Human-Agent Network

The Human-Agent Network in MixerFM is a decentralized, community-driven system that feeds data into our algorithm and AI models while also ensuring the data is accurate, unbiased, and “rich” (augmented with attributes and human context elements unique to our platform), resulting in a superior listening and discovery experience. The network includes ALL listeners whose searches, streaming logs, and interactions with the app contribute to user data that impacts our algorithm. However, a class of advanced users called “Human-Agents” plays a key role by actively editing, enhancing, and moderating data around individual tracks and track “mixes” and providing targeted feedback on algorithmic outputs.

Human-Agents take responsibility for deep tagging the 43 million (and growing) tracks in the MixerFM database with words and phrases that capture the essence of each track and describe the mix of a sequence of tracks. User agents also validate or assign a wide range of metadata like genre, mood, tempo, and even emotional tone to songs. This system contrasts with traditional, hierarchical tagging models using a dynamic and associative approach, where tags directly influence the discovery process based on real user actions.

For regular users, the Human-Agent Network operates behind the scenes. Regular user actions still contribute valuable usage and engagement data integrated into the platform’s recommendation engine. Our system's design ensures that every recommendation is based on richly tagged and verified data. This allows all users to benefit from a music discovery system that accurately surfaces music tailored to individual tastes while being community-curated.

The ranking system for Human-Agents in MixerFM is in place and designed to reflect their expertise, contribution, and influence on the platform. It motivates Human-Agents to consistently engage in curation, validation, and prediction tasks, rewarding those who provide high-quality input with higher ranks and privileges. The system also fosters competition, collaboration, and ongoing participation.