According to Mozilla’s new research, the ability to announce that users are not interested in certain video content on YouTube does not work properly, and even after sending negative feedback, similar offers are still shown to them.
Mozilla researchers, by examining the data related to YouTube videos suggested to 20,000 users, found that the buttons Not interested (not interested), Dislike (dislike), Stop Recommending Channel (Stop offering the content of this channel) and Remove From Watch History (deletion from viewing history) are largely ineffective in preventing the display of similar content that is not of interest to users.
Mozilla’s report shows that even in the best case with the use of the mentioned buttons, more than half of the suggested content is in conflict with the user’s interest, and in worse cases, there is no change in the suggested movies.
To collect data from videos and real users, Mozilla researchers enlisted the help of volunteers who use the RegretsReporter This company used The said plugin has a button called Stop Recommending adds to stop suggesting similar content to the user’s watched YouTube videos.
Behind the screen, users were randomly divided into two groups; So that in the experimental group, with each click of the user on the mentioned button, a different signal is included DislikeNot Interested, Don’t Recommend Channel And Remove From History was sent to YouTube and no feedback was sent to the control group.
Using data collected from 500 million suggested videos, the researchers created more than 44,000 pairs of videos, each pair consisting of a rejected video and a YouTube suggested video after it. They then evaluated the pairs either personally or with machine learning to see if the recommended YouTube video was similar to the user’s rejected video.
The research results showed that YouTube’s Dislike and Not Interested buttons have very little effect and 12 and 11%, respectively, have been effective in preventing inappropriate suggestions. Meanwhile, the Don’t Recommend Channel and Remove From History buttons were more effective in preventing bad offers by 43% and 29%, respectively. However, researchers believe that YouTube’s provided tool for removing unwanted content does not have an acceptable performance. Researchers say:
YouTube should respect users’ feedback about their experience and treat it as meaningful signals about how they want to spend their time on the platform.
Elena HernandezYouTube spokesperson, in response to this research, stated that these behaviors are intentional; Because the mentioned platform does not seek to block all content related to a topic. Also, he criticized Mozilla’s report, accusing it of neglecting the way YouTube’s control features were designed. He said:
Importantly, our moderation capabilities do not filter all topics and views; Because it may have negative effects like “echo chamber” for the viewers. We welcome academic research on our platform; for this reason; We recently expanded access to API data through the YouTube Research Program. Mozilla’s report ignores how our system works, and as a result, we can’t draw many meaningful conclusions from it.
According to Hernandez, Mozilla’s definition of “similar” doesn’t match how YouTube’s suggestion system works. The Not Interested option only removes a specific video, and the Don’t Recommend Channel button also prevents the proposed channel from being suggested in the future. According to his statement, YouTube does not intend to stop all suggestions related to a topic or opinion or person.
In addition to YouTube, other platforms such as Tik Tok and Instagram always introduce more feedback tools to show the content related to the user’s interest by training their algorithms. However, users have always complained about the inappropriate performance of the aforementioned features and believe that even by marking certain contents as inappropriate, they still see similar cases.
Becca Rex, a Mozilla researcher, believes that it is not clear what the various controls actually do, and the platforms do not clarify how feedback is taken into account. He says:
In my opinion, YouTube tries to create a balance between satisfaction and engaging the user with the content in order to create a balance between the suggested content with the ability to attract the user and his favorites. The YouTube platform has the ability to determine the importance of each of these items in its algorithm; But our research shows that user feedback may not always be the most important factor.