YouTube Tests Crowdsourced Annotations for Videos

YouTube is experimenting with a new feature that allows creators and viewers to add context notes to videos, aiming to enhance the accuracy of information and curb misinformation. This pilot feature, called "Video Context Notes," enables selected users to annotate videos with supplementary information.

Currently available on mobile devices in the United States for English language videos, the feature invites approved contributors to write brief annotations. YouTube's announcement explains the intended use of these notes:

“These notes could clarify when footage contains parody material, point out if a product review is outdated due to a newer version release, or confirm whether viral clips actually depict current events.”

The initiative builds on YouTube's ongoing efforts to provide context alongside videos, such as information panels and disclosure labels for altered or synthetic media.

YouTube acknowledges the potential for errors in the early stages of this feature, stating:

“We anticipate there will be mistakes – notes that aren’t a great match for the video or potentially incorrect information. That’s part of how we’ll learn from the experiment.”

YouTube Annotations - Availability and Moderation

Initially, a limited number of YouTube channels in good standing will be invited to contribute context notes to videos. Viewers in the U.S. will be able to see and rate the helpfulness of these notes. Additionally, third-party evaluators, who are the same contracted personnel providing feedback on YouTube’s search and recommendation systems, will assess the quality and accuracy of the notes.

These evaluations, combined with viewer ratings, will be processed through a “bridging-based algorithm” to determine which notes are published broadly. YouTube explains:

“If many people who have rated notes differently in the past now rate the same note as helpful, then our system is more likely to show that note under a video.”

As the pilot progresses, YouTube plans to explore having contributors rate each other’s notes to further refine the note-publishing system. This step aims to enhance the reliability and accuracy of the annotations, ensuring that viewers receive helpful and precise information.