Background and Vision

I have long thought that the music industry is a $100 billion business in $10 billion clothes and it's all because nobody can find any music they like. That is the basic problem that everybody faces... It's a problem for consumers; it's a problem for musicians because they can't find their audience, and it's a problem for [distribution channels] because they have a hard time introducing their customers to new music in their stores or online.

Tim Westergren, 2006

Founder of Music Genome Project and Founder/CEO of Pandora Media ($2.5BN IPO in 2011; $3.5BN acquisition by $SIRI in 2019)[7]

We firmly believe that innovations in music discovery alone can unlock billions in value for all parties in the music industry ecosystem. Moreover, we see the path to the future advancement of music discovery mirrored in the broader evolution of the internet and in solving problems that existed in earlier paradigms:

Web 1.0 - the earliest music retailer and streaming platforms set the foundation for modern discovery by inventing and implementing recommendation systems built on:

• Collaborative filtering (Amazon, iTunes): predictions based on popularity “people that bought x also bought”

• Artist similarity (Last.fm, Rhapsody): large relational databases enabled webs of data interconnections between artists as a basis for recommendations

• Song “DNA” (Music Genome Project): 32 musicologists cataloged 800,000 tracks, breaking down music into 480 attributes to feed into Pandora’s algorithm, focused on musical detail as a foundation for discovery.[8]

Music discovery in this era was hampered by the availability of online inventory and manual editorial and analytical effort (difficult to scale), which also limited the amount of music that went into discovery models.

Web 2.0 - dynamic, interactive, and social/ user-generated media unlocked a level-up in terms of discovery:

• User-generated content (8tracks): crowdsourcing of playlists, metadata, descriptions, tags, and comments proved immensely popular and useful for mood, emotional, and human-context-based search and discovery

• User interactions at scale (Apple Music, Deezer, Spotify): machine learning on billions and trillions of data points, including likes, skips, adds to playlists, adds to offline listening within algorithmic feeds

Music discovery in the Web 2.0 era has improved based on the sheer amount of digital tracks, metadata, and social cues available to refine past discovery methods (collaborative filtering, artist similarity) when augmented with user interactions at scale. However, with the top players in music streaming having removed human musical analysis, human conversations and context, and UGC (by eliminating or castrating core social layers within their platforms and apps), discovery remains hampered. Additionally, we remind readers of the insider influence on algorithms and record label pressure mentioned above.

Web 3.0. - Decentralization, user control, and transparency are key tenets for the future of discovery: This doesn’t yet exist meaningfully. But it should. And it will.

We see an opportunity to dramatically improve modern music discovery and remove the double-dealing and compromises inherent in the current status quo by shifting control of the discovery algorithms from centralized Web2 platforms to music fans and listeners.

Therefore, we propose decentralizing the verification, collection, cataloging, and labeling of key algorithmic inputs, crowdsourcing, and verifying these elements and parameters at scale through a Human-Agent Network in a dedicated version of our music streaming application. The algorithm will be open-sourced, first for use in our MixerFM streaming application and eventually as a data offering and protocol that can be implemented into third-party applications. This will help a new generation of music listeners discover their favorite artists and once again skyrocket music industry revenue.

Our decentralized approach to music discovery, as outlined further in this whitepaper, offers a revolutionary shift from static systems (like the Music Genome Project) and also from dynamic systems built on machine learning trained mainly on the interaction of large numbers of users with generic metadata (i.e., Apple Music and Spotify). By leveraging crowdsourced intelligence and feedback in our Human-Agent Network, our proprietary “chain of context” human sentiment parameters layered onto a graph-based architecture, blockchain transparency, and privacy-preserving technologies, our system creates a more personalized, context-aware, adaptive, and transparent music discovery experience. MixerFM will create a superior music discovery experience and the future of sound in Web 3.0.