Page - (000155) - in Knowledge and Networks

148 I focus on one specific kind of relationship: joint session recordings. In joint session recordings, musicians are in their closest contact with the commercial side of the jazz art world, and their final product success has a direct impact on their trajectories. To gather the album data, I consulted the Crazy Jazz website. Crazy Jazz is a compact disk (CD) seller specializing in jazz titles. The universe of music albums at Crazy Jazz com- prises 5572 LPs (long play albums records) produced between 1930 and 1969.4 The ideal would be to consider only those titles that reflect original recording sessions as relational data. However, many of the titles available are collections and compilations that could interfere with an accurate interpretation. Collection titles could establish relationships that did not exist (e.g., the compilation All Star Swing Players could wrongly link Benny Goodman to Duke Ellington). On the other hand, compilations may cover a lengthy period of the artist’s production and include changes of style, making it difficult to identify when relationships with other play- ers were established, and when the artist in question developed a certain style (e.g., Dizzy Gillespie’s biography could mistakenly relate swing to Afro-Cuban jazz). In order to minimize these effects, I eliminated collection albums from the data- base. Nonetheless, I could not simply eliminate compilation albums, because sev- eral titles were originally released individually as singles and later reissued on compilation CDs. As a solution, I limited compilations to a maximum 5-year range, taking the release year as the initial year of the period covered.5 Preparation of the Network From the information taken from the LPs and their featured artists, I built three preliminary databases: a list of musicians, a list of LPs, and a list of relationships between LPs and musicians. The LP database was divided into 85-year periods: 1930–1934, 1935–1939, 1940–1944, 1945–1949, 1950–1954, 1955–1959, 1960–1964, and 1965–1969. For each period, I counted the number of titles recorded by all pairs of musicians. In order to build the database of ties, I considered only those pairs that had at least two titles recorded in common. Next, I dichotomized all relationships without consider- ing the strength of the tie. Table 8.1 shows descriptive statistics for each period’s network. 4 It is worth highlighting the consequences of building this database using currently available com- mercial data as its main source. The first consequence is that I have had to rely on Crazy Jazz’s criteria when it comes to defining what is relevant. In terms of revenues, low-selling albums might not be included, although they may well have reflected important relationships at their time. Additionally, Crazy Jazz’s commercial classification of what jazz is may differ substantially from the viewpoints of other members of the jazz community. These sampling constraints underline the exploratory status of this paper. See DiMaggio (1987) on classification problems in art, as well as proposed solutions using network analysis. 5 In future investigations I intend to turn to a direct analysis of recording sessions in order to elimi- nate the distortions I experienced with the LP titles. C. Kirschbaum