Is a vibe the same thing as a style?

Plus some info on my updated subscription model

Is a vibe the same thing as a style?

One scholar thinks so. In his 2017 article “A Theory of Vibe,” Peli Grietzer argues that the machine learning technology known as autoencoding produces “vibe” as an object of knowledge (8), and that this vibe is analogous to what literature scholars would call a style or an aesthetic. Let’s take each part of that claim in order. First, as Greitzer explains, “autoencoders...deal entirely in worlds rendered as sets of objects or phenomena” (3), i.e., as sets of data. They then churn through all this data to sniff out “structural relations of alignment” among the objects or phenomena (data) in that set. For example, music streaming services break songs down into subindividual data points like instruments, lyrical content, presence or absence of vocals, vocalist gender, geographic region, etc., and then look for songs that have similar alignments; they also do this with user listening data through a practice called collaborative filtering. As I’ve argued elsewhere, this practice of collecting a set of objects/phenomena and then calling whatever alignment that collection produces a “vibe” maps on to how social media users craft posts hashtagged with #vibe/s/z. The technical processes behind contemporary AI/ML/algorithms definitely map onto the vernacular discourse of vibes. So, when machine learning algorithms are trained on, say, canonical works of literature, they learn to reproduce its vibe. Or, as Greitzer puts it, “suppose when a person grasps a style or vibe in a set of worldly phenomena, part of what she grasps can be compared to the formulae of autoencoder trained on this collection” (6). He’s arguing that when a person interprets a work as having a particular style, they’re parsing that work the same way autoencoders break down and model that work. For example, when I notice that a song is a dance remix of a rock song that is heavily compressed, I am breaking it down into its subindividual parts and noticing that these parts match up with the generic template for a bloghouse track like The Sisters of Mixing’s “Lucretia My Reflection” remix or DFA’s remix of Le Tigre’s “Deceptacon” - they’re all dance remixes of rock songs that sound highly compressed. So, for Greitzer, a vibe is a method of perceiving the way ML algorithms perceive the world, and a vibe is analogous to a style.

However, if vibes are vernacular versions of ML perception, vibes are not styles (or genres). Styles or genres classify works based on common aesthetic features or properties, like instruments, song structures, or timbres. A rock song traditionally has guitars, vocals, and drums (maybe keys or sax) and a verse/chorus structure. Hyperpop draws on lots of different genres in a way that takes genre-specific elements to their extremes. Acid house uses 303s. You get the idea. Greitzer thinks these qualitative features are what autoencoders pick up on as data points; as he puts it, “a ‘vibe’ or a ‘style’ [is what] we can sense when we consider all the myriad objects and phenomena that make up the imaginative landscape of the [literary] work as a kind of curated set” (3). A style focuses on the contents of that set--the collection of aesthetic features or properties that, put together, make a song a rock song or a reggaeton song. But that’s not what algorithms are perceiving--they’re not comparing the content of these data points (the common aesthetic features) but their alignment. As anthropologist Nick Seaver has shown, recommendation algorithms model data as vectors or orientations in space and then evaluate the similarity or dissimilarity of these vectors’ alignment to one another. The thing these algorithms perceive isn’t the qualitative content of the data set, but their form or orientation. So, algorithms aren’t perceiving styles or genres. They’re perceiving vibes, where the alignment among data points matters more than the contents of that data. So, it’s correct to say that algorithms perceive vibes, but it’s incorrect to treat vibe and style as interchangeable. To be a bit overly reductive, think of styles as describing the contents of a set of data points, but vibes as describing those points’ form or arrangement.

“Liminal space” is in many ways THE paradigmatic vibe because it clearly illustrates that vibes are common alignments of data points which themselves may be quite different in their contents. As Kyle Chayka describes this vibe in The New Yorker, liminal space is the vibe that unites such disparate locations as “Office-building hallway. Dead-end street. Loading dock. Nighttime hotel atrium.” The objects in each of these locations are all quite different, from the bright fluorescent lights, ceiling tiles, neutral paint, neutral, low-pile carpet of an office-building hallway to the concrete, metal, and whiff of rotting trash of the loading dock. Though each of these settings are populated by quite different objects, the alignment of these objects is the same across each instance--nothing’s moving, nobody’s moving, there’s a void of activity, and the objects highlight this void. That’s what makes these spaces liminal--as Chayka explains, in 2021 “liminal space” has come to mean “pretty much anything empty and weird.” So, to give off liminal space vibes, it doesn’t matter what kind of objects you have collected; what matters is whether or not those objects are aligned in a way that evokes emptiness and weirdness. Vibes are alignments or orientations that make some perceptual contents more palpable and some perceptual contents less so.

This shift from style to vibe is perhaps very evident on music streaming services such as Apple Music and Spotify, who curate playlists for vibe in addition to genre. I talk about this in the paper I linked to earlier in this post, but I’ll expand a bit more on that discussion here. “Chill” is a common vibe across both services. The relevant thing about chill is that any genre can be chilled. This screengrab of Spotify “chill” playlists includes pop (“Chill Hits”), country (“Chillin on a Dirt Road”), jazz, and singer-songwriter focused playlists--different genres, but all of them chilled. 

Other Spotify-curated playlists with “chill” in the title span all sorts of genres, from country to classical to R&B to 80s music. That latter “80s Chill” playlist includes songs by Madonna, Ryuichi Sakamoto, U2, The Durutti Column, Miami Sound Machine, and The Neville Brothers that have nothing sonically in common except the fact that they are toned-down versions of these artists’ otherwise widely divergent genres. All those artists write songs with varying aesthetic properties, but this playlist collects songs from their otherwise divergent catalogs that all share the same “chill” alignment of features. Thought The Neville Brothers, Miami Sound Machine, and Sakamoto all work in quite different genres and expressive traditions, Spotify’s algorithms can easily sniff out which songs in their catalogs share the same vibe.

Vibe is a Thing nowadays because it’s a pop culture version of the mathematical and technical tools tech such as autoencoders use to perceive us and our world. It’s a way for people to perceive themselves and their world the way algorithms perceive things. And though I don’t think genre or style is going to go away completely, vibe will increasingly become the language or discourse through which we negotiate the tensions that have been more traditionally negotiated through genre or style…tensions such as the race/gender politics of rockism and poptimism, or the racial politics of rock-vs-R&B or hip hop-vs-country, etc. Think about it this way: people still use secular and sacred as ways of talking about and categorizing music, but whereas Westerners used to get extremely upset by music that violated conventional boundaries around the secular and sacred, nobody really…cares that much anymore. Transgressing these boundaries don’t matter much to mainstream audiences these days. That’s because we use other musical boundaries to negotiate contemporary social tensions--boundaries like genre (remember “Old Town Road”? It wasn’t that long ago!), and boundaries like vibe. Or, more precisely: the purported boundaryless-ness of vibe, its ability to draw together what more traditional boundaries like genre kept apart (as in those chill playlists), is presented as more progressive and forward-thinking than the adherence to rigid, traditional boundaries. Vibe is one of the vectors through which we’re negotiating post-identity patriarchal racial capitalism.

I updated the subscription model. Last year I started this newsletter as a way to raise money to bring guest speakers to my classes. I am not bringing any guests this spring, so I don’t need raise money for that. So, I lowered subscription rates to the lowest levels Substack allows, and from now on the main difference between paid and free subscriptions is that paid subscribers get to comment on posts and participate in any community features/activities. Everyone, paid and free, gets to read the posts. I’ll use any revenue I generate to replace my reduced research and travel budget—paid subscriptions will directly translate into more content to read here.