When you are enjoying videos on YouTube, it is common for the YouTube algorithm to send you notifications of suggested videos. If you don’t like one, surely you click on the “I don’t like” button and you think that the matter is solved, but no.
Recently, Mozilla researchers discovered that the YouTube platform’s algorithm does not learn from that “Dislike” button. Or at least he doesn’t learn as much as expected. So it is valid to say that it is difficult to make the YouTube algorithm untrain so that it stops showing us these types of suggestions. So it’s not just your perception, it’s that it was shown during the study. The “I don’t like” button turns out to be very ineffective against determining user preferences.
The report states that these types of buttons “prevent less than half of unwanted algorithmic recommendations.” To reach these conclusions, Mozilla used data collected from RegretsReporter. This is a browser extension that allows users to “donate” their recommendation data for use in these studies. The report was based on millions of recommended videos and anecdotal information from thousands of people.
It may interest you: Humans, not algorithms, will review content on YouTube Kids
In total, Mozilla tested the effectiveness of four YouTube controls such as: thumbs-down “dislike” button, “not interested”, “don’t recommend channel” and “remove from watch history”. Although each one showed a different degree of effectiveness, in general the four have a “small and inadequate” level. The best spared of the four was the “Do not recommend channel” button, which prevented 43% of unwanted recommendations. While the “I’m not interested” only avoided 11% of the recommendations.
The same study highlights the efforts of users to avoid unwanted recommendations. Likewise, this result demonstrates that the company must develop more effective and proactive controls for users to determine the content they want to see and teach the YouTube algorithm.
Image: Pixabay on Pexels.com