I don't believe any of this is true, and i have 4 reasons:
1) Either its not true or i am a very weird exception.
I am subscribed to 108 Channels and at least 2 of them i ignore for ~95 or more % of the time. So basically i watch about 1 video in 20-35 or about one video in 1-2 months. I also have channels i have not watched in about half a year but i still get all their updates.
2) Missing sub-vids always happen in phases and when it happens all my creators are affected, even those from whom i watch every video (about 3-5 vids a week).
3) A Sub-Box that does not deliver would be a nonsensical notion, as there is no algorithm that can safely determine why I subscribed to someone in the first place. It would be even bigger nonsense if i missed the one format i am subscribed for just because i don't enjoy the rest.
4) Google has no reason to do it as i (the user) can unsub any time i want if a channel spams my inbox too much. No-one will be sad or angry. Also google is interested in you watching as much content as possible so they will show you everything on the off-chance you watch it. (veritasium2 has a video (called 'The Problem With Facebook') explaining how the interest of youtube differs from that of facebook in that regard better than i can.)
Now I know that point 3 and 4 are only my reasoning and google could apply a different logic and i know that 1 and 2 could be dismissed as anecdotal evidence but in the hard sciences one exception is often enough to disproof a theory. As an algorithm like the one you described would never arbitrarily exclude users I thinks the rather extreme examples in 1 and 2 should suffice to completely disproof your theory.
Furthermore if one is not wowed by your eloquence or the music picking up when you present your arguments, then they are pretty weak and include multiple fallacies. The data you present does not support burned subscribers as much as it does burned-out subscribers and everyone knows the psychology behind human value-assessment, so 'less videos = more views' is hardly a mystery. Also while you might have argued your case very well concerning facebook (note that a huge portion of your argument depends on this part), facebook is hardly suitable example (see video in 4).
Now while i might have some errors in my reasoning and i could be wrong, this and some other gametheory videos have led me to my own gametheory-theory: You are lately more and more often fitting the facts to your theory and not the other way round. While this would never diminish the entertainment-value of your videos, it still saddens me a little. It saddens me because i know you are capable of better and it saddens me because i already hate it when wrong rumors and conspiracy-theories haunt the public image and/or understanding of the sciences and it would pain me if half the 'public knowledge' of my hobbies would also start to consist of scaremongering and superstition.
All that said Theories are there to argue over them and i love your theories of little stuff which are blown totally out of proportion, which is why i replied with this antithesis which is blown totally out of proportion.
Best Regards,
your fellow theorist