PCM, ASCII, and straight RGBA bitmap encodings aren't going anywhere. By extension, derived formats like WAV, UTF-8, and word processor files and webpage HTML are mostly fine too. The formats are structurally simple enough that even if the associated file extensions were somehow to be forgotten, all you'd need to do to invent them again is hand the file to a bored nerd over the weekend.

I think you kinda got the BBC and NASA problems backwards. The BBC's had a couple of prominent incidents where digital "preservation" that was supposed to be eternal couldn't even be opened anymore after a couple of years, like their Domesday Book/Project application thingy. They've also lost a bunch of old shows, like early Dr. Who episodes, I think. NASA didn't just forget how to read the Apollo tapes; they overwrote them to reuse the tapes, as was their standard practice at the time. The original signal and tapes were very HD (or analog), but most of the videos we have today are from the TV camera that they pointed at their own TV screen last-minute when they realized they didn't have an adapter for broadcast— The equivalent of a grainy cell phone photo of a screenshot, basically.

The BBC and NASA incidents happened in an era before computers were a ubiquitous commodity product. So, everyone and their cat was basically inventing their own obscure single-implemention proprietary file formats at that time. Nowadays we have established technical standards, as well as formats that have already sorta stood the test of time based on their utility and simplicity— and millions of people who already know how to read them— so that particular vector for bitrot isn't really as much of an issue anymore.

…That said, I think I sorta missed your point. What you're really saying is that stewardship of digital records is much trickier and riskier than stewardship of physical records— and that results in stuff being lost. And that is absolutely true.