A few weeks ago, we posted a page preview feature on English Wikipedia. This deployment marked the completion of the biggest changes we have made to the desktop version of Wikipedia in years. Since then, we have received many questions about our process, motivation and documentation. Given that a lot of our work is already online, we thought it would be nice to share more.
As Tech Lead John Robson wrote in his blog post, the page preview took us a very long time to complete. This was due in part to our long-term technological complexity as well as our best efforts in terms of instrumentation, testing, and in part to fully interact with our communities to facilitate longer stays. Here is a list of the way we are assembled:
main MediaWiki page for a general project includes an overview of the project and functionality, links to research and requirements, an overview of the feature’s history, and a deployment plan.
Detailed description of product product requirements and workflow and functionality: We withdrew our code before we deployed any wikis. As this is somewhat of a new beginning, we made sure to collect all the requirements for this facility ahead of time. We included feedback from our communities to include all requests.
The initial blog post on launching it as a beta feature: the page preview story goes back a lot. The feature was initially offered to logged-in users as a beta feature, called “HoverCard” in 2014. Since then, it has changed significantly in appearance and functionality.
we designed page previews for Wikipedia and what they might do with them in the future: Even though page previews may seem simple at first glance, many complexities lie beneath the surface. This blog post goes into detail about why this is the case, releasing specifically for Wikipedia what we had to consider in terms of designing the feature, and how these ideas were translated into the final design Was
Design improvements since the first iteration of the feature: This is a list of design updates made since the initial iteration of the page preview in beta.
Records of Vastu Decisions: Here, you can see that we make the way of page building the way we do, and track the Vastu decisions we make.
API Specification: Initially, previews were in plain. Our communities were rightly interested in presenting page previews in HTML to allow content to be true to its original form. It makes sense – we don’t want weird bugs and missing formulas to distract from the benefits of page preview. However, we did not want to process all of this HTML within the page preview client.
The less work the client has to show the preview, the better. Thus, we created a new API that could generate page previews as well as summaries for other similar features in the future.
Code Description: For those of you engineers, this page includes details on the code for instructions on setting the feature for page previews and other wikis.
This blog post highlights the process of identifying and replicating one of the most confusing bugs that have come before us. We were seeing duplicate events logging in for every link hover and it took a lot of effort to get us to the core of the issue – a bug in the Firefox browser. We worked with the Mozilla team to resolve the issue and were able to continue to measure the performance of the feature bug-free.
Page preview Front-end tooling: This series of blog posts talks about various technical decisions and options for technology and tooling for the front-end portion of extensions. The posts provide arguments, explanations, pros and cons, and our conclusions.
2017-18 A / B Test: Page Previews are designed to reduce the cost of searching for a link as well as to promote learning, by allowing readers to get references on the article they are reading Or quickly check the definition of an unfamiliar event.
Without taking away the idea, object, or word from their original subject. To estimate the success of the feature, we wanted to test these assumptions by performing an A / B test on the English and German Wikipedia. This page includes a description of our methodology, the hypotheses we have decided to test, and our results.