PERFORMANCE RETROSPECTIVES, increasingly common, often highlight everything but the performances. Ephemera, video and photographic documentation, reenactments of excerpts from works restaged in galleries: These souvenirs abound as means of presenting and reflecting on an artist’s body of work, as opposed to fully staged productions themselves.
“Algorithmic Theater: An Annie Dorsen Retrospective,” held at Bryn Mawr College in September 2022, was exceptional because it re-presented four of her performances made between 2010 and 2015. This allowed audiences to experience the force and meaning of her politically unsettling and artistically profound expeditions into computer-generated making, a practice with deep roots in visual art but few peers on the stage. Dorsen began collaborating with computers well before algorithm was a commonplace term in digital and social media discourse, using software programs to co-create works of theater that require an audience’s focused attention over time. Algorithms (both digital and analog) are procedures, and Dorsen’s works are accordingly procedural: events that unfold moment by moment with different results every night. In some cases, her performances produce millions of possible arrangements of text. For her, spectators must confront this methodical madness in the real space of theater, rather than in the hypothetical or the abstract. “It seemed important to be able to say: this is it. This is the event,” she wrote in the retrospective’s accompanying catalogue, Annie Dorsen: Algorithmic Theater, Essays and Dialogues 2012–2022. (Full disclosure: Two of my own previous writings on Dorsen’s work were reprinted in that catalogue alongside the work of other critics.)
Dorsen’s algorithmic theater merges human artmaking with digital decision-making, often by supplying a curated body of text (for instance: thousands of Reddit comments or the entirety of Shakespeare’s Hamlet) to a custom-designed computer program, which then uses algorithms to make creative decisions in real time including those regarding what text will be spoken or sung by actors or vocalized by bots. “I wanted to make possible that the viewer could experience, if only briefly or intermittently,” Dorsen continued in her catalogue essay, “some portion of that real and not-real paradox which distinguishes theater from other live art forms.” That paradox, the infusing of material and temporal realities with fiction, manifests differently in each of the pieces presented at Bryn Mawr: the trilogy comprising Hello Hi There (2010), A Piece of Work (2013), and Yesterday Tomorrow (2015), as well as the participatory performance event Spokaoke (2012), in which spectators do karaoke to famous political or pop-culture speeches, including iconic monologues from films.
Dorsen’s works have always doubled as technological microhistories, enacting the lag time between the means of production and the thing produced.
For audience members who had attended the 2011 New York premiere of Hello Hi There at the former P.S. 122—its linoleum hallways and public-school architecture now glossily renovated into Performance Space New York—seeing the piece again figured as a reunion of sorts, albeit a reunion with a beloved friend who does not care that we’ve come to see them or even remember that we’ve met before. Featuring no human performers but a great deal of human labor, Hello Hi There is a conversation between two chatbots embodied by a pair of MacBooks seated on a patch of Astroturf. Taking inspiration from Noam Chomsky and Michel Foucault’s infamous 1971 debate about language and human creativity, as well as from other historic inquiries into the ways technology challenges concepts of humanness, the piece unfolds as a series of exchanges in which the bots draw on vast databases of text, ranging from YouTube commentary about the Chomsky-Foucault debate, to snippets from Hamlet and other works of famous literature, to the hollow botspeak that Siri and Alexa dish up daily. Chomsky and Foucault themselves are visible via a video recording of their debate that plays on a boxy old monitor hulking in the stage-right corner, the kind of pre-flat-screen device now referred to as a “dumb TV” and nearly impossible to buy anymore.
Dorsen’s works have always doubled as technological microhistories, enacting the lag time between the means of production and the thing produced. In Hello Hi There, outdated computer code cascades downscreen, despite the fact that this visible display of “processing” is unnecessary to the actual processing happening within the performance. Twelve years after the work’s debut, obsolescence operates both as an aesthetic and as dramaturgy. Hello Hi There runs on decade-old software not for the sake of nostalgia or kitsch but because these particular systems—conserved over years by Dorsen—are the trained performers that “know” how to execute the show. According to Ryan Holsopple, one of Dorsen’s technical collaborators, new programs could theoretically do it, but the piece would be deeply, even if indiscernibly, different. “I can’t remember,” repeatedly lamented one of the bots in the performance I saw in September. This admission could break your heart, because indeed bots can’t remember, not the way humans do—though digital technologies can, at the same time, preserve information with a precision we mortals could never match. Or the bot’s confession could send you into gales of giggles, as it did the Bryn Mawr students, who were not yet in middle school when the show premiered.
For A Piece of Work, created shortly after Hello Hi There, Dorsen and her collaborators fed Hamlet through 2013-era computer programs to produce five abbreviated iterations of the play, which scroll across an upstage screen as computer-generated voices intone the words. Sounds and lights, cued by the text, reverberated and glowed in Bryn Mawr’s chapel-like Goodhart Hall. Fog occasionally issued forth through an open trap door in the wooden stage, which obliquely echoed that of Shakespeare’s Globe. Only one of A Piece of Work’s five acts features a human actor: At Bryn Mawr, this was Scott Shepherd, who alternated performances with Joan MacIntosh in the show’s premiere and subsequent presentations. Perhaps best known for his work with the Wooster Group, including the title role in their technologically haunted Hamlet of 2007, Shepherd enters in act 3 of A Piece of Work, vocalizing computer-generated text fed to him through an in-ear device. Shepherd’s own voice produces an acutely different resonance from the digital modulations we hear in the other four acts. He isn’t Hamlet, exactly, but he isn’t not Hamlet, either—and the implicit question of just who he is, which is embedded in the original play’s philosophical interrogation of human identity, was one of the reasons Dorsen selected Hamlet as her source material. Shepherd’s monologue is different every night, and even if a given audience can’t spot the differences, we can still witness the theatrical labor that new language combinations require: the intense listening and vocal control necessary to enact an uninterrupted rush of computationally created commands. (In the published text, for instance, Hamlet’s most famous monologue becomes “to be, And not to be, such is the faith: to be, and not There be: These is the body.” The rhythms are recognizable, the phrases askew.)1
The brief presence of a human actor, combined with his abrupt exit, underscores the experience of loss that is at Hamlet’s core—as well as the loss of communal feeling that performances without people can produce. Nine years after its premiere and in this postpandemic moment, A Piece of Work offers a means of contemplating the myriad ways humans can be absent or abstracted from one another, distilled into statistics or collapsed into scrolling social media posts. In the performance at Bryn Mawr, we heard the phrase “Enter Gertrude,” and saw the words ENTER GERTRUDE flash on the upstage screen. But no Gertrude appears. This gap—between the command for Hamlet’s mother to arrive onstage and the failure of an actor to do so—invites us to receive the name as the character herself, or at least as a stand-in, a stunt double in a world populated by avatars. It conjures a very present kind of absence, and it resonates (if audiences let it) with the large-scale loss of contact the pandemic necessitated, the reframing of ourselves inside of Zoom boxes, the deaths of loved ones and others reduced to infographics, and the collective struggle to maintain social and public spaces.
Addressing questions of social and public space directly (though created well before the pandemic), Dorsen’s Spokaoke, part party and part performance, is both a joyful communal event and a meditation on the nature of public speech. For the retrospective, its restaging was held in the college’s student center and vigorously emceed by Philadelphia-based drag queen Martha Graham Cracker. Catalogues that listed famous speeches in alphabetical order were strewn across couches, and audience members filled in request slips for the oratorical address of their choice. As I flipped through the C’s, I noted a listing for “Hillary Clinton, ‘Concession Speech,’” dated to the early hours of November 9, 2016. A twinge of devastation, brought on by the memory of that moment, erased my desire to participate, to find pleasure in briefly ventriloquizing, say, Cher from Clueless. (Luckily, Martha Graham Cracker was persuasive and the college audience seemed less affected.) Many other speeches were performed that night: Uma Thurman telling the press about anger and #MeToo; Richard Nixon avowing that he was not a crook. Though only one of many possible selections in the Spokaoke repertoire, Hillary’s concession speech didn’t just provoke sadness. It testified to the necessity of the work and of Dorsen’s project as a whole. “The transformative ideas—whether ameliorative or destructive—are being embodied and delivered to attendant spectators in real time,” writes scholar Jacob Gallagher-Ross in his essay for the retrospective’s catalogue. “Ultimately, Spokaoke argues that the true meaning of any moment of political speech can only emerge in relief, in the critical juxtaposition of then and now, word and reception, ideology and human consequence.” Dorsen’s collected speeches here mark many historical ruptures and seismic shifts in our collective awareness about, among other pressing questions, the rise of digitally disseminated untruths.
As a whole, Dorsen’s practice interrogates the ways in which information is gleaned and mobilized (for capital, for war, for art) on emerging digital platforms.
As a whole, Dorsen’s practice interrogates the ways in which information is gleaned and mobilized (for capital, for war, for art) on emerging digital platforms. The algorithmic trilogy—as well as works like The Great Outdoors (2017), which examines the internet as a landscape of sublime scale, and The Slow Room (2018), a meditation on chat-room culture—has been praised for its technological prescience. In retrospect, though, Dorsen’s algorithmic and digital works are more than prescient. They use obsolete technology but refuse outdatedness by demanding a quality of attention that is neither anxious over technofuturist danger nor seduced by technoglamorous possibility. One can understand how her performances initially functioned as warnings of sorts: She provides algorithms with a level of agency that echoes their frequent manipulation of our daily lives—their unquestionable dominance over, as Shoshana Zuboff phrased it in her 2019 The Age of Surveillance Capitalism, “the right to the future tense.”2 Theater, on the other hand, layers multiple kinds of temporalities (“real time” and fictional time), and the spectator’s awareness of these layers offers the potential to resist the unconscious forward creep of algorithmic control.
Linear time is addressed most directly in Yesterday Tomorrow, which premiered at the Holland Festival in 2015 and at La MaMa as part of P.S. 122’s COIL Festival in Amsterdam in January 2016. Three performers sing their way from the Beatles’ “Yesterday” to the musical Annie’s “Tomorrow.” While Dorsen chose the songs, the algorithm produces the dissonant thicket of melodies that leads the singers from one to the other. The singers follow algorithmically generated scores one note at a time, creating a form of music that is legible to the computer system but less so to human ears, until the moment when “Tomorrow” begins to audibly emerge. As bars of music scroll across screens suspended above the performance space, singers and audience members alike may witness the algorithm’s decision-making, but none of us can, in the moment of performance, control it. Though everyone knows “Tomorrow” will arrive (“come what may,” the song lyrics cheerfully insist), attending aurally to the algorithm’s process demands a fundamentally different kind of attention than the usual focus on visual interfaces that we have, as a culture, come to expect from art created under the sign of AI. Deepfake, NFT, and GAN culture often invite panic over our relationship to images: See, for instance, whichfaceisreal.com, created as part of a project that, as developers Jevin West and Carl Bergstrom have noted, is intended to help viewers quickly distinguish photographs of people from AI-generated images.3 Anticipating the AI-driven loss of reality can understandably produce fatalism and crisis-speak: Which candidate is real, which election is real, which war is real? Dorsen, by contrast, welcomes panic into the theater, along with beauty, boredom, and heartbreak, because she has known for years that anticipation of these losses was beside the point: We are already there. Her pieces briefly disentangle audiences from the ever-shifting algorithmic creation of new and isolating unrealities by placing us in the temporarily shared reality of the theater.
Her works also invite the possibility of a distinct and less starkly dualistic understanding of the relationship between AI and art. The performance-studies scholar Lindsay Brandon Hunter argued in her 2022 essay “‘This Person Does Not Exist’: Performing Resemblance and Representation in Generative Adversarial Networks” that worrying over the so-called loss of reality produced by artificial intelligence is often irrelevant and also potentially dangerous. Art, she points out, has always produced faces that aren’t real. In AI-generated imagery, what’s more important to note is that the artist—and therefore the labor and the embedded politics—is obscured or rendered illegible.4 A GAN-produced image that appears photographic but lacks a photographer reveals more about our cultural understanding of what photography purports to do than about the terrors of AI. Meanwhile, the exploited human labor that allows AI to function too often goes unrecognized and unregulated.5 Dorsen’s works, by contrast, foreground process and assert the possibility of multiple outcomes, including ones that we—as artists, and as inhabitants of an algorithmic world—can and do still shape.
Miriam Felton-Dansky is a scholar and critic of contemporary performance based at Bard College.
NOTES
1. Annie Dorsen, A Piece of Work (New York: Ugly Duckling Presse, 2017), 83.
2. Shoshana Zuboff, The Age of Surveillance Capitalism (New York: Hachette, 2019), 331.
3. Jevin West and Carl Bergstrom, “Which Face Is Real? Seeing Through the Illusions of a Fabricated World,” whichfaceisreal.com, accessed December 8, 2022.
4. Lindsay Brandon Hunter, “‘This Person Does Not Exist’: Performing Resemblance and Representation in Generative Adversarial Networks” (conference paper for International Federation for Theatre Research, Reykjavik, June 23, 2022.
5. See, for instance, Julian Posada, “Why AI Needs Ethics from Below,” Medium, September 21, 2022, medium.com/a-new-ai-lexicon/a-new-ai-lexicon-labor-2db30b6b2349.