Language as Gameplay: Toward a Vocabulary for Describing Works of Electronic Literature


by Brian Kim Stefans


Creators of electronic literature are progressing toward a more pervasive employment of the “ludic” — of the spirit of play inhabiting not just the writing, and not just the programming, but both in an elaborate, symbiotic combination. The tradition of “ludic” writing is well-rehearsed in criticism of electronic literature, for example in the magisterial anthology The New Media Reader, edited by Noah Wardrip-Fruin and Nick Montfort. The tradition starts somewhere with the Oulipo, a group of writers and mathematicians who sought to delete chance and subjectivity from their work by the employment of extreme constraints,[1] but moves backwards to encompass any writer who employed structurally challenging elements (Lawrence Sterne in Tristram Shandy, Robert Browning in his Rashoman-shaped epic The Ring and The Book, Lewis Carroll in Alice in Wonderland, or Arthur Conan Doyle with his Sherlock Holmes stories) and forwards to include authors such as Robbe-Grillet, Nabokov and Cortazar who had nothing to do with the Oulipo, or authors of physically indeterminate narratives such as Marc Saporta with his Composition No. 1 and B.S. Johnson with The Unfortunates, both of which are novels that are published as boxes containing pieces that can be read in any order. Programming itself is, by definition, “ludic”: it favors the paradigmatic dimension of language – that which can be parsed, broken into sets, arranged by classes, and reconfigured into syntactically coherent sentences via algorithm – over the syntagmatic, which is that quality of a sentence that exists uniquely in space and time (that is, the moment it is spoken by an individual). Lev Manovich elaborates on the paradigm/syntagm distinction and its importance for databased language in “Database as Symbolic Form”:

[T]he elements of a system can be related on two dimensions: syntagmatic and paradigmatic. As defined by Barthes, “the syntagm is a combination of signs, which has space as a support.” To use the example of natural language, the speaker produces an utterance by stringing together the elements, one after another, in a linear sequence. This is the syntagmatic dimension. Now, let’s look at the paradigm. To continue with an example of a language user, each new element is chosen from a set of other related elements. For instance, all nouns form a set; all synonyms of a particular word form another set. In the original formulation of Saussure, “the units which have something in common are associated in theory and thus form groups within which various relationships can be found.” This is the paradigmatic dimension.[2]

More recently, electronic writers have resolved certain problems inherent in hypertext works – how long should the user spend with it? when does a user know when they are done? how does a user learn the often bizarre navigational conventions of a new work? – by drawing on the conventions of human/computer interaction as they’ve developed for decades in video game culture. If older interactive digital text pieces could be compared to interactive art in which the user is invited to engage in free play – Camille Utterbeck’s “Text Rain” is an obvious example of this, as it is in fact an installation, but any hypertext narrative that doesn’t have a strong plot element is essentially a form of interactive art – then more recent works can be called “task-based interactive art” in that the user is given a set of goals to achieve while navigating the work.

Evidence of this recent turn toward game elements can be seen in works of digital art that are still intended, at least partly, to be read — Jason Nelson’s “game game game and again game,” Noah Wardrip-Fruin’s 3D Cave-installation “Screen,” and Daniel Howe/Bebe Molina’s “Roulette” are three examples – and sometimes only heard, as in more elaborate audio/visual projects such as “Façade” by Michael Mateas and Andrew Stern which places the user in a 3D video game space with the unique challenge of trying to make his/her party hosts, a warring couple, kiss and make up via her actions and inputted text (this interactive fiction is intended to be able to assimilate anything typed by the user, though this often seems less than the case). The good news for critics and theorists is that it appears that the range of interests for electronic writers — as distinct from new media artists, a much larger group of creators who threaten to simply absorb the “electronic writing” concept — can now largely be triangulated between three conceptual nodes: screen space (against page space), algorithm (against author), and this most recent entrant, gameplay (against the more open-ended “interactivity”). The value of such a triangulation has become clear in recent years as video games have themselves advanced to the level of high-brow “literary” or “painterly” (or simply) art. The language for describing these new works of digital literature has had to move further from even advanced literary critical vocabularies, especially since the case has been made recently that long-form narrative video games — premised on the interplay of data, algorithm and player within the eschatological structure of “narrative” — and electronic literature (as opposed to new media art) have a family resemblance.[3]

In this paper, I’d like to examine a work of non-“electronic” literature that suggests reading itself as a form gameplay, Christian Bök’s hyper-Oulipian work Eunoia, and several works of electronic literature that attempt, on some level, to employ gameplay themselves, including Stuart Moulthrop’s “Pax: An Instrument,” Talan Memmott’s “Self Portrait(s) [as Other(s)]” and Judd Morrissey’s “The Jew’s Daughter.” I will introduce a critical vocabulary for examining these works, grouped around the following concepts: “The Holy Grails of Electronic Literature,” “Six Varieties of Crisis,” and the “Surrealist Fortune Cookie.” Respectively, these describe: the contradictions inherent between paradigms of science and paradigms of literature and how they have shaped motivations by creators; the manner in which writers of electronic works can provide “non-trivial” reading experiences in the absence of standard literary paradigms premised on apocalyptic (or simply “plotted”) narrative; and a concept of the basic unit of the sentence in an algorithmically created work as a “genre” of its own. I employ these lists of qualities and objects because it’s my general belief that all successful works of “electronic literature” are sui generis, in that they invent new genres unto themselves – their peculiar combination of data, interface and algorithm makes them appear largely unrelated to other digital literature works – and so they must more or less be mapped between sets of gravitational poles attached to no single work or genre. A digital literary object will be characterized by a set of texts (even if one extensive, core text), an algorithmic procedure that operates upon it, and an interface with the user (which needn’t be “interactive,” and needn’t be electronic). Of course, this requirement that a successful work of electronic literature be sui generis might not be the case in perpetuity, and certain works, like “Dakota” by Young-Hae Chang Heavy Industries and “Project for a Tachistoscope [Bottom Pit]” by William Poundstone, do seem to circumscribe a single genre – text movies in which words are replaced on the screen outside of user control – but this is the way I see it presently.[4] However, I do believe that unsuccessful electronic literary works can be primarily defined as those that did not make a strong case for the economy, beauty and playability of its combination of text, algorithm and interface – they fail on one of those levels, and hence fail to create their own genre of digital literature.

The Holy Grails of Electronic Literature

A holy grail is, of course, a now-colloquial symbol representing an object of sacred quest, by implication one that is unobtainable. The search for the holy grail is not necessarily Quixotic – our seekers are more alchemists than fools or dreamers – but it does seem to stand outside of the standard economy of desire and goal-fulfillment by having a quasi-visionary[5] element to it. I call these aesthetic goals “holy grails” because they exist in a no-man’s land between achievements that satisfy rubrics of science – concrete conceptual and technological breakthroughs that can exist on the continuum of paradigm-shifting discoveries like those of Copernicus or Pasteur[6] – and achievements in the arts that are of lasting historical value for their “humanistic” content – how these discoveries, put loosely, help us to explain the human condition. Art works that negotiate this apparent divide are actually not very scarce; architecture operates within this space all the time, for example, as do works of land art such as Walter de Maria’s “Lightning Field” and Robert Smithson’s “Spiral Jetty.” Electronic devices such as the iPhone (which many see as an architectural innovation more than anything else[7]) have done as much to alter paradigms of sociological or even biological thinking as they have transformed inter-human engagement and creativity. However, the conceptual purities of my “holy grails” are so extreme that they exist more as gravitational pulls rather than achievable goals; they simply cannot be obtained, since upon attainment the very aesthetic bases of the quest are destroyed – the work moves purely into science. Of course, in both cases, but more so with art, notions of “success” cannot always be determined strictly quantifiably or with any immediacy, as it is history itself that often provides the context highlighting the work’s achievement.

Writing Without the Author: to create a piece of literature that can be read several different ways – none predetermined by an “author” – which will provide distinctive, compelling reading experiences each time.

The primarily quest here is to displace the “author” entirely onto the algorithm and its operations on a database, ideally a communally and passively authored, infinitely expanding database such as the internet. Conceptually, this concern has been seen (especially by early hypertext critics such as George Landow) as a response to the poststructuralist preoccupation with the “death of the author.” Italo Calvino lovingly savages this perverse antisocial tendency quite comprehensively in his novel If on a winter’s night a traveler, in which works of meaningful literature are created variably by computers, visionary prophets and plagiarists. Christian Bök, in his writing on robotpoetics,[8] creates a very compelling, if still conceptually-flawed, portrait of a time when computers, having acquired a sense of intellectual vanity, will write poems purely for consumption by other computers: “Once a literary computer can analyze the formal limits of its own prior poems in order to revolutionize its output, anthropic culture may have to compete with an automated culture, whose spambots are already better equipped to overwhelm us with an enfilade of computer messages.” The anxiety permeating this holy grail – Oedipal in some sense, if we think of Harold Bloom’s famous articulation of artistic succession in The Anxiety of Influence – is with making a departure from certain forms of classic hypertext and its embedded link model as these feel more like an elaborate version of a “choose-your-own-adventure” story than a work of beautiful engineering. This anxiety is marked by a consciousness of the triviality of the toy, which only seems infinitely variable in the mind of a child, and its stark contrast: the sublime machinations of artificial intelligence.

Corollary goals of works that aspire to this holy grail include:

• that the work be entirely syntactically coherent, not to mention semantically coherent. One might think that all works that aspire to be “authorless” want to attain this goal, but in fact several existing works of algorithmic literature – inspired, perhaps, by the print example of William S. Burroughs and Brion Gysin’s “Cut-Up Method,” and, later, the Language Poets – are quite happy to be syntactically un-parsable and semantically indeterminate. These works take advantage of the inherent poetic quality of “broken” language.
• that the work, if it claims to be a “narrative,” provide in addition to point-by-point interesting text the clear lineaments of a “story arc” – set-up, inciting incident, rising action, denouement, etc. Stern and Mateas’ “Façade” might be most elaborate example of such a piece; the creators describe the issues they had with narrative form exhaustively in their essay “A Preliminary Poetics for Interactive Drama and Games.”[9] Open-world video games, such as Grand Theft Auto, permit the player to engage without slavish adherence to a story arc, though the arc is there if one should choose to stick to it.

• that the work, if it claims to be a “poem,” have a sort of lyric integrity that we associate with poems, not to mention a metrical integrity that we associate with traditional forms of poetry. Charles Hartmann’s experiments with computer generated poetry (which he describes in Virtual Muse: Experiments in Computer Poetry[10]) involved training a computer to parse works by poets such as Samuel Taylor Coleridge and, after such self-education, to write metrically stable works in that style. Most efforts at computer-generated poetry (such as Racter) have not bothered to deal with meter, and leave poetic subjectivity up to the chance encounter of an “I” and a verb.
• that the source text is entirely derived from a corpus of texts that were not written consciously as literature, such as the Wall Street Journal or the internet. Noah Wardrip-Fruin’s “NewsReader” and “Regime Change,” derived entirely from web texts, are versions of this, though little attention is paid to syntactical integrity in these works. “Get a Google Poem” by Leevi Lehto is a playful attempt at achieving some of these goals, though the piece as been entirely destabilized by changes in textual production – the predominance of PHP and XML over HTML as describing page architectures – in the Web 2.0 era.
that the reading object is experienced in “real time,” like a song or movie, rather than as an artifact that could very well appear on a page. An aspirant to this grail might be David Rokeby, whose “The Giver of Names” experiments with computer generated poetry that derives its inspiration from an electronic eye, and which learned language by parsing classics like Moby Dick and Lolita. “Façade” is, again, an example of this work, as it attempts to incorporate real-time user input into a story that unfolds largely like a film.

Reading Beyond the Page: to provide a visual-textual experience that exists in a three-dimensional, dynamic, entirely processed space – that is, moving as far away from the physical, “static” page as possible.

Since Marinetti’s experiments with Futurist typography and the elegant early masterworks of visual poetry such as Mallarmé’s Un Coup de Des and Blaise Cendrars’ La prose du Transsibérien et de la Petite Jehanne de France (in its book incarnation by Sonia Delauney), poets have aspired to create poems that flouted the conventions of page space – a goal seen as a “liberation” of the word from the page. Some of the most sophisticated theories of concrete and visual poetry can be found in writings by the Noigandres group in Brazil (especially in the theory of Haraldo de Campos), Charles Olson in the United States (whose “Projective Verse” is as much a theory of the body/mind as the page), and the writings of the Toronto Research Group (Steve McCaffery and bpNichol, collected in Rational Geomancy: The Kids of the Book-Machine[11]) who sought, in general, to reconcile the theories of the earlier generation of poets to new trends in post-structural thought. The real near-contemporary masterwork of visual poetics and theory is not a book of poetry at all: The Medium is the Massage, a collaboration between Marshall McLuhan and Quentin Fiore, is increasingly being seen as a milestone of verbico/visual poetics, and is the basis of my own theories of the “ludic book.” The real excitement of this “grail,” however, comes from outside of literature: the elaborate and innovative movie titles of Saul Bass and the several brilliant movie titlers who have exploited his breakthrough; the integration of textual elements into architectural works (as beautifully described in the seminal Learning from Las Vegas, by Robert Venturi et al.); the visual poetics inherent in the political theory of the Situationists, whose concept of the “detournement” sought to employ the aporias and breakages of text as cudgels against the spectacle; the musical experiments in indeterminacy by the composer John Cage that easily spilled over into visual poetics, such as the gorgeous mesostics – in which font, spacing as well as words were determined by chance – collected in M.

Any number of Modernist experiments with typography starting with the Futurists – surveyed in such works as Marjorie Perloff’s Radical Artifice and Johanna Drucker’s The Visible Word – to conceptual artists who work with language in physical spaces, including Lawrence Weiner, Barbara Kruger and Ian Hamilton Finlay, inspire the artists associated with this “holy grail.” The key break, though, is that digital artists work with what N. Katherine Hayles has dubbed “flickering signification,” which “extends the productive force of codes beyond the text to include the signifying processes by which the technologies produce texts, as well as the interfaces that enmesh humans into integrated circuits.”[12] An author who aspires to this “holy grail” hopes to amplify, rather than suppress, the visibility of these behind-the-screen dynamics.

Corollary goals of works that aspire to this “holy grail” include:

• that the work is immersive, such that the user is present diegetically – as an active role in the physical universe – in the text. Camille Utterbeck’s “Text Rain” is a basic, two-dimensional push in this direction, while Jeffrey Jones’ “Legible City,” which incorporates a bicycle as the main physical interface, is probably the most accomplished of such pieces, though in both cases the text is quite trivial or not engaged deeply with the algorithm. Several experiments in Brown’s Cave environment have attempted to utilize more absorbing texts, most notably Noah Wardrip-Fruin’s “Screen.” Daniel Howe’s experiments in Java, such as “Recube,” another 3D textual environment, are also angling toward this “grail.”

• that the visual output of the piece has a visceral efficacy on the level of film or theatre, and yet remain legible as “text” at all or most times. John Cayley’s experiments in Brown’s Cave have explored this property, though one must look at the work of filmmakers, such as nouvelle vague maestro Jean-Luc Godard, or players in the net art community, such as Young-Hae Chang Heavy Industries, for works that have most achieved this effect. The Flash version of “Bembo’s Zoo,” a child’s abecedarian created by the Roberto de Vicq de Cumptich, and my own “Dreamlife of Letters” might be other examples. The viral text video, “What Does Marcellus Wallace Look Like” created by Jarratt Moody, which synchronizes animated text with audio from the film Pulp Fiction, is a powerful example of this tendency.
• that a 3-dimensional space not only seem accommodating of text, but in fact creates situations in which hitherto unknown properties of text becomes revealed. Peter Cho’s exquisite series “Wordscapes” and “Letterscapes,” which are minimally interactive, reveal new properties of the visible word with each mouse swipe. Once again, the experiments in the Cave come closest to achieving this goal, as one encounters therein text from various angles, at varying pseudo-depths. Speed and proximity are the primary variables at play, as is richly described by Robert Venturi et al. in Learning from Las Vegas[13]; billboards, for example, have created a new genre of textuality, properties best exploited in the paintings of Ed Ruscha.
• that, for all of the variations of the text’s appearance on the screen, the textual image always maintains what can be called high standards of graphic design. One deficiency in electronic literature is the failure to deal with the history, standards, affects and ambitions of graphic design,[14] though recent works, such as David Clark’s “88 Constellations for Wittgenstein” and Sharon Daniels’ “Public Secrets” (designed by Erik Loyer) have made great strides in this direction. The iPad application Flipboard, which collects data from any number of sources such as the users’ social networking sites, RSS feeds, and blogs and reformats them in a sexy, consistent-but-variable graphic design – such that status updates, trivial or not, are set off like pull quotes in a high-end magazine – shows the powers of algorithmic graphic design in granting the air of cultural capital to ephemera. The video game “Shadow of the Colossus” is a third person game characterized by dynamically generated camera angles (rather than the fixed, over-the-shoulder perspective offered by most games), hence carrying over, successfully I think, the standards of good cinematography – shot composition, panning and tracking, even editing – into an algorithm.

Writing/Reading as Gameplay: To create a programmed object that serves equally as a piece of literature – something that can be read – and which serves as a “game” with all the fun implied in such a designation.

What differentiates a video game from interactive art is that the former instills in the user a lusory, or playful, attitude, or with what Bernard Suits terms “pre-lusory” goals – not just the desire to experience something different, but the desire to achieve a game state in which one can say one has “won.”[15] The pre-lusory goal in a game of soccer, for instance, is to have the ball appear, if even for the briefest second, in the opponent’s goal (“kicking the ball” is part of what Suits would call the “lusory means”; not touching the ball with the hands would be a “constitutive rule”). The key difference with the above holy grails is that the algorithms producing the text will have a quasi-cybernetic relationship with the user: the game state of the textual apparatus will adjust according to the user’s ability to “play” it. The user engages with the piece with a competitive motivation rather than merely the desire to continue an indeterminate aesthetic experience. In the first holy grail, writing without an “author,” all elements of gameplay are subsumed in the programs producing the text – the user herself never enacts algorithm in her input. In a true video game, as in a truly interactive fiction such as “Façade,” the user learns and adopts the set of inputs and acquires a sort of visceral feel for their effects – an illusion of “control” over an animated/simulated field – whereas in standard “hypertext,” this sense of control is largely absent because there are limited expectations on the part of the user that their input will do anything, certainly not change the game state – it is largely “indeterminate.”[16] Consequently, by inviting the user into engagement with the algorithm, “algorithm” itself is demystified – one’s fascination with the operations of an invisible, exotic algorithm (the “Orientalism” of the continent of code, as I’ve termed it elsewhere[17]) is reduced to the degree that the flesh-based player gains some sort of visceral mastery over it. One is not left in a state of fascination over the operations of pseudo-AI code, as in the first “grail,” but enacts algorithm him/herself. As Manovich writes:

[I]n games where the game play departs from following an algorithm, the player is still engaged with an algorithm, albeit in another way: she is discovering the algorithm of the game itself. I mean this both metaphorically and literally: for instance, in a first person shooter, such as “Quake,” the player may eventually notice that under such and such condition the enemies will appear from the left, i.e. she will literally reconstruct a part of the algorithm responsible for the game play. Or, in a different formulation of the legendary author of Sim games Will Wright, “Playing the game is a continuous loop between the user (viewing the outcomes and inputting decisions) and the computer (calculating outcomes and displaying them back to the user). The user is trying to build a mental model of the computer model.”[18]

I refer to this process as pseudo-cybernetic to the degree that the computer is not actually engaged in active observation of the physical world, but is responding to information the user volunteers in the form of thumb-twitching or typing. Nonetheless, both computer and player are engaged in constructing “mental models” of the other’s operations (the player herself behaving “algorithmically”) and adjusting accordingly.

Corollary goals of works that aspire to this “holy grail” include:

• that the desire to “win” is entirely centered around textual consumption, rather than the seemingly unrelated process of, say, getting your character to escape from a maze or to avoid falling bricks. Jason Nelson’s “game game game and again game” is flawed because of the lack of the presence of the text in the diegetic world of the game – the text itself changes nothing the user or non-player characters do. My own “Kluge” was an attempt to create a video game in which the desire to read was the inspiration for superior play, but it is largely a failure on this level (too much text, simply). Wardrip-Fruin’s “Screen,” during the last phase during which the user attempts to reposition falling words in a text, is a version of this, though the user has already read (and heard) the text by this point in the program.
• that generative processes as described in the first “holy grail” are the direct results of successful gameplay, which is to say, the creation of a textual game that literally creates its own new levels – new chapters of a fiction, new stanzas of a poem, new acts of a play. A holy grail in the video game industry itself is a game that can generate its new levels on-the-fly; several games have implemented procedural generation to some degree,[19] but not of a quality as interesting and engaging as the meticulously crafted levels that are hardwired into the program. Once again, “Façade” might be the closest approximation of this possibility. 

The Six Varieties of “Crisis”

Frank Kermode, in his influential book Sense of an Ending: Studies in the Theory of Fiction,[20] makes a compelling argument that the power of narrative lies in its movement toward an apocalyptic moment: the time at which everything we have known (in the narrative, in life) is submitted to judgment, a comprehensive trial that is, at the very last, highly uncomfortable. A reader of a text, much like a person in life, is caught in the “middest,” a place in space and time from which you can’t see past or future; one’s only recourse is to create arcs that anticipate future events, most likely based on your knowledge of the past. Some of these arcs are already provided for us when we become acculturated when growing up; for example, some forms of Christianity – the religion most associated with apocalyptic thought – actually provides a concrete description of a future cultural apocalypse. However, even a perfectly secular story about a single individual will, if it is properly plotted, provide the basic rubric of apocalyptic thought, a large part of which has to do with anticipating finality, and then revising this prediction when the moment passes or when facts and experience points elsewhere. These moments of failed apocalypse are moments of “crises”: the life pattern according to which one was living – the very pattern providing meaning to what would otherwise be disordered, atomized, ephemeral moments of experience – falls away, and we are only left with the ticking of the clock to chart time’s passing.

Most works of electronic literature, especially those that aspire to the above “holy grails,” will seek to avoid closure, and in fact advertise this very fact; but with no perceived end to an artistic experience, no anticipated “summing up,” the strong urge to apocalyptic thinking is discouraged. What you are left with, then, is a series of more or less disparate experiences extending in an endless chain before and behind you. To this degree, it is important that the creator of a piece of electronic literature provide other types of significant milestones in the course of the reading – “crises” that don’t rely on failures of apocalyptic thinking but which are nonetheless non-trivial. Implicit in any of these milestones or crises is a lack of knowledge (since one is, in fact, still in the “middest” when engaging with electronic literature) as to what significance each moment of play has in our framing of the underlying algorithm. One asks if point X – an eventful engagement with the electronic literature piece – is a moment of “consonance” with one’s view of the paradigm, a node of the learning curve in trying to determine the algorithm, or if it completely thwarts any mental construction of what is occurring in the digital other. The relative success or failure of these nodes of “crisis” would contribute to our understanding of whether the reading experience is “non-trivial.” These crises – free-floating events not linked to paradigmatic narrative structure – are, of course, experienced in time, and the play of these crises upon each other must be understood temporally.

The variations of “crises” that I list below are rudimentary, and I won’t, for the most part, try to link them to existing works; the case studies that follow this essay will play a much greater role in suggesting the value, or lack of value, of these “crises.” (See Appendix II.)

1. Crisis of ESCHATOLOGY – we are not sure where, in the standard narrative paradigm, poetic paradigm, or essayistic (syllogistic) paradigm, we are located nor can we, for the moment, imagine the end. Works that don’t advertise themselves as endlessly variable will often suggest that some forms of narrative convention are still in play (for instance, early hypertexts such as Michael Joyce’s afternoon: a story, which starts with an inciting car crash and death, and hence anticipates a moment of reckoning). Successful works of electronic literature will often be agnostic on the issue of whether normative narrative conventions are at play.
2. Crisis of SIGNIFICATION – something has occurred in our understanding of conventional relationships between word and thing, or even letter and word; language seems to be becoming pure scription and “non-referential.” Many works of electronic literature will draw from datapools of text organized in some fashion, but we are never quite sure how “smart” the algorithm is in choosing to utilize this text; thus, a chain of signifiers can either be pure garbage output – like a piece of spam – or a conventionally interesting piece of rhetoric in which a higher-level wit is at play. The works of Toadex Hobogrammathon – such as “Dagmar Chili,” a blog, and “Name: A Novel,” which appears, after a brief narrative introduction, to just be a list of words – are examples of this.
3. Crisis of SYMBOLISM – something seen to have a merely contingent value is seen to have a role in a symbolic universe. I often point to the novels of Thomas Pynchon as the great example of the paranoiac imagination in narrative fiction; a more difficult, perhaps accidental, form of this is the play of particular and symbolic (or allegorical) in Ezra Pound’s Cantos.[21] Algorithms are particularly good at instilling the paranoiac imagination, in which accidental occurrences are signs for larger forces, since the combination of purely mathematical thinking utilizing particulars from the human universe suggests a power over signification that can only be termed god-like in its indifference. The meta-fictional works of Donald Barthelme play with this possibility quite wonderfully.
4. Crisis of SUBJECTIVITY – the narrative “I,” whether of third or first person, has shifted. The basic test-case for this sort of crisis is displayed in the story of “Eliza,” a simple interactive text piece created by Joseph Weizenbaum that imitated the activities of a Rogerian therapist, one trained to merely ask questions that are interrogative versions of the patient’s previous statements. Weizenbaum’s secretary, who it must be assumed had some serious problems, spent hours with the chatbot not knowing anything about the programming and was apparently brought to tears in the end. This crisis of subjectivity can be brought into play in any case where it is not clear whether an algorithm or an “author” is behind the “I”; any frequenter of MMORPGs, Second Life or chat rooms will confront this possibility regularly.
5. Crisis of GENRE – we have slipped from a narrative event to a poetic one, or more critically, from a non-fictional, documentary mode to one that seems colored by the imagination of an “author.” Werner Herzog is notorious for injecting in his apparent “documentaries” moments of pure fiction – seemingly candid interactions being entirely staged, for instance – and works such as the novels of W.G. Sebald and the Museum of Jurassic Technology in Los Angeles demonstrate the power of keeping the user in a state of tension as to truth and fiction (which, after all, is the constant state of the internet).
6. Crisis of MORALITY – something in the flow of words has forced us to question our own place in the social network due to the “danger” of assimilating these words into our experience – i.e., will I choose to “own” this reading experience or not? This basic crisis is endemic to any interaction with the internet, most of which occurs in private but which tends to leave some digital trace. A piece like the Pornolizer, which is a simple program that replaces the verbs and nouns of actual webpages with (usually very antiquated) language from pornography, enacts its operations with sublime indifference to morality.[22] Any program that creates text based on randomization creates the potential to create inflammatory language, so unless the user is completely versed on the operations of the algorithm, he/she is not sure whether the production is hate speech and/or pornography or merely the comedy of robotics gone awry.

The Surrealist Fortune Cookie

A sentence by nature points to other sentences – they never exist out of context, even if there isn’t another sentence on the page, or even if the sentence is spoken in isolation from other sentences. Wittgenstein describes in the Philosophical Investigations how a one word sentence, “Slab!,” implies all sorts of corollary semantic (and syntactic) material. The word “Slab!” can be seen as a condensed form of the sentence “Bring me a slab!,” in which case it is elliptical; or it can, accompanied with a pointing finger, represent the simple sentence “That is a slab,” though in that case, we are not sure if the sentence in fact is saying “That is gray” or “That is hard,” our imaginary auditor presumably having no previous experience with English vocabulary upon which to draw to make the proper deductions.

It’s possible to think of some sentences pointing outward more than others. For example, “He sits in her car” seems to beg the questions: who is he? who is she? if he is sitting in her car, presumably there is a reason – did she invite him? did he break and enter? etc. Another sentence, “Bill fumbled with the keys he had stolen from Natalie’s dresser, clumsily inserted the largest and turned until he heard the familiar, dull click! denoting access, swung the door wide and savagely, and entered the dilapidated Nissan” on the face of it contains more information, though questions persist: why were the keys stolen? why did he fumble (what made him nervous)? how did the car reach this state? More importantly, this latter sentence allows us to drop a sentence like the pronoun-laden former after it with several of the holes filled in: “He sat in her car” now seems quite over-determined in meaning rather than the nearly abstract, unspecified image of a man sitting in a car.

These linkages between sentences are at the heart of the algorithmically generated text project. The problem confronting a writer of algorithmically generated literature is how to construct sentences that have a dynamic relationship to others when brought into some proximity, but which can also resonate with suggestions when seen in isolation. This is actually a problem that any writer of character-limited status updates or tweets confronts: how to make a sentence that can operate in a field of other sentences, appreciating a wide variety of contexts while making a substantial aesthetic (or affective) impact that is relatively undiluted – i.e. true to “intentions.” Electronic writing has elevated any number of minor forms of writing – subway advertisements, comments on Facebook or Myspace, cursory emails, and even marginalia scrawled in books as publication of scanned pages images becomes the norm among scholars – into modes that acquire the feel of “genres,” most likely because our sense of scale has shrunk since our reading has become more “linked,” one small lexia leading to another with no sure order.

I like to think of this genre of micro-lexia as “surrealist fortune cookies,” because they contain some of the prophetic qualities that we associate with fortune cookies while acquiring an air of the uncanny, mostly due to the unresolved narrative elements implicit in them. A “Surrealist fortune cookie” would also touch off some element of the various “crises” noted above, particularly as relates to the possibility of there not being a human “author” behind the sentence. Certain lines from the poems created by Racter – “They have love, but they also have typewriters. That is interesting.” – point to this possibility. John Ashbery’s poem “37 Haiku,” really a series of single line poems, play along these lines, as does much of Ashbery’s writing on a sentence-by-sentence level. Note the subtlety in which each of these sentences (the first five of the “37 Haiku”) just pushes over into syntactic indeterminacy:

Old-fashioned shadows hanging down, that difficulty in love too soon

Some star or other went out, and you, thank you for your book and year

Something happened in the garage and I owe it for the blood traffic

Too low for nettles but it is exactly the way people think and feel

And I think there’s going to be even more but waist-high[23]

Quite often, the details in his poems suggest a higher symbolic order, as if the poem would eventually resolve into a coherent meditation on reality like in an essay-poem by Wallace Stevens, but in fact the details more often acquire, and cling hard to, the enigmatic element of a surrealist object right out of the world of Isidore Ducassse, the Comte de Lautréamont, who wrote in Les Chants de Maldoror of the beauty of “the chance meeting on a dissecting-table of a sewing-machine and an umbrella.” Enigmatic juxtapositions such as these produce the “tick” that, in Kermode’s view of narrative, set apocalyptic thinking in motion since they create situations in which hermeneutic activity seems to become necessary. Barthes’ writing on hermeneutic and proairetic codes[24] is also apposite here: “Surrealist fortune cookies” create situations not unlike that in detective novels, in which the placement of things in proximity combined with the plot-element – the big question, Whodunnit? – engages the reader in a forensic quest for closure. The Language poets, particularly Charles Bernstein in such poems as “The Klupzy Girl,” create situations where the linkages between sentences are always troubling, but cannot quite be resolved. The Language poets in general have, more than any (since Whitman, at least) elevated parataxis – the list form – as a major poetical device, one that can dominate large works to the exclusion of all syllogistic or narrative structure. Parataxis is perhaps the 20th century’s greatest contribution to the history of poetic forms, though as of yet it hasn’t quite taken a hold in “mainstream” poetry in the same way that, say, the royal couplet (Chaucer) or the sonnet (Wyatt) did in their time.

A robust database of phrases, sentences, even words, created for the purposes of a recombinant piece of digital literature would be aware of these very properties of sentences. Sentences would be classed according to their paradigmatic nature (how many of the words are mere filler waiting for concrete “content” from other sentences) or syntagmatic (how many words seem to relate to a very particular situation of time and place). A higher-level programmer of recombinant works would have a sophisticated, detailed and aesthetically nuanced sense of how sentences can engage in playful relationships with each other, thus raising the literary possibilities above that of the merely random. Talan Memmott’s “Self Portrait(s) [as Other(s)]” has approached this level of aesthetics, but quite often, programmers have been content with approximating the effects of a Mad Lib or something like Raymond Queneau’s suite of “A Hundred Thousand Billion Sonnets,” which, for all of their aesthetic value, somehow seems to be more of a demonstration of the potential of recombinant literature rather than an enactment of an immersive, perhaps even sublime, work.


As I argue in Fashionable Noise, generative digital literature, especially works depending entirely on computer algorithm, starts as “noise,” the undifferentiated assemblage of pieces of language, and from there must approximate humanistic conventions of writing/reading literature in order to activate that spirit of playful reading that all “good” literature seems to be able activate. If electronic literature is ever to move from the “noise” on the fringes, where it thrives more or less as a coterie art, to something approaching the “center,” creators have to make peace with a few of the basic premises of literature, rather than referring to a narrative of technological progress that has already retired the book to the ash heap of history. The allure of apocalyptic narrative will not disappear with the absence of the book; after all, it predated it, and it persists in several cultural situations today in which books play no great role in the dissemination of stories, such as popular movies. The potential for a wider readership is there, even if I’m not optimistic it will realized, and don’t necessarily think it has to be to justify the venture as a whole. Alain Robbe-Grillet, author of some of the most extreme nouveau roman novels such as The Voyeur, saw himself strictly within the realist tradition of Balzac and Hugo, and lamented, in his short essay “The Use of Theory,”[25] that he was “not satisfied to be recognized, enjoyed, studied only by specialists who had encouraged me from the start; I was eager to write for the ‘reading public,’ I resented being considered a ‘difficult author’.” He saw himself addressing the questions of literature starting with the birth of the novel through the 19th century, through Flaubert, Kafka and Joyce to himself. Electronic literature will not progress until its creators also try to situate the particulars problems of their art within the larger literary tradition (not to mention the traditions of graphic design, visual art, sound design, film, etc). It’s not easy, but that’s the challenge.

Ludic writing offers a way of injecting this sort of “spirit of play” into a literary object without compromising the ambition to satisfy, on some level, the calls of the “holy grails.” Whatever vestiges of Romanticism that exist in literature have to be either contained (through irony) or simply deleted from the textual component of a digital literary object; the chance for bathos and absurdity is just too great simply because the power of the computer algorithm will always be greater than the authorial power of the “author” of the texts. As I argue in “Privileging Language: the text in electronic writing,” [26] the creator of a recombinant digital literary object has to “privilege language” rather than imagine it on the level of other digital objects – sound and image files, nodes of interactivity, even the shapes and colors that define how those words will be presented – since it is language that has the potential more than anything else to render a user’s experience entirely banal despite the combined powers of sound, color and movement. The theater of “meaningful text” being manhandled by algorithm – Wallace Stevens’ “13 Ways of Looking at a Blackbird” turned into an interactive, dynamic text environment, for example, which I’ve seen assigned as an exercise to learn Flash – implies a certain nihilism: death itself becomes pointless, as the vulnerable texton loses, again and again, to the actions of a scripton. The text must be prepared to face the constructed contingencies of the program, and by doing so offer some solace to our own sense of being in the “middest” of a personal narrative increasingly defined by the stretching of the human body over several databases, all of them banal: online banking sites, credit rating reports, Amazon recommendations, Twitter feeds, the entire panoply of methods by which Faceboook seems to want to reduce you to choices that can, themselves, be exploited for commercial gain. The confrontation with the triviality inherent in the dissolution of longer, potentially sublime texts into fortune-cookie size bites is the social context in which advanced electronic literature objects can operate, should its creators decide to accept this mission.

Piece originally posted at Electronic Book Review |


[1] George Perec, Raymond Queneau, Italo Calvino and Harry Mathews are probably the best known authors to an English reading public.

[2] Lev Manovinch, “Database as Symbolic Form,” Convergence, June 1999 vol. 5 no. 2 80-99. For a fuller version of this argument see chapter 5 of The Language of New Media (Cambridge: MIT Press, 2001).

[3] The set of MIT anthologies edited by Pat Harrigan and Noah Wardip-Fruin, titled First Person: New Media as Story, Performance, and Game, Second Person: Role-Playing and Story in Games and Playable Media, and Third Person: Authoring and Exploring Vast Narratives, state this case quite comprehensively.

[4] Another numerology-inflected essay that I’m working on, “The Comedies of Separation,” approaches this question from an entirely different angle, and actually doesn’t mention this distinction.

[5] Florian Cramer has examined this mystical nature of codework in his book, freely available on the web, Words Made Flesh: Code, Culture, Imagination.

[6] As described in Thomas Kuhn’s The Structure of Scientific Revolutions (Chicago: Univ. of Chicago Press, 1962).

[7] “Architecture bloggers Stephen Becker and Rob Holmes winkingly named the iPhone as one of the most important architectural works of the first decade of the new millennium, arguing: ‘urban systems are defined most fundamentally not by structure and infrastructure, but by practice, action, and thought-process; what act has more significantly altered the practices and thought-processes of urbanites in the past ten years than the mass distribution of smart phones?’” Joshua Noble, “Exploding Space: Conceptions of Space and Network in Interactive/Dynamic Architectures”, June 23rd, 2010.

[8] See “The Piecemeal Bard is Deconstructed: Notes Toward a Potential Robopoetics” ( in which he concludes: “We are probably the first generation of poets who can reasonably expect to write literature for a machinic audience of artificially intellectual peers. Is it not already evident by our presence at conferences on digital poetics that the poets of tomorrow are likely to resemble programmers, exalted, not because they can write great poems, but because they can build a small drone out of words to write great poems for us? If poetry already lacks any meaningful readership among our own anthropoid population, what have we to lose by writing poetry for a robotic culture that must inevitably succeed our own? If we want to commit an act of poetic innovation in an era of formal exhaustion, we may have to consider this heretofore unimagined, but nevertheless prohibited, option: writing poetry for inhuman readers, who do not yet exist, because such aliens, clones, or robots have not yet evolved to read it.

[9] In First Person: New Media as Story, Performance, and Game, edited by Noah Wardrip-Fruin and Pat Harrigan (Cambridge: MIT Press, 2004).

[10] Middletown: Wesleyan University Press, 1996.

[11] Vancouver: Talonbooks, 1992.

[12] “Virtual Bodies and Flickering Signifiers,” October 66, Fall 1993.

[13] Cambridge: MIT Press, 1977. Revised edition.

[14] Alan Liu charts this history and its implications for algorithmically-generated design in chapter 6, “Information is Style,” of The Laws of Cool: Knowledge Work and the Culture of Information (Chicago: University of Chicago Press, 2004).

[15] See the appendix to this essay, “What is a Literary Game?”

[16] Interactive fiction, or IF, is of course the exception here; the first video games were, in fact, interactive fictions. The key theoretical and reference text here is Twisty Little Passages: An Approach to Interactive Fiction by Nick Montfort (Cambridge: MIT Press, 2005).

[17] Fashionable Noise: On Digital Poetics (Berkeley: Atelos, 2003).

[18] Manovich, ibid.

[19] A list of these games can be found on Wikipedia in the article “Procedural generation.”

[20] New York: Oxford University Press, 2000. Second edition.

[21] Pound, of course, actually dissuaded the reading of “symbolism” in his work, and sought to elevate the particulars of his “poem containing history” to the status of myth through his “fugal” method of repeating motifs.

[22] I often point to the “u” chapter of Christian Bök’s Eunoia (Toronto: Coach House Books, 2001) as the prime example in print of an algorithm operating on a closed set of texts – the few English language words containing only the vowel “u” – and creating, almost accidentally, a hilariously overblown, vivid pornographic vignette. (Toronto: Coach House Books, 2003).

[23] Selected Poems (New York: Penguin, 1985).

[24] Described in S/Z: An Essay (New York: Hill and Wang, 1975). Translated by Richard Miller.

[25] Robbe-Grillet, Alain. For a New Novel (Chicago: Northwestern University Press, 1992).

[26] Electronic Book Review, November 5, 2005.

About the Author:

Brian Kim Stefans is the author of three books of poetry, including Free Space Comix (Roof, 1998). His most recent book is Fashionable Noise: On Digital Poetics, a collection of poetic essays. A new essay will appear in New Media Poetics: Histories, Institutions, and Audiences, published by MIT Press in 2004. He edits the website Arras, devoted to new media poetry.