Topic on Talk:Growth/Personalized first day/Newcomer homepage

Jump to navigation Jump to search

An important thing missing is a custom to-do list

5
197.235.52.125 (talkcontribs)

One of the most important things that has been missing in mediawiki since its inception is a way to organize one's work and decide what to tackle on next. It may be a good idea to draw inspiration from similar systems such as wikihow's task list, Extension:CollaborationKit and wikia's public dev:special:community as well as its private admin-only Special:GoToInterwiki/wikia:c:Help:admin_Dashboard .


The latter in particular is a pretty good model to follow as it basically serves as a tool to help users familiarize themselves with wiki tools, and it also helps guide the administrators to make a wiki successful (in wikia's terms). Although some of the tasks therein apply only to admins, others are helpful to just about any contributor. In a perfect world, both a private and a public dashboard would be available as that creates a sense of collaboration.


Currently wiki users must use awful hacks such as creating lists of pages in the user namespace, risking vandalism, reverts, and people belittling their tasks. The creative process is certainly most effective when it is made in private without any interference.


Also this feature is basically the request in https://phabricator.wikimedia.org/T91655.

MMiller (WMF) (talkcontribs)

Thank you for these tips on other efforts to check out. We've been looking at a set of platforms in our "comparative review". That work is here, and we're going to be summarizing it on that task this week. We had looked at Wikia, but not at wikiHow -- that's a great idea. I just went through the site and gained a lot of perspective. When you talk about wikiHow's task list, are you referring to this page?

I also think it's a good point that a good profile/homepage situation can help create a sense of collaboration.

197.235.61.79 (talkcontribs)

When you talk about wikiHow's task list, are you referring to this page?

Yes, exactly. As far as the task list is concerned, there is certainly a lot of value in automatically generated ones, but a good system would offer the ability to create a custom list (https://phabricator.wikimedia.org/T48103) by simply adding pages.

I also think it's a good point that a good profile/homepage situation can help create a sense of collaboration.

Yes, wikis lack a sense of collaboration because of that. I mean if one wants to even to contribute to something simple like meta:List_of_articles_every_Wikipedia_should_have, they'd need to discover that page somehow, create their own list, and track their progress manually. It is hard to tell if even a big wiki like english wikipedia contains most of these topics in a good readable state.


In pure terms, good wiki growth will be linked to page completion, which could be estimated by determining if articles have:

  • References (for fact based wikis)
  • An internal link
  • A category
  • Illustrative image (optionally)
  • More than XXX words

A good dashboard would contain such a measure instead of simply listing pages one has edited. Of course it may or may not be currently feasible.

MMiller (WMF) (talkcontribs)

What do you think of the article rating system (Stub, Start, C-class, etc.) as a way to approximate page completion? One of our planned components of the newcomer homepage is a task recommendation system, and one way to recommend tasks is to surface articles that are classified as stubs and could use improvements.

197.235.61.79 (talkcontribs)

Article rating systems are good for established wikis only for articles with quality higher than stub because the way they classify stubs is incredibly inconsistent. The article La_Loma_Bridge is certainly small, but is not really that much of a stub. My guess is that editors are quick to add a stub template to an article, but they aren't as diligent at removing them, english wikipedia alone has 2 million stubs. Also, the wikis that would most benefit from growth are those that are too small and unlikely to have an established system. Automation might be the only feasible alternative in those cases.


To sum up, the article rating system alone is insufficient to offer good recommendations. It needs two extra variables, impact (e.g. page views, important topics) and / or personal interest (topics / categories). The fastest way to seed the recommendation list is to simply ask the user what areas interest them, or alternatively offer recommendations using those top level topics, e.g. science, social , geography, history, etc. Another interesting area is hot spots ( topics or areas) where most contributors are currently working on, this is a tactic often used in forums to draw responses to new posts.


Personally, I think that wikimedia doesn't benefit from readers as much as it could. For instance, one easy way to create task lists is to surface red links to readers and ask them if the specific title deserve a topic of its own, and that could generate a good task of "popular" red links. It could even be enriched or filtered by wikidata linked interwikis. Another way would be to ask readers to correlate article quality, e.g. evaluate whether these stubs are really stubs, and eliminate these from certain task lists.


A good system will be a balance of the above, surfacing topics interesting to the user that may have high impact.

Reply to "An important thing missing is a custom to-do list"