They'd be discoverable by search or linking. The distributed web is compatible with feeds. It's what's happening on this site too. Except the content would be packaged and redistributable and wouldn't disappear when the original service goes down or such.
Sure, but (gen-pop) search itself has been degraded / hijacked beyond recognition compared to two decades ago, and walled gardens heavily discourage linking. Factor in that most people access internet from devices designed to consume in very specific ways from a controlled list of sources and we may as well be lamenting that nobody's finding the one physical copy of your book that's sitting on a shelf in a brick and mortar library in Omaha.
It feels like a chicken and egg problem to me -- realistically you need to meet people where they're at and appeal to the current channels' algorithms to get your content discovered, but in doing so you also reinforce the strength of the walled garden itself and participate in the diminishment of both search and the "wild" internet at large.
Mhh dunno, I don't think it helps to think about the unfortunates stuck to their media feed at this stage. Might as well make the dweb about something us nerds want. It's gonna be good... and then it'll be about keeping the parasites out that live off that cultural capital.
Something durable, censorship-resistant, network-agnostic, optionally trust-based. Maybe a bit of IPFS + i2p + content-adressable, composable, cross-linkable, multimedia documents + federated services? I feel the components already mostly exist actually...
That's a missing piece of the puzzle for the distributed web; curation and recommendation are ad-hoc and don't scale.