Maybe the mobile client is a step in the right direction? I can hope! But the fact that I have to tell people "okay, so sepiasearch is kind of like the youtube front page...ish?" is already just infinitely dumb. Make a damn client whose name indicates, in some way, that it's a video website. And then shows some damn videos on the front page. Randomize them if you really can't stand "algorithms", but honestly, just put some videos on a page with "videos" in the url (or something similar), and you can cut down on most of the confusion I've seen.
Engineers get so lost up their own asses about this stuff because they can't see that UX is entirely divorced from functional processes. The user needs to do thing X, and the computers can only provide processes Y, Z, etc; forcing the user to reconcile with Y and Z just because they want X is the definition of "programmer design". It's refusing to engage with the very real ways in which users understand and interact with services, for whatever sake the engineers want to make up ("I don't like to obfuscate what is happening", "this is not complicated. users should be able to understand", "it would waste resources to provide a more streamlined experience", etc. These are all terrible reasons to not bridge the interaction gap between developers and users). Bluesky is my favorite example of people abstracting away the complications of this stuff. Yes, they had to centralize some parts to start with, yes they had to compromise on features - but the damn thing is instantly recognizable to anyone familiar with microblog social media. That's all peertube has had to do for years now, and they have just staunchly refused to do it.
Like I said, hopefully the mobile app is their first steps in the right direction with this stuff. They've been doing the dev stuff - made it work, made it fast, made it good! Now they just need to do the user stuff - make it simple, make it familiar, make it accessible.