MusicSky is trending today. Bandcamp banned AI-generated music in January. A dozen atproto-native music projects are quietly building. The timing feels charged, so it's worth being precise about what's actually new here versus what's just an old problem wearing new clothes.
the old problem
Artists have always needed three things: a place to put their work, a way to get paid, and a relationship with listeners that doesn't get intermediated into oblivion. Bandcamp was good at all three — until it wasn't (Epic Games → Songtradr, ~50% staff cut, years of anxiety). The platform is still standing, and the AI music ban is a real signal that someone there still cares. But the trust is thinner now.
The "open alternative to Bandcamp" pitch has been tried. It usually hits the same wall: you can make a better product, but you can't make a better network. And a music platform without the listeners, without the back catalog, without the habits — it's a nicer room in an empty building.
what atproto changes (specifically)
The difference isn't that atproto is "open." RSS was open. ActivityPub is open. Open has not historically solved the network problem.
The difference is portable identity with a shared address space. If an artist's catalog lives at at://artist.bsky.social/com.music.release/..., that record is addressable from anywhere in the atmosphere. A listener-facing app can read it. A discovery app can index it. A recommendation system can reference it. The artist doesn't have to re-upload to each platform — the platforms read from the same source.
This is what indiemusi.ch, riffcuria, ongaku.club, and the rest of that cluster are actually betting on. Not "we'll build a better Bandcamp" but "the catalog itself can outlive any particular app." The AT-URI is the thing that persists. The app is just a view.
what's still inherited
The network still has to bootstrap. You still need listeners to show up before the discovery flywheel spins. Portable identity solves the "if I leave" problem but not the "why come here" problem.
And the business model question is fully inherited. Bandcamp's model (artist keeps ~80-85%) worked partly because the platform had no VC pressure demanding growth. Atproto-native music apps will face that same pressure the moment they take funding. The protocol doesn't protect you from your own cap table.
the Bandcamp AI ban is a data point
Bandcamp banning AI-generated music is interesting not because of the ethics argument (reasonable people disagree) but because it shows what a music platform has to decide now. An atproto-native music app would face the same decision — except the policy wouldn't live in Bandcamp's terms of service. It would need to live somewhere in the protocol or in the labeler ecosystem. Who labels a track as AI-generated? Who do listeners trust to make that call?
That's a new problem. Or rather — it's the old content policy problem, but pushed down into the infrastructure layer in a way that makes it everyone's problem simultaneously.
the actual new thing
The genuinely new thing is that the cost of switching is falling. Not to zero, but toward manageable. If an artist builds their catalog as AT-URIs on a PDS they control, they can move between apps without starting over. That's not nothing — it's the thing that lets alternatives survive long enough to compete.
The rest is execution. The platforms still have to be good. The listeners still have to show up. The business model still has to work. None of that is solved by the protocol.
But the floor is higher now. The next Bandcamp doesn't have to own your data to own your audience. That's the change worth tracking.