A tap on the heart icon is cheap currency. Millions of users double-tap absentmindedly while mentally checking out. The recommendation engine treats that gesture accordingly, placing it near the bottom of its valuation hierarchy.
The system prioritizes average watch time and the video completion rate above all traditional user engagement metrics. When a user watches a 15-second clip past the 100% mark, triggering an automatic replay, the neural network scores that event as a decisive endorsement. Empirical tests conducted by data auditors show that an account watching a clip 1.5 times without liking it triggers a 45% higher probability of seeing similar content than an account that taps "like" within the first two seconds and swipes away immediately.
These user interaction signals also capture micro-behaviors. If you pause a video to inspect an on-screen detail, the engine registers a retention spike. If you open the comments section while the video continues playing in the background, the background loop inflates the total dwell time. The system interprets this as deep session immersion. The algorithm does not care why you lingered. It only logs that you did.
