Page MenuHomePhabricator

[Suggestion] Prompt people to revise lead sections that surpass MoS length requirements
Open, Needs TriagePublic

Description

https://en.wikipedia.org/wiki/Wikipedia:Manual_of_Style/Lead_section#Length

This task proposes a suggestion that would prompt people to revise lead sections that surpass MoS length requirements.

Story

As a volunteer who is motivated to ensure articles offer readers a concise overview of the topic they're interested in, I'd value knowing when the lead section of an article I'm editing surpasses the length guidelines defined in my wiki's content guidelines, so that readers, especially those on mobile, can quickly grasp the article's key points.

As a reader who wants to get summarized information quickly (internal WMF link), I would value the lead section remaining short.

Meta

Requirements '

Meta

  • Default configuration
    • maximumEditcount:
    • minimumEditcount: 25 edits
    • Type: Suggestion and/or Check: Suggestion
  • Detection heuristic: Lead section length (words)
    • Note: I can see a future where the heuristic is more comprehensive and includes a range of features. E.g. length, existence of unique information, reading level, etc.

User experience

  • Card design
    • Title: Improve readability
    • Description (≤2 sentences): This section is longer than Wikipedia guidelines recommend. Consider making it shorter by doing things like moving information into other sections or replacing long phrases with simpler ones.
    • Calls to action
      • Revise: Activate the bounded text-editing state the ToneCheck.js introduced
      • Dismiss: Activate "dismiss survey" that invites people to share why they think the lead section is acceptable as it is.
  • Success toast

Instrumentation
As with all Edit Checks and Suggestions, we will want to know:

  • Any time a Suggestion of this type is activated within an edit session
  • Any time someone views a Suggestion of this type within an edit session
  • Any time someone engages with a Suggestion and how they engage with it

References

Event Timeline

This looks good! A few notes:

  • I've seen "four paragraphs" used as a heuristic for a good lead length by some editors. Maybe not worthwhile for the MVP, but it could be a useful heuristic.
  • Another useful heuristic could be the percentage of total article length, which is one way to capture the fact that more complex topics may need a longer lead than others.
  • I know that WP:MOS/Film#Plot advises that film plot summaries be 400-700 words. So if we're able to build it to be broader, it could be used for more sections than just the lead.

Change #1347878 had a related patch set uploaded (by DLynch; author: DLynch):

[mediawiki/extensions/VisualEditor@master] EditCheckAction: Keep the revising state in an ephemeral tag

https://gerrit.wikimedia.org/r/1347878

Change #1347879 had a related patch set uploaded (by DLynch; author: DLynch):

[mediawiki/extensions/VisualEditor@master] EditCheck: Add a suggestion for long lead sections

https://gerrit.wikimedia.org/r/1347879

What the appropriate length for a lead is, is a topic of debate. Policy geek hat here, but the guideline at LEADLENGTH does not make a recommendation, it gives a description of what typical featured articles lead lengths are (250 to 400 words).

For the description, I would make it something like: "This section may be too long to read comfortably. Consider making it shorter by moving information into other sections or removing unnecessary details.

Typically, you can shave off around 5% of words with pure readability fixes, which is how I interpret "replacing long phrases with simpler ones.". What is almost always necessary to shorten a lead is summarising better (removing details) or moving unique information into appropriate sections.

Leads can be a sensitive topic. We need them to be shorter for readers, and we want to teach newbies not to stuff them full of crap. So I imagine that this could have both an edit check and suggestionmode version, with slightly different thresholds:

  1. For edit check, this might be enabled when a user gets over 500 words, and for quite new accounts (0 edits? 10 edits?)
  2. As a suggestionmode, I'd put the threshold a bit higher, at 550 words. The FA climate change is 539 words for instance, and I don't want people messing with that one. To ensure that the editor has some clue at this point, maybe start from edit 50?

Change #1347878 merged by jenkins-bot:

[mediawiki/extensions/VisualEditor@master] EditCheckAction: Keep the revising state in an ephemeral tag

https://gerrit.wikimedia.org/r/1347878

Change #1347879 merged by jenkins-bot:

[mediawiki/extensions/VisualEditor@master] EditCheck: Add a suggestion for long lead sections

https://gerrit.wikimedia.org/r/1347879

^ that's experimental, so people can try it out.

Is the message editcheck-leadlength-descriptionlink supposed to be {{notranslate}} or {{optional}}?

That is, is it better if it's translated on translatewiki or on each Wikipedia?

Curently, the qqq file says {{notranslate}}, but it's not actually defined as ignored on translatewiki.

If there's a possibility that this thing will be used on non-Wikimedia wikis, I recommend:

If this feature is definitely only for Wikimedia, then perhaps it's OK to get it translated on translatewiki, but then maybe it shouldn't even be optional. But maybe @Nikerabbit has something to add.

(Also, I'm curious: Did a human or an LLM add {{notranslate}} to the qqq file?)

@Amire80 It's an example of a pattern found in these checks -- we have a bunch of messages that're foo-descriptionlink that're documented as {{notranslate}}. It should never be translated by translatewiki, because it's not about the language, but should instead be customized on each wiki because the goal is for it to link to that wiki's policy related to whatever check is being documented. We do normally link to a non-enwiki page for the default value, but we also don't bother to clean up all the copy for checks that're still behind the "experimental" flag.

What's involved in flagging something as "ignored" on translatewiki? I'd have assumed that the presence of {{notranslate}} would be enough, but...

(Also, I'm curious: Did a human or an LLM add {{notranslate}} to the qqq file?)

I did it, but in the sense that I just copied the template we're using for all the other checks.

@Amire80 It's an example of a pattern found in these checks -- we have a bunch of messages that're foo-descriptionlink that're documented as {{notranslate}}. It should never be translated by translatewiki, because it's not about the language, but should instead be customized on each wiki because the goal is for it to link to that wiki's policy related to whatever check is being documented. We do normally link to a non-enwiki page for the default value, but we also don't bother to clean up all the copy for checks that're still behind the "experimental" flag.

What's involved in flagging something as "ignored" on translatewiki? I'd have assumed that the presence of {{notranslate}} would be enough, but...

Yes, unfortunately it requires another step: update the relevant translatewiki configuration file in the translatewiki Gerrit repository. Visual Editor and other Gerrit extensions are in this file:
https://gerrit.wikimedia.org/r/plugins/gitiles/translatewiki/+/refs/heads/master/groups/MediaWiki/mediawiki-extensions.txt#3652

translatewiki maintainers (like me) often update these files, but patches from extension developers (like you) are very welcome, too. The file format is very easy.

(Core MediaWiki and non-Gerrit extensions are configured in other files in the same repo.)

(Also, I'm curious: Did a human or an LLM add {{notranslate}} to the qqq file?)

I did it, but in the sense that I just copied the template we're using for all the other checks.

👍️