[Superlecture]: Nobody Listens to Music Anymore (Marek)

This is the great report, come. Hey, happy new year. Here's the little bonus episode with just me. I thought I'd throw in a recording of a lecture that I've been giving recently called "Nobody listens to music" and for her. To kind of game your interest a little bit, the first time I gave a condensed version of this talk, the former head of "Generative AI" and Spotify literally heckled me and it was super embarrassing and amazing. So hopefully that gets you excited about this. This is a more recent version of the talk I gave at Columbia earlier December, which was tailored for a conversation on the future of music. I don't have the slides up here, which like if you know, you know, we're doing a section like this, it's kind of the sparrage of quotes and citations. So just throwing in right at the top that like all of these ideas are obviously intellectually indebted to many people with a lot of room for disagreement but a lot of big shout outs and a lot of love. Specifically to little internet of new models, but you know, Carly Basta as well. To Holly Herndon and Matt Dryhurst's work to our interviews with Jennifer Walsh, Daniel Falsted and Jen Lee Young, Tika Brain, Rafique Anadol and Alex Reisner, and to recent conversations with Diana Rodriguez, DMR, Sysx Club, Micah Silver and the ontological turn to aka Auntie. Actually, you know, just pulling up the deck here, I got to kick this off with this quote from Mackenzie Works, Anthropocene Lecture Series in 2017. I see Utopia as intensely practical. Actually, it's the practical taken to extremes. That's a nice, that's a nice way to kick this off. Okay, here's the recording. Thank you so much for having me. It's always fun to be invited to talk about the future. Maybe, you know, to start with like a little bit of a caveat, there's obviously no one future, and the future isn't a destination. It's more like a kind of infectious mutating disease with different incubation patterns and symptomologies. So, like when I talk about the future as a thing, I'm talking about the more extreme tendencies of the present as symptoms, or maybe potentially fertile mutations. I'm here to talk about the future of music in a couple of different ways. And ultimately, I think hopefully it's clear to most people that music is changing, and it's changing in a really big way. And there's like a kind of boomerasm that always comes with saying things, like things aren't how they used to be back in my day. Things are different now, or even to kind of suggest that things might be worse, which is, I really want to make it clear that's not an argument that I'm making. So, I totally acknowledge the kind of boomerasm of the critique, or of the kind of thread here, but it also feels pretty reasonable to point at, you know, the incredibly volatile economic and technological landscape around us and say, like something is definitely happening right now, right? And I do think that a lot of the change around music has to do with how the musical economy operates and the incentive structures it produces for technology, which all kind of, you know, roll up together and flow for the most part downstream into culture, which, you know, reappropriates those structures in its own interesting ways, but is for the most part reactive, at least to the kind of forms that are, you know, provided intentionally or not, to culture to kind of play around with. I think this talk will start to get a little provocative when I really get to suggest that many mainstream understandings of what music is, like as a collection of individual pieces or songs made by individual artists or songwriters or composers or whatever, that then you have like another series of individuals that listen to you, and they build subcultures and identities around, like that this understanding of music is actually something very modern, very recent, and kind of, you know, technically incorrect. It's possible that music isn't these things that we've kind of retconned a history of music around a very Western late 20th century way of thinking about production, authorship, and identity, and this kind of misunderstanding about music is being torn down now, like super, super, super hard by a kind of economic and technological realism, where, you know, these economic and technological forces have realized, maybe ahead of us, or ahead of, you know, most of us, that all of these constructed beliefs about music are artificial and unnecessary, something that can be kind of streamlined away or optimized against. And the resulting, you know, disenchantment that we, you know, experience as musicians and listeners and just people can feel very painful, but the moment also represents an opportunity to really rethink what music is or what it might be, how it might work, who gets to do it, and why. So I've broken up the lecture into like eight, these eight points, and I start with the economic side of things, then I move through technology, and then I kind of emerge into aesthetics and some really kind of, you know, future lookingness. So the first thesis is that a piece of music isn't really a thing that's bought and sold anymore, you know, economic terms would call it a commodity, like a, you know, banana or a chair or things that have associated markets, right, you can, you can buy and sell them directly. And something particularly interesting about music is that basically every vehicle to buy and sell pieces of music has, has died. Instead, music is mostly delivered via other vehicles like streaming or live performance. Bandcamp, for example, was kind of the last platform built around the idea of music as a commodity and like, you know, even though it was beloved by so many people from the musical community, it floundered economically and ultimately was sold and gutted by epic games and later song trader. And you know, there's some exceptions here, obviously like the 34 different physical versions of Taylor Swift's tortured poet's department, the kind of secondary markets around collectible merch or were also like niche user specific tools like beatport for DJs. But there's a great anecdote that the musician and just kind of general genius person Jennifer Walsh shares about how she, she observes this younger family member who when she plays Taylor Swift on vinyl. She also play the same tracks on mute on Spotify so that the tracks appear as played in her wrapped and that you know, T Swift kind of gets the streams. So the priority has changed like the ledger of listening has moved to the cloud. And this is important because when you when you stream music, the way that your money moves isn't in a direct relationship individual things you're not buying anything per say. So you're not going to be able to sign up for subscriptions to Spotify or Apple music, which give you you know, temporary access to music library. Streaming services are basically landlords, right? They're collecting rent on your ability to access a space. But it's not just about library access because ultimately like this very little difference in the libraries that the different mainstream streaming platforms use. And ultimately their ambitions are to contain kind of the totality of everything, right? And then the question is like, what's the difference? Like if it's if it's not the libraries, like is the difference between Spotify and Apple music like a UI? Is it wrapped? Is it different integrations into you know Apple hardware or something like that? And and actually like the real difference is the algorithm, right? To the extent that any kind of real difference exists. So most streamed music is listened to in the forms of playlists served to you by a platform. So these these streaming platforms aren't really charging you rent for access to a library. They're kind of more charging you rent for access to those algorithms as you know, they're kind of points of differentiation to kind of playlists that prompt you with cost effective music in a way that feels unique and personalized. Because what does Spotify really want you to do? I mean ultimately it wants you to renew your subscription or ideally to kind of not be sufficiently annoyed by anything to such an extent that you'd actually remember to cancel your subscription. But Spotify doesn't care if you listen to music and actually it would kind of prefer that you listen to a little bit less music. Remember to Spotify music is a cost it's it's not the product to be clear it's a massive cost in 2023 Spotify paid out 9 billion USD in royalties which was more than half of their revenue at 14.3 billion USD. Spotify only you know recently approached profitability this year despite being publicly traded since 2018. So it's not just that music is a cost for Spotify. It's their biggest cost. It's an existential cost. The reduction of which is obviously the biggest possible business priority for them. So yeah, Spotify doesn't necessarily want you to listen to music instead really the intention is to get you to passively listen to a kind of sufficient amount of music that you're not going to cancel your sub and you know maybe induce you to advertise the platform by for example posting your wrapped and you know any returns that Spotify might get on their investment when you do actually listen to music are ultimately redirected into data collection exercises. So this is interesting because it means that the tools that most humans use to listen to music aren't motivated by things that have anything to do with actually actively consuming music. They're motivated by passive consumption by making their platform as invisible as possible. This is kind of like a general characteristic of software as a service economics like really the goal is to get you the kind of user or the customer to use as little of the platform as possible to keep the cost of goods sold down while at the same time retaining you through kind of very very easily invisible methods of attachment. So a few conclusions from this kind of first thesis like first it's really hard to buy and sell individual pieces of music like that's not really how the market works like we've been really told this loud and clear by the market like that's not how people want to consume music either. Music is is by large consumed as a free object by all of us. music. Streaming isn't really invested in the consumption of music as a product, but it is interested in having you renew your subscription, as I said. And then I think the kind of real clincher here is that the kinds of technologies that these companies are interested in don't really have anything to do with music production, but instead focus on the problems of music as a cost and not as a product. So those technologies are music distribution, music licensing, and music information retrieval. And so that's the future of music technology, and that's the ethnicity of the future of music. And so there's another side of this coin, right, which is like if streaming services don't sell music, and if this is the main vehicle through which music is consumed, that means musicians don't really sell music either, right? Instead, most musicians make money from other things, things that might be unlocked by making music, but for which music isn't the actual thing that you're selling. Music also becomes a cost to you, a musician, just like it's a cost for a Spotify or Apple music or whatever. Music is a thing you pay to do, which could be if you're a professional musician, a kind of tax deductible operational expenditure for the way that you actually make money, which is almost always advertising. Being a creative, so to speak, is essentially a full-time content marketing gig. Either for your own brand, selling your own merch or for tickets to live experiences, or more lucratively for affiliate partnerships with other brands, like partnerships with technology companies or partnerships with instrument makers, but honestly a whole lot of generic lifestyle partnerships more than anything. Being a creative in this respect is often about selling the experience of being a musician, not really selling music. Even in music academia, this is often literally the function of academic faculty, right? You're producing marketing collateral for an institution in the form of research, and if you're an academic musician who's not paid for research, but rather for teaching, to be clear, you're not a professional musician. Like this is another case where your work as a musician is an operating expenditure for your work as a teacher, right? And this might all sound weird or counterintuitive because we're surrounded by narratives people making money from music, like Takashi 6'9, making $6 million for one record, or Pink Floyd's sale of their catalog to Sony the summer for $400 million. But of course these deals represent the tiniest 1% of the 1% of the 1% of the professional musicians precisely because those deals represent speculative investments in future payout for streaming and sync licenses. But you need like 70 million streams a year on Spotify to bring a label back 40k and that's what the 5050 split. In 2024, Spotify estimated the total number of professional musicians on their platform as what seems like a crazy low 200,000. But indeed, that does feel like a pretty reasonable ceiling on the total number of professional musicians in the market today. So yeah, that's like 200,000 total professional musicians in the world. In comparison, there are 7 million full-time influencers in the US alone. It makes sense. And you can criticize Spotify all you want for constructing this archetype of the professional musician and then generating the unit economics through which that kind of full-time understanding of professional musicianship is constructed. But at the same time, that also doesn't mean it's not true, right? Everyone else for the most part, including those at the top, is also doing something else. And that's great. Like that's fine. The point I'm trying to make is that the economics of music making are trending in a direction where music becomes less and less of a professionalized or specialized activity and more of a thing that people pay to do in order to do something else. It's a cost and is therefore treated by musicians like a cost. This isn't to suggest that there was ever a situation in which the commodity form of music really reigned supreme. But really what I'm describing is the economic transplantation of a professional-class of musicians from different historical modes of money making into a much more standardized economic model constructed around a completely standardized vehicle of music consumption, which is streaming. So now I want to talk a little bit more about technology across a global survey. The average person is exposed to more music every year, now almost 21 hours a week. Most of that music consumption is passive and it's also indirect. For the most part, we're consuming music not through direct music streaming platforms, but through other platforms that stream audio music as non-diagetic background content paired with video. And of course, this includes TikTok, Instagram, YouTube, Netflix, basically anything. For proper streaming via Spotify or some other similar platform, made up about 32% of a generic American listener profile, but sync streaming via social media made up 36% of that same average person's music intake. That doesn't account for music that plays during long-form video, like YouTube or Netflix or whatever. I want to get at the question of what music is becoming through a bit of a provocation about what music is in the context of the various platforms that stream music. So generally speaking, it's kind of a little blob of data. But there's an audio track in there somewhere, but that's kind of really the most expensive way to store and think about music. Instead, these platforms think about music in the form of lower resolution, more kind of cost-effective data structures. That metadata represents that music to a platform more than the actual audio track itself. The audio track is just basically a link to data storage, but the metadata is digested. It's rich, it's usable, and it's valuable. So you could say that in the context of a streaming platform, that metadata is the music, or it is the information, not the actual audio track itself. That audio track is kind of a more of a raw thing, right? It hasn't been actually rendered legible as information yet. So oftentimes the metadata itself, in terms of what it contains, things like genre, emotion, mood, historical style, subculture, alongside basic informational tags, instruments, key, composer, and that's one kind of useful register of metadata. And even more useful register of metadata is like, how is that piece of music being played with what velocity, after what other things, in correlation to what, and then how those factors relate to those kind of prior more informational or affective descriptors. When you stream music, you're actually adding new metadata. You're literally writing into a database that this music is good. This music is bad. This music makes sense after this other music, that these are the most interesting parts of this piece of music, and that these parts are totally skippable or not interesting. Spotify has literally thousands of full-time software engineers who spend time understanding this massive relational construct using it to make informed decisions about how to deliver the next piece of music for you or for a generic listener based on particular kind of genericized listening histories. And in this way, the kind of ontology of music within any software platform is actually a huge network of relationships. On K-A-L-A-M-I-K-D-W-E would say, "It's more than a network. It's an embedding. It's a latent space. It's an absolutely massive and almost unnavigable constellation of relationships across millions of dimensions." And frankly, that seems a lot closer to what music actually might be than how we've been trained to think of it. Instead of individual products made by individual people, these algorithms understand music to be a deeply, deeply relational, flexible, social dynamic. No music exists in a vacuum. Instead, this is a giant web of all of music, and it kind of flexes and twists over time. AI gets involved here in a big way, obviously, dumping tons of money into understanding music as metadata. Some companies are devoted to building more sophisticated, more cost-effective ways of mapping musical relationships, helping advertisers, for example, associate just the right piece of music with just the right message. Others involve using musical metadata to produce more music. Sooner or UDO or Fugato don't exist to solve a creativity problem. They exist to solve a licensing problem. The goal is to build a world where a mood playlist could actually be more or less generative and license-free. And to be honest, they also solve that problem for creators out there. Making music is a cost for creators too. If it's not what you're actually selling, you want to keep that cost as low as possible. So if you're a guitar influencer on TikTok, why record or pay for samples of a backing track? If you're a spooky story influencer doing spooky story podcasts and ultimately, you know, rang an advertising business on top of that, why pay money for music that underscores what you're saying? You can accomplish something license-free really, really easily and quickly. Sooner or UDO and Fugato work by extracting audio features from pre-existing music, likely, massive annotated piles of audio provided on the sneak from one of the many large libraries out there who have an interest in reducing licensing costs. These tools use information from those libraries to generate new music without paying licensing fees to the artists within those libraries because, well, the legal apparatus to prosecute them for doing so doesn't really exist and probably never really will because the cost of doing so would be pretty disastrous to the contemporary American software economy. It feels very likely that two musical economies will develop, one based on using affective musical metadata to synthesize new license-free audio and the other based on using relational musical metadata to serve content to specific audiences. And then technology will follow. You have one type of music technology, you know, gen II optimized for background sounds, for sync that will, you know, pad algorithms and creator content with easy free stuff. And then the other type of music technology will work more with the social dimension of relational metadata. While every like, you know, misogynistic music critic could argue that Taylor Swift's music could easily be algorithmically generated, it's actually kind of the only music that really can't be, right? It's really all about the referential object of Taylor, right? Theatrix of Taylor Swift as a person, as an event. I mean, the same could be said of a large group of successful musicians from ice spice to bad bunny or diberial or tyla or dasha or NLE chopper. right? So the the generative license-free music I'll call Spotify Core, the kind of neutral low-fi beats to study to things that can be completely generated perfect fit audio. And the latter I'll call TikTok Core, music that consists foremost of its capacity to represent a social relation. Neither of them have so much to do with music as a thing on its own terms. I'm going to mostly focus on TikTok Core because Spotify Core is like a real black hole. I mean the most iconic Spotify Core story I can think of is Michael Smith who used AI generated music and an army of automated bot listeners to quote unquote steel 10 million dollars from Spotify Amazon music and Apple music. Honestly, that's a pretty great art project. And I can see the Spotify Core landscape getting bleaker and bleaker in this respect. Spotify Core really means removing musicians from the equation altogether. And in fact this kind of algorithmically generated low license overhead background audio takes up the vast vast majority of listened music. Not just producing backgrounds for YouTube videos or mood playlists but largely playing in empty buildings to nothing. Most played music isn't listened to by anyone at all. But on the other side of the house things are getting really interesting. In TikTok Core you still have musicians but the ethnicity and instrumentality and virtuosity of being a musician has shifted to other modes of social performance. TikTok Core musicians, you know, they barely make music, right? Their objective is largely to make it as influencers to such an extent that they could potentially leverage their status as influencers to retroactively construct the only kind of sustainable music career, one leverage around persona and representation for advertising. Most up and coming TikTok core musicians start primarily with music that lasts the duration of like an Instagram reel between 10 seconds in a minute enough time to really deploy the hook and get out of the way. The goal here always is to put the most engaging part of the track first since it's extremely reasonable to assume that one's average listening time will be less than five seconds. A track can then be retroactively constructed around those moments if the content gets sufficient attention. There's obviously a kind of characteristic compositional structure to TikTok Core, you know, the function is to deploy a hook that facilitates a transition. There's obviously a whole semiology of these transitions and a whole compositional aesthetics of how to make music to facilitate them. Like Benson Boone's beautiful things. I'm sure you've heard the hook. Please don't take those beautiful things that I've got. You know, it is this like three-phase transition moving to an ultimately triumphant place, ultimately like the real kind of affective TikTok core thus far is this kind of like post-imagined dragons triumphant or you kind of have a variance on this compositional structure with things like the you know, the renegade dance, right, where you know, this whole parallel remix architecture developed where people added more and more weird stuff into it so that one could articulate a whole language of sub-transitions. Live TikTok Core is amazing. There's a kind of like reversal of the proscenium. So the audience is actually in the back of the artist like the camera is on stage facing the artist and the artist's back is to the audience and the goal of live music in this context is obviously making more video content. I also see like a change to ensemble structures instead of like that band dominated scene that I grew up in. You know, you see a lot of you see a lot of sole proprietorship businesses up and coming TikTok core artists use creative partnerships for reach, right? So so why would you kind of restrict your ability to do so by you know, prefiguring a kind of ensemble? Instead, you know, you can always collab with someone externally and double or you know, and tuple your total addressable engagement. Many aspiring TikTok core artists without a network of collaborators will just straight up tag their up and coming music with more successful music on silent, leveraging the associative power of that tag to drive up views while playing their own songs. I always find that to be a kind of like really elegant tactic. The music produced from this new ensemble structure borrows very heavily from K-pop compositional structures. So instead of playing a single piece of music simultaneously like in a band, you get a cascade of individual features of each artist, each of whom performs their characterizing difference, including different genres, keys, styles, instruments, moods in you know, as memorable ways as possible. So this method easily unlocks a whole cascade of different reels where each transitional moment can be you know, leveraged as a content transition. TikTok core also spins off a totally new relationship to genre, one based much more on mood or identity than like geographic scenes or cities or individual technologies. An individual emerging TikTok core artist will generate music in as many genres as possible, hoping for a fit while preserving the persona. And like genre kind of like coagulates around a whole bunch of different persona types like different RPG archetypes. So one genre might be like small town back from the city family trauma, social anxiety, reclaimed girl boss era, self-care, love yourself core. Another genre could be like repressed, horny, post unholy body empowerment, bisexual lighting, bottom core. So delicious. A genre I'm really obsessed with is is cringe core. In cringe core, for example, you know, you have this out of tune singing or kind of laughably provocatively bad lyrics that our techniques used to kind of game engagement. You know, the more something is left out or criticize the more views you get. And what's kind of cool about this is like literally unlistenability is the goal. And this is really interesting in some ways. Like there are some situations where like I think it's pretty successful. Like someone like Cooke jerky or Mario judo, for example, or like even JoJo Siwa kind of like the ultimate cringe core musician. But it's also a hotbed of like actually mentally ill people whose public experiences of illness, TikTok, and Instagram reinforce and exploit. Something that really kind of characterizes the present moment as a really different moment is the fact that for the first time, almost the entire catalog of recorded music is available to us. And obviously there's this bias in like how much of that catalog has been preserved. And for what reason, but the fact remains that nobody has been able to work with music history like this before. Like with all of it in in a really a historical way, right? At the same time, we also have the ability to make music faster than ever before. So this like giant pile is getting bigger and bigger at a faster and faster rate. This is all happening at a moment when the spaces through which music moves are more compressed, tighter, and contained in fewer forms than ever. So the ways that we communicate and exchange music are owned by a handful of companies who are pretty opinionated about how we use their platforms. And the result is kind of like forcing a lake through a dam with holes the size of a cheese cheese grater, you get like extremely narrow high velocity stuff. So by thinking about music as relational metadata and essentially storing all of music within a massive data set, you make some pretty strong assertions about what is and what isn't part of music. And in doing so, we are memorializing, if not completing a centuries long projective cultural imperialism through which the world is only rendered as the world by its capture and retranslation into the universalizing languages of techno capital. Languages that have nothing to do with music, they don't contain music, they aren't really interested in music. They contain ideas about music, ideas whose parameters of description correlate directly to how useful they are to the technology platform that's deploying them. What's cool, I mean, like it's so complicated, like it's really easy to access many different types of music right now for many different parts of the world with many different histories and functions associated with them in the same way. But when it loses its context, like a lot of the music loses some pretty important stuff. Like, you know, religious significance, relationship to, you know, traditional cultural practice, relationship to certain communities or certain kind of edginess or intensity. It's so easy to flip through TikToks that take you from, you know, 90s Norwegian, black metal to YNW Milley to Ella Fitzgerald to Yundee Lee playing list to, you know, Gucci music to Oscar D'Alyone. Right, you can enter and exit into all those vehicles the same way, which means a kind of flattening of the topology of those spaces into the terms of their accessibility via the platform. And so what's crazy is that the more experimental music that you make, the weirder your music, the more niche your music is, the more you make music that tries to, you know, articulate the edges of what's in the data set, the more you actually smooth out the total space of the data set, your end order ironic Christian hyperpop project or your deconstructed G comm three step inflected post dance hall remix, they contribute to a project of mapping the total space of the possible, which has already been circumscribed by the parameters of the database itself. The more diverse music you make, the less differentiated the space of music becomes the less difference exists in the total archive. And you can look at this and you can say, you know, this is only true of the streaming ecosystem, right? Like this is this kind of problem of the total space and music being mapped is like a conceptual narrow first world kind of perspective. Visit intersects with this specific mode of listening. And you know, actually there is this or that cultural practice, which exists totally outside of the streaming system and like fine, like totally, like I don't disagree with that at all. But at the same time, be aware that this streaming ecosystem is praying on all of those communities, leveraging them for their difference and using that difference to create a smooth universal globe core, whether you like it or not. And that globe core plays almost everywhere in it, you know, admittedly in a kind of unevenly distributed way, but it will, it will influence the future of music outside of itself as much as it does that within itself. This is, this is violent, right? And this is the total total neutralization of music history by incredibly unsophisticated and ambivalent technological agencies. I'm also describing a kind of a horizonless mainstream, not just a globe core, but a kind of like mainstream core. The kind of latent space where the difference between the most extreme thing you can think of, the most avant-garde, crazy experimental thing, and like, imagine dragons, probably like the most important artist of the 21st century, is actually really small. The music from Shanghai's iconic hardcore, hardcore, hardcore techno label sub-cult, it could sit on a playlist next to Euro Bro jump style from Defcon 1. The music from the most avant-garde composer, exhibiting their weird stuff at the Darmstadt Ferry and Corsa, considering a playlist comfortably next to the 1980s music I've held with Lock and Von, or come on, the 1910s music of Anton Weber, and like, we know that, right? The playlist as a medium, the feed as a medium, it spills outward into meat space. If you know, you know, but knowing that this or that thing is actually really obscure or complicated or part of a very important cultural practice, matters less and less, and the difference is in positioning. The avant-garde is a kind of brand, right? I can think of no better example than Charlie XCX, who entered the mainstream as a representation of sub-culture, of PC music, and even like Bip Ling, UK, blog core, and she leveraged it girl, and if you know you know, as a brand, or as an image, very successfully. And so, while lots of this is sad, like not just the kind of loss of edginess, but the kind of colonial dilution of everything into a kind of de-differentiated platonic slush, some of it isn't so sad, like the annihilation of the avant-garde as more of a type of pretense than a privileged mode of music making. Rockism gives way completely to pop-dism, like it loses so fucking hard, like it's just gone. So we have Tik Tok core, right, which is this kind of referential play with music as relational social metadata, right? And then we have Spotify core, which is kind of AI-generated mood music. And both of them are characterized by a movement away from music since music is a cost and not a revenue generator. And now I want to talk about Roblox core. The artist, John Raffman, on new models. I love new models. If you're not a new models listener, please listen to new models. It's phenomenal, phenomenal podcast. So John Raffman, he suggests that music isn't really a major vehicle of youth identity formation anymore, and that makes a lot of sense to me. If you have a compressed, de-differentiated, global data set, and a few platforms that mediate access into and out of it, and you do this hyperfast hopping from track to track or whatever, you get a closet of different interchangeable costumes that are available for you to try on. Sam Cummins from Nymphit alumni. This was brought to my attention by Daniel Felsdad and Jen Leung on my podcast, The Cinegrator. But she has this amazing analysis of Roblox's dress to impress, which if you're unfamiliar as a kind of looks-mixing contest for your Roblox avatar based on a given theme. So Sam describes a kind of generational tension where now for Jen Alpha, who had been raised within the data set, anything is possible in response to a given theme. For example, an all-barby avatar could win a Goth theme, right? And I can see how that's Goth, and I can see how it's super interesting, how that could be Goth, how when you're free from the kind of rigidity of an assumed real world history, you can kind of create your own history, right? Anything could kind of be anything else, which is liberating and kind of cool. And if you're in Gen Z or older like myself, you're still for the most part exist within a world where history matters, where there isn't this kind of super flat topology of everything. And so this more kind of Jen Alpha totally unchained experience of kind of swimming within the data lake can feel really destabilizing. Yes, of course, generationality itself is like an arbitrary and kind of stupid way to think about things. I don't want to over index too much on generations, but there is obviously a useful concept from Michel Foucault called the Episteme. Really, the entire world is reconstructed periodically throughout history. I think that's probably extremely true. It's not that history is necessarily moving in a specific direction, but that there are periods where ideas like truth or history or identity or subjectivity really mean like completely different things. And if you look at the last few decades of identity and subject construction through a kind of materialist lens, like there was a kind of cultural production machine that emerged after mainstream counterculture of the '60s and '70s, a more segmented culture industry that could sell different groups, different versions of the same product, MTV and MTV Latino, post-punk in the UK, Otaku, Grunge, the like post-wall cultural explosion in Eastern Europe. There was a kind of like Western led neoliberal forward, but global moment where a lot of identity formation had to do with the type of music you liked and listened to. Shoutouts to little internet from new models as well. Shoutouts to new models for kind of really framing this, really elegantly to me. A lot of people forget that this kind of moment was kind of a new thing. This type of personhood where one defined one's selves primarily by the type of media that they consumed. And to be very clear, constructing who we are based on what we consume is like that very fundamental relationship between neoliberalism and identity. One of the economics of consumer culture and the culture of individual consumption, or individualizing consumption, one could say. And obviously, this is still happening. I mean, neoliberalism isn't exactly dead. It's kind of permitted. So people are kind of arguing that it's gotten much faster. Instead of constructing one self as one identity, why not construct yourself in a very fluid way, circulating through a spectrum of identities? Why not make an entire universe of consumer goods available to you by giving you the tools to cosplay as this or that on this or that day or in this or that context, right? This isn't to argue that certain aspects of identity aren't more or less immutable or that fluidity and identity is bad. That's not my intention. But rather that the velocity of identity construction has objectively accelerated considerably in a way that directly relates to the ways that we shop. The fashion thinker, Ruba Alsweel, will point you to Temu or Aushiyan or Ali Express. And note two things. One, contemporary goods are sold based on vibe, like dress to impress theme. For example, the vibe of self-care or the vibe of office siren and contemporary goods are disposable, right? Obviously, as she notes, cloud fashion is literally designed for single use and outfit from Shia and physically last as long as it takes to take a few pictures before it starts unravel. So the point I'm trying to make is around velocity. Music no longer serves these kind of longer time scale functions, like architectural functions of building people as people, but more functions as varying types of costumes that could be put on or deployed to reinforce this or that mood, this or that kind of message. The contemporary subject is more like a surfer, riding the waves of the all human music thing, the total archive, that relatively complete picture of peak content. So now, again, I want to like go back to the comment about boomerism at the beginning and be like, yeah, this isn't necessarily criticism, right? I'm not necessarily describing a state of affairs in which things are getting worse. There are winners and losers here for sure, but there's also just the transformation and a big opening of possibility. And I think what we have right now is actually a kind of, like, return to tradition, like, something much more similar to, you know, Gregorian religious chant or carnatic music throughout the first millennium, like, instead of being individual musicians doing individual things, it feels more like we're all contributing to a massive codex, right? And once in a while, something really crazy happens and a few people get celebrated for it, but there's this more like a big kind of anonymous wave of music makers who are just pumping music into a more or less anonymized archive. And this is both good and bad. Like the bad, you lose the context for music and you invariably get power imbalances that play themselves out at scale. So, you know, this person from this or that underrepresented class doesn't get celebrated for an achievement that has just been straight up stolen by someone for a more privileged class, right? And, but there's also the good, which is you kind of lose the romantic hero image of the musician as some kind of special angelic creature with superhuman powers who, by virtue of the indelible and definability of their mastery, is somehow, you know, always a white man, right? So, one of the things that we're losing now, which I think is both bad, but also very good is intellectual property. Intellectual property is not a thing that's always existed, right? Like, it's a, it's a recent thing. It's a thing that was constructed as a legal object with the encoded ideology that people need financial incentives to be creative. I'm not sure that's true, but regardless, like IP has become a major concern again with AI and also with the complexity of compensation when your music is taken and premixed endlessly on platforms like TikTok. So, like, how do you solve that problem? Like, how do you articulate your value as a producer or as an author in the face of, you know, massive total archive? And there are a few ideas which are referred to as like the three C's of copyright law, compensation, consent, and credit. And, you know, I'll start with straight up financial compensation, like, this is just never gonna happen. It's just never gonna happen because the value of a piece of music has been foreclosed to the public at basically a fraction of a fraction of a penny, if that, right? I'm like, to me, like, there's no further debate here or it's not productive to have a debate here. Reanimating a free good into a newly purchasable commodity is a really, really difficult thing to do. I think about like pornography, for example, which has also been kind of foreclosed as a sort of free object. There are, you know, subscriber models, there are advertising models that work to kind of support this and you could point to kind of only fans as being a kind of universe in which that particular commodity has been reanimated into something that one actually pays for. But in the case of the first, of only fans, like that commodity has undergone a serious transition from a first order good, you know, a video of people doing it, or a photograph of a naked person to a subscription service largely mediated by third party content houses. Because the demand for continuous content production is just so high. In this sense, it's actually a pretty similar transformation for music, like the good has converted into a kind of relational object, to a kind of parasocial exchange, and in doing so that kind of first order good, the video, for example, or the song, has mutated from a larger scale higher production enterprise to short, easily manufactured, I mean, iPhone videos in the case of only fans. Again, in both cases, the subscription model transforms the underlying good into a cost into something that directly affects your margin as a business. So I'm kind of passing on the compensation layer, I think the other two are a little bit more interesting, not necessarily if they're more fruitful. The next C is consent, where you see artists asking for kind of opt in or opt out relationships to having their music appear in libraries and data sets. And I think this is where the kind of pain of the moment is most acute, because we don't really have an infrastructure to produce or kind of instrumentalize this kind of relationship. There's like some cool things like like a Hollywood and a Matt Dryhurst have this very cool tool called have I been trained, which you know, can help you identify whether your work has been used to train a foundational model and even request your work to be removed from a given data set. But at the same time, kind of doing the set scales is really where things kind of struggle. You know, the relative impression of a single piece of work on a machine learning embedding is stored in really weird and irretrievable ways. And it's super hard to remove things from training data sets, which are often random, massive, I mean, unthinkably massive web scrapes. Like we're talking terabyte size text files. And these text files are, it's not like the things that they're containing are super well organized and annotated or even like the correct versions of anything. They are really just scrapes, right? You need a Python library in order to actually kind of move around them and get a sense of even what's in them. And once you're kind of unpacking the things that are there, you see that a lot of it is really kind of slushy. A lot of it just is really just parts of things. And this kind of brings me to the final C of the three Cs, which is credit. And like it's totally reasonable to want to see your name recognize for some inscrutable connection to a data set or you know, some kind of algorithm or remix of a remix of remix of remix of remix. But the drive does kind of become weirder when you start to think about the origins of the samples you use, the computer use, the software use, the teachers you've had, you know, the friends music you listen to, the influences to set within your brain. And more so the kind of like ambiguous messiness of all of that. The push for credit at scale and consent at scale, you know, really invokes a push for the world's most massive creative bureaucracy. The kind of human resources management layer for music making, one which will actually prove out what I really do believe to be the reality of the situation that all of us are co producing music as a massive all human project. So we arrive at a we arrive at a space where content is free. It's limitless. It's everywhere. And in this capacity content is basically dead. It's very little utility and adding more to it apart from the continued filling in of the map of music. And in doing so, like the kind of reduction of difference within the kind of total space of music. And for real, like I think there's very little positive work for music left to do at the level of content. So I totally acknowledge that there's a kind of morning period that needs to happen. But it shouldn't consist in romanticizing some prior fictional history of music in which actually everyone was somehow compensated outside of an already compromised economics chema or it shouldn't consist in demands for the repatriation of music into early 20th century authorship models. And moreover, this morning period can't take so long that it eclipses what feels like a moment of opportunity for actual engagement in the future. Many people, most notably in the wake of the discourse around Holly Herndon and Matt Dryhurst's The Call Exhibition at Serpentine this fall. A general shift in the art world most excellently summarized by Carolyn Busta and little internet over the past few years. But yeah, many people are kind of calling for a shift in the register of work that artists supply themselves to with the foreclosure of content as a site of active artistic interest or intervention. Everyone needs to move up a level to engage directly with the entirety of content as a kind of given as a resource and to move their practices into the protocols through which these resources are directed and the architectures in which they are deployed. In the case of The Call, this means not stopping one's musical work at the level of content production but really enrolling a kind of sophisticated musicianship at the level in which music is cast into training data, building new topologies in music distribution, rethinking what a license means. This does strike me as a model for the future of musicianship or artistry in a lot of respects and I like how herndon and dryhurst put this in the exhibition text for The Call, which full disclosure I haven't actually seen. But the text reads if all media is training data, including art, let's turn the production of training data into art instead. Game. The only point I'd really add here is that like this kind of work is very similar to the work that is Tiktok core and Spotify core, right? Both of those practices need to be understood as musical practices or artistic interventions, as translations of the artistic act on top of the massive all human corpus of neural media that is music, right? Tiktok core constantly dodges algorithms, gamified systems, deploys ravenous bot armies to gobble up revenue, innovates new posting tactics and engagement manufacturing strategies. The problem is that they really don't do this at scale, right? Because these universes are primarily motivated by small scale financial interests, they assume the model of like an employee doing time theft or working several full time remote jobs or like of an uber driver playing the warp of the algorithm to drive as little as possible. They are like micro scams that justify themselves through the larger scale, categorical scamory of the financial system as a whole. And when you try and do this stuff at scale in a way that avoids outright antagonism with the larger players in the software economy, what gets kind of interesting is you do very quickly move into bureaucracy land. So much contemporary art does look a lot like quaint digital right management schemes, like desperate attempts to build a better future by essentially cloning and inverting the types of structures one might see yet. Well, Disney when it comes to making sure that mid-journey can't actually generate an image of Jar Jar Binks. And I guess I do wonder if I want a future art world that consists in innovative bureaucracy, in the micro management of permissions and histories as opposed to building, you know, mesmerizing architectures or blowing up buildings have hazardly. And I just want to be super clear, like this isn't a critique of the call or of herding and dry hers work at all, which doesn't strike me as having any of these characteristics. But I do see a bureaucratic turn kind of everywhere. I do see so many artists of all generations hop from Dow to Dow, project to project, petition to petition, trying to be the first to re-innovate vehicles with which to re-commodify content. One thing that I am really interested in is this kind of momentary feeling of real vulnerability throughout the tech sector with respect to creative technology right now. I do get the sense that folks at Google or OpenAI are at this, you know, particular moment kind of obsessed with creative technologists, with people who can kind of explain the ramifications of what it is that these companies have gotten themselves into, to the point that they'll even risk quite a lot, like in the case of OpenAI's Sora. And I think a lot of this interest is honestly in pretty good faith. It's done almost with a kind of assumed understanding that music or video art is essentially worthless in 21st century terms and understanding that seems to be correct by basically every metric apart from, you know, the artist's own. Um, an art world that can sustain that kind of contact. And even like, you know, competitive tension with the technical community and actually ratchet it up, an art world that like bristles with technical capacity that actually has real utility and unique explanatory capability in the context of the technical community. But also one that can react unpredictably and even violently at a scale that does actually make it a threat to anyone who might foreclose against its agency. That's an art world I believe in. Not an art world that endlessly clamors for public funding through some argumentive transcendental necessity, nor an art world building a kind of tower of bobble of rights management. But an art world with teeth that prints its own weapons and art world whose virtuosities and musicalities consists in its ability to continuously undermine any attempts to compress its activities down to the level of content production. This art world absolutely has to be technical and even virtuosically so. Um, and another kind of element to this like the problem of the world stack of the D differentiated kind of globe core, this like colonial black hole of context loss, it really does require a counter force. We are losing something in this transition no matter what and what we, what we're losing, I think we really do need to document it. Like we do need to find some way to immortalize the contextual matrix that is missing when you listen to Billy Holiday or Lee Scratch Perry or even Skrillex on Spotify, right? And that really does require both an understanding of music as a historically situated process, but also as a thing with its own internal logics of appreciation. I really think one of the most revolutionary musical practices one can engage with at the moment is archivism, the application of musicalities that exceed those that are describable on Spotify or YouTube's terms to the documentation of all the different ways that we found ourselves where we are. If this one thing we have learned consistently this century thus far, it's that digital archivism is weapons grade stuff, not weapons against techno capital necessarily, but a kind of nomadic equipedness, an ability to be dangerous, an ability to push through a risk assessment, spines, or quills, or stink, or nails, defense mechanisms. There's a concept from Sun Raw that plays and Claire breaks out beautifully in the opening of becoming presses music as the second fire of Prometheus. I'll read the quote. The problems in this world are caused by the music that musicians are playing, that they are forced to play. The music we play guides the ritualistic building or upholding of our world, and if we are always forced to think in terms of capital, then it makes sense that we have summoned late-stage capitalism, or rather chanted it into being through our capital-oriented musical world building. We build this world, and we are still building it with an ongoing ritual. I'll be honest, not a particularly big Sun Raw fan, and honestly starting to associate his legacy a little bit more now with online ceramics trucker hats than music. But if music has some incantatory, hyperstitial power, if it can actually manifest, then perhaps it's time to redefine music and art making in general as technological weaponry and weapons craft, which needs to actually be real to actually contain within itself real threatfulness, which I obviously mean to be something very different than the kind of endless, facile, politicized identity play at the level of content, or instead, more tepid, community, drum circle, togetherness core. But instead, per Mark Fisher, cybernetic theory fiction, or simply put, cyberpunk, for real this time. So yeah, I don't know. It feels like this real opportunity to motivate the future. Music at the moment, but only if we step outside of our comfort zones when it comes to acknowledging tactical losses as losses, recognizing the situatedness of our understanding of what an artist and a musician is in modernity, and moreover opening one's mind to the idea that the newer permutations of music playing themselves out on Spotify or TikTok should actually be studied by music's most serious practitioners. Without a doubt, the future of music has far more to do with a cringe core artist blasting out daily posts of the same 15-second loop with different captions and skits than it has to do with grading string quartets funded by literally dying cultural preservation institutions in Europe, or by a spectrum of electronic noise that, despite Sophie's correct claim, contains, you know, unbridled possibility, somehow, wherever you are in the world, it sounds basically the same. While closed with this quote by Emily Carrier from the amazing piece technically man dwells upon this earth, "Art does not die of starvation, it dies of endless proliferation of cancer, an art that would survive the automation of art is as of now unthinkable," which is good, "Art is being tasked with the creation of the unthinkable, a new art has been made necessary." Or maybe this quote from Schumann Bessar from Flash Art in 2022, "I promise you it's nearly over, I promise you this time it's true, I promise you we're in the end core now, now, now." [MUSIC]