Growth/Personalized first day/Structured tasks/Community feedback/de
Recently some contributors have expressed concerns about "add a link" and "add an image" structured tasks.
These discussions have taken place on Arabic Wikipedia, Spanish Wikipedia, and Russian Wikipedia.
The main concerns are as follows:
- Communities want to be more involved in decisions, including how we measure edit quality.
- The number of structured task edits are a burden to patrol.
- The perceived edit quality is low and the edit value is low.
While the Growth team focuses on newcomers, we want to create tools that work for the community as a whole. Although the "add a link" structured task has proven to be effective at both increasing the probability that newcomers will make their first edit and return to edit again (experiment analysis), we also understand that these edits may be frustrating for other contributors.
We are listening to feedback, and planning ways we can improve Growth tools or other areas of the site to ensure our work benefits the movement as a whole. We hope to pursue some quick improvements immediately, as well as consider changes that will require more planning and coordination. This overview is a work in progress and will continue to expand as more ideas are suggested and discussed:
| Problem | Potentielle Lösung | Anmerkungen | Status |
|---|---|---|---|
| Communities want to be more involved in decisions | More outreach | This will be an ongoing effort. We will continue to start every new project with a period in which we discuss ideas with the community. We will then iterate on designs and even change course as needed based on feedback. The best way for you to get involved and make sure your voice is heard is if you watch Growth team updates or subscribe to Growth Newsletters. | ✅ Ready / Ongoing |
| Allow for more Community Configuration | Allow communities to adjust the "weight of underlinked" articles via Special:CommunityConfiguration/GrowthSuggestedEdits | ✅ Complete | |
| Allow communities to adjust the "Minimum required link score" via Special:CommunityConfiguration/GrowthSuggestedEdits | ✅ Complete | ||
| Patrollers are overburdened | Community configuration | The Growth team tries to make sure Growth tools are customizable because we realize each wiki has different needs. Privileged users can edit Special:EditGrowthConfig to set a lower limit to "add a link" and "add an image" structured tasks. The default allows newcomers to complete up to 25 of these tasks per day. If patrollers are overburdened, we suggest lowering this limit. | ✅ Ready / Ongoing |
| Improve patroller tools | Create a patroller tool that displays an easy diff popup for any "add a link" edit in Recent changes. If there are multiple links added, the tool could allow the removal of just one or two links, in a simple one-click interface. | 💬 Discussed | |
| Create "check link" structured task | Create a "check link" task for the suggested edits feed that can be accessed by more experienced editors. In other words, a patrolling task accessed from the newcomer homepage, perhaps for newcomers who have been successful with the original "add a link" task. | 💬 Discussed | |
| Bearbeitungsqualität ist schlecht | Better newcomer training | Both structured tasks include onboarding instructions, but newcomers currently have the option to skip it. Data shows that users that complete "add an image" onboarding have lower revert rates. We could further improve onboarding and provide more contextual instructions. | 🗓 Geplant |
| Verbessern von Algorithmen | Increase threshold for confidence score on link recommendation service suggestions | ✅ Complete | |
| User experience improvements | There are many different ideas we could pursue. For example, newcomers have trouble writing good image captions, so we could offer translated captions taken from other wikis and then newcomers would only need to copy-edit the translation. | 💬 Discussed | |
| Reconsider categorization of "Easy" tasks | Remove add a link from the list of easy tasks for people gaining experience | 💬 Discussed | |
| Quality gate | Remove access to "add a link" structured task for newcomers who reach a certain revert threshold, and link to onboarding material. This is planned as part of the Positive Reinforcement project. | 💬 Discussed | |
| Add link "approval reason" | Ask newcomers why they think the link is appropriate (we currently do the opposite). Add the approval reason to the edit summary, so it would be visible in the RecentChanges view. | 💬 Discussed | |
| Test tasks | Provide lengthier onboarding including needing to pass a test before accessing the tool. If the newcomer passes the test, then they have access to structured tasks. | 💬 Discussed | |
| Edit value is low | Level up newcomers | To learn successfully, newcomers need to complete some successful minor edits while they learn the skills to eventually produce higher value edits. The Growth team will soon work on strategies geared at "Leveling up" newcomers so they progress from easy to more difficult tasks. | ✅ Complete |
| Prioritize underlinked articles | We want to ensure articles aren't a "sea of blue links." Developers are working on a task to ensure we prioritize articles that are currently underlinked. | ✅ Complete | |
| Reduce the number of links suggested per article. | Communities already have the ability to customize Growth features in several ways. Privileged users can adjust the maximum number of link suggestions to display in each suggested task via Special:EditGrowthConfig. We lowered the default to 3 link suggestions per article recently. Communities can also exclude links from articles containing certain templates or categories. | ✅ Ready / Ongoing | |
| Teach newcomers when they make mistakes | At the moment, it is up to patrollers to tell newcomers what they've done wrong. Messaging newcomers after a revert or when their edit has been improved could help them understanding what they have to improve. | 💬 Discussed | |
| Revert rates aren't an accurate measure of quality | Improve quality metrics | For various reasons, sometimes good faith edits aren't reverted even if they are poor-quality. We have discussed other metrics to consider, and how to better define Edit Quality | ✅ Complete: Quality of edits |
| Better understanding of edit "quality" | Edit quality is actually not a well defined concept within the wikis. Discuss with wikis and publish results. | ✅ Complete |
We hope to address some quick improvements immediately, and then plan ahead for some of the larger changes. If you have any ideas that aren't listed here, or want to vocalize your support or opposition for any of these solutions, please let us know on the Discussion page.