Discogs.com is a community-sourced database of information about music releases (think: IMDB for music). It’s been around since the dawn of time [November 2000], and currently boasts 7.6M recordings, covering 4.6M artists, and has around 300k contributors.
Discogs is the de facto place to answer esoteric, nerdy, or niche music questions like: “Did Tony Levin really play bass on Paula Cole’s Where Have All The Cowboys Gone?”, or “How many lock grooves are on the D-side of Richard Devine’s Asect:Dsect?”
While its slapdash graphic design and World-Wide-Web-era vibe might be off-putting to modern sensibilities, the amount of detailed information available there is generally fantastic.
However, as a data source, it faces the same limitations as all community-sourced websites: occasional faults and inaccuracies. For the purposes of this project, I was faced with 2 primary issues:
Because of their importance in the process of shaping a song, I have elected to split songwriting and/or production teams (such as Dreamlab, Stargate, The Elev3n, The Smeezingtons, etc) into their constituent members. In all cases, these groups were only 2 or 3 people.
Musical Ensembles also presented a wrinkle, as they are credited as a single unit but likely include members from all genders, and exact member information may not be available if it’s a pickup group. I dealt with these groups on a case-by-case basis, mostly leaving them out of the data set.
As this project examines the music & audio industries specifically, credit types relating to visual art (photography, stylist, graphic design, layout, etc) were not included. I have included some unusual credits, such as Production Coordinator, Contractor, and Copyist, as these are positions which specifically facilitate the production and performance of music. Additional vinyl mastering credits were included if available.
Lastly, not everyone has a photo up on Discogs, so I occasionally used their LinkedIn or Twitter photos instead. There were a few cases where people appeared to be specifically avoiding having their photos widely circulated, which I decided to respect.
The decision to present this project in a mode of gender binaries came not from an intention to endorse or enforce this mode, but rather was made for two specific, process-oriented reasons:
While more philosophically equitable, utilizing a process which involved individually contacting each of the 398 people in this data set to personally confirm their preferred gender identification (binary or otherwise) would have required an amount of workhours effectively rendering this project too impractical to carry to completion.
As the ability to work with large data sets of individuals becomes democratized, one who chooses to work in these areas will find oneself encountering the sorts of unpleasant, lesser-of-two-evils choices once only the purview of statisticians, generals, politicians and the like; and one must likewise be prepared to answer to reports of harm done to individuals represented therein.
In the designing of this project, I conceded the possibility of accidental misidentification of persons in this data set; and made the decision to move forward regardless. Do I feel great about this? No. In essence, I could be “throwing someone under the bus” to further my own goals.
But I felt (and continue to feel) that this work was strong enough to warrant completion despite these potential harms; that it might be catchy enough to grow legs and maybe lead to some kind of concrete action. Perhaps this is just narcissistic optimism — but maybe not, you know?
So, to anyone in this data set who found xself depicted incorrectly with regard to gender identity: please accept this personal apology. I offer this not as a weightless token; but rather with a full and complete account of my thinking on this matter.