Music streaming as a protocol: why I was wrong about Audius

Last week I wrote about blockchain-based streaming platform Audius in this newsletter. I criticized the strategy of ‘becoming the anti-SoundCloud’ and its plans for dealing with takedowns while recognizing the exciting potential of a more complex music economy. The next day Holly Herndon and Mat Dryhurst released a new Interdependence podcast episode where they go in conversation with the Audius founders. It made me completely change my mind about the platform.

Don’t be too quick to dismiss Audius: separating the content & functionality layer is one of the keys to unlocking a new music economy.

Protocol or platform, not service

Audius should be thought of as a decentralized platform rather than as a startup or streaming service. While there is a team behind the project, it seems they mostly work to bring into existence the components outlined in their whitepaper (PDF). These aspects are open-sourced and governed by the user community of the protocol.

While a service’s success is dependent on whatever interface is slapped on to the technical infrastructure, a protocol is less dependent on any one interface. For example, a company poorly redesigning an email client isn’t going to kill email, but it may kill the app or the company behind it. The most exciting aspect of Audius is not the current interface; it’s that they’re trying to create a protocol for digital music that could have any number of interfaces plug into it.

Clipped from the Audius whitepaper linked above.

One of the most important elements of that protocol is a concept I was previously very skeptical about.

Artist-determined stream pricing

Audius wants to let artists set a per-stream rate for their tracks, so if someone wants $1 / stream, they could do so. If they want something close to whatever are the current average Spotify or Apple rates, that’s possible too.

Why I was skeptical about per-stream rates is that it creates a form of metered streaming: load up your wallet and watch the currency tick away as you listen to music. If you’re creating a SoundCloud lookalike (or: ‘like Soundcloud, but better’) to fit into the current landscape of streaming services, then metered streaming is a constraint that will inevitably scare users off. It won’t just scare people off in the sense that they won’t want to load up their wallet if they don’t know what they’re going to listen to yet, but more importantly it will have people make economical decisions about established micro-behaviours around music listening. I think in the end, that would be stressful and exhausting and users would stop coming back. However I’ve come to realize I’m wrong about this.

Differently priced music does not have to live in the same streaming interface. Lower-priced music might be what fuels background-listening type radio apps. Higher-priced music may be at the center of interfaces that connect you to new releases by your favourite artists. There are some pieces of classical music that I listen to maybe once a year, that I value more than some music that I listen to weekly. Per-stream pricing could help make up for the fact that that classical piece gets only 1 yearly stream from me versus 52 streams for a random track — whose creator I don’t even know — in a playlist I use to focus on writing.

What Mat and Holly communicate very well in their podcast with Audius is that ‘music’ is not one thing, yet the online landscape doesn’t reflect that fact well. To some degree, the compensation model embedded in the licenses dictates a lot of the user experience. In the past I’ve led teams at two music streaming services wanting to do things differently and while you can go quite a long way, in the end you have to exist in a landscape dictated by just one economic game: maximization of streams.

A change to that landscape would be refreshing and a welcome way to generate more revenue streams.

Content / interface-layer challenges

There is still a whole range of challenges to deal with in such a system. Who sets the per-stream price for example? Is it whoever has the highest degree of ownership of the sound recording? What if they change the price while it’s live and because of that the track is no longer playable through certain interfaces? Or if it all lives in the same interface: do you interrupt playback to give people a warning before an expensive track starts playing? What if their phone is in their pocket? These are just some of the design problems developers would have to think about in a decentralized system with such principles.

It’s possible that the decentralization of the functionality layer doesn’t go as far as I imagine. In any case, Audius or the apps using the protocol will have to deal with the existing national and international copyright regimes which inevitably dictate some of the economics.

In the end, it’s not about Audius versus SoundCloud; it’s about creating a new layer for music streaming. Imagining music streaming as a protocol, rather than a competitive space of services, is refreshing though. It allows for a rethink of the principles underpinning the digital music landscape without going through the arduous mental exercise of imagining small iterative improvements to solve streaming’s flaws.