and editing slows to a crawl (there's already a delay in editing pages now due to these filters, it's noticeable if you're trying to edit quickly, e.g. to deal with vandalism). Gurch (talk) 17:55, 18 April 2009 (UTC)
- There is also the problem that temporary versions are sometimes prepared in spaces like Talk:Foo/draft, and moving those can be legitimate. Dragons flight (talk) 19:03, 2 May 2009 (UTC)
- The problem with allowing "only seen it happen once, it might happen again" filters is that you end up with hundreds of them, and editing slows to a crawl (there's already a delay in editing pages now due to these filters, it's noticeable if you're trying to edit quickly, e.g. to deal with vandalism). Gurch (talk) 17:55, 18 April 2009 (UTC)
- I've only seen it happen once, but I don't see why it couldn't happen in the future.Smallman12q (talk) 21:55, 17 April 2009 (UTC)
History Edit link
- Task: To detect when a new user modifies someone elses "edit history link" (I am aware there is a template, but no one uses it).
- Reason: Vandals use this to cause confusion in debates.Smallman12q (talk) 00:34, 17 April 2009 (UTC)
- Smallman12q (talk) 00:34, 17 April 2009 (UTC)
- Presumably this is something you saw once which might concievably happen again? See above. Gurch (talk) 17:56, 18 April 2009 (UTC)
- Well, yes. Perhaps if it should happen a lot a year down the road, I'll be able to say that I did suggest a solution=P.Smallman12q (talk) 00:45, 20 April 2009 (UTC)
- I'm not even sure what you are refering to by "edit history link"? Dragons flight (talk) 01:11, 23 April 2009 (UTC)
- I think it has something to do with the "contribs" links in {{user}} and {{vandal}} and the "history" link in the {{ln}} family. --Damian Yerrick (talk | stalk) 20:11, 3 May 2009 (UTC)
Category, Template, and language breaker
- Task: To detect when the "{}" or "[]" are removed from categories and templates breaking them. Also to check to detect when people change the word category or take out the ":" or "|" from templates and categories. Perhaps also to check invalid languages.
- Reason: I haven't seen it much, but it doesn't hurt to propose. It's to prevent people from breaking templates, categories, and languages. Also to detect invalid languages.
- Smallman12q (talk) 22:21, 17 April 2009 (UTC)
- Not rejecting it outright, but if you haven't seen it much, the server overhead might not be worthwhile. Stifle (talk) 10:41, 20 April 2009 (UTC)
- Aside from the question of whether it is useful, I don't see any reasonable way to implement this given the complexity of valid template code, etc. Dragons flight (talk) 00:11, 23 April 2009 (UTC)
- How about the removal or languages and categories? (ignore the templates)
- there's fairly frequent need to remove languages when articles elsewhere get deleted. , it might be worth having a list of those to check the articles here against DGG ( talk ) 20:56, 18 August 2009 (UTC)
Cut and paste move filter
- Task: To prevent cut and paste moves
- Reason: Many new users move pages using the cut and paste method instead of the correct method
-- IRP ☎ 22:32, 17 April 2009 (UTC)
- This seems unlikely. Cut and paste moves are a combination of two different edits on two different articles, and as the filter works per edit, not per series of edits, there is unlikely to be any reliable way to detect them. -- zzuuzz (talk) 22:41, 17 April 2009 (UTC)
- Concur, not possible. Dragons flight (talk) 00:08, 23 April 2009 (UTC)
- Would it be feasible to filter for the really bad cut-n-paste moves that were copied from the article as viewed instead of the source? That would be, warn users trying to make edits containing the text [edit] Someguy1221 (talk) 00:16, 23 April 2009 (UTC)
Yes, for example adding /From Wikipedia, the free encyclopedia/. Rich Farmbrough, 00:14, 30 June 2009 (UTC).
Signiture Modification Detector
- Task: To detect when new users remove other people signatures or edit them.
- Reason: I have seen occassional vandals who edit and mix up people's signatures so as to create confusion in a consensus debate. Ideally, it would check to see if the user is editing established users signatures such as admins.
- Smallman12q (talk) 00:31, 17 April 2009 (UTC)
- This seems like it would be rather difficult and error-prone. Also, there's no way to get the user groups of anyone but the person making the edit. Mr.Z-man 00:50, 17 April 2009 (UTC)
- Not practical. Dragons flight (talk) 00:49, 23 April 2009 (UTC)
- and too rare. DGG ( talk ) 20:54, 18 August 2009 (UTC)
Protection Circumvention Prevention
- Task: Check recently auto-confirmed accounts contributions to semi-protected articles which barely have any edits, or mostly edits to user talk pages, and block them, or at least prevent them from making such edits.
- Reason: New accounts are going to user talk pages and creating a character, then deleting and creating it several times in succession to become auto-confirmed, and then using this status to circumvent the semi-protection of pages. The current, basic consensus, is to Block such accounts on sight.
- — Dædαlus Contribs 21:52, 1 April 2009 (UTC)
- No, it isn't. The current consensus is to block abusive accounts that make trivial edits to be more quickly able to edit semi-protected pages. If the edits they make to the semi-protected pages are constructive, then what exactly is the problem? -- Gurch (talk) 17:30, 17 April 2009 (UTC)
- If you do this they will just start doing the same thing on other pages instead.--Otterathome (talk) 14:11, 2 April 2009 (UTC)
- Then the obvious answer is to make it search for the pattern in more spaces than the user talk namespace.— Dædαlus Contribs 19:54, 2 April 2009 (UTC)
I'm probably going to deny this request as I think it overlaps too closely with other things we are now trying. Dragons flight (talk) 20:13, 2 April 2009 (UTC)
LEPS
For more information about this, see User:IRP/LEPS. I have created all but the filter. See User:IRP/LEPS for more information. -- IRP ☎ 10:53, 12 April 2009
Besides the reasons given in the box above, this would require doing (multiple) potentially expensive regex searches on every page save to catch things that only in 1 circumstance could really be seen as abuse.
- when a user attempts to disguise an external link as an internal link - This is the only thing that could possibly be abuse, but excluding other Wikimedia sites would make it a rather complicated regex
- an external link that actually links to an internal page, interwiki page, interwiki file, or internal file - this isn't abuse, and there's no harm done by people making links this way. The regex would also have to make sure the link could be done as a normal link as well. Diffs are links to internal pages, but can't be made using internal link syntax.
- an external, interwiki, or internal link to link to an article - People use interwiki links to link to articles here? Also, looking at your example, I still don't understand the difference between an internal link and an "article link" unless you're using a different definition of "internal".
- a user incorrectly links to the history of a page - So we should show big nasty errors to people to force them to use a template that almost no one knows exists and does nothing but make a slightly prettier link?
Not done Mr.Z-man 20:37, 13 April 2009 (UTC)
- So, then go ahead and remove the test edit filter because tests are rarely abuse. That's why I assume your point is not valid. -- IRP ☎ 20:58, 13 April 2009 (UTC)
- I figured that someone who was going to be reviewing my request would at least realize that this filter isn't necessarily used solely for abuse. If you believe the name ("abuse filter") causes confusion, then you can propose that it be changed to either "edit filter" or "undesirable edit filter". By the way, your parody of my request is not funny and makes me think you are being very frivolous when reviewing requests. -- IRP ☎ 21:10, 13 April 2009 (UTC)
- (e/c)Test edits are reverted however as they are unconstructive and they are a lot more common than the edits proposed here, and the test edit filter doesn't require any complicated regular expressions, just simple string searching (it also uses a friendlier message). In any case, the existence of one filter that doesn't strictly meet the rules is not justification to create more. I would also note that there's no consensus that history links must use the {{ph}} template (as evidenced by the fact that it has fewer than 200 transclusions). Also, just saying that I'm wrong isn't going to get me to change my opinion. You'll have to actually adreess the points I raised and explain how A) Such a filter is desirable by the community in general, as you've not even proposed this system anywhere but here and B) How it could be done without being an unnecessary performance drain. (And I didn't make the sarcastic box). Mr.Z-man 21:11, 13 April 2009 (UTC)
- There are three different link types (according to Happy-melon). See this comment. Those link types are there for a reason, and that reason is not for the link types to be used inconsistently, but to allow users to know if they are going to an article, a non-standard page (such as the history of a page), or an external web site. If you wish to complain about the error message, then we could just use a bot to fix the links, however, that would increase server load even more. This will just increase it initially until users learn to correctly create links in the first place and then we wouldn't need anything. If these links are going to continue to be used in a confusing manner (such as the internal links disguised as external links) then we should remove the different link type support from Wikipedia altogether and have all links look the same. -- IRP ☎ 21:27, 13 April 2009 (UTC)
- I think you're missing my point. My point is not that we shouldn't care about link types, but that its not really a big deal if people misuse links. Of all the things that could possibly be caught by the abuse filter, this ranks pretty low on the list of things that we really should catch. If this entire thing is based on Happy-Melon's comment, then you're really taking this way too far.
- There is a difference between the server load from this and from a bot. The abuse filter slows the time it takes for everyone to make an edit, a bot doesn't. A poorly performing filter could even cause edits to fail on large pages. Bots do almost all of their processing on a separate server or computer, so they can have as complex rules as they want.
- Also, most misuses of links are by new users, so the problem isn't going to go away after having such a filter for a couple months. I'm also somewhat confused by your comment about bots and the error message. Are you saying that if we don't use the error message almost exactly as you propose that the system would be worthless? I just don't see how one makes the jump from "complaining about the message" to "using a bot instead." Mr.Z-man 21:43, 13 April 2009 (UTC)
- I was making two points. One was about the bot and the other was about the filter. I think the link types should be used correctly so it isn't made to where the developer who implemented the feature has wasted his or her time on it. It is supposed to make it easier to predict where the link will lead to. Perhaps, propose that external links shouldn't even use the external link icon at all. Instead, a warning message such as "You are about to leave the Wikipedia website" that appears when one clicks on an external link may be used. -- IRP ☎ 22:03, 13 April 2009 (UTC)
- They are used correctly in the vast majority of cases, hence the performance issue with such a filter. It has to do complex regex searches on every edit, but it only triggers on a small minority to enforce a minor style rule. Mr.Z-man 22:47, 13 April 2009 (UTC)
- Conclusion: I propose that a developer should deprecate the plainlinks, as it should be superseded by a feature that should allow the software to determine whether the external link icon should be there. If it contains the "http://en.wiki.x.io" prefix or any other prefix in the interwiki table, it should not have the icon. If it has a prefix that is not in the interwiki table, then it should have the external link icon on it. I think that would be a better alternative to both the bot and the LEPS. I will now {{db-author}} the page User:IRP/LEPS and its subpages. -- IRP ☎ 23:19, 13 April 2009 (UTC)
- Also: If it has the "http://en.wiki.x.io/wiki/" prefix, it should be a bluelink. -- IRP ☎ 23:30, 13 April 2009 (UTC)
- They are used correctly in the vast majority of cases, hence the performance issue with such a filter. It has to do complex regex searches on every edit, but it only triggers on a small minority to enforce a minor style rule. Mr.Z-man 22:47, 13 April 2009 (UTC)
- I was making two points. One was about the bot and the other was about the filter. I think the link types should be used correctly so it isn't made to where the developer who implemented the feature has wasted his or her time on it. It is supposed to make it easier to predict where the link will lead to. Perhaps, propose that external links shouldn't even use the external link icon at all. Instead, a warning message such as "You are about to leave the Wikipedia website" that appears when one clicks on an external link may be used. -- IRP ☎ 22:03, 13 April 2009 (UTC)
That was opened almost two years ago and nothing has been done! I'm going to re-make the proposal at Wikipedia:Village pump (proposals). -- IRP ☎ 20:21, 14 April 2009 (UTC), modified 20:23, 14 April 2009 (UTC)
- Two years is about par for the course with bugzilla requests. There isn't really any other way to get things implemented though, short of coding them yourself. Gurch (talk) 11:24, 24 April 2009 (UTC)
Deletion of AFC submissions
Action = delete
article_namespace = 5
((title contains "Articles for creation")
- or page_in_category ((Pending Afc requests) or (Declined AfC submissions)))
If someone could help me figure out how to implement this, that would be great. Thanks. Someguy1221 (talk) 07:09, 7 April 2009 (UTC)
- Why does this need a filter? Aren't administrators supposed to know what they are doing? Gurch (talk) 08:27, 14 April 2009 (UTC)
- I wish they always did. Someguy1221 (talk) 00:18, 23 April 2009 (UTC)
- Declined. If an admin deletes a submission improperly, tell him not to. Stifle (talk) 19:26, 26 April 2009 (UTC)
- I wish they always did. Someguy1221 (talk) 00:18, 23 April 2009 (UTC)
Silent detection of Twinkle, Huggle, Etc installation
This may already exist, but I think it would be helpful to check for a very young account installing huggle first chance they get, would help supplement other behavior indicators for possible sockpuppetry. Less of a request, more of a musing.--Tznkai (talk) 15:04, 29 September 2009 (UTC)
- Seems a good idea to me. JamesBWatson (talk) 17:11, 14 October 2009 (UTC)
- I think a lot of users use huggle first thing after rollback. Because seeing (HG) in an edit summary is a lot more common then the normal rollback. New users at first wonder what the (HG) text that everybody has on their edit summaries is (I did), and click it. Then they want huggle, so they get rollback and then huggle straight away (since huggle is the reason they got rollback in the first place). So this would be pointless for catching puppets imo. Besides which, it would be very easy to look manually for a certain user. - Kingpin13 (talk) 17:17, 14 October 2009 (UTC)