I appreciate the out-of-the-box thinking. Seem like you might appreciate my random idea for an in-person media store app.
And sure, if a musician wants to go waldenponding, and opt out of putting their music online, that’s their business. If they only make their tunes available on 8-track reel-to-reel at the one surviving record shop in their town, or make a single copy on vinyl and bury it in a time capsule, that’s an artistic statement in itself.
But most musicians, especially those who want to make a bit of coin out of their creative work, are interested in being discoverable. That’s why record shops and music venues exist, instead of bands waiting in their garages for passers-by to stumble upon them.
Now I take your point, in that any federated search ought to be opt-out. Or even opt-in, depending on the participating service. If a musician wants to set up a FairCamp site and make it undiscoverable on our hypothetical artist-centric search federation, again, that’s their business. But …
And how do most people search that one place? Goggle, and other centralised, DataFarming general-purpose search gatekeepers. And if robots.txt is used to exclude that unfederated FairCamp site from those too, how is anyone going to find it? I mean, sure, they could put the web address on their merch, or a QR code pointing to it on their posters, but that means the only visitors to their website are people who have already discovered them.
Worth keeping in mind too that it’s a lot easy to be discovered by serendipity if you live in a big city than a small town. For the same reason it’s easier to meet a suitable person to ask on a date if you live in a big city.
Here’s hoping 
But I think what @icaria36 and I are driving at is that because of network effects, there’s always tends to be an “everything place” portal for each thing people want to do online. YouTub for video. GritHub for code. BandCamp for music sales.
It seems like the only way to avoid everyone being herded into one big place, controlled by corporate DataFarmers, is to interconnect many smaller places. The most fundamental piece of glue that makes many places into one place, is federated search, that can find our work regardless of where we choose to host it, accessible from whichever app we use to look for it.
OpenVerse is great, as is search.creativecommons.org, and a number of other CC search portals, including LibreFM and open.audio for music. These could all be linked into the federated search system, selecting only those results that are marked as being under CC licenses. Other music discovery sites like Discogs and MusicBrainz, Hype Machine and Libre.fm, could both feed into and get results from the federated search if they chose, regardless of license.
The fundamental idea here is not to create a new music search portal, but to use protocol plumbing to link as many existing discoverability systems as possible together. So they work better for both artists and audiences.