“I have songwriting credits, even though I don’t know how to write a song.”
--Oleg Savitsky of AI music generation app Endel (speaking with Dani Deahl at The Verge)
Once we had music. Now we have content. We have songwriters who have never written songs. What is going on here?
Content as a concept, as an asset class, has brought a deluge of contradictory, confusing economic developments for industries like music and publishing. Content’s advent is a format shift, but one that changes our relationship to knowledge and what we’re willing to pay or be paid for it. There’s more access to quality compositions and songs than ever, yet just as much dreck.
We’ve seen something like this before, when the notion of knowledge changed and accelerated wage inequality and the pace of social transformation.
I’m talking the European moveable-type press and the book world that sprang from it. The new format changed the way information was compiled and conveyed. It sped things up dramatically, for one: Gutenberg’s mix of better ink, paper, and moveable typefaces made it possible to print around 200 impressions an hour by the late 16th century, a rate it took nearly three centuries to double. That shifted the value of what was worth putting to paper. In turn, this transformed the knowledge market quickly and dramatically: In the span of a few decades, the price of books plummeted, compared to the rather stable price of manuscripts in the centuries before.
Books got cheaper, sure. But something else fascinating happened, according to the London School of Economics’ Jeremiah Dittmar and Skipper Seabold. A very specific kind of economic inequality emerged. Despite a relatively stable relationship between the wages of skilled and unskilled Europeans from the birth of printing up through the early 19th century, the wages of certain groups of highly specialized knowledge experts--natural history and mathematics professors at universities in places like Rome and Florence, for example--rose at a rapid, increasing pace, almost immediately after printing technology came to their cities. There was a jump in inequality between the specialists and their just-as-educated peers. Something they had was suddenly more valuable to people, and those people were willing to pay a lot more all of a sudden.
The internet era has spawned a similar dynamic. It has transformed the production and distribution of knowledge and art, depressed the cost of acquiring said knowledge, and increased inequality among the skilled. Suddenly, people are willing to pay a whole lot more for those who have mastered a very specific body of knowledge.
Economists have been pondering this and have their theories. In a 2019 paper, MIT economists Seth Benzell and Erik Brynjolfsson present a theory designed to explain the strange behavior of the US economy since the 1980s. The economy has seen a cluster of curious phenomena, such as slowing growth in productivity despite rapid technological innovation, the lower price of both labor and capital, and very low interest rates. Their solution: a mysterious “genius” factor puts the brakes on growth despite explosions in technology, as this element cannot be automated. This genius resides in particularly skilled individuals. They earn far, far more--increasingly more--than their also very skilled and educated, equally productive, non-genius counterparts.
Quick aside: The term “genius” has a ton of baggage, and maybe that doesn’t bother economists. It bothers some of us in the arts or in history, because extraordinary individuals are never alone in creating something. To give Benzell and Brynjolfsson credit, they admit that “luck” is often part of the “genius” factor. (Take that, Byron!) Indeed it’s a term that has been used to erase the contributions of and reinforce the marginalization of far too many people, on the basis of a purported genetic lack of said “genius.”
Terminological quibbling aside, the specific, un-automate-able skills these “genius” workers have are not all that mysterious. If you look at AI and machine learning, for example, you can home in on some of these skills. “Genius” workers have the ability, training, and means to turn their tacit knowledge into explicit mechanisms. This unlocks “artificial” expressions of human creativity or effort, outlining how technology can implement the dreaded “last motion” in complex, delicate intellectual, aesthetic, or manufacturing processes. As analyst Benedict Evans notes, “part of the challenge of machine learning is not just working out what problems to solve but working out how to surface that to the user.” If you can build a machine that mimics tacit knowledge well enough, you’ve got the genius factor. And that will earn you an order of magnitude more money than someone who’s a good accountant or music tech writer, say.
Specialists in the 15th and 16th century managed something similar. Printing allowed knowledge of a very specific kind to proliferate and spread. Books like Huygens’ work detailing his pendulum clock or translated compendia of sea voyage accounts forged a new opportunity for those with certain inclinations, knowledge bases, and the skills to digest, share, and increase knowledge. The men who did this, such as the Italian professors mentioned above, saw their incomes and clout rise.
Yet printing did something else to the knowledge economy by lowering the cost of books. It opened up a whole new world of knock-offs, fakes, shady collections of dubious materials, witchcraft manuals, and general garbage--all books, but of questionable value. Plagiarism and wretched prose formed the business model for printers, who became increasingly specialized. By the 17th century, the various aspects of the book trade splintered into separate professions, including booksellers, publishers, and the people actually running the presses. Many sellers and commissioners of books became wealthy as a mass market emerged (wealthier than the guys and a few gals writing the poems and heart-wrenching love tales), even as much of the population in Europe remained overwhelmingly illiterate.
The proliferation of both scientific knowledge and bullshit sounds awfully familiar, doesn’t it? It mirrors the ridiculous abundance of content, and more specifically music, online, from the sublime sounds you would never have heard under the old label system, to the absurd--the fakes and bedroom disaster tracks and re-recorded versions of old chartoppers and mediocre AI generated “chill.” Yet the spoils of all this economic activity are going to the bottlenecks in the system, the tech execs and VCs and programmers/developers/data scientists we often lionize. They control the “genius” factor that turns tacit knowledge--what music you’d like, what music you’d like right now--into an automated interaction.
That’s why a programmer with songwriting credits, though he bemusedly admits he has no ability to sit down and write a song, speaks to our time. Savitsky and company have captured a narrow body of tacit knowledge and automated it. Their annual income is likely an order of magnitude greater than many songwriters’. It’s the Gutenberg moment, when new formats and production methods breed an unstable world of brilliant knowledge and penny dreadfuls, of exquisite music and meaningless content.