Mapping controversies on wikipedia

Notabilia investigates the nature and shape of deletion discussions in Wikipedia. A pair of data visualizations represents the 100 longest discussions that resulted in deleting or keeping the respective articles. The award winning data art work combines data-driven analysis of the phenomena around consensus finding in ad hoc communities with an allegorical, artistic treatment of the topic.

Produced in collaboration with Dario Taraborelli and Giovanni Luca Ciampaglia

How it works

Each discussion is represented by a thread starting at a shared root node, with curvature and coloring representing the sequence of “keep” and “delete” votes during the discussion. Each discussion is represented by a thread starting at the bottom center of the graphic. Each time a participant recommends to keep the article, a green segment leaning towards the left is added. However, when a user argues for deleting the article, a red segment leaning towards the right is added. As the discussion progresses, the length of the segments as well as the angle increments decay.

Some background on Article for Deletion Discussions

Like a garden, an online encyclopedia needs constant weeding. Unlike a garden, an online encyclopedia has thousands of potential gardeners. Over the years, Wikipedia has developed guidelines and policies to help editors collectively decide whether topics are suitable for inclusion or not. All articles, especially new ones, are reviewed by the community to determine if they meet Wikipedia's notability guidelines. Any editor can nominate an article for deletion and, if this nomination is legitimate, a community discussion takes place, where any fellow editors have the opportunity to make their voices heard.

The usual process is to have a week-long discussion during which community members can discuss in favor or against keeping the article. At the end of this period, an administrator reviews the discussion and speaks the final verdict. What decides, whether consensus is reached, is the administrator clo sing the AfD discussion, not a headcount. As a result, the proportion of Keeps and Deletes may be at odds with the final decision, as illustrated by the above visualization. AfD discussions also take a variety of shapes depending on how they evolve over time.