Early collaboration networks may hint at which musicians succeed

A musician’s earliest collaborations may carry clues about their long-term career, according to a study published in Applied Network Science. Keegan Woodburn, Caleb Frey, Chad Petterson and Ian McCulloh constructed a large-scale collaboration network from MusicBrainz and Spotify data and investigated whether the structure of an artist’s early-career collaborations could predict long-term artistic success, measured by Spotify follower count.

Their analysis indicated that structural tendencies such as triadic closure, homophily and productivity-driven exposure shaped early collaboration. Triadic closure describes the tendency of people who share a collaborator to begin working with each other, homophily reflects a preference for connecting with similar artists, and productivity-driven exposure suggests that artists who release more work encounter more potential partners.

One notable pattern was a shortage of weak ties, which the study found to be underrepresented relative to what the structural models would expect. In network terms, weak ties are infrequent, low-intensity connections, often with people outside an artist’s usual circle.

Graph embeddings are compact numerical representations of a node’s position in a network, produced by algorithms that learn from the graph’s structure, while interpretable features are measures with clear meanings. The findings demonstrated that early-network connectivity added predictive signal beyond metadata baselines. In particular, weak ties, cross-community bridging and deviations from expected structural patterns contributed information that the simpler baselines did not capture. Cross-community bridging refers to connections that span distinct clusters of artists.

The study frames success as follower count on Spotify, a platform-based proxy for audience reach rather than a direct measure of artistic quality or income. The work is presented as an early-access, peer-reviewed accepted article that is citable and carries a permanent DOI; the authors note that this version is subject to further edits and will be replaced automatically by the final Version of Record. The article is open access, allowing other researchers to examine the modeling choices.

As the final version of record appears, the predictive value of early collaboration structures may be refined, but the central result stands: the shape of a musician’s first network appears to carry information about the career that follows.

Subject of Research: Predicting long-term musician success from early-career collaboration network structures

Article Title: Predicting musician success from early career collaboration structures

Article References: Woodburn, K., Frey, C., Petterson, C., & McCulloh, I. (2026). Predicting musician success from early career collaboration structures. Applied Network Science. https://doi.org/10.1007/s41109-026-00823-1

Image Credits: AI Generated

DOI: 10.1007/s41109-026-00823-1

Keywords: music, network science, collaboration, Exponential Random Graph Models, Spotify, MusicBrainz, weak ties, graph embeddings, triadic closure, homophily, career success, predictive modeling