Pastiche is back, if it ever went away. For the Dispatches column, Ryan Jeffery considers why Noah Baumbach would pick the present moment to adapt Don DeLillo’s novel White Noise for the screen, amid the clairvoyant promises of large language models and image-imagining software.
A film still of a toxic cloud erupting from a train wreck in the style of Noah Baumbach. Generated using DALL-E 2, April 3, 2023.
At the height of the Covid-19 pandemic, writer-director Noah Baumbach picked up Don DeLillo’s novel White Noise and decided it was time to film what many have deemed an “unfilmable” cornerstone of postmodern American literature. Why now? A Hollywood take on the book’s story of an overanxious, overstimulated bourgeois family seems obvious enough given the current state of global events. The plot’s iconic inciting incident, “the airborne toxic event,” is a mutable signifier, easily swapped for a viral pandemic, ecological collapse, conflict between world superpowers—or the displacement of reality by simulacra and representation altogether.
Published in 1985 at the dawn of the twinned regimes of financialization and the information economy, DeLillo’s dispatch from Reagan’s America offers some melancholic comfort for a Gen-X’er like Baumbach (or myself): despite how bad the current societal contradictions may seem, the crackup has been happening all along. In this sense, Baumbach’s filmic adaptation for the 2020s can be viewed as what Fredric Jameson deemed “nostalgia film,” engaging with the past only insofar as it desires to disappear into it, a time seemingly simpler than our own. Writing roughly alongside DeLillo, Jameson arrived at a similar diagnosis of a new type of social life emerging within a new economic order, and with it a new type of cultural production that he argued was characterized by pastiche and schizophrenia. This is an apt description of where Baumbach’s White Noise leaves the viewer in the end credits, with the entire cast dancing through the consumerist motif of the supermarket—an ambivalent pastiche of Jean-Luc Godard and Jean-Pierre Gorin’s Tout Va Bien (1972) and Randal Kleiser’s Grease (1978).
A film still of a toxic cloud erupting from a train wreck in the style of Jean-Luc Godard. Generated using DALL-E 2, April 3, 2023.
“Pastiche is blank parody,” writes Jameson, all that’s left “in a world in which stylistic innovation is no longer possible.” This notion of imitation in dead styles is central to Jameson’s nostalgia film, and to what he calls “the consumer society” more broadly. But if we widen our lens beyond the consumer societies of the late twentieth century, then pinning down what pastiche can and cannot do gets trickier. Scholar James F. Austin draws heavily on Jameson, but also Proust, Barthes, and a host of others, to recoup “a positive meaning of pastiche” that doesn’t just flatten history into “a representation that effaces political content, ‘pastness’ instead of the past.” To the contrary, argues Austin, “positive pastiche” has the power to reactivate or redefine works of the past through a form of textual criticism that folds writing and reading, producing and consuming, into one act. What pastiche is or what it does becomes historically contingent, responsive to societal changes and overturning forms of media.
What does a contemporary positive pastiche look like—one that actively engages with history in the present? Is Baumbach writing as he reads, or is he beckoning us to hide with him in the past? Film critic Bilge Ebiri astutely names Baumbach’s pastiche of “Spielbergian action fantasy,” pointing out how the movie feels like it could have come straight from an 80’s multiplex of the depicted era. This is particularly acute regarding “the airborne toxic event,” portrayed with spectacular sci-fi disaster FX fit for the big screen that also seem pulled straight from Netflix’s other Spielbergian sensation, Stranger Things. If Spielberg serves as a prime avatar for Jameson’s original nostalgia film, it appears that mimicry and self-reference have only been further engrained. The hyper-charged consumer society of the twenty-first century has fractured further into a matrix of private codes and languages. In the era of big data, today’s white noise is near deafening as compared to the time of DeLillo’s book, exponentially scaled according to Moore’s Law.
A film still of a toxic cloud erupting from a train wreck in the style of Alejandro Jodorowski. Generated using DALL-E 2, April 3, 2023.
One of the strongest signals piercing today’s static has been the hype and debate surrounding large language models, a new breed of so-called general artificial intelligence. Technologist Gary Marcus flatly describes the software as performing pastiche: “a glorified version of cut and paste” with no understanding of what actually connects A to B, or how meaning is made. The bad kind of pastiche is certainly being invoked. Nevertheless, this class of software has made its way from the business pages to meme culture and art museums, even scandalizing at least a slice of the cinephile world in filmmaker Frank Pavich’s New York Times op-ed about AI-generated images from a nonexistent Tron sequel by Alejandro Jodorowsky.
Pavich is intimately invested in the particulars here, having made a documentary about the psychedelic auteur’s heroic but vain attempt to adapt Frank Herbert’s Dune (indeed, another nonexistent Jodorowsky film). His mixed reaction to the uncanny, alluring “film stills” and “production shots” illustrates the disorienting mix of hype and threat the technology elicits. As a fan, Pavich describes the frustration of wanting the film to be real—a nostalgia for something that never was. But as a filmmaker, he is disturbed by the ease with which a computer can reproduce the auteur’s seemingly unmistakable style. “Has Alejandro been robbed?” he asks; “[how] much of art-making is theft, anyway?”
A film still of a toxic cloud erupting from a train wreck in the style of Stephen Spielberg. Generated using DALL-E 2, April 3, 2023.
An artist like Christopher D’Arcangelo might emphatically reply, “All of it!” For these generative AIs to work, they require vast data sets, including swaths of the internet—Jodorowsky’s unique style included. Several lawsuits have been filed, including by Getty Images. The claim seems to be that generative AIs, by design, simply connect, or pastiche, elements of past works of art (or intellectual property), guided by the logic of mathematical pattern recognition. But despite the name, nothing is actually being generated by this AI technology because nothing from the past is being reconsidered. In fact, it’s the opposite. Elements are recombined precisely by the predictive logic of what’s already been done by all the human labor and creativity that’s come before. For this reason, some have argued this is simply pure primitive accumulation, Silicon Valley’s latest digital enclosure: “selling us back our dreams repackaged as the products of machines.”
DeLillo in the style of Spielberg. Was that Baumbach’s prompt? More than anything else, what separates Baumbach’s film adaption from DeLillo’s book is the novelist’s relentless preoccupation with narrative authority and overturning media forms. The Gladney family in White Noise, like nearly all of DeLillo’s characters, must fumble their way through a deluge of television, news, commercials—they are “media haunted.” This haunting carries on today in the vast neural networks of new generative AI. They are, in a sense, mediums for our media, built to commune with dead languages and styles. Are they conjuring spirits or zombies?
A film still of a toxic cloud erupting from a train wreck based on a photograph of the derailment in Palestine, Ohio. Generated using DALL-E 2, April 3, 2023.
When a real-life train derailed in East Palestine, Ohio, delivering an actual airborne toxic event precisely where Baumbach staged the Hollywood version, much was made of the Baudrillardian rupture of simulation and reality. Less was made of the fact that rail workers saw it coming. Despite their capacity for spectacle, these new generative AI models would likely miss anything that distinguishes images of Baumbach’s simulated disaster from those of the real one. No matter the scale of the network or amount of data, the task of understanding our images still falls to us. x
Ryan S. Jeffery is a filmmaker who lives in Los Angeles.