Relatedness and innovation in urban music scenes: The evolution of symbolic knowledge spaces, 1970-2015

In recent years, the study of regional knowledge bases has benefited from the introduction of measures on the relatedness between its constituting elements (HIDALGO et al. 2007), allowing for detailed analysis of variety in regions and its visualization as knowledge spaces (KOGLER et al. 2013). However, these studies have focused on technological knowledge, despite the growing importance of symbolic knowledge (ASHEIM et al. 2011). This contribution aims at applying the analysis of knowledge spaces and relatedness to the context of symbolic knowledge by the example of the evolution of music scenes. It is used to test whether or not the degree of variety in a music scene is linked to innovation in music, asking: Where do new genres and combinations emerge: In specialized scenes such as Berlins or in diverse settings found in New York or London? In our database acquired from the website last.fm on 8769 artists originating from 33 cities and active from 1970-2015, social tagging data constitute music genres as bits of symbolic knowledge. Co-occurrences of genres that artists are tagged with allow for computing a relatedness matrix between genres (HIDALGO et al. 2007) that is used to compute knowledge spaces and average relatedness measures (KOGLER et al. 2013) of urban music scenes. Furthermore, by deriving a genre classification system from the data, we can characterize music scenes by their unrelated variety, semi-related variety and related variety (CASTALDI et al. 2015). Results show that semi-related variety, but not average relatedness is linked to innovation in symbolic knowledge.