This is an archive of past discussions with User:Citation bot. 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 talk page.
I was not referring to all those changes, only the first (sorry, I posted the wrong link - I thought I posted a diff, not the link to an article revision). That link is now dead, the bot is indeed right. (The doi though is still broken, so I just reinserted the doi broken date after your last edit).
Sorry for my stupidity and hubris, I really am a terrible human being.
Sorry to bother again, can you please look at the last change the bot made to the article? It deleted the inactive doi tag. But the doi is inactive. JudeFawley (talk) 22:45, 16 July 2025 (UTC)
thank you for the report. I actually check every good to bad and bad to good DOI and then double check. This DOI is in the "kinda works" category. DX.doi.org works, and the url resolves to something valid. BUT, the the final valid URL is a publisher/journal landing page. I have added it to the list of "trust me bro, it's bad" DOIs for the bot. We have similar issues with sites that have non-404 "404" pages and journals that have disapeared and now are websites the set herbal viagra or offer slot machines. And yes, the checking every changed DOI is a thankless but time consuming task, but it does not involve humans, so no ever complains. AManWithNoPlan (talk) 23:24, 16 July 2025 (UTC)
@GreenC: See the third box at the top of the page. I keep a list that can be CB-bot cleaned up at User:Headbomb/Sandbox. Dates/Invisible characters need a lot of manual love, and Bibcodes will always have like ~25 in the category because it lags in time. The others can usually be brought down to <5 with automatic cleanup and the rest done by hand. Headbomb {t · c · p · b}16:30, 17 July 2025 (UTC)
Headbomb, OK. One of my bots expands archive.today shorform to longform every 15 days (around 1,000) and this creates entries in the category (50-100), because it exposes archive timestamp mismatch with the |archive-date= which typically happens because of timezone mismatches eg. editor adds an archive-date based on local timezone not GMT. Thus you may see the category bloom in size every 15 days because of my bot. I try to get to it by the end of the day to fix those cases. Unfortunately due to the design I can't do both at the same time (fix the short form and fix the archive-date), they are separate programs. — GreenC17:11, 17 July 2025 (UTC)
You work tirelessly and never complain. Though your work is little known or appreciated now, your true value will only be realized when something prevents you from doing your job. I hope that that doesn't happen any time soon! GrinningIodize (talk) 17:33, 8 July 2025 (UTC)
It thinks cite web citations to bibliographic data are citations to books and replaces them with cite book links
What should happen
It leaves the cite web citations alone, as said in the RFC
Relevant diffs/links
When we did the WorldCat RfC I supported with the condition that the bot knew how to distinguish citations to WorldCat for their bibliographic data to book citations. It does not. Please stop it from doing this. [1]
We can't proceed until
Feedback from maintainers
This was my one request in the WorldCat RfC - that it does not make this extremely obvious error. The consensus was to keep WorldCat links when they are to bibliographic data, not the source book. The bot cannot tell these apart, when it was said in the RfC that it could. It does the extreme obvious error, against everyone's consensus in that RfC. It has done this on three pages I can see, please stop before it ruins more. if it can't distinguish this it should not be running this operation whatsoever. PARAKANYAA (talk) 02:22, 19 July 2025 (UTC)
Hi there - the Bot deleted two "url-status=live" tags from two Cite news templates without giving a reason (in this diff), while it was also adding a ProQuest "id=" tag. I've reinstated the "url-status" tags there, because both of the news articles are still live online. Is there any reason to remove those tags, or was it a mistake? Thank you, 101.98.24.129 (talk) 22:26, 19 July 2025 (UTC)
Lastly do not do this at the same time. Let one finish, or else the second one will either fail or blow away the first ones work. AManWithNoPlan (talk) 14:31, 18 July 2025 (UTC)
Greetings, bot runners. I'm curious about this edit. The edit summary refers to, among other changes, an "altered title," which, I assume, means the article's title. But I see no change there. Could you, please, enlighten, me? -The Gnome (talk) 17:38, 18 August 2025 (UTC)
The bot changed the quotation marks in an article title from the language-appropriate ones to the English ones. I do not want to get into an editing war with the bot and am wondering whether this is an oversight or part of some manual of style. I have not found anything in my limited search suggesting that language-appropriate quotation marks should not be used on the English Wikipedia in citation templates. Stefán Örvar Sigmundsson (talk) 19:46, 18 July 2025 (UTC)
Ten consecutive Citation bot edits over six months, only two minor visible changes
In [4], over ten edits, the bot changed the capitalization of a template name, added a class parameter to an arxiv cite, and added an issue parameter to a journal cite. This is the last part of a stretch of 38 edits among which only four were not Citation bot, and two of those four were Headbomb manually fixing something in a citation that Citation bot didn't do. Can we do something to throttle this never-ending churn? At a minimum these many consecutive edits could be prevented by a rule like: if the most recent edit was by Citation bot, and was less than three months ago, skip the article rather than making another edit. —David Eppstein (talk) 08:49, 1 March 2025 (UTC)
It is the underlying issue, if the JSTOR page wasn't taken down, the bot wouldn't keep this cycle of adding/re-adding it and it would just stay added. Headbomb {t · c · p · b}19:34, 1 March 2025 (UTC)
The underlying issue is that, whatever causes the bot to start cycling, there is no throttle against this cycling. The simple rule I suggested above could act as such a throttle. —David Eppstein (talk) 20:11, 1 March 2025 (UTC)
I disagree. My watchlist is overrun by Citation bot edits, on a long term basis (and edits that are too error-prone to ignore). I have been saying here for a long time now that a single run of Citation bot frequently makes big improvements, but that running the bot over and over and over and over on the same citations tends to amplify minor inaccuracies into total garbage, and it wastes the time and effort of everyone else who cares to check that the bot isn't garbaging the citations in the articles they care about. It needs throttling. If you don't want to hear that, then fine, but to me it's one of the biggest problems with the bot. From this point of view, this specific example was one of the better outcomes: the bot succeeded in being merely useless, instead of actively damaging, over its many edits. —David Eppstein (talk) 21:00, 1 March 2025 (UTC)
I was wondering why I almost never notice any Citation bot edits. Then I noticed that I filter them out of my watchlist. Johnjbarton (talk) 22:51, 1 March 2025 (UTC)
Not that. I don't know how it found this pmid but it's neither the right pmid for this reference nor a useful pmid to add to any article (because all it does is link to a broken doi).
I've reported the issue to Pubmed. First time doing so, so I don't know if I did it right. Support messages says to expect a reply within 2-5 business days. Headbomb {t · c · p · b}00:49, 7 September 2025 (UTC)
Even if the PMID was matched correctly, I'd skip it here though. It includes zero additional information not already in the citation. It's basically an empty link to the publisher's site, which doesn't really have any value for readers. –jacobolus(t)01:44, 7 September 2025 (UTC)
Agree. We should omit links whose only content is another link redundant with the ones we already include. That is often true of pmid, bibcode, and s2cid. (The ones that include an open copy of the article text, or someone else's review of the article, can be kept, of course.) —David Eppstein (talk) 02:18, 7 September 2025 (UTC)
My best guess is it sees that the publisher is Google and it assumes it's a mistake, like if a novice got the info via Google Search.
— W.andrea (talk) 11:39, 11 September 2025 (UTC)
The edit was correct. See Help:Citation Style 1 § Work and publisher: "The 'publisher' parameter should not be included for widely-known mainstream news sources, for major academic journals, or where it would be the same or mostly the same as the work." In this case, |publisher=[[Google]]is the same or mostly the same as|website=[[Google Fonts]].
Oh, OK! The bot should still explain that though, as opposed to just "Removed parameters." Something like "Removed redundant parameters" would be fine, or better yet "Removed redundant publisher". Thanks. — W.andrea (talk) 16:52, 11 September 2025 (UTC)
"Removed URL that duplicated identifier." except the removed URL links directly to the supplementary content being cited, while the DOI only goes to the document itself.
What should happen
URLs with more info than the DOI should not be removed.
