Yep, that is indeed the #YouTubeProblem - and it is going to be a problem for a while, and the reason why YouTube itself will stay relevant for years to come, unfortunately. That said, sometimes I wonder if Google is actually aware of the value they have there... But, that's not the point.
Right now, p2p file sharing and duplication across nodes is unfortunately the best the non-big-company crowd can do - be it torrent, peertube, ipfs, heck even sia. But you pointed out correctly that this also means that unpopular content dies, which is always a shame. However, at the current point, unavoidable. I ran into torrents before with zero seeders for months because it was, well, "unpopular".
Yet, I hope to discover a breakthrough solution at some point which will help these matters - because I believe that "there has to be" one, whilst knowing full well that we are far away from that.
That said, whilst discussing the pros and cons is important, we also have to try and find those new and better solutions. Yes, a big company with the apropriate resources is one way to tackle it - but even there, I don't see anyone "replacing" or let alone competing with Youtube in the forseeable future. Even if Tiktok tried to get into long-form video, I doubt they'll have an easy time (aside from the amount of attention span they already killed over the years in their audience).
This is, even though I am not very fond of what Sia writes and promises, I still want to give it a try - who knows, there might be something useful in there that could potentially be reused.
Whilst IPFS' content addressing isn't perfect since it depends on the chunking settings, it is definitively something I would consider "a good feature". Let's find more of this, combine this, and... maybe we have something good...? x)