Jump to content

Wikipedia:Wikipedia Signpost/2012-09-10/Op-ed

From Wikipedia, the free encyclopedia
Op-ed

Fixing Wikipedia's help pages one key to editor retention

Peter Coombe is a community fellow with the Wikimedia Foundation who is researching help pages. He is also an editor on English Wikipedia, Wikimedia Commons, and Wikinews under the username the wub.
The views expressed in this op-ed are those of the author only. Readers are invited to respond or offer critical commentary in the comments section, while those wishing to author their own op-ed may use the Signpost's opinion desk.


Peter Coombe, Wikimedia community fellow

Much like article content, the English Wikipedia's help pages have grown organically over the years. Although this has produced a great deal of useful documentation, with time many of the pages have become poorly maintained or have grown overwhelmingly complicated. There are several issues with the current network of help pages:

  • Too long. Wikipedia:Citing sources is one of the most viewed of all help pages, and it's more than 8000 words long.
  • Widely varying complexity. If you're a new user who wants to learn what the "edit summary" box is for, you're probably not interested in precisely which MediaWiki message defines the autogenerated summary for page blanking. But we told you anyway (at least we used to—thankfully someone has removed this now).
  • Accumulated cruft. Early on in my fellowship I found a help page that was just about how to draw chemical structures using ASCII art. So I thought, that's quaint, it's from 2003, obviously things were a bit different then. I checked with the people at WikiProject Chemicals, and it turns out ASCII art was never encouraged. This was just someone's random page, in Wikipedia space, telling people how to do something that had always been against guidelines. (It was deleted).
  • Navigation is poor. One thing people complain about is that after they arrive at a useful page, it's often hard to find it again. A pattern we see over and over is people adding links on their own userpage when they find useful pages, building their own navigation system, so that they don't lose the useful page. People really shouldn't have to resort to this: pages should be findable.
  • Fragmentation and duplication. We have at least four pages about how to make tables in wiki markup, and it's unclear which is most suitable for a given problem. And they all need to be kept up to date.

For some idea of the scale of the problem, the main help landing page – Help:Contents – now gets around 10,000 hits a day. It's my belief that improving Wikipedia's help system is one of the most important steps we can take to improve editor recruitment and retention. That's why for the past few months I've been working as a Wikimedia community fellow to research how we can improve help pages.


How disorganised is this? Partial map of Wikipedia's help pages, created by User:Ironholds. Full version (5 MB png)


Community project

There were already some community attempts to improve the help pages, such as the Help Project, but sadly they had become fairly inactive. Because improving and maintaining the help pages is such a huge ongoing task, there's no way I can do it on my own, and I really don't want the efforts to end with my fellowship. Therefore I've worked to revive the project: the homepage received a major overhaul, and there's now a monthly newsletter for members and a regularly updated statistics page covering all the pages within scope. (If you want to get involved, or just keep up with the latest developments, do sign up!)

Learning more

The next major step was a large survey, taking in both new and experienced users, to find out what they are looking for help on, how they find it, and what they think of the existing pages. The full results and conclusions are available. Unsurprisingly, "how to use wiki markup" and "how to start a new page" are the most popular topics among new users. What is surprising is that people rate the help on these topics as ok. It's still not great and could certainly use improvement, but it's better than the others. The help topics people really didn't like are how to add references and how to add images. This is fairly consistent across all experience ranges, from newly registered users with no edits to old hands with thousands of edits.

Recently we've also been able to deploy the article feedback tool to help pages. This should allow us to get extremely valuable feedback: until this deployment and the survey I conducted, we really had very little evidence on what users thought of them.

Tutorials

One thing the Help Project created in the past were a couple of "Introduction to ..." tutorials: Introduction to policies and guidelines and Introduction to talk pages. These focused tutorials have a friendly tone and don't overwhelm new users with details. They've been very well received by the new users who find them, so I decided to make the tutorials we do have more prominent, and developed new tutorials in the same vein on topics that the survey suggested would be valuable: Referencing, uploading images, and navigating Wikipedia. These are brand new, so please edit and improve them! But do try to avoid making them too long and detailed, or adding too many links.

The new tutorials make use of vertical tabs, which were well received in usability tests.


"Helped by people"

Probably the clearest finding in the survey is that experienced editors love the results they get from asking questions on another user's talk page, but new users aren't really aware of that as an option. The same is true to a lesser extent of asking questions at the Help Desk, or in IRC, as clinched by one respondent's comment: "I was helped by people, not help pages." The personal touch certainly seems to ease things along, and that's why it's great that we have new initiatives like the Teahouse, and more friendly warning messages that explicitly invite questions on the warner's talk page. Part of my work to redesign the navigation will try to make these question pages more visible to those who could benefit from them. This is particularly true of the Reference Desk, even though the article feedback tool has only been deployed for a few days we have already seen many factual questions appearing in the feedback for help pages.

Making help findable

Over the next month I'll be focusing on the final part of my project: redesigning Help:Contents, the main entry point into our help system. At the moment this page is a mess, with too many subpages, too many links, and not enough explanation. Many of the links that do exist are misleadingly labelled. This was borne out by usability tests I conducted, where people found it difficult to navigate and find the help they were looking for.

A serious problem with Help:Contents is that it has to speak to many different audiences:

  • Readers looking for info ("-1" edits). The biggest volume. Not a primary target of the fellowship, but we certainly need to better support them, help them find places like the Reference desk, and maybe we could entice them to edit with calls to action.
  • Readers who want editors to fix something. Similar to the above.
  • Brand new editors (0–1 edits) looking to get started, most likely with step-by-step tutorials and hand-holding.
  • New editors (10–100 edits) looking for specific help on a topic.
  • Experienced editors and admins (100+ edits) looking for more advanced help.

At the moment links relevant to these different groups are all mixed together. The aims of the redesign are to better funnel these different users to where they can get the right kind of help, and to better expose the personal help mentioned previously for those who want it—and, one hopes, to make the page look a bit more attractive too! Again I'll be doing usability tests to try to identify any potential problems with the new design, and to confirm that it's better than the existing one.

How to get involved

If you're interested in improving help pages, please do join the Help Project and the discussions on its talk page. There are also some open tasks you could get started on. It's going to be a long haul, but this work is something that could really make a big difference to the future of Wikipedia.