It has been 10 years since the current default Wikimedia skin (Vector) was deployed. Over the last decade, the interface has been enriched with extensions, gadgets and user scripts. Most of these were not coordinated visually or cross-wiki. At the same time, web design, as well as the expectations of readers and editors, have evolved. We think it's time to take some of these ideas and bring them to the default experience of all users, on all wikis, in an organized, consistent way. Over the next couple of quarters, the readers web team will be researching and building out improvements to the desktop experience based on research and existing tools.
Our goals are to make Wikimedia wikis more welcoming and to increase the utility amongst readers and maintain utility for existing editors. We measure the increase of trust and positive sentiment towards our sites, and the utility of our sites (the usage of common actions such as search and language switching).
Currently, on most wikis, only logged-in users are able to opt-in individually. On selected wikis, our changes are deployed for all by default, and logged-in users are able to opt-out. We are hoping to increase the set of early adopter wikis gradually, until our improvements are default on all wikis in 2021.
January 2021: New search widget and other general updates
Over the past few months, we have been working on the new search widget, which will optimize the current experience by providing context, such as images and descriptions, for searches. While we have experienced some delays with this deployment, we are currently scheduled to deploy some time in February.
We have also begun reviewing some of the data on the Desktop Improvements project as a whole, as well as on individual features. We have published a report on the usage of the collapsible sidebar. According to the data, logged-out users are more likely to keep the sidebar collapsed than logged-in users. These results gave us confidence that our eventual default will be keeping the sidebar open for logged-in users and closed for logged-out users.
In addition, we also have begun looking at the rates at which logged-in users are opting out of the desktop improvements on our pilot wikis. The average opt-out rates for all logged-in users were between 1.78% on euwiki and 4.09% on hewiki. For active editors, the range was between 5% and 14%. We will be looking at this data in more detail in the weeks to come.
Finally, we wanted to give a quick timeline of the next few months:
- February 2021 - new search widget deployed to pilot wikis. The pilot wiki list will expand to include Portuguese Wikipedia, Turkish Wikipedia, Korean Wikipedia, Serbian Wikipedia, and German Wikivoyage.
- February 2021 - Second round of prototypes for logged in users. We will be running a second round of prototypes for the sticky header and user menu with logged-in users over the month of February. We encourage everyone to review these and give us feedback!
- March 2021 - We are currently focused on improving our language switching capabilities. We plan on deploying the new functionality this March to our pilot wikis.
September 2020: New location of search bar now available on all wikis
We have deployed the new location of the search functionality to all projects. The new location is available by default for anonymous users on our early adopter wikis, and by preference for all other users.
We are also performing an A/B test of the new location with logged-in users on our early adopter wikis. 50% of logged-in users are seeing the new experience, while the other 50% are seeing the old experience. This test will last two weeks. Our hypothesis is that the group with the new experience will search more frequently. Results will be available in approximately one month.
Basque, Farsi, and Hebrew Wikipedias, as well as French Wiktionary and Portuguese Wikiversity have now received the new version of the Vector skin by default. For now, these changes include the collapsible sidebar, maximum width, and new header. For feedback and questions - please head to our talk page.
What is our objective?
Imagine a wardrobe
Currently, the interface…
…doesn't match the expectations. …is cluttered and not intuitive. …doesn't highlight the community side. …isn't consistent with the mobile version.
- The desktop interface does not match the expectations created by the modern web platforms. It feels disorienting and disconnected. Navigation and interface links are organized haphazardly.
- There is clutter that distracts users from focusing on what they came for. It is challenging for readers to focus on the content. It is not possible for them to intuitively switch languages, search for content, or adjust reading settings. New editors are unable to use their intuition to set up their account, open the editor, or learn how to use non-article pages for moderation purposes.
- A very small percentage of readers understand how Wikimedia wikis function. Many readers are not aware that the content they are reading is written by volunteers and updated frequently, or that they can potentially contribute as well.
- The large difference in experiences among our desktop interface, apps, and the mobile web, makes it difficult for readers to connect our products. There is a lack of unity in the concept of Wikimedia sites.
How the changes are made
We do not touch the content. We do not remove any functionality. We are inspired by the existing gadgets. We do not make major changes in single steps. We do not touch other skins than Vector.
- We are working on the interface only. No work will be done in terms of styling templates, the structure of page contents, map support, or cross-wiki templates.
- Elements of the interface might move around, but all navigational items and other functionality currently available by default will remain.
- We have analysed many wikis and have noticed many useful gadgets. Some of them definitely deserve to be surfaced and be a part of default experience.
- Though our changes are easily noticeable, we are taking an evolutionary approach and want the site to continue feeling familiar to readers and editors. Each feature is discussed, developed, and deployed separately.
- Skins other than Vector are out of the scope of our adjustments. We have frozen Vector to Legacy Vector, and begun deploying our features as parts of the new default Vector.
Deployment plan and timeline
- It is our intention to test our improvements in collaboration with a diverse set of volunteering early adopter wikis, both Wikipedias and sister projects.
- Both prior to development and after deployment, we collect data (via A/B testing, prototype feedback rounds, etc.). In the case of significantly negative results, we will roll back our changes.
- First three features (the new header, collapsible sidebar, and limiting content width) were deployed in July 2020.
- Our second large feature, an improved search widget, will be deployed to early adopter wikis in February, 2021.
- We are hoping to increase the set of early adopter wikis gradually, until our improvements are default on all wikis in 2021.
|Initially (in 2020)|
|the majority of wikis||early adopter wikis|
|editors or readers
don't see our changes by default
|editors and readers|
see our changes by default
|only editors can opt-in
(from user preferences, by unchecking Use Legacy Vector in the Appearance tab)
|only editors can opt-out|
(using Switch to old look in the sidebar, or directly from their users preferences)
List of early adopter wikis (test wikis)
First group of wikis (marked as ◇ on the timeline above):
- Basque Wikipedia
- French Wikipedia
- French Wiktionary
- Hebrew Wikipedia
- Persian Wikipedia
- Portuguese Wikiversity
- Office Wiki
Second group of wikis (marked as ◇◇ on the timeline above):
- Bengali Wikipedia
- German Wikivoyage
- Korean Wikipedia
- Portuguese Wikipedia
- Serbian Wikipedia
- Turkish Wikipedia
- Venetian Wikipedia
- Wikimedia Incubator
What features will be added
We have not created a complete, detailed product specification. Each feature is built and implemented separately, over time. Features which have been created are still being adjusted and improved. Our decisions are based on community feedback, user testing, and extracted API data.