Wikipedia talk:Short description/Archive 6

Archive 1

←

Archive 4

Archive 5

Archive 6

Archive 7

Archive 8

→

Archive 10

40 chars

Can we please make it emit a warning message and/or a maintenance category when this limit is exceeded? ―cobaltcigs 06:37, 15 December 2019 (UTC)

This page says "A target of 40 characters has been suggested, but this can be exceeded when necessary." Emitting a warning message is usually limited to clear-cut cases where there is an error. – Jonesey95 (talk) 14:48, 15 December 2019 (UTC)

Some way to find absurdly verbose ones like this would be good. The two strategies I'm aware of are (a) parse database dumps, and (b) watchlist several thousand pages, sit back, and wait. Surely a maintenance category Pages with long "short description" or somesuch would be more efficient. On the supply side, adding a char counter and a warning box ("Are you sure it needs to be [that long]?" [OK]/[Cancel]) to the gadget/script that many of these folks are using would also be helpful. ―cobaltcigs 15:25, 17 December 2019 (UTC)

I expect that it would be technically possible to put articles in a maintenance category, and display a warning message if desirable, if their short descriptions exceed a certain number of characters, but it would be unwise to start with a limit of 40 characters. The description you linked to was 108 characters, which is certainly longer than 40, but I don't have a way to know whether it is still reasonable. Can you please link to the discussion that determined that 40 characters was the right number for a suggested target? When designing a check on the length of the short description, it will be helpful to understand the technical, usability, and other concerns that led to the choice of that number. – Jonesey95 (talk) 17:10, 17 December 2019 (UTC)

I've gone ahead and added a category for pages with short descriptions longer then 100 characters at Category:Articles with long short description.It currently does nothing else but should make it easier for people to find. It will also be useful to see the scope of the issue. ‑‑Trialpears (talk) 18:17, 17 December 2019 (UTC)

What is the rationale behind the choice of 100 characters? Was there a previous discussion? I think that we should be more circumspect about making changes to a template that is used in nearly two million articles. – Jonesey95 (talk) 02:31, 18 December 2019 (UTC)

I commented at shortness of short descriptions on Pbsouthwood's talk. Jonesey95 is correct in that a template with such a large number of transclusions should not be boldly edited—propose a change and test it in the sandbox, wait 48 hours, then, if no objections, make the change. Johnuniq (talk) 03:07, 18 December 2019 (UTC)

100 characters was chosen mostly arbitrarily as where I expect that there will be basically no descriptions that couldn't easily be cut down without losing much information and where consensus to eventually add a warning would be relatively easy to obtain. I've seen a few descriptions at 80-90 characters where making it more concise would be difficult due to names or technical terms. Feel free to change it if you want. Regarding the edit I was working under the presumption that adding a tracking category not affecting the templates appearance after testing it in the sandbox could be made unilaterally especially since it falls under WP:TPE#Changes that can almost always be made unilaterally. It involved no risk and performance is no concern. Since you two have expressed concerns about it I will not implement similar edits for templates with very high transclusion counts unilaterally. ‑‑Trialpears (talk) 03:45, 18 December 2019 (UTC)

Performance is a concern for something like this. The norm is as I said. See "Please consider discussing changes on the talk page before implementing them" in the big box at the top of Template:Short description. In typical Wikipedian guideline style, that is stated very gently but it is actually obvious and should be followed unless something urgent is needed. Thanks for your agreement. Johnuniq (talk) 04:16, 18 December 2019 (UTC)

I like the idea of a category for long short-descriptions, as it may encourage people to consider the options and in some cases a better, shorter description may be provided. I don't consider it an urgent problem. A long short description is usually better than no short description, but sometimes it is beyond the competence of the ordinary editor to provide a useful and concise short description, in which case I would prefer useful. Johnuniq, how is performance a concern here? I ask out of interest. Cheers, · · · Peter Southwood ^(talk): 07:10, 18 December 2019 (UTC)

When a template or module is edited, every page using it is parsed again to update the cache of HTML shown to readers. That happens slowly, and in principle, editing a template ten times in a day might have only a little more overhead than editing it once. However, the overhead is so massive that it makes sense to wait to be sure that only a single edit will be needed. Johnuniq (talk) 07:17, 18 December 2019 (UTC)

I'd estimate that for about 50% of articles, a good short description could be formed by choosing, from the bottom of an article, the first category matching one of several patterns tested in some descending order of precedence, then converting said category's name from plural to singular. Perhaps there is a tool that already does this. That result will certainly be more uniform and concise than choosing, from the top of an article, the first whole (often run-on) sentence and subtracting '''{{PAGENAME}}''' [is/was/are/were] [a/an/the] from the beginning of it. The latter seems to be a popular strategy, and may or may not be an automatic effect of existing tool(s). ―cobaltcigs 18:50, 19 December 2019 (UTC)

A limit of 100 would find the worst of the worst, but it risks insinuating that 99 is acceptable, whereas it never will be. I see that the "Population genetics" example discussed in October had an ellipsis in place of characters 47–131. The naïve assumption would be that anything up to 46 is safe, but it's not that simple because we can't predict what distribution of wide vs. narrow letters are to be rendered in a variable-width font. Hence rounding down to 40. I don't have a link to any previous discussion, but I'm certain it went just like this. If we wanted to make this closer to rocket science, we could use a "weighted strlen" function that attempts to measure pixels, according to a mapping of each character to its pre-measured width in some particular but well-researched font and size (whatever the greatest number of end users see). Then compare this estimated total width against some value representing the usable portion of typical horizontal screen space (adjusted down by some percentage, for extra surety). Then determine error/category status based on the latter comparison, rather than number of characters. Otherwise, let's just commit to using as few words as possible. ―cobaltcigs 00:30, 19 December 2019 (UTC)

The linked discussion shows just two examples copied from just two example phone screens, which is quite a small sample size on which to base any choices about a template that is used on nearly two million articles. Interestingly, the second example description displays 90 characters before the ellipsis, so we can reasonably assume that a 90-character short description is fine for at least some viewers of short descriptions. So let's start with the "worst of the worst" (i.e. 100+ characters) and see if we can keep up with that population. Right now, the population of Category:Articles with long short description is approximately 2,137 articles, and it will presumably increase as the job queue catches up with null edits of a couple million pages. If we manage to catch up with that backlog, we can discuss reducing the character count to something lower in order to make more work for helpful gnomes and make short descriptions more useful for more readers. In the meantime, maybe the WMF programmers will make a helpful change of some sort that mitigates the current problems caused by long short descriptions. – Jonesey95 (talk) 00:48, 19 December 2019 (UTC)

There is not really anything that developers can do about displaying long short descriptions. That is easy to see if you try using the phone app to view a few topics of interest. It (almost) always starts with a search for the topic. The search displays a list of possible titles with the short description on the next line in a gray font. However, each title and each description is truncated to fit the width of the screen. It has to do that in order to fit a reasonable number of alternative matches on the screen. The description is intended to disambiguate titles, for example, by telling you something vital about the "John Smith" you are searching for so you can see if it is the one you are after. Johnuniq (talk) 22:02, 19 December 2019 (UTC)

Screen shot of mobile search for "Boston" (Firefox 71 for Mac; iPhone SE Safari looks the same). No truncation of short descriptions.

I am not experiencing this truncation. I just pulled up en.m.wikipedia.org in Safari on my iPhone SE (a small-screened iPhone). I clicked the search magnifying glass and typed "Boston" (without the quote marks). The first five suggestions showed as a two-column table on my screen, with a small, square image on the left of each row, and a cell on the right side of each row containing the article name (Boston, Boston Marathon bombing, Boston Red Sox, etc.) and a short description. None of the short descriptions ("State capital of Massachusetts, U.S.", "Deadly explosions during the 2013 Boston Marathon, and subsequent shooting and manhunt", "Baseball team and Major League Baseball franchise in Boston, Massachusetts, United States", respectively) were truncated. The second and third short descriptions are 86 and 89 characters; again, neither was truncated.

I believe people in this discussion when they say that some people are experiencing truncation of very long short descriptions, but I continue to question the basis for the choice of 40 characters as a reasonable limit for viewers using modern technology. Again, I wonder aloud how 40 characters was chosen as guidance for this page and whether any sort of actual usability testing or browser/device population survey data went into the decision to choose this number. So far, this discussion has yielded a link to a brief user talk page discussion with just two participants. At the risk of annoying the heck out of all of you in order to save en.WP gnomes a bunch of work and heartache, I will repeat myself: Can you please link to the discussion that determined that 40 characters was the right number for a suggested target? When designing a check on the length of the short description, it will be helpful to understand the technical, usability, and other concerns that led to the choice of that number. – Jonesey95 (talk) 00:35, 20 December 2019 (UTC)

The truncation occurs in the app. I don't think there was any formal discussion about a character limit, or pretty well anything else such as whether to start with a capital letter. There is some discussion in Archive 5 (search for "40 char") but it is essentially what we are saying here. Johnuniq (talk) 00:59, 20 December 2019 (UTC)

The logical follow-up, then is "What percentage of viewers who see short descriptions are using an app?" (and which app(s), specifically?) or more fundamentally, "What percentage of viewers who see short descriptions are seeing truncation?" The answer to that question will influence how much we should care about "long" short descriptions, and what "long" means. I imagine that we would all agree that 500 characters is just too long, but I wonder if 40, or even 80, characters is too short for a limit, especially if a small percentage of viewers are seeing truncation. I have edited en.WP for ten years and use the mobile interface occasionally, but it works fine in Safari, so I have never bothered with the app. I know better than to think of myself as a representative sample of anything, however, so some real data would be lovely.

Thanks for the link to the archive. In /Archive 5#Character limit, I see links to multiple app screenshots that show multi-line short descriptions with no truncation. I have yet to see an app screenshot showing truncation (though, as I said, I believe that it is happening to at least one viewer.) In an earlier section, I see an editor explain that 40 characters is a suggestion which is not always reasonably practicable while remaining useful. Sometimes no short description is necessary, other times 80 or 100 characters are needed to make any sense. I tend to agree, based on some of the 80-character short descriptions I have seen. – Jonesey95 (talk) 02:08, 20 December 2019 (UTC)

Truncation occurs when you search for a term related to what you want in the app. The short description is shown in full when you view an article in the app, but it's pretty useless there because you are looking at the actual article and should be able to quickly work out whether the topic is the one you want. The purpose of the short description is to disambiguate a list of terms when searching in the app. Someone, possibly at WP:VPT, would know where to find a list of how many articles are viewed with an app versus other techniques. Johnuniq (talk) 02:25, 20 December 2019 (UTC)

(edit conflict) It also occur at the bottom of the mobile site where 3 related articles are displayed. Testing at my iPhone 6S using Chrome showed truncation at just 42 characters due to it being displayed on one line with a picture. There seems to be space to display it on two rows significantly increasing the possible length without truncation. When using the iOS app the cutoff is around 45 with no possibility of making it multi-lined. I don't think the app is such a big deal however since only a few percent of readers use it. ‑‑Trialpears (talk) 02:30, 20 December 2019 (UTC)

Daily siteviews on the mobile app are between 4 and 5 million, compared with around 100 million on desktop and 150 million on mobile web. See Siteviews Analysis for details. That is indeed only a few percent of readers, but I'm not convinced we can ignore 4,000,000+ views per day. --RexxS (talk) 03:06, 20 December 2019 (UTC)

I've probably shortened about 100 of these and been reverted on over 20. Meanwhile the count has grown to 3,347. I think this is because people don't realize these are supposed to function as disambiguation labels, not a Cliff's Notes summary. ―cobaltcigs 09:56, 26 December 2019 (UTC)

Inevitably long short descriptions that are useless when 40 characters

Summary of points agreed in this discussion

Original aims (by SUM1) To get the community's view on: how lenient they are willing to be with the short description length "target" in certain scenarios how important the help page's emphasis on short description distinguishability over length is whether certain topic areas (e.g., genetic syndromes, genes) should accept longer short descriptions due to the difficulty in distinguishing between articles in that area a proposed diagram and whether it could be included on the help page to better steer new editors in the right direction when it comes to short description length whether policy or guideline should be drafted for short descriptions, as it exists for article titles, since they're being added en masse and conflicts/misinterpretations are constantly arising Points with support Feel free to add your username if you've made your support or disagreement clear. The diagram proposed, with tweaks if necessary, could be used to represent the length rule on the short descriptions help page. For:0000Jonesey95, TheDJ, Peter Southwood, Trialpears, Newslinger, SUM1 Against: On the help page, "The short description should focus on distinguishing the subject from similar ones rather than precisely defining it" should be changed to, "The short description should focus on distinguishing the subject from similarly named ones rather than precisely defining it". For:0000Galobtter, SUM1 Against: For articles that are both similarly named and similar in nature, the length limit does not need to be as strict in order to accommodate distinguishability. (Examples: Gerstmann–Sträussler–Scheinker syndrome vs. Gerstmann syndrome, Klippel–Feil syndrome vs. Klippel–Trénaunay syndrome, Lujan–Fryns syndrome vs. Fryns syndrome, BRCA1 vs. BRCA2) For:0000Trialpears, SUM1 Against: The above rule should be added to the help page. For:0000SUM1 Against: A whole new help page section for short description length and where and how it should be applied should be drafted. For:0000Trialpears, Peter Southwood, SUM1 Against: Points not yet agreed There is an absolute cutoff limit where no short description of that length is acceptable (not including hypothetical rare technical exceptions). For:0000 Against: Points contended For articles that are similar in nature but not similarly named, the length limit does not need to be as strict in order to accommodate distinguishability. For:0000 Against: TheDJ, Johnuniq, Galobtter

The Wikipedia page on short descriptions states that, "A target of 40 characters has been suggested, but this can be exceeded when necessary", and the WikiProject page on short descriptions states, "Preferably limit to about 40 characters, but function is important. A slightly overlong description is better than a wrong, misleading or useless one." But the Wikipedia page also states, "The short description should focus on distinguishing the subject from similar ones rather than precisely defining it", and that it may be used "as a disambiguator in searches".

These guidelines work well together in most cases, but they come into conflict when it comes to genetic syndromes (and many other cases). Almost all single-gene conditions could be described as "Autosomal dominant genetic condition" or "Autosomal recessive genetic condition", and those are 36 and 37 characters each, but they're useless for disambiguation. If we prioritised the 40 character target, almost all genetic syndrome articles would end up as either "Autosomal dominant genetic condition" or "Autosomal recessive genetic condition". Not only is the mode of inheritance a minor, non-distinguishing feature of the condition usually, but half of the text is redundant since the viewers are likely to know a "syndrome" is, most of the time, a genetic condition. The short description would need different or additional distinguishing information to be of value.

However, many syndromes cannot be easily described in few characters, because they involve multiple systems, and it's the combination of features that distinguishes them from related syndromes. I encountered this problem while trying to develop a description for some of my syndrome articles, including Okamoto syndrome, which is characterised by "hydronephrosis, neurological impairment, heart defects and facial features", in the shortest way I could put it. To give a comparison, CDK13-related disorder is characterised by "neurological impairment, heart defects and facial features", and Strømme syndrome is characterised by "intestinal atresia, eye abnormalities and microcephaly". Those would bring their total short description lengths to between 80 and 120 characters. I'm aware that a compromise could easily be made by merely stating "multi-system condition", but, even by emphasis, that would only distinguish it from maybe 10–30% of genetic syndromes, and that isn't very helpful and doesn't address what seems to be the main concern of the short description Wikipedia pages, distinguishability. It therefore doesn't eliminate the question at hand.

The question I put to this talk page is just how important is the 40-character limit, and where does the spectrum end where long short descriptions are no longer acceptable? Should it depend on the topic? I.e., should longer descriptions be tolerated more when it comes to genetic syndromes but less when it comes to topics with more easily distinguishable articles? Is there an absolute cutoff limit where no short description of that length is acceptable, and if so does this have any exceptions? Is the tolerance between 40 and say 100 characters linear or weighted, i.e., are 70-character descriptions closer in likelihood to 40 or 100 of being accepted? I basically want this to be put to bed once and for all, because it has to happen sooner or later, and people are already adding short descriptions en masse. Hopefully this can be the start drafting of policy or guideline.

I see it like this:

Feel free to comment on or suggest edits to that entirely hypothetical diagram.

Other posters on this talk page have already demonstrated that even 100-character-long short descriptions do not cause any overflow or usability issues; they just become distractingly long. SUM1 (talk) 21:32, 30 December 2019 (UTC)

I think that is a reasonable spectrum. Someone could propose summary text, or use of this figure, on the page in lieu of the minimalist guidance that we have now. Given that "long" short descriptions appear to affect a low-single-digit percentage of page views (according to the data given above), I think we should have a reasonable tolerance for short descriptions that are longer than 100 characters.

As also discussed above, there appears to be room in the WP user interface (UI) for an additional line of short description in places where it is currently cut off. All we have to do is persuade the WMF programmers to tweak the UI. – Jonesey95 (talk) 04:04, 31 December 2019 (UTC)

I agree. Sounds good. SUM1 (talk) 18:24, 31 December 2019 (UTC)

SUM1, i see no problem with that diagram. I call this: "the further down the line, the less likely people are to see and/or read whatever you have written" (the less useful it becomes).

I think that some things are hard to disambiguate no matter what. That's just the reality of collecting all knowledge. But the name generally disambiguates between things that are similiar (part of the same field). Descriptions mostly disambiguate between different things that have (accidentally) similar titles.

For me personally for instance 'autosomal' is completely useless information in the description. You might as well have said 'kluktuk genetic disorders' and it would have made just as much sense. For me the description should disambiguate for me when I type "okamoto" and it shows me results for Okamoto, okamoto syndrome, Okamoto Station (Tochigi) and Okamoto Station (Hyōgo), it should not disambiguate between syndromes, unless there are two Okamoto syndromes. Wikipedians keep thinking it describes the article/topic, but it doesn't that is what the article is for. —TheDJ (talk • contribs) 09:48, 31 December 2019 (UTC)

@TheDJ: I agree about people being less likely to read longer ones. However, about the name disambiguating, people aren't to know how "Klippel–Feil syndrome" is different from "Lujan–Fryns syndrome", "Hallerman–Streiff syndrome" or "Schwartz–Jampel syndrome" without some sort of differentiating descriptor. And that's where the struggle arises with genetic syndromes, what with them being so unwieldy in distinguishing features. About omitting "autosomal", that unfortunately would be "wrong, misleading", because autosomal dominant or recessive is not the same as X-linked dominant or recessive. But about distinguishing between things named "Okamoto", that's one of the ideas that need to be resolved. A similar argument could be made for things named "syndrome", and that happens to be part of my argument. Currently, the information page says that the short description "should focus on distinguishing the subject from similar ones", so I believe I interpreted that as similar articles rather than just articles with part of the same name. If misinterpretations like mine are so common, then the information page needs to be updated.

To be clear, I myself would not feel comfortable adding those 80–120-character descriptions, but my aim here was to get an idea of what other people think and how lenient they are willing to be with certain topics. I've seen extremely long user-added short descriptions in my time. Even the one of a major article like the One World Trade Center, as of this post, is 90 characters long. SUM1 (talk) 18:24, 31 December 2019 (UTC)

You know that colleague at the Christmas party that keeps talking about this fancy 8000 dollar carbon fiber, windtunnel tested, laser designed, built from argentinian magnesium and recycled plastic sports bike ? And then everyone goes "right, so... its a bike?" That's what the short description should be. A description for outsiders, not for insiders ;) The most dumbed down, non-precise description you can get away with. —TheDJ (talk • contribs) 09:56, 31 December 2019 (UTC)

@TheDJ: That's true, but that's still only one side of the coin. The point of this topic was to reconcile that with the stress on disambiguation the information pages seem to make. "The most dumbed down, non-precise description you can get away with" therefore seems to be so unwieldy when it comes to genetic syndromes. If all genetic syndromes should end up with almost the exact same short description (due to the so-called length "target", which itself is not even agreed upon how well it should be enforced), then the information page should be updated to reflect that. If not, due to being decided just too useless, then it should be updated to reflect that too. The point is it's conflicting and contradictory, and my mission is to get a consensus on what the rules actually should be so it can no longer be conflicting, seeing as people are already adding short descriptions and disputing each other's edits over this, and this seems like a topic that should be very possible to get consensus on.

And just to point out, disambiguation pages regularly make use of descriptions in the realm of the lengths I used up there, and that alone raises the question of whether shorter ones are useful enough in some cases (for some topics). SUM1 (talk) 18:24, 31 December 2019 (UTC)

Precisely. First people liked writing articles. Then they liked writing the leads in articles, with as many facts as possible. Now they want to put all the facts in the short description without knowing what the short description is for. Johnuniq (talk) 10:16, 31 December 2019 (UTC)

Well by all means offer your input on what the short description should be for and how it should be applied, in light of what I said and the content of the Wikipedia information pages. SUM1 (talk) 18:24, 31 December 2019 (UTC)

cobaltcigs Trialpears Pbsouthwood May be interested in this discussion. SUM1 (talk) 18:24, 31 December 2019 (UTC)

Short descriptions are also used as annotations in "see also" lists, and should make sense in that context too. A short description should not be misleading, so it should not contain wrong information. However, annotated links are currently excluded from use in disambiguation pages, though the text may be the same, for reasons I cannot recall, but were persuasive at the time. I quite often cannot compose a short description that makes sense in 40 characters or less. In those cases I use as many characters as I need, and leave it to someone more gifted in brevity to try to reduce the length. If they succeed, fine. If not, not a train wreck, it is no worse than not having a short description, and there are lots of topics which do not even need one, and quite a few topics which are so vague that even after reading them it is not clear what they are about. In those cases I usually tag them to request an improvement of the lead.
The graphic recommending amendments to short descriptions based on length looks like generally useful and sound advice. There will be places where it does not work, but it is nevertheless a good working principle that should improve many short descriptions. · · · Peter Southwood ^(talk): 05:50, 2 January 2020 (UTC)

@Pbsouthwood: About short descriptions also being used in See also lists, that's a good point. I had an icky feeling when I first saw that, but who knows, maybe short descriptions will evolve to the point of being descriptive and reliable enough to be consistently used for that purpose. SUM1 (talk) 04:15, 5 January 2020 (UTC)

If I might make a suggestion to clarify the graphic: The section currently labeled "ideal" could be amended to "ideal length if otherwise appropriate". Someone is likely to edit war over getting the length to ideal even if the function is impaired. · · · Peter Southwood ^(talk): 06:02, 2 January 2020 (UTC)

Unfortunately you're probably right. SUM1 (talk) 04:19, 5 January 2020 (UTC)

One thing we can be reasonably sure of is that there is unlikely to be a solution that will suit all possible situations. · · · Peter Southwood ^(talk): 06:08, 2 January 2020 (UTC)

I agree, but we can definitely do better, better than what is currently on the help pages. SUM1 (talk) 04:19, 5 January 2020 (UTC)

I think "The short description should focus on distinguishing the subject from similar ones rather than precisely defining it" should be fixed to "The short description should focus on distinguishing the subject from similarly named ones rather than precisely defining it". Disambiguation is for distinguishing between similarly named topics, not topics that are similar in type, and the goal of the short description, of disambiguating searches, is perfectly served by description of "Rare genetic condition" which gives well enough information for the reader to know they found the right article. As TheDJ mentioned above, the goal of the short description is to distinguish between the various things named e.g. Okamoto, not distinguishing various syndromes (since readers aren't going to search "syndrome" and somehow hope to end up at the right syndrome article). Galobtter (pingó mió)

@Galobtter: All very fair points, and I suggest you add that to the help page. There are still very many caveats to this discussion, such as the automated default short description for disambiguation pages being "Disambiguation page providing links to topics that could be referred to by the same search term" (95 characters long). Those aside, what do you think of the graphic, and could it be used (if edited as needed) to steer new editors of short descriptions in the right direction? SUM1 (talk) 04:15, 5 January 2020 (UTC)

In general I agree with the sentiment that short descriptions are for distinguishing between similarly named articles not articles with similar topics, but there are also quite a few that are both similarly named and similar topics such as Gerstmann–Sträussler–Scheinker syndrome and Gerstmann syndrome. I would probably distinguish these using a longer description listing symptoms. ‑‑Trialpears (talk) 10:57, 2 January 2020 (UTC)

@Trialpears: I agree with that too. What do you make of the use of the graphic? I am continuing to see humongous short descriptions being added, so by any stretch it may help to paint the picture more immediately to those viewing this page for the first time. I can remove the "depending on the topic" part until the precise topics of debate have been agreed upon, or indeed if they end up being only those that meet your rule that you just stated there. I would suggest you add that rule to the help page – there are tonnes of syndromes just like that (Klippel–Feil vs. Klippel–Trénauney, Lujan–Fryns vs. Fryns, just to name a few off the top of my head), and it encompasses the basic idea I was trying to get across. SUM1 (talk) 04:15, 5 January 2020 (UTC)

As time goes by we will have more topics that are tricky to disambiguate. At some point one just has to open the articles to see which one is best. If this only happens 1% of the time we would be doing well. · · · Peter Southwood ^(talk): 05:53, 6 January 2020 (UTC)

I wouldn't have any problem with adding the graphic. Perhaps a whole new section for description length would be appropriate since there clearly are a lot of opinions on the matter. Feel free to draft something! ‑‑Trialpears (talk) 20:32, 5 January 2020 (UTC)

Agreed. · · · Peter Southwood ^(talk): 05:53, 6 January 2020 (UTC)

@Peter Southwood, Trialpears: I'm seeing a strong need for WikiProjects to build their own standards for short descriptions. Just encountered a gene one, where ProteinBoxBot has gone around and added "protein-coding gene in the species Homo sapiens" or some variation ("mammalian protein found in Homo sapiens", "Human gene") to every one on Wikidata. I've been unable to import them for the sole reason of not knowing whether their seemingly redundant length would be acceptable and whether they should be further disambiguated. I was thinking they could do with some distinguishing, i.e. for a notable role like BRCA1 and BRCA2 in cancer or the nature of the gene (e.g., "transcription factor"). By virtue of gene symbols being a set of letters and numbers, all genes will have "similar names", it could be argued, especially those of the same family. This help page could link to the individual WikiProject standards for each topic. Until those are agreed upon, I'll summarise what has been agreed here in a box at the top. I'll probably take this to WikiProject Medicine and let them be the first to establish a standard, building on the more general rules we've agreed upon here. SUM1 (talk) 01:08, 8 January 2020 (UTC)

WikiPojects can write guidance that is helpful for topics within their scope, but it should not conflict with the basic uses of short descriptions, and experts should bear in mind that the readers who will use the descriptions are generally not experts. My guess is that there will be a few areas where special guidance will be useful, and many where it will make little difference. · · · Peter Southwood ^(talk): 04:40, 8 January 2020 (UTC)

The proposed guidelines in the chart look good to me. I appreciate the flexibility offered by the proposal, as a strict 40-character limit would be less effective for more technical subjects.

In the past, there was some opposition to extending short descriptions past 40 characters due to compatibility concerns on smartphones. However, the current versions of the Wikipedia apps are able to handle lengthy short descriptions with no issue. For example, the default short description for disambiguation pages, "Disambiguation page providing links to topics that could be referred to by the same search term" is 96 characters long and is shown in the apps with no issue on the Test article. On the Wikipedia apps, lengthier short descriptions are displayed on multiple lines. — Newslinger talk 07:01, 6 January 2020 (UTC)

In the app, what happens if you search for a name like "Brian Johnson"? I see a list of possible articles and can view the title of each article and roughly the first 40 characters of the short description. I can see the full short description by viewing the article, but the description is pointless while viewing the article! The description only helps disambiguate the wanted page in the list of search results. Johnuniq (talk) 07:17, 6 January 2020 (UTC)

Mobile views of search page

• 
  Wikipedia Android app search page for the query "Brian Johnson"
• 
  Wikipedia mobile website search page for the query "Brian Johnson"

​Here's how short descriptions are displayed in the search page on an average-sized Android smartphone. The short description for Brian Johnson, "English singer, songwriter and television show host of car shows", is 65 characters long. The interfaces can accommodate short descriptions with seemingly no length limit, using as many lines as necessary to display the entire text. However, the Wikipedia apps (on both Android and iOS) impose a 90-character limit when you are editing the short description in the app. — Newslinger talk 08:30, 6 January 2020 (UTC)

Thanks. I must have an outdated app. I tried to update it before my earlier post to be sure I was using the latest but was left confused about the result (I don't use apps much). Johnuniq (talk) 08:36, 6 January 2020 (UTC)

No problem, I can produce more Android/iOS screenshots if they would be helpful. — Newslinger talk 08:54, 6 January 2020 (UTC)

What is displayed on smartphones is the choice of the app developer. We should not be constrained by their arbitrary choices. If a longer description is necessary for it to make sense, we can make it longer. The app devs are also free to make the display useful or not. They write apps to display our content, we do not write content to fit their displays. Maybe in the next version the short description display is 100 characters, or they scrap it altogether. It should still be useful on Wikipedia. · · · Peter Southwood ^(talk): 04:20, 8 January 2020 (UTC)

I think I'm probably missing the issue here. You write whether certain topic areas (e.g., genetic syndromes, genes) should accept longer short descriptions due to the difficulty in distinguishing between articles in that area and in other places mention "disambiguation" a few more times. Short descriptions, as designed, are not used for disambiguation. In your examples CDK13-related disorder and Okamoto syndrome would never need disambiguation between each other. When disambiguation is needed say between things called "Titans" then you have Titans (2018 TV series) and Titans (mythology) disambiguating the titles (sorry I don't have examples from your field). Since it isn't used for disambiguation, it doesn't matter that all syndromes use "Autosomal genetic condition", as that is supposed to give a very brief info on the topic for anyone searching that phrase. --Gonnym (talk) 18:49, 13 January 2020 (UTC)