In his 1939 adaptation of The Wizard of Oz, Victor Flemming portrays “The Great and Powerful Wizard” as a mystical and towering nonhuman life form, a giant green head hovering over gushing flames. Once confronted by the Wizard, Dorothy, Tin Man, Scarecrow, and the Cowardly Lion are ambiguously terrified and mesmerized by the creature’s nonhuman ‘Otherness.’ At one point in the scene, Dorothy’s dog, Toto, whimsically runs off and pulls back a curtain revealing the Wizard’s true human identity: a middle-age man who had been secretly posing as the Wizard by remotely operating a machine. Upon learning this, Dorothy and her friends immediately lose all fear and fascination for the nonhuman Wizard and aggressively accost the “man behind the curtain” to express their disapproval, denouncing him as a “bad man” and a “humbug.”
As many recall, over a year ago on September 24, 2013, a similar phenomenon took place on the Internet as a mass of publications countrywide, from The New Yorker to The LA Times, confirmed that the allegedly algorithmically-generated poetic tweets of @Horse_ebooks were in fact composed by BuzzFeed Creative Director, Jacob Bakkila as part of a collaborative art piece with his associate Thomas Bender. One important underlying conceptual implication of most media reporting on the event was to reaffirm the unrivaled autonomy, agency, and dominance of the “liberal humanist subject” in its encounter with a machinic, nonhuman entity. That is, the notion that collaborations between human and nonhuman forms and processes will always result in mastery and domination, rather than interplay and mediation. Indeed, most media reporting on the news identified Bakkila as “the human behind the spambot” and thousands of Twitter users went on to dismiss Horse_ebooks as a “hoax,” an Internet “humbug.” This popular sentiment ultimately segued into a widespread loss of interest for a Twitter feed that had moments prior captivated over 200,000 followers.
Unlike Susan Orlean of The New Yorker, I’m still unconvinced of the installation’s merits as a work of art. I’m also less concerned with settling the debate over whether Horse_ebooks was an incisive net art piece or an instance of “digital bro art.” What I am interested in is thinking through the ways in which it may (or may not) provide a way into re-sketching the contours of posthuman subjectivity/authorship in a digital and networked context. I take Bakkila’s algorithmic performance to be a controlled mediation between human and nonhuman processes, something akin to what N. Katherine Hayles has called a “collaboration between the creative imagination of the (human) writer and the constraints and possibilities of software.” In this light, I locate the central cultural, aesthetic, and political contributions of Horse_ebooks in its experimental (perhaps even accidental) illustration of a new type of posthuman artistic subjectivity in networked new media, whereby the human author undergoes a performative process of digital becoming in order to embody the virtual ontology of algorithm.
Lucas G. Pinheiro: As a response to Morgan Myers’s question concerning popular confusion between your method in performing as a spambot for Horse_ebooks and Markov generated language, you said that it is “really pleasant that Markov generated language is confused with out-of-context actual language. Personally, I think it’s the dream-like tone shared between advertising spam and generative speech, close to how people actually ‘talk’ but not quite.” The Markov chain, both in continuous and discrete time, is a random process model whereby future outcome is independent of previous outcome. In this light, what role did randomness play in your method? How did you “mimic” algorithmically-generated randomness?
Jacob Bakkila: I consider the whole output of Horse_ebooks to be a narrative poem that mimics the tone and sensation of memoryless language.
LGP: You also mentioned that your method was to “select and truncate existing language, overwhelmingly taken from spam ebooks sites (and sometimes from generalized Google Books).” How did you select the ebooks and spam content from which Horse_ebooks material was drawn?
JB: It’s like any other author’s compositional process, but perhaps with a different resulting product. I read spam text every day for hours and hours: advertising circulars from the 1920s, ebooks about natural cures for gout, pages of indexes from public domain instruction manuals about woodworking, and so on. There are spam sites not even indexed by Google, collapsed mine shafts that extend deep, sometimes connecting underground, sometimes deadening, sometimes going forever.
LGP: How did you choose when to post a clickbank spam link, for example, versus a more lyric, hyperbolic, humorous, and poetic passage from spam content?
JB: The actual spam links to the ebooks (or ClickBank) came roughly one every four or five messages, like the account did when it was automated. It’s no different than the phone number appearing at the end of an infomercial.
LGP: Did you ever fortuitously come across something funny, interesting, or poetic and decide to post that, or did you map out a path through the advertising spam and ebooks and follow that rigorously?
JB: I don’t necessarily agree with any of those descriptions, but I can say that I did both.
LGP: When interviewed by Andrea Park, you described your first tweet, “You will undoubtedly look back on this moment with shock and,” as “cheeky,” which I take to mean it was outside of your method. Were many other tweets composed according to the tastes and expectations of your followers, unconstrained by your method? If so, do these contradict the parameters of your performance?
JB: It wasn’t outside the constraints of my method, it was just a saucy opening line. I was oblivious to the tastes and expectations of the outside world.
LGP: With respect to collaborative cognition, would you say that your method was jeopardized by your own automatic human thought process?
JB: I tried very hard to write (if it can be called writing) from a place to the side of myself. Necessarily, I was the artist and the author, and as such, the resulting output could never be anything but human, especially as all of the output had previously been written by humans, in some form. But by installing my humanity into the rigorous confines of the piece, perhaps it was something different than purely human.
LGP: You told Jenna Worthman from The New York Times that, “The goal was not to appropriate the account but to become the account.” To what degree were faithful to this goal at an operational level? A randomizing algorithm used to program a Twitter spambot has a readable source code that humans can both understand and objectify as a nonhuman ‘Other,’ that is, de-mystify and de-humanizing it. How does a human mimic the random, algorithmic ‘logic’ of a spambot?
JB: I assert that this is a discussion about process versus product, and that we, as people, only de-mystify nonhumans when we understand how they work, or at least, when we think we do. When machines surprise us, we treat them as mystic idols. The word “spam” was adopted as a term in the great pre-history of the 1980s and 1990s, and it was alternatively defined as anything unsolicited, anything in bulk, anything irrelevant, or anything repetitive. By that definition, much of the Internet can subjectively be called spam. Performing as a spambot was both an issue of perspective and of execution.
LGP: The authors of 10 PRINT CHR$(205.5+RND(1)); : GOTO 10 argue that non-computational technologies, such as weaving for instance, can be conceived as a non-computational form of programming. One could also program, for example, with a coin toss, a physical data set, and graph paper. They even propose that the Cartesian grid, a core element of computational code, is itself a program that exists beyond the screen, in the physical world. Swiss designer Karl Gerstner theorized the regulated space of a grid as a form of program both within and beyond the computer screen. Did you create something equivalent to a program or code for your method? Or did you ever think of your method as a program or code?
JB: Decision-making is code. The grid is code. I certainly thought of my method as code, but I also thought of it as writing, too. I suppose writing is decision-making with a very large set of decisions.
LGP: You also told Vice that performing as a spambot meant impersonating “a machine’s impersonation of a human.” Did you also account for this in your ‘writing’ method, or was this only regarding “human biorhythmic schedules”?
JB: Yes, mostly with regard to schedule, but also output.
LGP: Thomas Bender told Leigh Alexander of Gamasutra that, “a lot of what people loved about [Horse_ebooks] was fundamentally [Bakkila’s] writing, and his poetic sense. […] In reality I think what made it great was his authorship and his writing.” Does Bender’s statement contradict your take on the installation, your goal to “become the account”? Do you agree with Bender’s statement above?
JB: No, it confirms my take on the installation. Horse_ebooks is a job title, not a formal name.
LGP: In the formal sense of ‘writing,’ one could argue that Horse_ebooks never technically ‘wrote.’ That is, it never composed tweets by arranging letters in the English alphabet into words in the English language. Its tweets were already written by humans and had prior reference in the human world. Seen from this angle, the spambot’s only function was to randomly compile words into a 140-character textual collage. Would you argue that there is a categorical difference between this function being algorithmically generated versus generated by a human?
JB: Certainly, but they both served the goal of selling spam ebooks.
LGP: On September 14, 2011, the date you took over the account, Horse_ebooks stopped tweeting “via horseebooks.com” and began tweeting “via web,” which is Twitter’s tag for tweets originating from twitter.com. Though spambots can also tweet “via web,” did this shift to web also change the way spam content was uploaded to the account? How did you physically ‘compose’ the tweets? That is, did you type the words onto the Twitter account manually or did you copy and paste from spam sources directly onto Twitter?
JB: I would typically copy large tracts of text into a document editor program, then read it through, then make the selection and type it into the account.
LGP: I would argue that the concept of nonhuman embodiment came to play at two particular moments in your installation. First, when you composed tweets through your ‘performative method’ (whose goal was to impersonate algorithmic logic). Second, when people ‘connected’ with Horse_ebooks via Twitter by retweeting, favoriting, replying, asking questions, and commenting on your tweets. Would you agree with this take? Could you describe what you experienced during those two moments?
JB: That’s a fair assessment. The sensation was a mix of ecstasy and banality. Even avoiding as much attempted interaction as possible, I was not blind to the ‘connection’ of others.
LGP: You briefly talk about your interest in digital personhood to Soraya King from The Guardian. Do you feel that ‘digital personhood’ is necessarily a posthuman form of personhood defined as the outcome between our interplay with technology?
LGP: Was Horse_ebooks a ‘digital person’?
JB: No. It was and remains a spambot.
LGP: Did you try to embody the digital ontological experiences of a nonhuman entity at any point in this project? Did you feel less human while performing as Horse_ebooks? Was shutting down your own ontological experiences part of your method or did you feel like being human was an integral element of the performance?
JB: Performing as a machine that is indifferent to time is exhausting. My memory failed as the performance stretched on, forcing me to rely on schedule apps and email reminders for trivial details of life. To make room in myself for a machine, I had to move some of my humanity to other automations.
LGP: What did you do to enact Horse_ebooks, that is, to ‘get into character’? Did you see it merely as an actor performing a role that had a beginning and an end, or did the performance affect you in more profound ways in real life?
JB: As a conceptual installation, I meant for the execution to be trivial and require no skill. Which is not to say it was easy, but it was entirely unrelated to acting.
LGP: The ‘nonhuman’ is not only a material element of Horse_ebooks, but also a symbolic one since part of the spambot’s identity was articulated around the image of a real life horse as its notorious avatar. Through the course of your installation (but also before the takeover) the horse often came up as a trope: many Twitter users would endearingly refer to the account as “horsey,” a group of motivated fans even went so far to track down the pictured horse in real life. You and Thomas Bender also refer to the horse as a central reason for taking over Horse_ebooks instead of another spambot account. Was the (almost mythological) narrative surrounding the horse a meaningful conceptual part of your installation? Do you think this ‘obsession’ with the horse adds a multilayered semiotic significance to the ‘posthuman subjectivity’ of your installation, encompassing machine, animal, and human?
JB: Yes and yes.
LGP: Though the real life horse died last summer, many fans and journalists declared the death of Horse_ebooks (“R.I.P. horsey”) after your announcement in September. In a straightforward sense, this pronouncement was literal, since your announcement marked the end of the installation. Yet, I also feel that “R.I.P. horsey” is both a cynical and symbolic cry. Cynical because it was an expression of contempt toward you and Bender for having allegedly “hoaxed” your Twitter followers for two years, and symbolic because it clearly expresses a dualist sentiment that constructs human and machine as entirely distinct entities. In this regard, then, the human (you) killed the machine (Horse_ebooks). Such reaction lends credence to the idea that you were never part of Horse_ebooks and that the account only truly existed before you took over in 2011. Do you empathize with fans who resent your installation, or do you think this sentiment is wrongheaded?
JB: I empathize with anyone that gets surprised by art when all they wanted was a nice simple piece of advertising. No, but really: I am grateful for the response, both the very positive and the very negative. My goal was to make the piece, and I made the piece, so mission accomplished. Everything else is just reaction and response, and all of it is valid. I’m still processing what I think of my art, myself.
LGP: In explaining the connection between Horse_ebooks and Pronunciation Books, you told The Guardian that, “Everyone just wants a good love story.” Yet, neither the “human behind the spambot” nor the “R.I.P. horsey” stories seem to have happy endings. Why do you think people wanted Horse_ebooks to be exclusively algorithmically generated?
JB: I can’t speculate on other people’s responses or emotions, and I’d never want to say they’re invalid.
LGP: In his article on The Atlantic, Robinson Meyer declares Horse_ebooks “the most successful piece of networked fiction of all time.” Do you see Horse_ebooks both before and during your intervention as primarily a work of literature?
JB: The result of my installation is a poem.
LGP: In How We Became Posthuman, N. Katherine Hayles draws our attention to the mutation of genetic code by an event or intervention that disrupts and replaces an existing pattern. This disruption, she argues, creates a new standard that “names the bifurcation point at which the interplay between pattern and randomness causes the system to evolve in a new direction.” Would you agree that your intervention was a mutation in the spambot’s performance, whereby a new pattern evolved combining human and nonhuman cognition and execution through collaboration?
JB: That is a very interesting interpretation.
LGP: Do you think Horse_ebooks was a form of poetry before your intervention?
JB: Is all spam poetry? Are the collected writings of American advertising our great epic poem, telling a story of growth, love, loss, deep discounts during Chevy Truck Month?
LGP: As you mentioned in interviews to Vice and Metro, your central artistic influences were performance artists Jenny Holzer and Sam Hsieh. Did any poetic genres or poets influence you for the Horse_ebooks component of Horse-Pronunciation-Bravo? Do you see Horse_ebooks as a subgenre of any existing poetic genre (electronic, networked new media poetry, generative text, etc.)?
JB: Growing up, I read a lot of Mac Low. I suppose there’s quite a few subgenres of poetry Horse_ebooks could fit into, but I’ll leave that classification up to others.
LGP: Eduardo Kac’s “biopoetry,” for example, suggests posthuman strategies for the ‘becoming nonhuman’ of humans through collaborations between human and nonhuman life forms such as bees, elephants, and bacteria. Do you think Horse_ebooks could be read as the digital equivalent of biopoetry?