@Certes and GoingBatty: This follows from our recent wouldn't-that-be-a-nice-idea convo at User talk:GoingBatty#Misdirected links (28 December 2020). I've drafted a project page; improvements, suggestions and derision (OK, derision not so much) will all be welcome. It's an open draft, so just go ahead and edit it.

Do you know of anyone else who might be interested in getting this project up and running, who could contribute at this early stage?

My idea is there would be basically three project areas:

1. Project page, in two parts. (1) A stable description. (2) #New investigations, where anyone could add links which might be worth looking at.
2. Talk page. Slow-moving, for discussion of the project as such.
3. Archive.

The Archive is currently empty, and could take some time to get running efficiently. My idea is, that it should be a sortable wikitable with 3 columns: (1) bluelink, as primary field (2) date last looked at, in ISO format (yyyy-mm-dd) as being (a) the most easily sortable and (b) language-independent, and (3) freeform comments, which could be multiline. As a start, we could simply populate it with links to relevant User and Talk pages we know of.

Categorisation will be important. I'm not at all sure my suggestions are ideal.

A {{userbox}} and/or {{topicon}} for members would add an air of professionalism.

If we agree on something I could tidily move the whole shooting-match, history and all, into Wikipedia-space. If it gets to that point, we should advertise. I suggest WT:DAB, WT:DPL and WT:SIA. Is there any other WikiProject worth notifying? Or, should we advertise before going live? Narky Blert (talk) 14:39, 3 January 2021 (UTC)

@Narky Blert: Thanks for setting this up!

I suggest also inviting User:GreenC, who created the backlinks functionality I'm enjoying. Other people might want to use the same functionality, which might require some work on his end to implement.

A link to H:INSOURCE with some examples would be helpful. For example:

• Bull insource:/\[\[Bull\]\]/
• ford insource:/\[\[ford\]\]/
• "The Crown" insource:/''\[\[The Crown\]\]/
• "Back to Black" insource:/\"\[\[Back to Black\]\]/ (looking for instances where the link goes to the album but should go to Back to Black (song)

I agree categorization is important. Disambiguating New York helped me find a link that should have been New York GAA, which led me to articles that needed London → London GAA and more. There may be GAA fans who want to run queries with GAA in them.

Thanks again! GoingBatty (talk) 16:07, 3 January 2021 (UTC)

@GoingBatty: How about another subpage for the nitty-gritty of the techniques? called, ooh maybe, Wikipedia:WikiProject Bad bluelinks/Search tools or something? (I could never get my head around grep, so count me out.) That would warn casual visitors, Here Be Dragons - stay in the well-lit part of the cave. Narky Blert (talk) 17:03, 3 January 2021 (UTC)

Splitting the static text from workings which will change frequently should make content easier to find and prevent important changes from getting lost in a long edit history. Certes (talk) 17:18, 3 January 2021 (UTC)

Or, to steal an existing name, Wikipedia:WikiProject Bad bluelinks/ToolForge? Snappier, and less offputting than "Beware of the leopard". That itself could have subpages. Narky Blert (talk) 17:36, 3 January 2021 (UTC)

Yes, thanks for setting this ball rolling. I would also find the ability to add titles to a "new backlinks" report useful. User:Certes/misdirected links#Finding links has some useful search examples. Certes (talk) 16:20, 3 January 2021 (UTC)

@Certes: Your every desire is my whim. So long as the general-editor part of the project is straightforward and welcoming, there is every opportunity and reason for clans of dwarven smiths to excavate their own subpages. Narky Blert (talk) 17:14, 3 January 2021 (UTC)

@GreenC: It's been suggested that you are another editor who might be interested in this idea. Narky Blert (talk) 16:51, 3 January 2021 (UTC)

Prompted by the preceding discussion, I suggest a fourth project work area in addition to the three I listed at the outset (which remain unchanged):

4. Workshop, WP:WikiProject Bad bluelinks/Workshop

which could of course have subpages. I've tweaked the Project Page to suit. We don't want to scare possible participants off by intimidating them with screensful of code (if it isn't in C or 80x86 assembler, you've lost me). Narky Blert (talk) 18:20, 4 January 2021 (UTC)

My "asdf" key hve gone on trike, or jut ie :-( I won't be poting here gin until my new PC h rrive (promie iy). Narky Blert (talk) 18:44, 6 January 2021 (UTC)

Sorry to hear that. It's probably just something stuck underneath. If it's a keyboard you're about to discard (rather than a laptop with a fragile disk) then turning it over and giving it a good shake and bang may do the trick. Certes (talk) 20:40, 6 January 2021 (UTC)

3yo lappy - weak contacts - nothing happened or occasionally a double hit - postit note under the keys did nothing. Oh for a lovely late-80s IBM clacky kbd.

I'm in the middle of the usual nuisance of copying bookmarks across; a good opportunity to get rid of the ones I'll never look at again.

& I'm going to have to rebuild my boilerplate edit summaries in two browsers. grrr Narky Blert (talk) 18:12, 8 January 2021 (UTC)

Ah well, at least you're up and running on something with a greater life expectancy. It's always a tough call between copying the browser profile across (complete with junk) and starting afresh. Certes (talk) 18:48, 8 January 2021 (UTC)

Unnecessary ping to @Certes, GoingBatty, and GreenC: I opened this discussion on 3 January. Unless any of you have anything to add, I propose to move this whole shooting-match into Wikipedia-space once 7 days for discussion are up.

Repeating for convenience, the structure will be:

1. Project page, in two parts. (1) A stable description. (2) #New investigations, where anyone could add links which might be worth looking at.
2. Talk page. Slow-moving, for discussion of the project as such.
3. Archive, WP:WikiProject Bad bluelinks/Archive WP:WikiProject Bluelink patrol#Archive, which I will create, empty.
4. Workshop, WP:WikiProject Bad bluelinks/Workshop WP:WikiProject Bluelink patrol#Workshop, which I will create, empty.

Once that is done, please add links and move old stuff (including Workshop stuff on this page!) into the Archive and Workshop subpages. Feel free to add subpages to the Workshop.

I will advertise the project at WP:DAB, WP:DPL and WP:SIA. I'll try to categorise it as best I can.

(When was the last time anyone set up a maintenance WikiProject? There's a sort of lyocell effect here. In the middle of a patent battle, the two competing manufacturers agreed that it would be a Good Idea to register a generic term for this fibre, and that it should be lyocell. So far so good. The problem was, that although all the regulatory authorities had procedures for registering new generic fibre names, all the people who knew the procedures and were still contactable (other than by séance) had retired.) Narky Blert (talk) 21:11, 8 January 2021 (UTC)

All I can find is Wikipedia:WikiProject#Creating and maintaining a project. The related Wikipedia:WikiProject Council/Guide suggests that we may wish to start a task force under an existing WikiProject, but I'm not clear which one. The nearest fit is Disambiguation but these links aren't ambiguous, just wrong. Certes (talk) 22:31, 8 January 2021 (UTC)

One more thing: what's the best name? Although I tend to be candid rather than politically correct, perhaps we can pick something that emphasises improvement over defects in our inputs, along the lines of the defunct Wikipedia:WikiProject Red Link Recovery. Certes (talk) 22:40, 8 January 2021 (UTC)

Thanks to both of you! I agree with Certes' suggestion to emphasise improvement. How about WikiProject Bluelink detection and correction? "detection" is the use of queries and Backlinks to find links to be changed, and "correction" is the act of manually or programmatically fixing them. GoingBatty (talk) 23:39, 8 January 2021 (UTC)

A pedant might point out that bluelink detection is trivial for those of us lucky enough to enjoy decent colour vision. How do we convey detecting bad links without using the B word? Perhaps we need a term like checking or verification. Words used in other projects include review and patrol. Certes (talk) 23:55, 8 January 2021 (UTC)

You make a good point about the name; you've reminded me of the Society For Indecency To Naked Animals, whose founder readily admitted that the name gave the wrong impression. Narky Blert (talk) 00:50, 9 January 2021 (UTC)

Update - IMO there's no point in going live until we here are happy with the name. Narky Blert (talk) 17:04, 10 January 2021 (UTC)

I'll tentatively weigh in with "Wikiproject Bad bluelink detection and correction". Narky Blert (talk) 18:18, 10 January 2021 (UTC)

@Certes: Good point. "Wikiproject Bluelink patrol" ? GoingBatty (talk) 18:37, 10 January 2021 (UTC)

"Wikiproject Bluelink patrol" sounds good to me. Narky Blert (talk) 06:26, 12 January 2021 (UTC)

WP:BPA would be available as a shortcut. Narky Blert (talk) 20:59, 12 January 2021 (UTC)

Further unnecessary ping to @Certes, GoingBatty, and GreenC: I've tweaked the main page so that the title will be "Wikiproject Bluelink patrol", as suggested above. I have not retitled this user page, so as to avoid unnecessary clutter and breaking links, even at this discussion stage. (I have the WP:EM privilege, and should be able to do inconspicuous cleanup when we reveal ourselves to the world.) I've suggested WP:BPAT as shortcut; WP:BLUP is also available.

Thoughts? When we go live, I would be happy to set up first stabs at the Archive and Workshop subpages and to do the advertising. Narky Blert (talk) 21:48, 23 January 2021 (UTC)

Yes, let's go, and thanks for offering to coordinate the launch. I'm happy to help where I can. In the interests of being collegiate, I should probably move a few pages from userspace to the project. User:Certes/misdirected links should probably be split into a general how-to and a workshop list. (Despite its name, User talk:Certes/sandbox, mainly written by you, could also be turned into a useful guide. Although mainly about disambiguation, many of its tips apply equally to bad blue links.) Certes (talk) 23:18, 23 January 2021 (UTC)

I've got bits of User talk:Certes/sandbox in one of my own sandboxes, and keep hoping that someday I'll turn that into a WP:ESSAY. Narky Blert (talk) 17:25, 24 January 2021 (UTC)

Looking forward to seeing Wikipedia:Wikiproject Bluelink patrol Wikipedia:WikiProject Bluelink patrol become a blue link! Thanks! GoingBatty (talk) 23:19, 23 January 2021 (UTC)

There's a draft Archive at User:Narky Blert/Wikipedia:WikiProject Bluelink patrol/Archive Wikipedia:WikiProject Bluelink patrol/Archive. Comments/tweaks welcome. Narky Blert (talk) 06:38, 26 January 2021 (UTC)

I've moved everything into Wikipedia space, and advertised at WT:DAB, WT:DPL and WT:SIA. I've moved the Backlinks discussion into Wikipedia:WikiProject Bluelink patrol/Workshop (it ain't exactly for general readers). "Happy editing". Narky Blert (talk) 22:18, 27 January 2021 (UTC)

Thank you! I've created the userbox {{User WP Bluelink patrol}} and added it to Category:Wikipedia:WikiProject Bluelink patrol, but should that be simply Category:WikiProject Bluelink patrol to match most other WikiProjects (which are all in Wikipedia: namespace)? Certes (talk) 23:25, 27 January 2021 (UTC)

I'm no showman but we may want to blow our trumpets more widely. Ideas include:

• Solicit a mention in Signpost. I think one is imminent, so it may have to await the end February edition. Certes (talk) 14:39, 31 January 2021 (UTC)
• Edit summary suffix. I've seen brief piped links appended mentioning other WikiProjects such as Disambiguation and tools such as AWB and Twinkle. I'm using edit summaries such as improve link: Orchid and it may be worth having a standard suffix to copy and paste where appropriate, or simply use something like improve link: Orchid. Certes (talk) 14:39, 31 January 2021 (UTC)
• How about creating subpages or page sections by topic and asking WikiProjects to assist? For example, maybe WP:WikiProject Comics might be interested in all those Marvel characters. GoingBatty (talk) 15:23, 31 January 2021 (UTC)

  (1) I've never read Signpost, so can't comment. (2) My boilerplate summary is "Bad link repaired"; force of habit, I suppose. (3) Some WikiProjects are extremely helpful if presented with individual DABlinks (I'll single out Classical Greece and Rome, Judaism, Mathematics, and Medicine for special praise); others a lot less so. I bookmark pages, and offload them about annually; often, the majority have been fixed in the interim. I'm unsure as to how they would react if given "please check the bluelinks on this page". Narky Blert (talk) 15:20, 1 February 2021 (UTC)

  I find it helpful to see from my edit history what I've changed, though it does take time to fill in. Checking the bluelinks on a page is something I'd expect a good author, GA reviewer, etc. to do routinely (but if they did, we could disband). "Please check all bluelinks on these 50 pages" might be a less welcome request. I'm still hoping for a linksfrom: search option, because linksfrom:"List of Foo" -Foo could find lots of errors. WP Comics has supported previous exercises. Certes (talk) 15:46, 1 February 2021 (UTC)

  Sorry, I wasn't suggesting asking a WikiProject to check an individual article. I was suggesting creating a list of links (e.g. Cyclops, Wolverine) that a particular WikiProject could feel ownership for checking their articles for those links and fix them. GoingBatty (talk) 17:28, 1 February 2021 (UTC)

Hi, I'm not sure if I have the right idea about what the exact goal of the project is, but if I'm on the right track, then I think maybe I ran across a couple of candidates to be looked at by someone who knows how to see if they need fixing or not. I found Systemd might easily be confused with System D. Is this the kind of stuff the project is looking for? Huggums537 (talk) 11:03, 8 February 2021 (UTC)

@Huggums537: Hello and welcome! Yes, I think it's worth checking what links to Systemd and System D to see whether any links should be improved. (List of trance genres looks suspicious.) So far, the wikiproject has mainly been looking at links to X when X (y) or X, Y was intended, e.g. Prince instead of Prince (musician), because they're easier to find. However, the sort of error you describe is certainly a good area to expand into. Certes (talk) 12:59, 8 February 2021 (UTC)

Ok, thank you for the links. I'll check them out when I get a chance and see if I can find anything. Huggums537 (talk) 14:18, 8 February 2021 (UTC)

@Huggums537: Yes, that's very much the sort of thing we look at - bad links which might make a puzzled reader wonder why on earth they've landed on the page they're looking at; or, worse, mislead them. (I've just fixed seven links to a C18 opera which were actually intended for an extinct species of clam; and one which related to a rock musical...) Narky Blert (talk) 11:41, 12 February 2021 (UTC)

OK, great. I just fixed one link in the trance article that Certes pointed out, and all the other links check out in the other search results. Now, I guess I can do a "what links here" for systemd and check all those results, but there are a whole shipload of them. Any advice on how to go through articles more quickly, rather than one by one? Huggums537 (talk) 18:38, 12 February 2021 (UTC)

@Huggums537: One way is to filter your search to exclude a term likely to be on the page if the link is good (e.g. add -Linux); another is to include a term likely to be on the page if the link is bad (e.g. add economy). You could also check the case of the text: systemd usually has a lowercase S, System D uppercase. Further suggestions: User:Certes/misdirected links#Finding links. Certes (talk) 18:45, 12 February 2021 (UTC)

Ok, thanks for the help! Huggums537 (talk) 19:02, 12 February 2021 (UTC)

Now that we have some experience of completed investigations, what's the best thing to do with them? I've just tabulated another batch but that loses some information and may not add much. It's also labour intensive. In future, should we just archive the conversations? Certes (talk) 16:22, 23 February 2021 (UTC)

For me, the key data are "This link was looked at on this date". The purpose is to avoid duplication of effort. Anything else is commentary.

I recently remembered your investigations into PTOPIC elements - Lead, Silver, Palladium and so on. Those might be worth archiving in that way, because they'll come up again. Narky Blert (talk) 21:09, 23 February 2021 (UTC)

I have posted an invitation at WT:WikiProject Classical Greece and Rome#WikiProject Bluelink patrol. As I said there, I'll be happy to deal any posts made here. Narky Blert (talk) 21:13, 23 February 2021 (UTC)

Wikipedia talk:Disambiguation pages with links#'Hidden' disambiguations is discussing bad blue links to certain types of page which are not dabs. Certes (talk) 23:15, 27 February 2021 (UTC)

Following on from Wikipedia talk:Disambiguation pages with links#'Hidden' disambiguations, do we want to start a unified list of terms liable to attract erroneous (and perhaps suboptimal) wikilinks? If so, what are our inclusion criteria? We certainly can't hope (or want) to list them all: the 260,000 dabs called "Foo (disambiguation)" each suggest that links to Foo may mean something on the dab instead, and there are other cases such as WP:ONEOTHER hatnotes. How do we make the cut? Do we count links to the alternative topic(s), on the basis that if 10,000 articles link correctly to Apple Inc. then there are likely to be a few bad Apples too? Would it be better to go by the number of bad incoming links actually fixed, assuming we can count those? Do we take the number of correct links into account, delisting anything with too many false postives? (A few links to London will refer to London, Ontario, but checking all 114,000 would not be the best use of our time.) Any other thoughts? Certes (talk) 23:52, 2 March 2021 (UTC)

I certainly think it's worth creating a unified list of known serial offenders; even if some have to be labelled as "too much effort for too little reward".

In new cases, you don't know what the problem might be like until you look. Many won't deserve looking at again for some years. Narky Blert (talk) 10:54, 3 March 2021 (UTC)

A search for "london linksto:ontario linksto:london -england" reduced the number to a more manageable 415; which I got down to 409, on the BOLD assumption that no-one commutes on the train which takes a detour between Toronto and Sarnia. It could be worth listing that kind of filter which people have tried, both for future reference and as a prompt for other cases. Narky Blert (talk) 06:45, 17 April 2021 (UTC)

There's a list at User:GoingBatty/misdirected links. I also have a mental list which includes London, Ontario but never got around to writing it down. Basically it's User:Certes/misdirected links minus User:Certes/Backlinks with suitable filters like "Ontario -England" added. ("London, ON" is also worth looking for.) I had toyed with the idea of enhancing GreenC's bot to search for such cases. (It could already list new links to London, but 99% will be good text about England.) Certes (talk) 11:13, 17 April 2021 (UTC)

Hi, I'm not sure where to put this (investigations section of this page or the list of misdirected links) so I'll just post it in this talk page. I've found and removed many bad links to Peterborough (supposed to be Peterborough, Ontario), and smaller amounts of other mistakes such as Whitby, Ontario / Whitby and Cambridge, Ontario / Cambridge (which is on the list, but only the MA one). 20UF6 (talk) 00:16, 1 May 2021 (UTC)

Update: I think I have cleared out most of these. 20UF6 (talk) 01:05, 1 May 2021 (UTC)

Welcome, 20UF6 and thank you. I worked through city lists last August before BPAT kicked off). List of cities in Ontario prompted about 50 fixes for Aylmer, Brighton, Cobalt, Erin, Grimsby, Hanover, King, Niagara Falls, Renfrew and Uxbridge but new errors will have crept in since. I didn't spot the links to Peterborough, and I skipped towns, though Whitby does look large enough to merit a check. New links to Cambridge are monitored but some old ones will have slipped through. London will acquire bad links, as it's impractical to monitor such a widely linked page. A good place to document this work is Recently completed investigations, or Investigations in progress if you think there are still more bad links to be found.

We also have links to Dalton, Ontario, some of which relate to the second item in the set index which has no article. Certes (talk) 09:59, 1 May 2021 (UTC)

Hi, thanks for the information and help. I'll go check them out and fix what I can find. 20UF6 (talk) 14:10, 1 May 2021 (UTC)

Detailed progress records have helped us evaluate the nature and scope of the task and prove that this is an active project worth retaining. Now we're a few months in, how much practical use is the full archive? We certainly need a way to record likely perennial offenders – perhaps searches where we fixed 20+ links and expect enough new errors to make regular checks worthwhile – and sources of inspiration – Top 1000, works list, etc. – but is there much value in archiving run-of-the-mill activity? Certes (talk) 10:36, 2 June 2021 (UTC)

I think it's worth maintaining, possibly as an unstructured dump. Like the backlinks lists, it's a place to check before starting what might be a fiddly investigation. Narky Blert (talk) 06:01, 10 June 2021 (UTC)

@GoingBatty: Any comments on the Works lists? My normal practice is to work through those with over 200 links then cherry-pick from the 100–200 range, investigating any where I've heard of (or feel I should have heard of) the qualifying title. I also widen the search by repeating it without requiring quotes for the few cases where the base link is a redirect from title case to sentence case, e.g. check all links to Gingerbread Men in case they intend Gingerbread Men (album) rather than Gingerbread man. Any suggestions for improvements? Certes (talk) 20:09, 6 June 2021 (UTC)

@Certes: I've been working through the "H" list and posting my results. When the base doesn't have a lot of entries, I review them all. When the base has thousands of entries, I focus on those where there are italics or quotation marks before the link, and then do searches for the disambiguator in the text. GoingBatty (talk) 23:51, 6 June 2021 (UTC)

@Certes I just went through the entire "H" list. It went much faster by doing searches such as Hook insource:/(''|\")\[\[Hook\]\]/

It looks like your Quarry query now sorts the list alphabetically, which would be much easier to use. Could you please post some new lists? Thanks! GoingBatty (talk) 18:48, 29 May 2022 (UTC)

@GoingBatty: I run a simple script off-wiki to postprocess the query results, add useful links and sort by link count. I've archived the completed G–J, refreshed K–L which were a year old and added M–N. I've made the tables sortable so anyone who prefers an alphabetical list can click the "Qualified" header. Yes, the search on italics or quotes will eliminate a lot of false positives but may also skip real errors like Smith wrote for Time where the author didn't bother with punctuation, and citations like |publication=Life where the italics are added by a template. The '' pattern also matches ''' bold markup, which may or may not be helpful. Certes (talk) 20:31, 29 May 2022 (UTC)

@Certes - Thank you! Your new table format is much more convenient, and the search link works well. I realize our insource searches may skip the occasional real error, but the insources searches should be the best use of our time. (I have Time and Life on my Backlinks.) Thanks again. GoingBatty (talk) 23:21, 29 May 2022 (UTC)

@Certes I noticed your search link for Kannagi (newspaper) appears to be malformed. You may want to double check your query. Thanks! GoingBatty (talk) 16:24, 30 May 2022 (UTC)

I don't see the problem. For me, the search seems to run correctly but finds no results. Is there a page that should appear? The canned search is just for the title in italics (or quotes if a song, etc.) or in a |publication= or similar where the title of a work is expected. Other incoming links may be wrong but devising a search for them would require more skill than my query can muster. I've added a new column with a full search: this shows (e.g.) all links to Kannagi. In this case there are only four, and all seem to refer correctly to the legendary woman who is the primary topic; it looks as if we need just the 275 existing links to the newspaper. Certes (talk) 18:59, 30 May 2022 (UTC)

@Certes - Oh, OK. The "newspaper" search links are different the "album" or "film" links, and I mistook that difference as an issue to be fixed. I don't understand the "newspaper" search, but shouldn't has presumed that it's wrong. I apologize for that, and thank you for all your hard work. GoingBatty (talk) 00:54, 31 May 2022 (UTC)

@Certes Also, you include Karate Chop (song) in your list, even though Karate Chop is a redirect to Karate Chop (song). Should situations like this be excluded? Thanks! GoingBatty (talk) 16:35, 30 May 2022 (UTC)

I've removed Karate Chop and six similar cases. Certes (talk) 18:59, 30 May 2022 (UTC)

@Certes Thanks! GoingBatty (talk) 00:54, 31 May 2022 (UTC)

@Certes Have you considered changing your search links so they will also identify section links with a "#"?

I think you could do this by changing insource:/[^']''\[\[ *Foobar *[\]\|]/ to insource:/[^']''\[\[ *Foobar *[\]\|\#]/. Thanks! GoingBatty (talk) 17:06, 1 June 2022 (UTC)

@GoingBatty: Done. Getting the # to pass through is tricky, as it usually introduces an anchor, but I think it's working. Certes (talk) 21:59, 1 June 2022 (UTC)

@Certes Works great, and I immediately see additional links to fix. Thanks! GoingBatty (talk) 00:29, 2 June 2022 (UTC)