When I first joined YouTube, you could upload a crappy video with distorted audio, totally pixelated, and expect to get a few hundred to a few thousand views. The viewership didn't just last a few hours, it was days, weeks, months. So it was fairly easy to actually get people to see your videos, pretty much if you posted, people would see it. But now, it seems like YouTube uses an ML-based system that does periodic checks on engagement rather than basing suggestions on the actual content.
So basically, they're missing the forest for the trees. By focusing on engagement rather than the content, they are actually reducing interest on the content they serve to you. How does that work? It means YouTube doesn't suggest videos to you based on what your interests align with, it suggests based on the probability that you will engage with it (whatever that means). It should be noted "engaged views" are not views that watch the entire video. That's retention rate and it's a separate meteic