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.
FP
here (that's the 69 filter); and although
this was proper vandalism, it probably needs to be investigated further because it's possible there will be actual FPs with phrases like "is suspect[ed]" or things like that.
RandomCanadian (
talk /
contribs)
03:42, 20 February 2021 (UTC)
Glad to see this - there seems to have been a spike in this silliness recently. I imagine some TikTokker or YouTuber or other spurred it...
ƒirefly (
t ·
c )
17:52, 23 February 2021 (UTC)
[moved from section below --SoY] This character: ඞ should be added to the filter as it looks like an Among Us character. It is used by some IP vandals. NASCARfan0548↗19:43, 3 March 2021 (UTC)
It looks (if you squint a bit and zoom in / enlarge the character) like one of the Among Us spacemen. I’d say add it to the filter but only when it’s surrounded by word boundaries to limit false positives.
ƒirefly (
t ·
c )
21:34, 3 March 2021 (UTC)
[6]... (repetition did not disallow that? or is that another mainspace-only filter?). @
Suffusion of Yellow: If you can do something about this. The more radical solution would be semi-protecting this talk page, but usual practice is to shy away from that if it's relatively low volume like here so not sure. Cheers,
RandomCanadian (
talk /
contribs)
13:30, 12 April 2021 (UTC)
Add the word "sheesh" to the list of vandal meme phrases. It is often used in vandalism and has very few good users.
aeschylus (
talk)
17:49, 1 June 2021 (UTC)
"eat kack" LTA
Task: All pages, any addition of "kack", which isn't a commonly used word and should have few false positives, should be disallowed by a filter. Perhaps include the ! at the end of kack in the filter.
Reason: A bunch of different IP addresses come along and simply add "eat kack!" with no variation to articles until blocked.
Diffs:
[8], also every contribution by that IP. There's older ones too on other IPs.
@
Suffusion of Yellow: Good luck with whatever secret filter you're doing the work on! Though I enjoy racking up edits by spam-refreshing the contributions page of a fast vandal, I enjoy more knowing vandals are being stopped in their tracks! Is there a LTA page for them or are they
WP:DENY'd? Gatemansgc (
TɅ̊LK)
06:11, 26 May 2021 (UTC)
LTA pages are only made when there is a reason to know them. For example, serious harassment, long-term spam/advertising of a certain person, making certain types of sleepers, good-faith accounts, and other. This is just like any other vandal - we don't need to give them attention, as there is nothing useful to know about them.
aeschylus (
talk)
23:11, 4 June 2021 (UTC)
Disallow large swathe of romanized Hindi in articles
Task: Disallow addition of large swathe of romanized Hindi to article space by an IP hopping editor geolocating to Odisha, India.
Reason: I’ve noticed since February, swathe of text written in what may be romanized Hindi appearing in articles, often BLPs. The text appear to have no meaning and appears to be straight up vandalism. Today’s effort resulted in
this.
Ullu is another that suffered badly earlier this month. Perhaps there’s a few common phrases that can be added to an appropriate filter?
The text appear to have no meaning This is not true. The text is offensive, and extremely so. One of the sentences translates to "I'll rape all the women in your house" and that's like one of the milder parts. All offensive words have been altered with diacritics, presumably to bypass existing filters. –
SD0001 (
talk)
03:23, 26 May 2021 (UTC)
And probably to foil anyone's attempts to simply paste into google translate, so that people will think it's simple vandalism not horrible vandalism? Gatemansgc (
TɅ̊LK)
06:08, 26 May 2021 (UTC)
@
Malcolmxl5, I don't think so, as the Dalit LTA had some specific numbers and places that could be filtered. This user just places highly offensive text with no personal info and floods it into articles. Maybe we could make
Special:AbuseFilter/965 antispoof? The user appears to spoof the profanity in order to let it pass.
aeschylus (
talk)
14:06, 26 May 2021 (UTC)
Merged changes into disallow filter (965) owing to the high volume nature of the vandal and nature of the abuse, and the fact that there's no FPs in the day I ran it in filter 1. Will monitor for any FPs that arise. Some tweaks for performance may be necessary, since the diff delta check was dropped.
ProcrastinatingReader (
talk)
13:20, 27 May 2021 (UTC)
Performance seems fine. Out of interest,
Suffusion of Yellow, the same code as in filter 965 ATM (taking 0.2ms to run) but with a & check below it (to make sure it didn't match the then-existing 965 filter) seemed to take 1.5ms on average to run (at the time I disabled it). Why would this be? Surely this filter matches a minority of actions in the first place, and even on the ones it does match the extra check (at best) should add another 0.2ms run time?
ProcrastinatingReader (
talk)
13:34, 27 May 2021 (UTC)
@
ProcrastinatingReader: Filters are run in order. The first filter to look at any lazy-loaded variables (e.g. user_groups and added_lines), gets "blamed" for the time taken to compute those variables. And function results are cached, so your filter was also blamed for the ccnorm(added_lines) call, too, both in condition count and runtime.
Suffusion of Yellow (
talk)
21:04, 27 May 2021 (UTC)
Or maybe it's similar to
this? That's pretty weird if so, it looks like maybe ccnorm isn't sanitising the Ü (as it seems the filter should be matching that diff if that letter is changed to U)? As an aside, it seems the vandal is technically sophisticated to some degree, and a good chance that they're reading this.
ProcSock (
talk)
02:21, 28 May 2021 (UTC)
They are hopping between IPs and ISPs in the Odisha region, always mobile web edits using the visual editor. I’ve done a lot of rev del in the last day or so and can go back over my logs to compile a comprehensive list of IPs to look for ranges. I did notice a rangeblock in in the 117.x.x.x area (which I can’t find now). --
Malcolmxl5 (
talk)
16:37, 28 May 2021 (UTC)
I'm not sure but it's a good question (one I've wondered before). Also I think Huggle has IRC functionality, and
MusikAnimal has a bot that can listen to certain filters and post changes into an IRC channel, so that could be a possible way too.
ProcrastinatingReader (
talk)
23:01, 29 May 2021 (UTC)
Can we just prevent additions which consist of the same sentence (broadly construed) repeated multiple times, whether in Hindi, English or gibberish?
Certes (
talk)
17:20, 31 May 2021 (UTC)
Friendly reminder this is not the ideal venue to discuss how to stop long-term abuse, as the vandal may very well be watching and will adapt accordingly. I recommend moving this discussion to the
mailing list. — MusikAnimaltalk18:31, 2 June 2021 (UTC)
Large typo edit
Task: Tag edits with "typo" in the edit summary that change >100 bytes. This would most likely have some false positives and may catch good-faith edits so tagging would be appropriate, maybe also exclude auto- or extended-confirmed?
Reason: Large edits that claim to just fix typos are occasionally (but not always) done to avoid getting spotted, and are regardless an inaccurate edit summary.
Seconded, but we should probably limit it to summaries consisting only of "typo", "Fixed typos" and similar. (/^ *([Ff]ix(ed|ing)?)? *[Tt]ypos? *$/ would be my first stab). There's nothing wrong with adding 5000 bytes with "Add history; fix typos", though it might have been better split over two edits. We could also include other canned or common summaries such as "Fixed grammar", "Added content" when the page has shrunk, etc. See also
Special:AbuseFilter/633.
Certes (
talk)
14:14, 7 June 2021 (UTC)
I have noted a series of IP addresses indiscriminately change "Israel" to "Palestine". Is anything possible for that?
aeschylus (
talk)
23:13, 4 June 2021 (UTC)
The problem would be that if we had someone incorrectly changing "Palestine" to "Israel" (it has happened in the past) then the filter would prevent reverting it. If such a filter is to exist it should prevent both switches.
Black Kite (talk)00:19, 5 June 2021 (UTC)
I've been noticing these too over the past weeks while fixing links to disambiguation pages. It seems to be just a bunch of unrelated IPs. A good thing is that since
Palestine is a disambiguation page, new links there will always be noticed. But that doesn't help when links are not involved, of course, and it is indeed worrying that some of the changes wouldn't have been noticed if they hadn't involved a new link to a disambiguation page.
Lennart97 (
talk)
13:40, 5 June 2021 (UTC)
Tag for unexplained numerical changes, with some exceptions
Task: Detect when numbers/figures in an article are slightly changed without explanation (with the exception of sports and TV articles, which seem to change frequently)
Reason: I'm seeing a lot of hard-to-detect vandalism in which numbers in an article are changed. It would be nice to have some sort of flag or tag so these are easier to spot and vet. I should mention that I recently
made an idiot of myself on ANI about this because I thought some numerical updates were vandalism that weren't since people sometimes legitimately update sports scores, numbers of TV episodes aired, and so on.
GiantSnowman, a lot of this kind of stuff happens on your turf, so maybe you have some thoughts. Personally I strongly dislike those unexplained updates (if that's what they are--it's hard to tell unless you're watching the match or whatever), but we're probably talking about thousands of edits per week, if not per day.
Drmies (
talk)
15:46, 20 June 2021 (UTC)
Yeah it's a real chore sorting through unexplained stats updates on soccer players, working out which are AGF (but possibly incomplete i.e. game is updated but not date, or a player scores without playing etc.) and which are vandalism.
GiantSnowman16:43, 20 June 2021 (UTC)
@
Cheerful Squirrel, however, there are good number filters. For example, we can filter changing the birth date to the future or over 150 years ago (100% used in vandalism). You can also filter the numbers 69 and 420 as a tag (possible vandalism) which will allow good faith edits to stay and vandals to get reverted faster. @
GiantSnowman and @
Drmies, any thoughts?
aeschylus (
talk)
18:23, 20 June 2021 (UTC)
You also get edits like
this - what does this mean? Are you saying he has left the current club? If so, has he signed for a new club? or is it just vandalism/test?
GiantSnowman10:50, 21 June 2021 (UTC)
Jake Reeves has just left (or at least arranged to leave) Notts County,
[37] though it would be nice if the article and/or edit summary said so.
Certes (
talk)
11:32, 21 June 2021 (UTC)
More generally, sports articles are unusual in that Wikipedia seems to have become a news feed. There are few other areas where people add statistics expecting to update them weekly – maybe just TV episode lists and ongoing news such as COVID numbers.
Certes (
talk)
11:37, 21 June 2021 (UTC)
I'm aware that there's a warning that pops up when people try to add references in-text to The Daily Mail. I'm wondering if we'd be able to also add one to the use of Press TV, a recently
deprecated source that publishes Holocaust denial, provided that this is something we typically do for deprecated sources. I know that only administrators have the ability to add edit notices, so I am making a request here that a similar filter be put in place for Press TV, which has
673 articles using its .com TLD and
1219 articles using its .ir TLD at this time. I'm not sure if this is typical of all deprecated sources, though if it isn't then please feel free to take this with a grain of salt.—
Mikehawk10 (
talk)
02:58, 25 June 2021 (UTC)
Task: All non-userspace pages, any addition of "presstv.ir", "presstv.tv", "presstv.co.uk", and/or "presstv.com" should be met with a warning, similar to what happens when we use The Daily Mail. This should be enough to catch the use of the URLs for
Press TV while remaining narrowly tailored enough to avoid tagging edits involving other entities that are similarly named.
Reason: As I note above,
Press TV has been deprecated. The site's ".com" ending and ".ir" ending are cited in an approximate 2000 articles currently, indicating that a warning may be useful in assisting editors in not using the deprecated source.
It's probably just as well to note that, on Tuesday 22 June, 33 sites including PressTV.com were seized by the US government, meaning that their content cannot be accessed at their original domain names any longer. See
this Slate article: "PressTV was the best known of 33 websites whose domains were seized by U.S. authorities on Tuesday. According to the Justice Department’s statement, 30 of these sites were controlled by the Iranian Islamic Radio and Television Union, which is under the control of the Iranian Revolutionary Guard Corps, the powerful military force that exudes significant control over Iran’s foreign policy and economy. Others sites seized included Al-Masirah, the news service of the Iran-backed Houthi rebel movement; and Palestine Today, a pro-Hamas news outlet. The U.S. also seized three websites linked to Kataib Hezbollah, the Iranian-backed Shiite militia group that has been blamed for attacking U.S. troops in Iraq. The legal basis for the IRTVU seizures came from a 2018 Donald Trump executive order, which authorizes sanctions against foreign entities involved in interference in U.S. elections. The U.S. has seized Iranian sites accused of spreading disinformation before, but none as prominent as PressTV, the country’s flagship international broadcaster launched in 2007." ←
ZScarpia16:28, 25 June 2021 (UTC)
Before I saw this request, I put presstv.com and presstv.ir in the filter anyway! presstv.co.uk doesn't work, and presstv.tv has been grabbed by a domain squatter; neither have any article hits -
David Gerard (
talk)
11:01, 27 June 2021 (UTC)
Reason: There is a sockpuppeteer who keeps on removing all images from the article. He does it only once every 2 or so weeks but it's now annoying and it would be better to block those edits altogether.
I did consider that but he is a once-in-a-fortnight vandal so I don't know if protection is warranted. Besides there are still some other IPs and new editors who are productive (besides, if we protect the article, inexperienced editors may not know that they can help by creating an edit request, though I do concede that non-productive edits by newcomers amd IPs outnumber productive ones). Tube·
of·
Light10:14, 28 July 2021 (UTC)
Task: Block edits with the phone number "0240961557", "0240042887" or similar
Reason: A spammer or a group of spammers from Ghana are spamming said numbers into articles especially those pertaining to Ghanaian food and beverage products. Perhaps a regex of these would at least deter them from attempting to slip them again into articles, based on
this number scheme.
Noting
793 for the record. I think I tracked this for a while but was either missing too much or had too many FPs and didn't get around to narrowing it down; may look again if nobody beats me to it.
ProcrastinatingReader (
talk)
12:37, 29 June 2021 (UTC)
The following discussion is closed. Please do not modify it. Subsequent comments should be made on the appropriate discussion page. No further edits should be made to this discussion.
Task: Disallow or tag additions of .*jew?ish.* by non-autoconfirmed users in mainspace (if I got the regex right ;) )
Reason: IPs have been perennially stating that BLPs are jewish without RS, suggested by David Eppstein at [47]
The discussion above is closed. Please do not modify it. Subsequent comments should be made on the appropriate discussion page. No further edits should be made to this discussion.
Long-term addition of nonsense pages
Task: Prevent repetitive spamming of "word salad" pages that has been going on for years.
Reason: Someone, for years, has been repeatedly re-creating the same nonsense pages and spamming them across Wikipedia and other user-editable sites. The content is always the same copied and pasted strings; someone took the time to list them
here. For history, see
[50] and
[51]. It was suggested
here that an edit filter be given a shot.
Task: Catch issues with overlinking, catch test edits in article space, catch edits where common words are liked to unexpected targets.
Reason: There are a number of very common short words in the English language which should not ordinarily be linked in articles. I suggested this filter in response to an ANI post I made (
here) where an IP range had been vandalising by adding inappropriate links to articles, some of which included things like linking the word "The" to the article on the holocaust, or linking the word "In" to the article on India.
Certes Then followed up by saying that they monitor incoming links to a number of pages on common words like "A" as adding links to them is a common test edit for newcomers. Checking for links added to these type of common words could also spot issues related to overlinking, especially the worst kind of overlinking where every word in a sentence is converted to a wikilink. In other cases an editor may have added a legitimate link to an article, but have done it in such a way that it is attached to a common word that is likely to surprise readers when they click it.
I propose that a filter should look for additions which match the forms [[foo]] or |foo]] where "foo" matches one of the common words that should not ordinarily be linked. Checking for strings beginning with both double brackets and a pipe is necessary to catch piped links.
Not sure on the exact list of words to check, but a good starting point would probably be something like: "a, an, and, as, at, be, by, if, in, it, its, of, on, or, so, the, to". This will of course require some experimentation, words that generate a large amount of false hits should be removed, and there may be other words that are worth adding.
Diffs: Vandalism edits that would be caught:
[59],
[60], Test edits that would be caught
[61],
[62],
[63],
[64] Overlinking issues that would be caught
[65],
[66],
[67]
The first few 10 or 20
common words should catch most cases. New editors occasionally add good new links to the letter
A but it's rarer than testing and vandalism.
Certes (
talk)
20:04, 10 July 2021 (UTC)
"Julian Williams" LTA
Task: Flag the addition of "Julian Williams" to articles. It may be enough to flag only the
Thinking Out Loud article as that is a regular target but it spans music related articles
Reason: An IP hopping LTA going back at least to 2017 is adding "Julian Williams" (presumably them) as a song writer, composer or the like to articles, most prolifically to
Thinking Out Loud and various country songs but just found today they had added themselves to the
Cheers TV show as the theme song composer in the infobox (not sure when, but
I removed it). You can see in the current example, they added themselves to
Where Everybody Knows Your Name, the article about the Cheers theme song along with
Thinking Out Loud.
This needs to be some sort of filter that does not disallow. Rather, it silently logs it and can only be patrolled by helpers. This is an LTA that targets the pages
BS,
Konjac,
74, and similar. They make subtle changes that are hard to decipher unless you are familiar with them (ask
Zzuuzz). The summaries often ask wierd questions and the edits often mention Michelle Enrile, who died of choking on konjac, Kristoffer Hebert (often shortened to "Kris"). An LTA page may be useful, as this user is hard to spot and the edits are not blatantly vandalism.
aeschylus (
talk)
01:42, 21 July 2021 (UTC)
Task: This filter should prevent the addition of such misleading templates. I see some of them are already tracked via Z numbers (see the list at
Template:Z1#List of assigned z number templates) - I'm not sure if that includes everything, but it's a good start.
Reason: There's no reason for anybody besides admins to be adding block templates, and this should be rather easy to implement by checking for Z numbers (or the unsubstituted templates, which usually contain "block" at some point).
A quick check found no other inappropriate use of the word, so this problem may not be significant.
Certes (
talk)
16:16, 30 July 2021 (UTC)
Filter for edits with edit summary "typo", but which are actually extensive
Task: Flag edits with edit summary if typo/fixed typo that where more than a couple-few characters are added or subtracted.
Reason: "fixed typo" and "typo" are common edit summaries (I think one of them is a listed choice). There's essentially never a good reason for an edit with that summary to have +- character count of more than a few characters. If so, it's usually vandalism, and if not that then multiple changes of "color" to "colour" and that sort of thing, or at any rate a mendacious or at least misleading edit summary. There're verrry few cases where "fixed typo" and (say) the addition or subtraction of 15 characters go together.
Dif:[69] (It was a couple sentences of libelous nonsense which was in place for over a month.)
OK. Well there was a very similar one just last June. OK canned edit summaries are flagged (I don't know where or to whom, but I'm assuming that works). But this is a little different than that, it'd be tagging just one canned summary -- "fixed typos" --, and then if and only if the actual edit added or subtracted n characters -- like say eight. (Obviously this excludes a lot of misuse of "fixed typo", but at least it'll flag some, and perhaps the worst ones). I'm sure other canned summaries are compatible with large character-number changes, so I'm just concerned with "fixed typo" (and any others incompatible with large character-number changes, if any).
There's no action item from the June request -- no checkmark or X or whatever you guys use (if you do). It's unclear where it stands (I'm not familiar with the process). But I trust you guys, I'll leave it to, and if it doesn't seem worthwhile that's fine. Also thanks for your work on these filters, they rule.
Herostratus (
talk)
17:54, 2 August 2021 (UTC)
Warning for attitude -> altitude
Task: Warn people who want to "fix"
attitude to
altitude - with any meaning, but it's probably most common in aerospace articles because altitude looks like a plausible option there. This is typically a good faith edit from users who don't know attitude is a word and it's wrong with ~100% probability. A warning would avoid these edits.
Reason: If the edit goes unnoticed we get an error in the articles that's hard to spot.
Diffs: Most recent example
here, I have seen multiple other edits like that in the past.
Don't know if it's common enough to warrant a filter rule. If there is another way to find these edits that's fine, too.
mfb (
talk)
00:20, 3 August 2021 (UTC)
(duplicate post from WT:RPP) I have quite boldly implemented a new JavaScript form for the "Increase" side. If it looks OK, could we use it for the Decrease and Edit workflows, too?
Enterprisey (
talk!)
09:44, 18 August 2021 (UTC)
Sounds like a plan. It looks to me like people are successfully using the new form (
example), so later today I'll roll it out to all three workflows (Decrease, Edit, and the "submit edit request" button).
Enterprisey (
talk!)
00:47, 19 August 2021 (UTC)
A single-character file name would likely either be too generic, or just be absurdly annoying (e.g. "⌬.svg" for a diagram of a benzene molecule). Create protecting every title seems like it'd be really difficult, so I'd think it'd be better to just use an edit filter. InvalidOStalk17:26, 14 September 2021 (UTC)
@
InvalidOS: Do you have any evidence of this being a problem? Edit filters have finite resources so we tend not to use them speculatively. Additionally, since this would impact good faith edits, I think it warrants consensus to implement, per
WP:EF. Has this been discussed anywhere?
Sam Walton (
talk)
17:28, 14 September 2021 (UTC)
Task: This edit filter would prevent users, or at least non-autoconfirmed users, from inserting emojis associated with profanity into any page on Wikipedia.
Reason: Edit filters already exist to prevent editors from inserting profanity into articles. This filter would be similar, as it is unlikely that an edit adding these emojis would be constructive. Though, honestly, it seems there are few valid reasons for adding an emoji outside of one's own userspace, signatures, and articles about emojis. A couple diffs are provided below, but
this page gives a few more examples of the sorts of emojis in question.
TornadoLGS (
talk)
20:44, 27 September 2021 (UTC)
I'm being deliberately vague here, but (a) how easy would it be to create a filter which would block a certain IP range from editing any article which started with a particular string? (i.e. block 999.999.0.0/16 from blocking any article that started with the digit 1?) and (b) could this be done for a wider reach than a /16?
Black Kite (talk)22:22, 4 August 2021 (UTC)
ProcrastinatingReader The edits themselves are very random (some are actually OK) but
here's a typical one, (persistently) re-adding people to a year page where it has been decided they shouldn't be despite multiple warnings (at previous IP addresses). Having said that, the content of the edits aren't an issue anyway, as it's simple block evasion (every time one IP in the range is blocked, they simply switch to another).
Black Kite (talk)16:04, 26 September 2021 (UTC)
But since it's kinda like enforcing a block via filter, it'd be better if you (or another admin) created & set it to disallow. Probably worth notifying EFN in advance too, to ensure there are no objections.
ProcrastinatingReader (
talk)
19:46, 26 September 2021 (UTC)
Reason: See
Wikipedia:Village pump (technical)/Archive 191#tel URI. Although usage of these is sometimes around an actual phone number (e.g. at
List of suicide crisis lines), when using VisualEditor some browsers are mistakenly adding these links around other numbers including date ranges and ISBNs. It seems like there has been an increase in frequency of these recently. An edit filter to at tag them would help with fixing individual mistakes, and determining whether further technical action should be taken.
We currently have
343 links, though only 62 are in articles. A few seem to be actual telephone numbers, but it's still debatable whether they should use the tel: syntax or be included at all.
Certes (
talk)
23:47, 7 October 2021 (UTC)
CounterPunch is Deprecated
Task: The filter is supposed to warn users that CounterPunch is deprecated.
Around sporting events, when one player or team bests another, vandals often go to their pages to say that Player X "owned" Team Y, or Player A is Player B's "daddy". I'm adding a few diffs here for you to see:
[74][75][76]. Is there a way that a filter can be written to prevent some of this overdone pattern of vandalism? –
Muboshgu (
talk)
18:32, 13 October 2021 (UTC)
@
Muboshgu: Probably preventing strings such as "adopted", "owner", "daddy", "son", "father", and other such strings from being added to BLPs. This could probably be added to
189, which already covers BLP vandalism. InvalidOStalk11:36, 14 October 2021 (UTC)
"\badopt(ed|er|or)\b|\bdad(dy)?\b") seems reasonable to add. I've realized it'd be really hard to cover every case without causing a lot of false positives. This might not work well, and I'd definitely recommend testing first. InvalidOStalk13:16, 14 October 2021 (UTC)
189 is already set to tag-only. Then again, a separate filter might be better, as 189 seems to be for more libelous edits. I haven't really worked with edit filters that much, and don't even have
EFH, so I'm not really too experienced here. InvalidOStalk14:10, 14 October 2021 (UTC)
I mean no? I second the motion that there ought to be a filter for this -- why not? An edit summary of "typo" with like -300 characters is obviously a false edit summary. I think that false edit summaries are pernicious and I almost always roll back even good edits when the summary is mendacious. Whether there's a rule about this I don't know.
It ought be be easy to code, providing you restrict yourself to a list of default and common false edit summaries. It doesn't help with blank edit summaries. A blank edit summary with a change of like +800 characters is sus, but not illegal, as actually lying is or should be. It's going to generate some situation where the person was just being lazy tho. But so? A deceptive edit summary, whether on purpose or just not giving a rat's ass, is a pretty bad look.
Herostratus (
talk)
21:35, 17 October 2021 (UTC)
@
Certes: I added plain "typo" and "typos" to 970 (
hist·log). Don't know why I left that out; it matched on "fixed" and "fixed typo". Can you suggest a better tag than "possibly inaccurate edit summary"? That's kind of wordy, but I don't want to accuse the editor being intentionally misleading.
Suffusion of Yellow (
talk)
21:45, 17 October 2021 (UTC)
Thank you! That seems to be working well. Checking the top entry at random shows a very unwelcome and possibly libellous edit (already reverted). A tag such as "summary mismatch" would be more concise but slightly cryptic; on balance it seems better to be clear than short.
Certes (
talk)
23:32, 17 October 2021 (UTC)
@
Certes: Tagging now. Just went with "possibly inaccurate edit summary". As tempting as it would be to set the filter to warn, I don't think we should do that. There's only so much we can warn new users for before they just start ignoring everything. Plus there's
Mediawiki:Mobile-frontend-editor-summary-placeholder which I suspect leaves some people with the impression that "Fixed typo" and "Added content" are the only allowable summaries. Or maybe they're just having a
here have code moment.
Suffusion of Yellow (
talk)
01:07, 30 October 2021 (UTC)
Thanks. I had another look at 970's log last night and reverted a few; most of the bad ones had already been caught. I'd forgotten about
mobile (especially
app) problems.
Certes (
talk)
11:50, 30 October 2021 (UTC)
Request Blacklist
I am requesting a black list about this word "any body can get on wiki and type liesssssss Bye Barney b", the same wording has been spammed and used by an LTA to vandalize multiple articles for several months until now
[77].
220.100.66.173 (
talk)
22:54, 29 October 2021 (UTC)
Talk page refactoring by IPs
This isn't a formal request since I don't quite know what the most effective way to go ahead this would be. See for example the edits of
Special:Contributions/36.80.242.214. I think having something to prevent refactoring of talk page comments by others would be helpful (for example: if the edit does not add a signature but alters a comment which contains a signature from a different user: I don't know if this is too hard to code for, or if any potential solution would lead to false positives or to it being easily circumvented), especially since most vandalism and hence vandalism patrol happens in mainspace (and it is through the sheer luck of having this on my watchlist that I noticed). Cheers,
RandomCanadian (
talk /
contribs)
21:55, 10 November 2021 (UTC)
I think I played around with a similar idea in 1142 (
hist·log) ("Possible impersonation") some time ago. Ran into too many FPs for it to be viable, and too many weird edits that made no sense but didn't seem to be intentional impersonation either (eg
[78]). There were also FPs related to IPv6s that often seem to use a different signature on their /64 (perhaps maybe
[79]); I presumed some write another signature manually so queries go to a single talk page.
ProcrastinatingReader (
talk)
22:04, 10 November 2021 (UTC)
Filter 869: Republic TV
Task: To warn users against the use of Republic TV. The url is "republicworld.com"
Reason: There is community consensus that Republic TV should be deprecated.
Requesting a filter that warns users if they mention Eddy Chen in their edit on Absolute pitch and its talk page
Task: To warn or deny new/IP users from saving their edit request if it includes 'Eddy Chen' or 'TwoSet Violin' on their edit on
Absolute pitch and
Talk:Absolute pitch.
Reason: This has been an ongoing issue, with too many unregistered IP editors requesting to add Eddy Chen of TwoSet Violin onto the actual page using self-published sources. Many of the new/IP editors also are failing to read the warning on the talk page to not request any edits of Eddy Chen, as there are still requests about once every week or two, and we have been getting tired of having to continually deny such requests as they are unnecessary and clog up useful requests.
@
Jeuno: Is this disruption limited to only those two pages? Probably better just to go with protection, then. There are only a limited number of filters, and they generally aren't used for single-page issues.
Suffusion of Yellow (
talk)
21:28, 8 December 2021 (UTC)
@
Suffusion of Yellow: Sorry for the late reply, I was busy finishing off my final exams for the past week. The reason why I decided to request this edit filter was because the admin that responded to my RfPP request to have indefinite semi-protection had said that they will consider investigating into whether an edit filter might be a better option instead of having to renew the semi-protection every few months. They have just extended the protection for another 6 months while they investigate.
Jeuno (
talk) (
contribs)
00:28, 19 December 2021 (UTC)
@
Jeuno: It look like
Ohnoitsjamie did indeed say that.
Ohnoitsjamie, with only two pages affected, this doesn't seem like a good candidate for a filter. It's not something super-critical like
WP:AIV that needs to stay unprotected, and it's long-term issue, so the filter will need to stay enabled indefinitely. Plus if the requests aren't all deliberate trolling, the filter will need a custom message, so there's no way to merge it another filter.
Suffusion of Yellow (
talk)
20:02, 19 December 2021 (UTC)
Prevent manual addition of automatic mediawiki categories
Task: Prevent manual addition of categories listed at
Special:TrackingCategories, which should only be automatically added by the mediawiki software.
Reason: These categories should never be manually added, as adding them manually clutters up tracking categories even after the issues with the page are resolved.
Reason: This should be an easy one: there are a lot of IPs making talkpage posts to
Talk:Federal Bureau of Investigation,
Talk:Alphabet Inc. and others with identical mentions of "Nathaniel meskimen". I've placed a /48 block, but they're getting around it. I've left the last IP unrevdel'd, the edits are all identical to this one. Can we make or modify a filter to deny these edits?
Task: The urls being counterpunch.org and counterpunch.com
Reason: There are significant concerns regarding the validity of the previous RfC, which appears to have attracted at least 6-7 sockpuppets, found after its close. It has led to this discussion at
Wikipedia:Reliable sources/Noticeboard#Rerun Counterpunch RFC?, seeking a new one and started by the closer of the RfC themselves. In the meantime, the filter is inappropriate for this site.
I've opined in that discussion, but with the closer of the discussion (David Gerard) also feeling the discussion isn't robust due to sockpuppetry this seems fairly clear-cut I think. So DoneProcrastinatingReader (
talk)
15:19, 11 January 2022 (UTC)
I have no clue what's going on at this point. There seems to be an edit war ongoing at
WP:RSP/
WP:DEPS etc, and I don't want to have the dispute carry over to an edit filter's contents which would be quite disruptive, so I'm going to hold off on any further edits to the filter personally. I will say the situation is highly confusing at this point, with everyone seemingly agreeing the previous discussion is tainted, but not agreeing on whether to still class the source as deprecated, and now a new RfC is started re whether the source should be deprecated. I'd have suggested figuring out the status of the source at
WP:AN, but with the new RfC I don't know if that's a suitable path anymore.
ProcrastinatingReader (
talk)
23:11, 11 January 2022 (UTC)
I've not seen any of this recently, and not much before this report, besides which, if I'm right about this, next time if you poke a checkuser I think you might get some results ;) --
zzuuzz(talk)23:34, 15 January 2022 (UTC)
New users requesting edits on "Name of page you are requesting an edit to" at Wikipedia:Requests for page protection
Task: This filter task is to warn and tagged new users of making edit requests.
Reason: This filter is needed because there are many newcomers and IP making too many requests to edit "Name of page you are requesting an edit to" at
WP:RPP.
(I'm not too familiar with edit filters, so this may have been brought up before; if so, apologies, and just give me a pointer.) One of the most common forms of inappropriate editing we see from new editors is the addition of hyperlinks within body text. Would it be possible to use an edit filter to track this and maybe someday help guide users to use a reference instead?
In technical terms, we would want this to tag all mainspace edits that introduce a URL to a non-Wikimedia site and that do not fall into one of these exception buckets:
Anything within a template (includes stuff like {{External media}} and infoboxes that link a website)
Anything within the last section of an article, or a section that contains any of "Reference", "Source", "Further", "Reading", "Work", "Publication", "Citation", "Cited", "External", "Link", or "Note"
@
Sdkb: I don't think we should warn people for this. It's basically a formatting mistake to write Some fact.[https://some.ref] instead of Some fact.<ref>https://some.ref</ref>. We can't warn newbies for everything, or they'll just ignore everything we say, or worse give up and go to a more user-friendly site. As to creating a filter at all, even a log-only or tag-only filter, well I won't say "impossible", just "incredibly hacky if possible". If I can think of a clever way, I'll try something.
Suffusion of Yellow (
talk)
23:21, 15 January 2022 (UTC)
@
Suffusion of Yellow, thanks for the reply. Re We can't warn newbies for everything, yeah; I think the main issue is that our tools for warning are too blunt—all we can do is have them encounter a big notice when they try to publish, rather than having a friendly "want to turn this into a reference? We don't allow inline external links" prompt pop up in a box next to the paragraph as soon as someone tries to add an inline external link. See
WP:Making editing easier 2021. {{u|Sdkb}}talk23:29, 15 January 2022 (UTC)
These sockpuppets will spam the user talk of as many recently active editors they can find with requests to improve
Akane Yamaguchi and the spam is getting to be quite disruptive. I just blocked
Semwq, then immediately after that one
Zasjd. Would it at all be possible to disallow new users from doing this rapid spam? I don't believe I've ever made a request here before so apologies in advance if this is not feasible or worth the time to create.
Sro23 (
talk)
07:09, 10 January 2022 (UTC)
I've been spammed twice over the course of two years, very annoying. I was thinking about proposing this myself. The easiest way to implement this would be to create a filter for new users using the string "Akane Yamaguchi" (The spam for Maureen Wroblewitz seems to be historical and no longer relevant) that would either disallow the edit to be made or would be logged to alert admins.
Hemiauchenia (
talk)
08:58, 10 January 2022 (UTC)