Is the Music Genome Project…Racist?

There are a lot of reasons to headdesk over this 538 video about Pandora’s Music Genome Project and its application to music therapy. There’s the video itself: some people in my twitter TL found its cinematography too precious. There’s the project it details: a big data project that uncritically draws assumptions about music from 18th century European music theory (CPE Bach actually wrote the book on tonal harmony), and assumptions about structure, organization, and relationships from genetics.

In addition to the technical problems with the project (that is, its uncritical reliance on Western music theory…whiiiiiich is also racist, in the sense of normatlively white supremacist), the Music Genome Project is, I want to suggest, racist. Its genetics-based approach is too too resonant with 19th century race science, and its therapeutic application (the second half of the video is entirely about this) is pretty clearly biopolitically racist.

As actual biologists have argued, the MGP isn’t genetics in any technical scientific sense: the “atomistic” bits of musical data it records are not genes, in large part because they are not heritable (that is, genes are causal in a way that the musical characteristics the MGP breaks down into “genes” are not). [1] As David Morrison points out, “ the study of musical attributes is clearly a study of phenotype not genotype,” Phenotype is outwardly observable characteristics–races are phenotypes, for example. Genotype is genetic structure or organization. The move 19th century race science made was to conflate phenotype with genotype, to think that people with different visible characteristics (skin color, hair texture, skull shape, nose shape, etc.) were of different genotypes or “races” (race in the genetic sense is a subdivision of a species, basically like a breed of dog). This conflation of phenotype with genotype is, we now know, bad genetics, and, uh, racist. (The lack of genetic differences among race-based phenotypes is generally what analytic critical philosophers of race referred to in the 1980s and 1990s when they talked about race not being “real.”)

So, the study of phenotype itself is not necessarily racist–it’s white supremacy that makes phenotypical differences racist, not the phenotypical differences themselves (or, it’s power, not objective features). So the MGP isn’t racist for studying musical phenotype–it’s racist for studying phenotype in a way that privileges white phenotypes.

How does the MGP privilege whiteness and white musical phenotypes? There are any number of ways. First, as Nick Seaver pointed out in our conversation about this video, it uses the Western record industry’s genre categories as its basic “phenotypes”–classical (which means Western art music, not, say, Indian or Chinese art music…the art music of non-Western cultures gets folded into the vernacular category ‘world’), pop/rock, jazz, hip hop, world, and so on. These are phenotypes as white culture hears them; they are modeled on white supremacist racial categories. Genre has always, always been tied to race and class, just as radio format is unabashedly tied to demographics (there’s an interesting question here about radio:demographics::streaming:psychographics, but I’ll leave that aside for now). Let’s not forget that for a good part of the 20th century, record stores were divided into two genres: “music” and “race music.”

Not only are the genres identified along white supremacist articulations of the color line, but the musical “genes” and traits the MGP identifies as “important” reflect European understandings of what music is and how it breaks down into fundamental elements. CPE Bach, cited in the introduction as an influence, literally wrote the book on 18th century tonal harmony (Jean-Phillippe Rameau wrote the other). Melody, Harmony, Rhythm, Meter, Timbre, Texture, Form–these are what every gen-ed music appreciation textbook lays out as the fundamental elements of music. But just as genetic research identified two sex chromosomes (X and Y) not because binary sex is an objective feature of DNA as such, but because binary gender/sex is a key organizing principle of Western thought (which organized how we made sense of genetic information), music research hones in on these elements not because of anything about the objective properties of sound, but because of centuries of practice and convention. As far as I can tell, the music theory behind the MGP seems fairly art-music focused. This matters because, as many pop music scholars argue, Western art music theory (its concepts and methods) aren’t really adequate to analyze Western pop music, let alone anything else. Pop music, for example, is a multimedia art in which things like the video impact how we hear the music; so, an analysis that just focused on the musical elements would miss key features of that work, features that impact how the music works and is interpreted.

The misapplication of the genome and genetics metaphors allows the MGP to naturalize white supremacist racial hierarchies as objective features of sounds themselves. That’s how the MGP itself relies on old-school racism. But then in the video Gasser, the head of the MGP, talks about how he applies this “genetic” research to music therapy, to create, as it were, eugenic music, music “that can make our lives healther,” as Gasser says in the video. (This discussion begins about three minutes into the video.) This is where the MGP starts to get biopolitically racist.

As Sylvia Wynter explains in her “No Humans Involved” essay, “our present epistemological order…is now that of securing the material well being of the biologized Body of the Nation” (64). Biopolitics establishes an ideal of “life” or “health,” and then uses race to distinguish between those whose lives it’s in society’s interest to promote, and those whose lives it’s in society’s interest to quarantine and or eliminate. In other words, it separates out eugenic from dysgenic populations. And it does this by using what Wynter calls the “technocultural fallacy,” which is the fallacy of mistaking “imperative of securing the overall conditions of existence (cultural, religious, representational and through their mediation, material) of each local culture’s represented conception of the Self” (49) for “the imperative common to all organic species of securing the material basis of their existence” (49). Biopolitics naturalizes what white bourgeois culture takes as the conditions of its own existence and flourishing as the basic biological conditions of “life” itself: whatever sustains white cisheteropatriarchal bourgeois culture is thought to be the basis of all organic life.

The MGP naturalizes what white bourgeois culture (the culture of institutions like the academy, medicine, etc.) understands as the conditions of the existence of its musical practices and its experience of musical pleasure as the basic empirical features of “music” itself–namely, the “genes” the MGP tracks. Gasser’s search for the “healthiest” music studied not every musical genome/genre, but only three–classical, pop/rock, and jazz. These three genres are the ones most embedded in ‘official’ (white, patriarchal, bourgeois) institutions like symphonies, museums (the Rock n Roll Hall of Fame, the EMP museum, etc.), and state-funded or supported “public” institutions like PBS, NPR, school music curricula, and so on. These are the most culturally and institutionally “white” genres. He didn’t even include R&B, hip hop, trance or ambient or new age (three genres that seem obviously relevant to some sort of therapeutic project, right?), techno, country, bluegrass, Indian film music, J pop, any one of a bazillion Latin genres, reggae–none of these genres even made it into the original study. So this wasn’t a study of music, it was a study of the three most institutionally white genres of music. There’s a racialized distinction between eugenic and implicitly dysgenic genres/phenotypes of music built into the very framework of the study.

And analyzing these three most institutionally white genres, Gasser found that they had some particularly eugenic phenotypical traits: “slow, heartbeat-paced tempo. Consonant harmony. Lyrical and sustained melody. Occasional bursts of rhythmic energy. The use of strings…Classical harmonies with a rock groove.” [2] These were what the MGP found to be the most “healing” and therapeutic sounds. But they’re also, uh, descriptors of most of the white canon, be it ‘classical’ art music or ‘classic’ rock. Conveniently, the most eugenic musical phenotypical traits are the whitest ones! This reminds me of 19th century phrenological charts that claimed to show correlations between skull size and shape and intelligence; of course the whitest people had the most intelligent skulls and blacks the least. In both cases, “science” tells us that white phenotypical traits are the most eugenic. That last phenotypical feature, “classical harmonies with a rock groove” exhibits a specifically neoliberal approach to whiteness, which doesn’t treat white as a category to be kept pure, but which treats whiteness as a healthy, indeed, “multicultural” and “diverse” mix. Jared Sexton talks about this as a shift from a white/non-white binary (where whiteness is exclusive and pure) to a non-black/black binary (where the privileged/white category is mixed with everything but blackness, which is seen as dysgenic). [3] According to Gasser’s research, the most eugenic or “healing” music is that which feels and sounds most white. (The opening chords of the piece are really neo-Coplandesque, there’s some straight up harmonic resolution at the end of the video…).

The fallacy of “supraculturalism” works in tandem with the technocultural fallacy. This “second fallacy…mistakes the representation for the reality, the map for the territory…narratively instituted culture-specific discursive programs” for the “genetic programs specific to its genome” (49-50). The supracultural fallacy treats culturally and historically local accounts of human life for objective biological fact. The MGP does just this–it takes a historically and culturally local account of what “music” is (in fact, the very IDEA of music as such, as ‘art,’ that’s not a universal idea!) and naturalizes it as a genomic fact. Because it naturalizes a white, bourgeois (patriarchal, too) “map,” the MGP concludes that the kinds of music most “therapeutic” for people who embody white/bourgeois/patriarchal norms are in fact the most eugenic kinds of music for all people.

All people–so now we’re back to fallacious generalizations. For someone working for a streaming service that treats taste as absolutely individual (the algorithms are supposed to taylor the stream specifically for you), it seems weird to assume that Gasser bases his therapeutic project on a generalized, universal “we” with a consistent, coherent taste. The MGP (and not centuries of music criticism and musicology) is what has finally explained what Gasser calls “our musical taste.” But who’s this “we”? Well, Gasser seems to generalize from himself and the people he interacts with. Gasser begins by talking about how his favorite thing about music is how it communicates, and then quotes/paraphrases CPE Bach: “You can’t expect to move the audience unless you yourself are moved.” This reveals a presumption of commonality with his audience, that musicians and audience are of similar enough background to be moved by the same things. (Shit, even the idea that music is expressive is a pretty culturally specific idea.) So we really need to be asking: What music is most eugenic to whom? (I’m reminded here of Mara Mills’s work on the way that test subjects’ bodies become built into audio technologies…)

The video also includes its share of the supracultural fallacy. It mistakes one kind of musical cartography for the “territory” of music itself. Throughout the video there are images of traditional Western notation–either handwritten or digital. What’s this doing? On the one hand, it’s appealing to the respectability of Western Art Music. On the other hand, it’s naturalizing the data-derived findings of the MGP—we never see any analytics, we only see staves and notes and accidentals and key signatures. The constant references to traditional Western notation also reveal the project’s privileging of the kinds of music that can be notated this way. Plenty of Western pop and non-Western genres aren’t best notated in traditional Western terms. (I mean, it’s probably possible to notate something like Skream’s “Midnight Request Line” or a PercTrax release–stuff originally crafted on a DAW–but traditional Western notation just isn’t equipped to deal with all the variables at work in, say, an Ableton session.)

Wynter argues that “the ostensibly evolutionarily determined genetic organizing principle of our Liberal Humanist own as expressed in the empirical hierarchies of race and class (together with the kind of gender role allocation between men and women needed to keep these systemic hierarchies in place) is as fundamentally secured by our present disciplines of the Humanities and Social Sciences” (53). Based on what just went down with the MGP, we can update her claim to include data science. The MGP is just one tiny bit of evidence that “the ostensibly evolutionarily determined genetic organizing principle” of Neoliberal Biopolitics “as expressed in the empirical hierarchies of race, class,” and gender is “fundamentally secured by our present disciplines” of knowledge, including data analytics. So yeah, the Music Genome Project is racist; the white supremacy is built into the very DNA of the research methods, so to speak.


[1] Genes are heritable: they are transmitted directly from one generation to the next. Musical influence obviously happens (Missy sampled Cybotron’s “Clear,” Pitbull constantly references “Rapper’s Delight,” the first B-52s record sounds a lot like Gang of Four’s Entertainment!, Ministry’s With Sympathy sounds like a survey of then-current British post-punk…), but influence, sampling, and remixing even aren’t heritability. Sampling comes closest to genetic heritability, because there is concrete material that gets lifted and reworked, but it’s still not inheritance in the genetic sense for a number of reasons: it’s not the result of sexual reproduction; there are no necessary and/or sufficient number of allelles or genes that must be transmitted–a song can, for example, simply not have harmony or melody at all; samples are not expressions of traits, samples are formal elements that don’t necessarily express anything. And to be honest, using sexual reproduction as a model for musical influence would import a TON of crappy (hetero)sexist assumptions into our understanding of musical influence and composition.


[2] A larger-scale feature Gasser names is a slow build to a climax: “striving toward apotheosis,” or “climax. And you can’t get there right away; there has to be a gradual unwinding. Because if you’re going to get to the point of having those actual chills, it has to be earned.” This idea of delayed gratification (uh, totally Freudian; see: Civilization & Its Discontents), of building dissonance to a climactic resolution, that’s central to tonal harmony, to some kinds of pop composition (though not all–some pop songs are really repetitive)…and as plenty of music theorists and musicologists have shown, often done in really racist and misogynist ways (I’m thinking here especially of McClary’s Feminine Endings). The idea that music should slowly build to a big climax–that’s just one way to build climaxes! The recent spate of EDM-influenced pop usually climaxes in the first chorus (early and often, like voting in Chicago, lol). Also, DJ sets generally build to a climax, but in ways that are different than the way classical, pop/rock, and jazz pieces do. So again, this is just a really narrow way of thinking about (a) compositional form in general and (b) how to climax a song/how to handle tension-and-release devices.

[3] Wynter puts that non-black/black binary in these terms: It is a “differential…between the suburban category of the owners and job-holders on the one hand (of all races including the Cosby-Huxtable and A Different World Black Americans, and the Black non-owners and non-jobholders on the other. Consequently, since the Sixties, this new variant of the eugenic/dysgenic status organizing principle has been expressed primarily by the growing lifestyle differential between the suburban middle classes (who are metonymically White), and the inner city category of the Post-Industrial Jobless (who are metonymically young black males). Where the category of the owners/jobholders are, of whatever race, assimilated to the category of ‘Whites,’ the opposed category of the non-owners, and the non-jobholders are assimilated to the category of the ‘young Black males’”( 53).