While watching the DM's Lair video, I had an epiphany: I don't think high-levels are now (or have ever been) intended to be played. It's like buying a Powerball ticket when the prize has reached $500M. It's aspirational. It's the story of the American dream - "if you just work hard enough, you too can become Jeff Bezos."
Realistically, it's never going to happen, but it's an extra power fantasy grafted on to your existing power fantasy of playing D&D.
As usual: Except for 4e. Which had multiple full 1-30 adventure paths (including some
very highly regarded ones from ENWorld!), as well as publishing multiple books specifically containing adventure possibilities and hooks for Epic-level characters.
Personally, I don't think it's true that high level isn't intended to be played. Instead, I think it's that the people who make the game keep making the same mistake, over and over, because it doesn't have much negative consequence for them in the short term...and by the time the long term rolls around, they're switching to a new edition or updating the current edition (as we see with 5.5e, regardless of what they choose to call it.)
That mistake is, quite simply,
they don't test high level play. Or, at least, they do no more than a worthless token effort and presume that patterns which existed at level 1 will continue to exist at level 15.
Unfortunately, game design is significantly more of an effort than a lot of people realize, and it requires more than just the
hope that patterns of player behavior will remain consistent when the numbers informing that behavior change. I wish more designers--professional and amateur alike--understood the critical importance of actually doing real (if basic) statistical analysis of the system. For a game built around a probabilistic, mathematical framework, its creators have a tendency to be unbelievably
blithe about how probability actually works and what math can actually tell you about things!