Audius – the decentralized platform for music

In a recent post we discussed the pros and cons of decentralized platforms (enabled by blockchain) relative to “traditional” centralized platforms. We noted the key benefits of decentralization include a credible commitment to not exploit platform participants and better aligning incentives with all participants, while the key costs include the problems of decision-making via voting with a large number of participants, and the legal uncertainty surrounding decentralized structures. We gave three examples of decentralized platforms that were trying to take on existing centralized incumbents (Rarible, Braintrust and Minds).

In today’s post we further illustrate these tradeoffs with another decentralized platform, Audius. It raises a number of interesting issues that we didn’t discuss in the previous post.

What is Audius?

At a high level it is a decentralized music platform developing a community-owned and artist-controlled music sharing protocol. One can think of it as a blockchain-based SoundCloud alternative, providing a way for artists to distribute their music directly, and to cut out intermediaries (such as music labels) that take the vast majority of the revenue streams created by artists. Founded in 2018, today it has more than six million unique users a month and over 100,000 artists.

The key defining characteristic of Audius from our perspective, compared to, say, the earlier version of SoundCloud, which also tried to cut out intermediaries in the music industry, is that it is a decentralized platform. As explained in its whitepaper, its main stakeholders are:

• Artists (musicians): Using Audius, artists can generate timestamped records for their creative works and register them to the Audius network. Because the network is decentralized and secured by blockchain technology, the idea is that the metadata attribution remains safe, verifiable and immutable.

• Fans (users listening to music): For a user that isn’t into crypto, the music player works like any other music player. You can create an account, follow your favorite artists, and listen to their tracks.

• Node operators: these are professional operators of two types. The first type of node operators host content and make it accessible on behalf of artists, whereas the second type index the content and provide an easily searchable interface for retrieving metadata.

• Apps: Audius is by design committed to open and free APIs, which third-parties can use to build apps on top of the network, allowing users to interact with the content in all kinds of ways on all kinds of devices (e.g. a music racing app). Audius (the company) has its own music streaming app, but the majority of listening on Audius is now via third-party apps, including third-party music players.

• The Open Audio Foundation: a foundation formed to steward the long-term growth of Audius. It is endowed with a fixed supply of tokens for incentives and growth programs.

The Audius token ($AUDIO) serves to align the incentives of these stakeholders by motivating them to take actions that increase the value of the entire platform. For instance, artists can earn tokens by uploading tracks that cross a certain level of engagement, and users can earn tokens by curating playlists that become very popular or by building API integrations. And, as with other decentralized platforms, the token is also used for governance. For a governance proposal to pass, at least 5% of all staked $AUDIO must vote on the proposal and more than 50% of the votes must be in favor of the proposal.

Tradeoffs

A fundamental benefit of decentralization is the commitment power to not exploit platform participants – or more generally, to not change the rules on them without giving them a say. This means decentralized platforms are most likely to thrive in industries where participants have been burned by centralized platforms (through increased fees, changes in algorithms, being de-platformed etc.).

Audius is a good illustration, as it was designed to stand in sharp contrast to Apple Music, Spotify and SoundCloud. The difference with Apple Music and Spotify is quite obvious since those services do not allow any meaningful direct interaction between artists and their audience (e.g. Spotify only allows artists to send one email to their audience a year, in a standardized format). The comparison with SoundCloud is more interesting. As we noted earlier, SoundCloud also initially set out to provide artists a way to directly connect with users. The difference is SoundCloud was a centralized platform, that could change its design and governance rules at any time – and it did. Some of the changes SoundCloud made over the years that upset its users and artists include:

• It revamped its website in 2013 to emphasize usage and the interests of mainstream artists over the interests of its original (indie) artists who had helped grow the platform in the first place.

• In order to appease music labels so it could offer a paid subscription service that could rival Spotify, SoundCloud went to great lengths to reach agreements with the major music labels. It allowed certain copyright holders to unilaterally remove paid subscribers' content without recourse. For instance, Universal Music Group had the right to take down any file or any account (premium or free) on SoundCloud, regardless of whether it violated copyright laws.

• SoundCloud phased out groups in 2016, upsetting many indie artists, for whom groups were the most effective way to share music on SoundCloud.

The primary driver of these changes was SoundCloud’s decision to try to compete with Spotify and Apple Music in the mainstream music market, given the much higher revenues and valuations these services were attracting. To do so, it catered to the interests of labels and investors rather than those of its original artists and users.

Audius’ decentralized structure is designed to avoid precisely this type of situation. After the launch of the Audius main service in October 2020, Audius the company has no longer had any ability to make unilateral changes to the service. Any change must be voted on by the network according to the governance system briefly described above. In this sense, Audius is clearly committed not to behave opportunistically.

On the other hand, decentralization also carries some downsides. One is that decentralized platforms cannot be very nimble in making design improvements due to the need to get stakeholders to vote on each decision, including some relatively minor ones. For example, in the case of Audius, the user experience leaves a lot to be desired: something as simple as resetting a password is quite confusing and not secure, and Audius’ own music player is rather limited. As mentioned above, Audius gets around this limitation by enabling other developers to produce their own music players on top of the Audius database, some of which are much better than Audius’ own player.

The other downside we discussed in our previous post is the lack of legal certainty. Could Audius be sued if there is a copyright dispute over the illegal distribution of music on one or more of its distribution nodes, or would the operators of the nodes be sued? Similarly, who bears responsibility for commercial deals between Audius and other companies? For instance, Audius signed a deal with Tik Tok in 2021 to allow artists to easily upload tracks from Audius to their Tik Tok accounts. This seems to have been negotiated by Audius the company without requiring stakeholders to vote. What happens then if TikTok and Audius get into a complicated commercial dispute? Having a centralized authority to do the negotiation and to enforce the terms of a commercial deal might make more sense.

Implications of the openness of Audius’ model

As noted above, in addition to being decentralized, Audius is also an open platform in that it allows third-party apps to freely build on top of its network. Its long-term focus seems to be on building the underlying content ledger and governance infrastructure, and allow user-facing apps to be built on top.

It is worth emphasizing that it is not openness per-se that distinguishes Audius from centralized platforms. Clearly, centralized platforms could choose to be open to a certain extent, like OpenSea and Android have done. Audius is different in that it is credibly committed to remaining open --- indeed, changing the openness of its APIs would require the approval of its stakeholders (which is unlikely to the extent they benefit from openness). By contrast, there are many examples of centralized platforms that started with open APIs to encourage the building of their networks, only to close them off later (most prominently, Facebook and Twitter).

Of course, a platform may want to close APIs if the apps building on top of it become too powerful and it fears losing critical control points. This can be an issue for Audius’ open design at some point in the future. Imagine a for-profit music player built on Audius that becomes very popular and dominates usage by artists and fans. Such an app could (at least in principle) start convincing artists to host their music exclusively on its service and not on Audius, thus eventually diverting both users and artists away from Audius’ network. We have discussed a similar risk in the context of turning products into platforms by opening the door to third-parties. If this were to happen, the Audius founders and executives would be unable to unilaterally close off access to such an app. However, if they could garner the support of a voting majority, they could still do so. Thus, the possibility of dealing with such threats is not ruled out by decentralization. Rather, under decentralization, it would not be up to the whim of the CEO or a board of directors to decide the fate of third-party apps.

Concluding thoughts

Despite the apparent benefits of a decentralized platform for music distribution over existing centralized platforms, we don’t think Audius will disrupt the major music labels or services like Spotify and Apple Music anytime soon. Since Audius leaves it to artists to build and manage their own audiences, it does not work well for those artists that want a lot of support and upfront capital to try to become the next big star. Similarly, the next big manufactured band trying to follow in the footsteps of AKB48, Backstreet Boys, BTS, and the Spice Girls is unlikely to be born on Audius. And to the extent there is popular demand for such music, there will always be a whole industry built around music labels and streaming services to support them. Rather, Audius will likely co-exist with such services, by addressing a different market segment – artists that cater to more niche audiences and who previously found it difficult to make a living from their music.

Finally, while Audius’ white paper paints an appealing vision of how artists’ and users’ interests can be protected via decentralization, many details remain to be put into practice and stress tested. Currently, on-chain voting seems to be restricted to node operators and the Open Audio Foundation, although anyone holding $AUDIO tokens can delegate them to a node operator for the purpose of voting. This means that for now individual users and artists can only affect Audius governance decisions indirectly via off-chain discussions in Discord and by delegating their tokens to particular node operators (a form of representative democracy). It remains to be seen how robust the Audius network will be in cases where different stakeholders have conflicting interests. Looking through the record of 48 on-chain proposals to date, there has yet to be a single proposal where there were votes both for and against. Democracy works just fine when everyone shares the same view.