Train the YouTube Algorithm Getting better video recommendations is a common practice for many users. Removing a video from your story, using the dislike button, or requesting a channel not be recommended all serve this purpose. Curious, none of these actions are as effective as they seemaccording to Mozilla research.
Does this button work? is Mozilla’s latest study on the effectiveness of YouTube controls for training the recommendation algorithm. The foundation has collected data from over 22,000 users who use a variety of methods to monitor the video recommendations they receive on the platform. The result shows that the dislike button, as well as other options offered by YouTube to improve the experience, are ineffective.
Using over 500 million recommended videos, the researchers found that The YouTube algorithm suggested content that the user didn’t want to see. Among the alternatives by which people can control their experience, the “I don’t like” button and the “I’m not interested” option are the least reliable. Removing videos from history or asking YouTube not to recommend a channel are more effective at filtering out unwanted videos, but they are not reliable.
Through our analysis of the data, we determined that YouTube’s user controls are inadequate to prevent spam. […] In general, there is a persistent idea that some of these tools do little to improve recommendations, but are inadequate as tools for meaningful control.
According to Mozilla, “Don’t recommend channel” option has a positive effect in similar videos on other channels. However, this feature does not prevent the algorithm from suggesting videos from a channel that you have explicitly marked as uninteresting. The researchers observed the behavior for a week and found that unwanted videos leaked into recommendations.
YouTube personalization controls are ineffective
Source: Hiper Textual