Growth/Personalized first day/Newcomer tasks/Maintenance template tests

The first version of newcomer tasks will recommend articles to newcomers based on the presence of maintenance templates, which indicate that articles need attention with respect to copyediting, adding links, expansion, adding references, etc.

Example of maintenance template on English Wikipedia

Maintenance templates tend to contain only a short explanation with few links to find out more, and so we anticipate that newcomers will struggle to understand what the maintenance templates are asking for and to figure out what edits to actually make. To address this, our team plans to add clear guidance to the newcomer tasks experience in V1.2, which will use the help panel to explain to newcomers how to complete the suggested edit. As we design this guidance, during the week of October 21, 2019, each member of our team attempted to complete edits to two different articles that have maintenance templates. We recorded our challenges and ideas from that experience, and they are listed below.

Ideas for guidance edit

  • One small change: It is critical to make newcomers understand that they only need to do one small change -- they don’t need to perfect the whole article by completing all copy edits or referencing all unsourced claims.
  • Multiple templates: It is confusing when articles have more than one maintenance template, because it's unclear which task to do.  How to address this?
  • Use a search engine: For adding references, updating articles, and expanding articles, we should tell the user to search the title of the article in a search engine.
  • See examples: It would be useful to see examples of what complete versions of each task should look like.  Or see example articles that are similar to the one being edited. This might be an even harder than the challenge of finding article of similar topics -- it would mean finding another article about a short story collection (not just “literature”), or another article about a rock band (not just “music”).
  • Balanced message: How to strike the right balance of “mistakes can be undone” with “but do a good job because this is real Wikipedia”? We want people to take these edits seriously, but not be so afraid of making a mistake that they don't even try.
  • Internal links: Need to be clear that “adding links” means internal links to other articles, not links to external references.
  • Jargon: Remember that templates have jargon in them, like “inline citation”, that needs to be decoded for the user.
  • How to add links: For adding links, we need to tell the user the actual technical steps: select text and then click the link icon.
  • Original user: It might be motivating to tell the user which other user added the template, to show that the request comes from an actual person, and maybe give them the opportunity to check in with that person to ask questions.
  • Follow links for sources: A possible recommendation for finding citations is to follow links in the article so as to see potentially good sources in similar articles.
  • Highlight inline templates: It would be illuminating if there were a particular way to highlight inline templates (like “citation needed”) so that they are easy to see on the page.
  • Templates on mobile: We need to look into what maintenance templates look like on mobile, to see how they differ in visibility and clarity from desktop.

Other ideas edit

  • Shorter articles: Shorter articles may generally be easier for newcomers to deal with.  Could consider filtering recommended tasks based on length -- could also analyze results to see whether shorter articles do lead to success.
  • Communities check templates: Some maintenance templates are very unclear.  We could ask communities to take a look. But it may not be good to encourage them to put help links in the template, since we intend to deliver the guidance through the help panel -- we don't want newcomers to have too many links to follow and end up being confused.
  • Filter by age of template: Some templates have been on articles for many years.  Maybe we should sort or filter suggestions to newcomers based on that, so that they get newer ones that are more likely to still be issues. This has been filed in Phabricator here.