Jump to content

Template talk:Annotated link

Page contents not supported in other languages.
From Wikipedia, the free encyclopedia

Capital letters

[edit]

Is it possible to use this template in a way that doesn't violate MOS:CAPS, i.e. doesn't introduce unnecessary capital letters in words that aren't proper nouns, sentence starts, or acronyms? – Arms & Hearts (talk) 18:45, 15 July 2022 (UTC)[reply]

Can you give an example? AFIK, it just repeats the article name as given, then appends the short description from that article. I have discovered rather too many horrible SDs when using this template. --John Maynard Friedman (talk) 19:30, 15 July 2022 (UTC)[reply]
The example that I noticed was Great Filter#See also. The issue is that the short description is appended with a leading capital letter, when there's no MOS or common sense reason for there to be one. – Arms & Hearts (talk) 20:23, 15 July 2022 (UTC)[reply]
Not possible because the good folks over at WP:SDESC have decided that short descriptions on the English Wikipedia should all start with a capital. The template could force the first letter to lower case but that doesn't work because the first word in a significant number of short descriptions is a proper noun. ~Kvng (talk) 13:53, 19 July 2022 (UTC)[reply]
Thanks Kvng. Your last point hadn't occurred to me. That would put me in the "let's not use this in articles ever" camp, for what it's worth. – Arms & Hearts (talk) 18:32, 25 July 2022 (UTC)[reply]
A possible solution is to create short descriptions with a lowercase first word unless it's a proper noun. This is the Wikidata convention. It is easier to create a capital versions from this. There are still some confounding examples like, "iPhone accessory". ~Kvng (talk) 14:18, 26 July 2022 (UTC)[reply]
@Arms & Hearts: Are you questioning cases like Black swan theory – Theory of response to surprise events, that there is the second "theory" has a capital T? Since it is essentially a bullet point, surely that is a trivial technical breach that is completely inoffensive? --John Maynard Friedman (talk) 18:55, 26 July 2022 (UTC)[reply]
The MoS is there because it reflects the consensus of the community; while there are times when we might want to make exceptions to it, hardcoding them into templates that can't be context-sensitive is, if not exactly the end of the world, not quite "completely inoffensive" either, in my view. I'm not saying that I'm going to remove it from every article I see it in, but I'd probably object to it in any article I've worked closely on and certainly won't be adding it anywhere myself. – Arms & Hearts (talk) 19:42, 27 July 2022 (UTC)[reply]
Just noticed extsiance of this template. Text shouldn't be automatically capitalised. Eurohunter (talk) 17:13, 11 August 2022 (UTC)[reply]

The caps look stupid on the Relish artilce. 2404:4404:27B3:6500:C480:79C0:6BBA:1 (talk) 08:01, 22 August 2022 (UTC)[reply]

Since I'm apparently not the only one irritated by this, I wonder if anyone with appropriate levels of template clue could look into a fix? From what others have said, it looks as though the best option would be to add a case-determining parameter to this template, such that, for example, {{Annotated link|lc=y}} would change the case. That way the MoS issue could be averted without breaking things elsewhere or needing widespread changes to short descriptions. (I appreciate it's probably no one's top priority, but worth a shot.) – Arms & Hearts (talk) 20:31, 24 August 2022 (UTC)[reply]
Good idea! {{as of}} has an |lc= parameter. Not sure I have the chops for this but could learn. ~Kvng (talk) 16:11, 27 August 2022 (UTC)[reply]
But this is a bit of a sticking-plaster solution. Suppose you use the lc option to convert a description of "Television show" to "television show". Then six months later someone rewrites the description as "TV show" and that will be converted to "tV show".  Dr Greg  talk  16:34, 27 August 2022 (UTC)[reply]
That's definitely an issue, but isn't it in some sense an issue with the template itself rather than the proposed fix? It's already the case that someone could change a short description in ways that negatively impact the description used in a "see also" section or similar elsewhere. That could be via subtle vandalism on an unwatched article or just a case of a description that's suitable for transclusion elsewhere being changed to one less suitable. This would be just another case of that broader problem, which would be a reason to avoid using the template rather than to avoid making a change which would otherwise be an improvement. – Arms & Hearts (talk) 19:43, 1 September 2022 (UTC)[reply]
"The perfect is the enemy of the good". There are a huge number of articles with a See Also list of terse article names that are meaningless except to cognoscenti. Yes it would be great if all these were annotated by hand but it doesn't happen. {{anli}} achieves a good enough result for the rest. Serendipitous information discovery is a key objective of the project and if a tiny number of articles get trivial collateral damage in the process, too bad. Vandalism is a fact of life, hacking SDs is among the least of our problems. You are entirely at liberty to annotate the See Also of your favourite articles manually if you prefer. --John Maynard Friedman (talk) 00:07, 2 September 2022 (UTC)[reply]
Yeah, but liet's not shut down discussion of possible improvements. I think adding a |lc= to the template would be an improvement. A bigger improvement would be starting descriptions with lower case as is done as WikiData. Making that change at this point will produce pain. ~Kvng (talk) 17:02, 4 September 2022 (UTC)[reply]
If I gave that impression, it was entirely unintended. My objection is to those who seek to deprecate the whole template because of this less than perfect side effect. Clearly a change to the way that the template works (so as to remove the source of friction) would be the best outcome. --John Maynard Friedman (talk) 19:09, 4 September 2022 (UTC)[reply]
@Kvng: I don't suppose you got any further with thinking about this? There seems to be a consensus in favour of a change but I'm afraid it's beyond my know-how. – Arms & Hearts (talk) 19:49, 15 January 2023 (UTC)[reply]
@Arms & Hearts, It does look like we potentially have consensus to add an |lc= parameter. I don't have a lot of template experience but am interested in learning. I have just looked and have not found an example for how to lowercase the first letter of a string. Closest I found is how to lowercase the whole string. ~Kvng (talk) 21:33, 15 January 2023 (UTC)[reply]
@Kvng: If you were able to look into it that would be terrific. If not, I'm sure there are others who've posted on this talk page, and who've worked on this and similar templates, who'd be able to lend a hand (and who are welcome to weigh in here). No huge urgency of course. – Arms & Hearts (talk) 19:43, 20 January 2023 (UTC)[reply]
I will admit to scanning the discussion, but looking at the Great Filter#See also example given at the start and picking out the important part that short descs should start with a lowercase letter; this template is not at fault and shouldn't be responsible for tidying up other people's mess i.e. the short descs need fixing at the source. Fred Gandt · talk · contribs 06:41, 21 January 2023 (UTC)[reply]
Might I suggest a tracking/maintenance category so interested editors can find and fix the problems instead of hiding them? Yes, I think I might. Fred Gandt · talk · contribs 08:08, 21 January 2023 (UTC)[reply]
Per WP:SDFORMAT, short descriptions should begin with a capital letter. This makes sense in the context of the search (which is where I assume readers most often see them), but not in the context of this template. This is why the template, not the SDs themselves, is the issue. – Arms & Hearts (talk) 10:27, 21 January 2023 (UTC)[reply]
Ah ha! I had it the wrong way around and sit corrected; thank you. Yes, so, the concern would be the incorrect application of lowercasing. An initial uppercase letter is rarely going to be wrong, in terms other than those defined by the MOS; but incorrect application of lowercasing for the MOS might often create a mess (demonstrably the Preview step is frequently skipped). Perhaps a tracking category for cases where the |lc= has been applied? Fred Gandt · talk · contribs 16:31, 21 January 2023 (UTC)[reply]
Someone has documented a |desc_first_letter_case= parameter. Does this work? Should we use it? ~Kvng (talk) 20:48, 7 June 2023 (UTC)[reply]
I think the capital letters look fine. Not unlike how the first letter of an item in a bulleted list is capitalized. -- Beland (talk) 03:50, 6 July 2024 (UTC)[reply]

Should this template be removed?

[edit]

I find this template horrible.

Unfortunately, it is has started to become used on many See also sections of pages. But the annotation supplied is often not very suitable for all the different contexts that the See also links are used in. This means that the annotation is not very relevant much of the time, and it is hard to make quick edits of the text in the relevant context. I mean this goes against the principle of a wiki, where text can be continuously improved on in various contexts. Many people probably don't think of this when they use the "Annotated link" template, but in reality it creates a lot of more work for those who come afterwards and want to contribute to improve the text for the given context. Therefore, I propose that this template should be abandoned. It creates more headache than it is good. Sauer202 (talk) 07:36, 26 November 2022 (UTC)[reply]

Oppose. This template has has resolved the issue in so many articles of a cryptic list of "See also" topics. Article names are terse by design which means that they can be meaningless to readers who are not already familiar with their topics. A key attribute of Wikipedia is that it provides access to new information and broader perspectives. This template provides a quick way to address that problem, by exposing the WP:short descriptions. Of course it is true that the ridiculous 40 character limit means that the default SD is going to be inadequate in some cases – but it is better than no explanation at all, which is what would happen if your proposal were to be accepted.
As for your specific complaint, you are entirely at liberty to provide an explanation of a related topic that is more tailored to the the container article. You aren't obliged to use the template, nor are you obliged to retain it where it is already used provided that you supply a description that is better in the context than the default SD. --𝕁𝕄𝔽 (talk) 09:18, 26 November 2022 (UTC)[reply]
I agree with this analysis, and add that while a short description is expected to be verifiable through the page where it originates, a local annotation is content subject to the normal Wikipedia requirements for verifiability, and in many cases, probably not all, will need an in-line reference. I have seen many local annotations that were just plain wrong, with no reference cited. · · · Peter Southwood (talk): 17:39, 5 December 2024 (UTC)[reply]
Oppose I generally find that SA entries using this template are an improvement over the bare wikilinks they replace. Not perfect but better. Better is better. Let us know if you have an even better suggestion. If you install WP:SDHELPER, the ability to update a description is two clicks away. When you update a description you get two birds with one stone. ~Kvng (talk) 14:44, 29 November 2022 (UTC)[reply]
Oppose per better is better (chuckle) Fred Gandt · talk · contribs 19:00, 18 January 2023 (UTC)[reply]

Book titles and major artworks per MOS:NAT

[edit]

@Fred Gandt: I wonder if it might be easy to add an option disp=it[alic]? Or something similar? (combine with current quote=yes ?). Meanwhile I've added a simple example to the template doc. (Mona Lisa – Painting by Leonardo da Vinci). Not a show stopper, just a nice to have. 𝕁𝕄𝔽 (talk) 10:48, 3 April 2023 (UTC)[reply]

Adding parameters to do simple formatting changes seems more complicated than simply adding it to the second parameter in the manner you documented. Though one thing that would make things easier for editors would be to add automatic detection of {{DISPLAYTITLE}}, which would enable automatic application of not only italics but other special formatting. Though apparently many infoboxes like Template:Infobox book add this indirectly, so that might not be a straightforward programming task. -- Beland (talk) 04:02, 6 July 2024 (UTC)[reply]

Anomaly

[edit]

Hypothermia – Human body core temperature below 35 °C (95 °F) is not the expected result. Cheers, · · · Peter Southwood (talk): 11:39, 18 July 2024 (UTC)[reply]

Again, the SD needed to be fixed to follow the guidelines. – Jonesey95 (talk) 13:37, 18 July 2024 (UTC)[reply]
Jonesey95,:Which guideline was it not following? I see that you removed decimal parts, is that it? If so, why? Cheers, · · · Peter Southwood (talk): 16:07, 18 July 2024 (UTC)[reply]
The SD was attempting to use a template. Wikimarkup of any kind is not allowed in short descriptions. See WP:SDFORMAT. – Jonesey95 (talk) 16:08, 18 July 2024 (UTC)[reply]
Thanks, I managed to miss that somehow. Eyes not what they used to be, I'm afraid. Cheers, · · · Peter Southwood (talk): 09:34, 19 July 2024 (UTC)[reply]

Failure to return SD from infobox?

[edit]

Annotated link for Broughton Island (New South Wales) returns – 'island in Australia' and notice "Pages displaying wikidata descriptions as a fallback", but there is a short description, 'Protected area in New South Wales, Australia' presumably via the infobox ({{Infobox Australian place}}), which the template apparently does not find. · · · Peter Southwood (talk): 07:09, 21 August 2024 (UTC)[reply]

Not a bug (or a feature). Code to support falling back to Wikidata was added some time ago because there were so many articles without SDs, it was relatively easy to do [which is easy for me to say, as I didn't write it ] and it was a good return on investment. So you would have to request an enhancement showing that there are a goodly number of such cases. Meanwhile, it would be easier to give the article a proper SD.
Does {{Infobox Australian place}} provide SDs? Are there other infoboxes that do that? 𝕁𝕄𝔽 (talk) 11:56, 21 August 2024 (UTC)[reply]
There are a few such templates. A message box in this template's documentation states his template adds an automatically generated short description. If the automatic short description is not optimal, replace it by adding {{Short description}} at the top of the article.. older ≠ wiser 12:08, 21 August 2024 (UTC)[reply]
I do that, but the automatic short description is often good enough and in this case it is better than the Wikidata fallback. Cheers, · · · Peter Southwood (talk): 12:14, 21 August 2024 (UTC)[reply]
It is not falling back when there is no short description, it is usurping the short description generated by the infobox. To me this is a bug, but I don't know if the bug is in the infobox or the annotated link template, and I have no idea how to find out. Yes, {{Infobox Australian place}} provides SDs, and as far as I know others do too. I think it is quite common. Most of the time it is not a problem and the generated SD gets passed back by annotated link. (actually I have no idea how often this happens) · · · Peter Southwood (talk): 12:10, 21 August 2024 (UTC)[reply]
Are you aware that the whole purpose of short descriptions was originally to prevent the automated use of Wikidata descriptions in Wikipedia? I think that consensus probably still holds, as I have never seen an RFC to overturn it. Automatically generated short descriptions were also developed specifically to prevent automated use of Wikidata descriptions. It was quite a heated debate as WMF was forcing their use at the time and there was a lot of pushback. Not sure we want to reopen that can of worms, but you never know, consensus can change.· · · Peter Southwood (talk): 12:22, 21 August 2024 (UTC)[reply]
This limitation is documented on the template's documentation page: Some pages are assigned short descriptions by automatic methods; templates that generate short descriptions include infoboxes that use the data provided to its parameters to create a suitable short description, which may overrule other short descriptions that exist for or on the page. The module responsible for fetching the description is currently not able to detect or determine this type of dynamically created short description. – Jonesey95 (talk) 13:44, 21 August 2024 (UTC)[reply]
Thanks again, Jonesey95, I was not expecting such a severe limitation. I would think that it should not revert to the Wikidata description for these cases, and should rather leave out the short description altogether until consensus has been reached to overrule the decisions made when short descriptions were first applied, as that was a very widely discussed RFC. Overriding Wikipedia content with Wikidata content by an automated process seems contrary to accepted guidance. Cheers, · · · Peter Southwood (talk): 16:34, 21 August 2024 (UTC)[reply]
I think that whether other templates have started using code to automatically generate dynamic SDs does not have any direct relevance in how this template works. If those dynamic SDs are a violation of the RFC, it seems discussion about that should take place in some location more relevant for the templates that use such code. older ≠ wiser 17:19, 21 August 2024 (UTC)[reply]
It is not the generation of dynamic SDs that violates the RfC. They were specifically accepted at the time if I remember correctly. The constraint was the usual policy that infobox generated SDs are the reponsibility of the editor who codes the infobox, and must be possible to manually override, which is the case in all that I have inspected. Using a Wikidata description automatically is bringing content into Wikipedia from Wikidata without personally checking that it is appropriate. Using Wikidata description when a Wikipedia description exists is as far as I can make out, still a violation of that decision to prevent Wikidata descriptions from being automatically published in Wikipedia, and the responsibility for whether they are appropriate is specifically laid on the person who imports them. In this case, that would be the coder of the automated system. Therefore, as this is the talk page for that code, it seems to be the right place to discuss it. Cheers, · · · Peter Southwood (talk): 18:45, 21 August 2024 (UTC)[reply]
I'm confused. It is possible to override the SD generated by these templates. Or are you suggesting that this template should suppress the display of SD where is it derived from WikiData? While I did not follow the original RFC all that closely, I don't recall that the mandate was to prohibit the use of SDs from WikiData. older ≠ wiser 18:56, 21 August 2024 (UTC)[reply]
The mandate was that descriptions from Wikidata may be manually imported at the discretion of any editor, who takes personal responsibility that each imported Wikidata description is appropriate, but Wikidata descriptions must not be automatically imported without scrutiny. Cheers, · · · Peter Southwood (talk): 19:05, 21 August 2024 (UTC)[reply]
Thanks for clarifying. My recollection of the RFC details was poor. But if this template does fetch SDs from Wikidata, then something has changed in how it functions since Template talk:Annotated link/Archive 1#Documentation clarity this comment in 2021. Pinging Pppery if there has been some further changes made. older ≠ wiser 19:16, 21 August 2024 (UTC)[reply]
Hmm, I wonder if it was the module update made by Fred Gandt in 2023 that did this. See Template talk:Annotated link § Module above. older ≠ wiser 19:28, 21 August 2024 (UTC)[reply]
Yes, looks like that did it. * Pppery * it has begun... 21:20, 21 August 2024 (UTC)[reply]
Technically, Wikidata descriptions should not be displayed automatically even if there is no local short description, automatically replacing a good local short description with an unchecked Wikidata description is worse. Cheers, · · · Peter Southwood (talk): 19:10, 21 August 2024 (UTC)[reply]
So much speculation when the RFC is freely available to read! (Scroll to the top for the outcome of this question, which was "Show no description where the magic word [later implemented as the short description template] does not exist") Unless there has been a new RFC on the matter since 2018, the current consensus is that if there is no local short description, no short description should be displayed. I believe that means that this template should never pull text from Wikidata. – Jonesey95 (talk) 00:16, 22 August 2024 (UTC)[reply]
I concur. I am not aware of a later RfC on the matter. · · · Peter Southwood (talk): 04:38, 22 August 2024 (UTC)[reply]

I have found more instances of the same problem with other infoboxes, requiring changes to a perfectly adequate short description generated by the infobox to prevent a Wikidata description from being displayed.

It looks to me like this problem is an artifact of a bigger problem, which is that the template is returning content from Wikidata which is explicitly in contravention of the community-wide RfC linked above by Jonesy95, and it should not have been coded to do that in the first place. I see two legitimate options.

  1. The code is reverted to not returning Wikidata descriptions under any circumstances, and refraining from doing so in future, accepting the community decision as remaining valid.
  2. The code is reverted in the same way until a new RfC has been run, overturning previous consensus, and allowing the use of Wikidata descriptions where Wikipedia short descriptions are not available, and ensuring that where Wikipedia short descriptions exist in any form, that they are not usurped by Wikidata descriptions under any circumstances.

Either way, this template is used on thousands of pages, and on some pages, hundreds of times, so it must be kept efficient to avoid excessively long loading time and crashes, which detract from its intended function. It may be necessary or desirable to split Template:Annotated link into two versions to keep it usable, in which case the original version with minimal options and minimum overheads should keep the original title.

If anyone sees other acceptable options, please list them below. Cheers, · · · Peter Southwood (talk): 03:40, 25 August 2024 (UTC)[reply]

Minimum intervention would be to no-op the only= parameter (I've never seen it in the wild, though?) and hardcode only=explicit.
But the policy really does strike me as excessively "patch protecting". Are Wikidata entries really that oppressive? especially when the option to resolve any perceived issue by adding an SD. To my mind, this
  • Passenham – Village in Northamptonshire, England
is a lot more friendly than this
What ever happened to WP:Think of the reader?
I strongly advise that we do nothing. It is not broken, it doesn't need fixing. I have never seen any "long loading times and crashes". If we must satisfy the wikilawyers, then let's repeat the RFC first before spending any time on such low-impact coding work. 𝕁𝕄𝔽 (talk) 15:13, 25 August 2024 (UTC)[reply]
It is not relevant whether you have seen a thing or not, what is relevant is whether it is happening, and repeating the claim that it is not broken in the face of evidence that it is broken is unpersuasive. Look, and you will find. Doing nothing is not an acceptable option. Display of a Wikidata description without checking that it is appropriate when no Wikipedia short description exists is explicitly not permitted, and that is the least bad effect. The template is suppressing actual Wikipedia short descriptions and replacing them with unchecked Wikidata descriptions, which is totally against the letter and spirit of the recorded consensus. Denigrating the people who decided that unmoderated Wikidata descriptions are not acceptable as wikilawyers is unlikely to persuade them to change their minds. You are free to start the RfC as soon as you have formulated the appropriate question. Until then, the template should be reverted to a condition where it does not import unmoderated Wikidata descriptions at all, as that is the correct way to do it. I will not do this myself as my coding skills are not up to it and I consider myself involved. · · · Peter Southwood (talk): 17:34, 25 August 2024 (UTC)[reply]
Fred Gandt, As the person who has done the most recent edits to the template, I would like to give you the opportunity to make any comments or suggestions you think are appropriate at this point. You are probably in the best position to advise on what is possible, what is simple, and how we might withdraw from this position of non-conformity with established consensus with the least disruption. Cheers, · · · Peter Southwood (talk): 17:53, 25 August 2024 (UTC)[reply]
It seems to me that the purpose of this template is being forgotten in a nit-picky sprawl about what constitutes a proper short description; this template is for annotating links, by pulling a description that is short from a convenient and generally reliable location, with more ways to influence the result that I can be bothered to list (again; see docs), and importantly, is an entirely optional alternative to writing the annotations by hand. I think you're looking for a fight that doesn't exist and I have no intention of doing anything about it. Fred Gandt · talk · contribs 18:26, 25 August 2024 (UTC)[reply]
Fred, the issue apparently is that there are some infoboxes that generate a short description but don't use {{SD}} to do so and such SDs are invisible to this template. So it is not really about 'proper' short descriptions but rather about how they should be generated. Thus if we force only=explicit, then any invocation of this template for an article that has an infobox generated SD will show nothing. Which apparently is The Right Thing To Do. 𝕁𝕄𝔽 (talk) 19:20, 25 August 2024 (UTC)[reply]
Maybe Habst would like to have a crack at it? Fred Gandt · talk · contribs 21:33, 25 August 2024 (UTC)[reply]
Thanks for the ping, I made {{anbl}} for a specific use case but don't have experience with the actual extracting short descriptions part.
It looks like the problem lies in Module:GetShortDescription, which it seems like surprisingly (to me) uses a regex on the wikitext to find short descriptions, so it doesn't catch transcluded {{short description}} calls like the one provided by {{Infobox Australian place}}. Surely there is a better way than regex here to get the SD? Until I just discovered this, I thought it would have been fetched via some Mediawiki API method, like how you can use mw.title.getCurrentTitle() to get the page title without manually parsing the wikitext.
If such an API method doesn't exist, I think a Phabricator ticket should be created to add it. It seems like there is a method to fetch SD from Wikidata, but not one to fetch the Wikipedia SD used on the mobile site. Once the Phabricator ticket is implemented, Module:GetShortDescription should use that method. That would fix this issue. --Habst (talk) 22:05, 25 August 2024 (UTC)[reply]
Article short descriptions are available on the "Page information" page as the "Local description". Can this template just pull that item somehow instead of parsing the page text? If so, it would probably be a lot faster and simpler than this this template currently is. And Wikidata can be ignored easily, per the RFC. – Jonesey95 (talk) 04:20, 26 August 2024 (UTC)[reply]
Good suggestions. · · · Peter Southwood (talk): 05:44, 26 August 2024 (UTC)[reply]
When I wrote the original template in September 2018 I did not use Module:GetShortDescription as it did not exist (first edit 06:21, 20 January 2023‎), so it is possible to extract the short description by another method, probably without all of the module's bells and whistles, and probably significantly faster, but as displayed on mobile etc (I think this is from API). This worked acceptably for several years and tens of thousands of annotated links. I have no objection in principle to broadening its usefulness, as long as the default remains fast and does not conflict with consensus practice. If a user chooses to select an option which pulls in a description from off-wiki, that is on their head, and the source should be mentioned in the annotation per WP:Verifiable. Cheers, · · · Peter Southwood (talk): 08:18, 26 August 2024 (UTC)[reply]
The purpose of the template was originally to extract the "Wikipedia:Short description" of a page as defined by WMF and the magic word they coded for it, and to display it as an annotation to a link in a list in an article. It appeared to be quick and efficient. I know this because it was my intention when I created this template in the first place. If that purpose has changed, has it been explained somewhere? Who made the decision? Was it discussed somewhere? Since then the template has been used for the original purpose in a large number of articles, some of which, like index and outline articles, have a large number of annotated links, and some of which have become extremely slow to load, to the extent that I am having to split lists repeatedly to get acceptable load times. · · · Peter Southwood (talk): 04:53, 26 August 2024 (UTC)[reply]
Another useful side effect of the original template was that it drew the editor's attention to articles where there was no WP:Short description, which encourages the creation of short descriptions for those articles, thereby increasing the number of articlea with a short description. · · · Peter Southwood (talk): 05:04, 26 August 2024 (UTC)[reply]
@Pbsouthwood, I'm confused about how you were able to get the Wikipedia short description without using regex or textual analysis on the wikitext. In my research there is no API method to do that. Can you provide a minimum working example or hint? --Habst (talk) 15:21, 26 August 2024 (UTC)[reply]
Take a look at the old version of the code, Last edit by User:Pbsouthwood. I took it out and tested it and it still works. See User:Pbsouthwood/Annotated link/test. I also did a test using the current code which is about half the speed if I read the stats correctly, but the new code has more functions and checks so probably not a very fair comparison. Cheers, · · · Peter Southwood (talk): 08:25, 27 August 2024 (UTC)[reply]
PS: Don't ask me how it works, I have no idea. I probably got a suggestion from someone and messed around with it until it worked. Cheers, · · · Peter Southwood (talk): 08:29, 27 August 2024 (UTC)[reply]
I added one of the problem links to the test pages. The old code also does not appear to see the short descriptions generated by some infoboxes. · · · Peter Southwood (talk): 09:12, 27 August 2024 (UTC)[reply]
@Pbsouthwood, thanks. So I looked into it, and the old code uses Module:Template parameter value which ultimately (if you look at that module's source code) just uses a regex search as well. So it's still susceptible to the same problem as the OP, until we get an API method. --Habst (talk) 12:11, 27 August 2024 (UTC)[reply]
Yes, I have found that out by experiment too. (see subsection on statistics below). It affects about a quarter of all short descriptions, so worth fixing. Cheers, · · · Peter Southwood (talk): 12:57, 27 August 2024 (UTC)[reply]

When would Wikidata be appropriate

[edit]

<brainstorming a bit> I have been trying to think of a circumstance in which fallback to Wikidata description for an annotated link would be a good thing. So far none come to mind. If the Wikidata description is good enough for use as a short description, it should be imported. If it is the same, use the Wikipedia short description, If there is no Wikipedia short description, and the Wikidata description is not suitable, either omit or create one for the topic on Wikipedia. If there are other cases I have not thought of that would benefit from an imported Wikidata description, please list them here so I can understand why people would want to use them. From my researches it seems that it may be possible to pull a description from other Wikimedia projects using this template. It this correct? Is there a notice identifying the source for verification? Could we see an example? Using a description imported from Wikidata is relatively clearly not permitted, I would guess per WP:Verification, any source other than English Wikipedia would need a reference of some kind. I would also guess that an interproject link would be generally acceptable, and incidentally also provide attribution, though for such short statements attribution might not be necessary (probably should check this with WMF copyright lawyers).</brainstorming> Cheers, · · · Peter Southwood (talk): 10:26, 26 August 2024 (UTC)[reply]

I can see the logic of that. If I recall correctly, the reason to introduce the "fallback to Wikidata" was because it seemed that the vast majority of articles lacked any form of SD. I have an impression, no more, that the problem (while still substantial) is no longer quite so embarassingly terrible. As Peter said at 05:04 UTC, using {{AnnotatedListOfLinks}} on a SeeAlso has been a good prompt for me too, to fill in the blanks. And yes, also to correct any silly or verbose pseudo-SDs imported from Wikidata.
So maybe it is time to bite the bullet now and no-op all the Wikidata import code and see how loud the screams are? 𝕁𝕄𝔽 (talk) 12:06, 26 August 2024 (UTC)[reply]
@JMF, before this is done, the code must be changed to switch from the regex parser to an API-based parser to get the Wikipedia short description. I don't know if this is possible currently or requires a Phabricator ticket, but that would actually resolve the parent question while simply making a no-op would not. --Habst (talk) 15:22, 26 August 2024 (UTC)[reply]
Sorry, I was being lazy. I meant "take it out of use", without any idea of how significant or otherwise that would be. And before we get into the means, we have to have consensus on the ends. Though I don't see how else we can comply with the existing RFC (I really can't see any likelihood of a countervailing RFC succeeding). 𝕁𝕄𝔽 (talk) 15:59, 26 August 2024 (UTC)[reply]
I think we will find that a very large number of short descriptions are currently the type that are generated by infoboxes and therefore ignored in favour of Wikidata descriptions. My guess is in the hundreds of thousands, possibly millions, but it should be possible to find out. Cheers · · · Peter Southwood (talk): 07:47, 27 August 2024 (UTC)[reply]
Also agree about a new RfC having little chance of success.· · · Peter Southwood (talk): · · · Peter Southwood (talk): 08:15, 27 August 2024 (UTC)[reply]
Template Annotated link used to get short descriptions without a regex parser (apparently not), and the old code still works (which I think uses API). I ran a test on it a day or two ago, and it was also faster, but has no fancy functions and only three parameters. I have no idea how this could be integrated with all the additional functions Fred Gandt and possibly others added as I don't write Lua and am not great with templates even in Wikicode. Cheers, · · · Peter Southwood (talk): 08:12, 27 August 2024 (UTC)[reply]
I looked into this, and it seems like the old code used Module:Template parameter value which still uses regex, but just offloads it to this other module and so it's still susceptible to the same problem. We're still in need of an API method or Phabricator ticket to solve this issue. --Habst (talk) 12:13, 27 August 2024 (UTC)[reply]
Update: I can see a possible use for Wikidata descriptions as annotations when they are unsuitable for use as WP:Short descriptions, but are appropriate as annotations in the specific use cases. When this is done it must be a deliberate action of the editor, so definitively not a fallback, but an informed and conscious choice, and should include a link for attribution and verification. I do not have any examples, but can see the possibility, though it may generally be better to just manually annotate, with a normal citation for verifiabiity. This would still be importing content that could change without notice from outside of English Wikipedia, so it is a bit of a grey area, and may well meet with opposition, however we do allow links to Wiktionary, and some articles import data from Wikidata, so there is precedent. Cheers, · · · Peter Southwood (talk): 03:08, 28 August 2024 (UTC)[reply]
No, that would just create needless complications in the template code, potentially conflict with the RFC decision, and delay resolution of this issue. On the few occasions it might arise, it is just as easy for the editor to manually copy the wikidata info. --𝕁𝕄𝔽 (talk) 11:06, 28 August 2024 (UTC)[reply]
I agree that it would probably not be worth the effort, and may well slow the execution excessively, but I do not know that, so I mentioned it for completeness, usually a desirable thing when brainstorming. Whether it would conflict with the RfC decision is debatable, as it would be an informed choice by the editor, and referenced, but as you say, could cause delays for very little gain if any. Cheers, · · · Peter Southwood (talk): 11:53, 28 August 2024 (UTC)[reply]

Stats

[edit]


  • There are a total of 6,873,711 content pages in the main namespace, of which 360,409 are disambiguation pages.
    Out of 6,513,302 articles, 5,619,969 have a short description and 893,333 do not have a short description.
    5,619,969 / 6,513,302 = 86.284% complete (quite good?)
  • Of the 5,619,969 pages with a short description, About 4,243,936 pages include the template {{Short description}} These should(?) produce a short description accessible to {{Annotated link}} (Search string: insource:"Short description" hastemplate:"Short description"), which leaves 5,619,969-4,243,936 = 1,376,033 short descriptions produced by other means, possibly all inaccessible to Template:Short description. (about 25%, a significant fraction). It is possible that all or most of these are using Wikidata descriptions when used in Template:Annotated link. At this stage we do not have an absolute number.
  • There are currently 1,861 pages in category "Pages displaying wikidata descriptions as a fallback via Module:Annotated link", some of them will have multiple instances, so this is a minimum for instances. This number will grow as the use of Annotated links grows, and could peak at about 25% perhaps.

Redirects

[edit]

There are also about 2676 pages in category:Pages displaying short descriptions of redirect targets via Module:Annotated link. This should perhaps also not be by default. While some short descriptions of redirect targets are totally appropriate, such as {{R from alternative name}}s/synonyms, etc, others, like {{R from subtopic}}s may be quite confusing, and some will be plain wrong. It may be reasonably feasible to filter for good types of rcat, but it may be easier to leave the choice to the editor while this filtering feature is being considered, and I don't know all the rcats well enough to say which ones will always or even usually be OK. Also a lot of redirects have no rcats. The absence of an annotation is generally not a major problem as it is the original default condition. If anyone is sufficiently concerned by the lack, it is usually not a big job to fix at source by manually adding a suitable short description or just use a manual local annotation, and doing so will help expand the number of useful short descriptions in our articles and redirects. However, re-using short descriptions from R to synonym and the like does seem appropriate, as adding those to each of those redirects would be a right pain unless automated. Cheers, · · · Peter Southwood (talk): 08:14, 28 August 2024 (UTC)[reply]

Peter, can you dig into this more? I have certainly added SDs to redirect articles where it is a {{Redirect to section}} or {{Redirect to anchor}}, typically because the SD for the whole target article is inappropriate, unhelpful or both. So it is essential that the facility continues to exist and be supported. Or have I misunderstood your concern?
I suspect that there are very very few straight redirects with their own SDs (why would anyone bother?). If this template is used with an SD-less redirect, it derives it from the target article, which seems to make sense. --𝕁𝕄𝔽 (talk) 15:54, 28 August 2024 (UTC)[reply]
Adding a short description to a redirect where the SD of the redirect is sub-optimal works just like any other local short description, I do just what you say you do, and have not had problems. When there is no local short description at a redirect, the Annotated link template currently fetches the SD of the redirect target, which as you say is not always appropriate, and one must manually fix these by going to the redirect page and adding a suitable SD. It should be possible to filter by Rcat which SDs from the target are likely to be good, like R to synonym, R to short name, etc, and which are likely to be bad, like R to section and R from subtopic. Using such a filter to decide which target article SDs should be used and which should not would be a useful feature, though not essential. Problem is there are a lot of Rcats and I do not know which ones indicate that the target SD will be good or bad. Currently there appears to be no filter, and Annotated link just gets the target SD in all cases, good or bad, and one has to manually check if they are appropriate, then fix as needed. I hope this helps, but if not ask again. Cheers, · · · Peter Southwood (talk): 06:28, 29 August 2024 (UTC)[reply]
Time for another hand-waving reply , since I don't know the significance of this idea. Surely it would be better to have a bot that goes round collecting a list of such cases for attention (as happens in a number of cases I have seen, such as bare URLs), rather than build that function into this template? Especially given the concern about it load. 𝕁𝕄𝔽 (talk) 22:26, 29 August 2024 (UTC)[reply]
Maybe, but the bot should preferably fix the problem by adding the SD to the redirect. Human attention should not be wasted where it is not needed, we have better things to do. There is already a list of similar cases generated by the template, so it is partway there. A bot might eliminate the need for this function, which could streamline the template a bit, and might, as you suggest, reduce the load and speed up the performance. Cheers, · · · Peter Southwood (talk): 05:17, 30 August 2024 (UTC)[reply]
Do you use the CSS to make the Annotated link maintenance category warnings visible? It is quite useful.· · · Peter Southwood (talk): 06:35, 29 August 2024 (UTC)[reply]
No, I haven't come across that? Where do I find it? 𝕁𝕄𝔽 (talk) 22:27, 29 August 2024 (UTC)[reply]
It is conveniently listed in the top matter of the category pages (Category:Pages displaying short descriptions of redirect targets via Module:Annotated link and Category:Pages displaying wikidata descriptions as a fallback via Module:Annotated link), (kudos to Fred Gandt), and can just be copied and pasted into your CSS. You will suddenly be very aware of the cases when you encounter them in articles. · · · Peter Southwood (talk): 05:20, 30 August 2024 (UTC)[reply]
I was about to suggest another maintenance category for ANLI links to redirect-to-section or redirect-to-anchor articles but I suspect it would be huge. . 𝕁𝕄𝔽 (talk) 19:23, 30 August 2024 (UTC)[reply]
It would be huge, and I don't see the point, as there is no problem with R to section and R to anchor that needs maintenance that would remove those links from the category, so the maintenance category would just keep getting bigger and more useless instead of smaller. As a general principle, performing the targeted maintenance should automatically remove an article from a maintenance category. Redirects without a short description would need a short description to remove them from such a category, but there are a lot of redirects which should never be used in an annotated link, like from misspellings, foreign languages etc, that do not need a short description, and they would waste editor time for no useful effect. However there may be some maintenance categories we have not thought of yet that would be useful, so don't stop thinking yet.
A category for redirects to a section or anchor, or from a subtopic which also do not have a short description would be useful, but may require a bot to populate and update. In this case most of the redirects should have a short description, and adding one should remove the redirect from the category eventually. Cheers, · · · Peter Southwood (talk): 07:08, 31 August 2024 (UTC)[reply]
There is already at least one other maintenance category, for short descriptions without a space, which basically means they are one word or hyphenated. I will try to find it again and leave a link here. There may be others I have not found yet. They may even be listed somewhere, but if so I don't know where. There is so much on Wikipedia that one only discovers by luck. · · · Peter Southwood (talk): 07:25, 31 August 2024 (UTC)[reply]
Category:Pages displaying short descriptions with no spaces via Module:Annotated link, and they are listed in Category:Wikipedia maintenance. I don't know if I found them all, and some are more useful than others. · · · Peter Southwood (talk): 08:06, 31 August 2024 (UTC)[reply]
I have also found them listed at Template:Annotated link in the documentation, which is the logical place to look once one has assumed they may exist. Hindsight 20/20 again. Cheers, · · · Peter Southwood (talk): 08:19, 31 August 2024 (UTC)[reply]

Very poor annotations when annotating redirects

[edit]

I have just fixed two cases at Sanewashing where this template provided incorrect descriptions for Prebunking and Steelmanning, both of which redirect to a page with a nearly opposite meaning:

  • Prebunking – The process of debunking lies, tactics or sources before they strike
  • Steelmanning – The opposite of a straw man argument

While I can see the value of not having to write these annotations manually it seems to me like this is quite a dangerous template in its current form. Would it perhaps be reasonable to display no annotation at all when redirects are linked? Even checking the output manually at the time of an edit adding this template looks to me like it wouldn't be sufficient, since an article might be redirected at any time. I'm not sure how useful the tracking category is given that it is 2900 pages strong (and if the intent is to ignore correct cases it can only grow).

I see there has been discussion of this issue above but it doesn't seem like that was ever resolved. Tollens (talk) 06:56, 29 November 2024 (UTC)[reply]

So if someone searches for "Prebunking", the short description that they will see is "a nearly opposite meaning". So in fact this template is doing us all a favour my making the error obvious. Don't shoot the messenger, correct the message. Redirects should have SDs too.
But if someone has time, maybe it is possible to emulate the code that detects and flags 'fallback' SDs taken from Wikidata? --𝕁𝕄𝔽 (talk) 12:36, 29 November 2024 (UTC)[reply]
Sorry – I'm not sure where else any reader would be shown the short description of a redirect target without simultaneously being shown the title of that target? As far as I'm aware, the search tools that display short descriptions don't place them next to the title of the redirect being searched, but rather its target. (If this template were to replace the link provided with a link to the target in that same way, I would see no problem there either, but I wouldn't have assumed anyone would want that.) Certainly if this is an issue that's broader in scope than this one template I agree that this template isn't the problem, but I wasn't under the impression that was the case. Tollens (talk) 17:06, 29 November 2024 (UTC)[reply]
I think only a few cases of {{annotated link}} are at issue here, not redirects or short descriptions in general. It appears that while most of the time, redirects point to an article whose topic matches the redirect, like Trevor Tahiem Smith, Jr. (American rapper (born 1972)), occasionally, a redirect points to an article that is about something quite different from what the redirect's title describes. We may need to come up with a solution for that case, which I am guessing is rare. Maybe adding short descriptions to some redirects would work? I was unable to find redirects with short descriptions, but maybe my search was not right. – Jonesey95 (talk) 22:14, 29 November 2024 (UTC)[reply]
There is Category:Redirects with short description – adding them to the incorrect cases does fix the issue. My concern is primarily that there seems to be no way to tell where those incorrect cases are. The tracking category Category:Pages displaying short descriptions of redirect targets via Module:Annotated link looks like it was created to try to find these cases for review, but as above, it isn't useful given that correct cases aren't removed from the category (and even if they were, they may become incorrect at any time, with no way to detect that change). Tollens (talk) 22:49, 29 November 2024 (UTC)[reply]
To clarify, I have absolutely nothing against redirects nor short descriptions. My original suggestion above was just to have this template return an un-annotated link when a redirect is being linked, not to do away with redirects, short descriptions, or this template – my apologies if I communicated poorly. Tollens (talk) 22:52, 29 November 2024 (UTC)[reply]
I agree that this appears to be a tricky problem to locate. Maybe you could ask for help at Wikipedia:Request a query, but I think that analysis of what you want would require someone looking at a database dump rather than constructing a database query. I think what you want is something like "Find and list all transclusions of {{Annotated link}} contained in pages found in Category:Pages displaying short descriptions of redirect targets via Module:Annotated link, limited to transclusions of {{Annotated link}} for which the page listed in |1= is a redirect page. Create a table of entries, with the redirect page title in the first column and the output of {{Annotated link}} (or the short description of the redirect page's link target, which can be pulled from the target page's Page Information) in the second column, and the transcluding page title in the third column." – Jonesey95 (talk) 23:50, 29 November 2024 (UTC)[reply]
It happens, but I don't know how often or what percentage of redirects. I think it mainly happens when a redirect is to a subtopic or related topic which is rather different to the target topic, and therefore should have its own short description, and should not use the target's short description. I don't know how the target short description is fetched, but it may be a "feature" of Template:Annotated link, which was not well thought through before implementation. Cheers, · · · Peter Southwood (talk): 17:31, 5 December 2024 (UTC)[reply]

Thinking about it further overnight, I have reluctantly come round to accept the validity of Tollens's argument. If the redirect article does not have its own SD, we should definitely not use the SD from the target article if the redirect is to a section or an anchor, because the probability of error or misdirection is just too high. If that means (and I suspect for practical reasons it does) that this template stops the practice of looking ahead from all no-SD redirects, then that's how it has to be. --𝕁𝕄𝔽 (talk) 13:57, 30 November 2024 (UTC)[reply]

I concur. · · · Peter Southwood (talk): 17:32, 5 December 2024 (UTC)[reply]
@Fred Gandt:, can you be persuaded to give some time to making this adjustment, please? --𝕁𝕄𝔽 (talk) 18:04, 5 December 2024 (UTC)[reply]
Agreed, The correct way to manage this is to provide an appropriate short description to redirects to related topics or subtopics as they should be different to the short description of the target article. This has worked well in practice in my limited experience. See Index of underwater diving for an extensive list of examples. · · · Peter Southwood (talk): 06:48, 6 December 2024 (UTC)[reply]

Returning warning when there appears to be a short description on the page

[edit]

California Federal Savings and Loan Association v. Guerra appears to have an infobox generated short description, but Annotated link returns a warning that it is a Wikidata description used as a fallback. The short description in this case is identical to the Wikidata description, so it is not clear what the template is actually returning.

  • California Federal Savings and Loan Association v. Guerra – 1987 United States Supreme Court case (California Federal Savings and Loan Association v. Guerra – 1987 United States Supreme Court case Pages displaying wikidata descriptions as a fallback for those who do not have the CSS to see the warning)

Whatever is happening is wrong, but it is not clear whether the fault is in the infobox or the annotated link template. Cheers · · · Peter Southwood (talk): 04:32, 8 December 2024 (UTC)[reply]

At Cleveland Board of Education v. LaFleur the Wikidata description differs from the short description, and is categorised as such,

  • Category:Articles with short description
  • Category:Short description is different from Wikidata

Yet we get:

  • Cleveland Board of Education v. LaFleur – United States Supreme Court case (Cleveland Board of Education v. LaFleur – United States Supreme Court case Pages displaying wikidata descriptions as a fallback for those who do not have the CSS to see the warning)

In this example the annotation displayed is the wikidata description, so it looks like the short description produced by the infobox is not visible to Template:Annotated link. I think this is a common problem with infobox derived short descriptions. Cheers, · · · Peter Southwood (talk): 04:50, 8 December 2024 (UTC)[reply]

The infobox does add a default short description within the included code (right at the top). It appears that the Template:Annotated link can not see this. · · · Peter Southwood (talk): 05:58, 8 December 2024 (UTC)[reply]

Without wishing to detract from your main point (that the template as it stands does not have the capability to read infobox-generated SDs and wont necessary code), we should acknowledge that the red warnings are not displayed to most readers. That doesn't mean that there is no problem – there is – but it is not as "in your face" as it might first appear to the uninitiated.
The problem is not generally a critical one: specific cases can be resolved by either adding explicit SDs to the egregious cases or simply not using this template to reference those articles (writing a description in each case instead). 𝕁𝕄𝔽 (talk) 09:19, 8 December 2024 (UTC) revised to strike out extraneous phrase generated by to the vagaries of the mobile editing interface. --𝕁𝕄𝔽 (talk) 10:22, 8 December 2024 (UTC)[reply]
Agreed, it is not urgent. The biggest problem is actually that many acceptable short descriptions are not available, and wikidata descriptions are being used contrary to general policy on short descriptions. As a local short description may be challenged for verifiability (I have found a lot of erroneous and suspect local annotations), which would call for a citation, I prefer to add explicit SDs. The red warnings are actually useful as they inform the reader that the annotation may well be inaccurate. This affects a very large number of articles, so a proper fix would be best. Cheers, · · · Peter Southwood (talk): 09:51, 8 December 2024 (UTC)[reply]
Retrieved from "https://wiki.riteme.site/w/index.php?title=Template_talk:Annotated_link&oldid=1261941348"