by Andrew Piper
Counting has become a part of reading like never before. In a world of computation, writing is fundamentally numerical, just as reading has become a punctuated engagement with algorithmic process. When we read a digital text we are not reading a static object. We are reading one that has been generated through a set of procedural conditions that depend on our interaction with them. Digital texts are never just there. They are called forth through computation and interaction, whether by a human or a machine. This is what makes them dynamic, not static objects. It is this feature that marks the single strongest dividing line between the nature of books and that of their electronic counterparts.2
For many today, mathematical computation has become the negative pole of humanistic knowledge.3 The “objectivity” produced by computation is not only deterministic, it no longer even needs us. Humanistic inquiry, by contrast, one premised above all on reading books, concerns the production of subjectivity, what it means to be a person in the world. The logical (and technological) determinacies of computation, so the argument goes, stand in stark opposition to the linguistic indeterminacies of reading books.
And yet the history of literature and its attachment to number, sequence, procedure, process, and form tells us a different story. Reading literature is indebted in profound ways to the world of the numerical, just as the history of mathematics is far from the deterministic horror story many make it out to be. Symbols like ∞, π, e, −2, √, ≈ or ideas like “plane,” “circle,” or “parallel line”—are these any more precise than when Mrs. Ramsay reads, “Nor praise the deep vermilion in the rose”? They are both signs of the necessity of form, of our need to model, approximate, represent. Whether as fiction or theorem, they are means of understanding a world that at bottom always seems to elude our grasp.
Can computation be thought of as part of reading? As part of the humanities? Must the humanities stand for human only? Are there ways to think in more elementary terms about the relationship between number, procedure, and words? Can we bridge this expanding gap between the computational and the literal and instead posit their intersection as a new core of intellectual life? Will this not be a prerequisite for the literacy of the future?
Part of understanding the future of reading will entail knowledge of the history of textual computation, the way we have done things to texts and the way texts have done things for us. We require a clearer sense of the notion of process as it relates to reading. Such thoughts move us from the nighttime of reading with which I began (dreaming in books), to reading’s daytime (work and play). They return us to reading’s numerological origins.
. . .
When Augustine opened the Bible at random and found the passage that changed his life, he was engaging in a time-honored practice. Readers routinely sought meaning from randomly accessed texts, one that in Augustine’s day went by the name of sortes Virgilianae or Virgilian lots (so named because Virgil was its most popular source). The practice later came to be known more generally as “bibliomancy,” divination by means of books. Like many readers, Augustine was combining chance and reading to arrive at a fundamental truth, one that was already there, but that could not be directly accessed. It required constraints, rules, and luck. It was a game. Augustine was playing his text.
With the advent of video games, game theory—ludology—has become a bedrock of new media studies.4 In the often messianic rhetoric of reformers (beware the prophets), gaming will change our lives. “Games aren’t leading us to the downfall of human civilization,” writes Jane McGonigal in her gamer manifesto, Reality Is Broken. “They’re leading us to its reinvention.”5
This of course is nothing new. Games are some of the oldest forms of human practice, the idea of “Homo ludens” one of the recurring ways we have tried to understand our nature. “Man is only truly himself when he is at play,” writes the playwright and sometime philosopher Friedrich Schiller in his epoch-making aesthetic theory written at the close of the eighteenth century. For Schiller, as for later philosophers like Johan Huizinga, “culture” was play’s outcome, not something we created once we grew up.6
But there is also nothing new here in terms of reading. As the eminent Canadian critic Northrop Frye illustrated some time ago, the riddle is one of the oldest forms of literature.7 It is thought to be a cognate of the verb to read, which derives from the Germanic root (rādan) for council, deliberation, but also conjecture. In taking time with language, in treating it as a game to be played, we learn to read. In European languages, riddles are some of the oldest forms of literature, from the tenth-century Exeter Book, which was an important source for the epic Beowulf, to the Icelandic Hervarar Saga, which utilized a popular form of question and answer as part of its narrative.8
The riddle depends upon a sense of intrigue, of a truth that is already there but when discovered feels new, sudden, miraculous. As André Jolles has suggested, the riddle is more about process than product.9 Chance, not reason, reigns over such insights. I cannot deduce my way into the solution to the textual game, I can only stumble upon it. This truth, the one that appears by accident, is more profound, perhaps truer.
The history of how we have played with our books is a long one. In the thirteenth century, the Majorcan theologian Ramon Llull devised a system of overlapping rotatable discs that corresponded to a list of moral attributes and that drew heavily on an Arabic astrological device known as a za’irja. By rotating the discs, one could arrive at a number of fixed combinations of truths about the world drawn from sacred scripture. “The subject of this Art,” writes Llull in a companion to his ultimate system, the Ars Magna, “is the answering of all questions, assuming that one can identify them by name.”10
In the seventeenth century, the German poet Georg Harsdörffer produced a massive compendium of philosophical and mathematic games (like how to write on egg whites or calculate using letters).11 One of his diversions was a so-called thought ring (fig. 7.1), which consisted of five separate layers of phonemes that allowed users to generate all possible combinations of the German language—and of course many more (97,209,600 in all). Language and the ideas that emerged from it were understood in a combinatory sense according to Harsdörffer’s acutely baroque sensibility. Breaking down language into parts and then playing with these units was the condition of bringing new words, and presumably new ideas, into the world. Harsdörffer’s thought ring was an instrument, to be played in much the same way as the popular harpsichord was in his day. Unlike Llull’s discs that always brought you back to the same scriptural truths, Harsdörffer’s had the potential to deliver you into a world of novelty, the unthought and the unsaid, and of course the unsayable—nonsense was always a possible outcome of the “ring.” Nonsense was the flipside of novelty.
[FIGURE 7.1] The “thought ring” from Georg Philipp Harsdörffer, Deliciae Physico-Mathematicae (1651), 517. Courtesy of the Klassik Stiftung Weimar, HAAB/G4:14.
However much Harsdörffer’s thought ring was an idle pastime for the salons and drawing rooms of the well-to-do, it also drew upon the more popular genre of “counting-out” rhymes that one finds across the world’s cultures and that depend on the use of meaningless words: eeny meeny miny mo, hakara bakara, essike tessike, pico pico masa rico, and so on. At the very basis of computation, of “counting out,” is linguistic nonsense. But so too is death. Not only are these rhymes related to the process of drawing lots, they are often interchangeable with incantations recited over the dying.12 “Ena, mena, bora, mi,” begins one English rhyme that ends with the words, “Stick, stock, stone dead.” Counting out, deciding who is “it,” is the symbolic rehearsal of our own mortality.
At the opening of the twentieth century, two of the founders of the Dada movement, Francis Picabia and Tristan Tzara, sat down in front of each other one evening to write spontaneous, and nonsensical, prose poems on opposite sides of a page.13 They were arguing (once again) for chance and play as the core principles of creative writing. As Tzara asked in his manifesto, “How
can one expect to put order into the chaos that constitutes the infinite and shapeless variation of man?”14 In Tzara’s hands the page was the imaginative equivalent of the spinning wheel.
Less than twenty years later, the mathematician and key founder of the modern computer Alan Turing would reflect on how a text could become a machine. His insight was to translate writing from the two-dimensional space of the page to the one-dimensional line of tape. “Computing is normally done by writing certain symbols on paper,” writes Turing in one of the most important papers in the history of computing. “I think it will be agreed that the two-dimensional character of paper is no essential of computation. I assume then that the computation is carried out on one-dimensional paper, i.e. on a tape.”15 The combinatory arts of the wheel were united in Turing with the serial procedures of the tape, redeploying the age-old textual metaphor of the “spool” (incidentally one of Samuel Beckett’s most beloved words from Krapp’s Last Tape). At the dawn of computing the scroll returns, an ironic bookend to the end of the book.
The term we use for such computational procedures is “algorithm,” named after the ninth-century Arabic mathematician Mohammed ibn-Musa al-Khwarizmi. As David Berlinski tells us, “An algorithm is an effective procedure, a way of getting something done in a finite number of discrete steps.”16 On the one hand, algorithms can be very precise, like the Google search engine that more often than not returns exactly what you are looking for. But algorithms can also be the means of producing surprise, of something unexpected. Harsdörffer’s thought ring is one such example. Critics have recently begun to call for a more creative relationship to the use of algorithms, to expand our sense of the “query” or “search” beyond the highly restricted outcomes of Google or its competitors.17 Other than the “feeling lucky” option, there is remarkably little play when we search online.
A number of creative writers have begun experimenting with computational algorithms, what Noah Wardrip-Fruin calls “textual instruments” or Kenneth Goldsmith “uncreative writing,” in order to rediscover a sense of linguistic surprise. “The world is full of texts, more or less interesting,” writes Goldsmith. “I do not wish to add any more.”18 Instead of writing books or texts, writers now write the rules that make books or texts. They write writing machines. As a set of rules, the algorithm is imagined to be the tool that will paradoxically rescue us from the overly rule-based nature of computational writing, to reintroduce chance and play into our highly structured textual universe. In this, I can’t help but think of Goethe’s play Tasso, whose poetic hero clings at the close of the play to the worldly prince on whom he depends for his well-being. In his last moments, he remarks, “And so at the very last the helmsman clings fast to the rock on which he foundered.”19 The algorithm is the rock that rescues us from the shipwreck of computation.
There are a variety of new projects on offer whose aim is to scramble our overly formatted textual flow. In Wardrip-Fruin’s News Reader (2004), for example, an algorithm synthesizes semantically related pieces of news to create entirely new articles. By clicking on headlines drawn from RSS feeds, his instrument takes stories from other sources and gradually builds new pieces, so that reading is also a form of writing. The articles that emerge are, in a very ancient sense, woven together. The more we play, the more they entwine, the further we are from “the news” and the closer we are to “the new” (and also nonsense, which is never far away from play).
In a more literary vein, the new media artist D. Fox Harrell has created GRIOT, a computational narrative program named after West African storytellers that is designed to produce haibuns, or prose haikus. These short linguistic snapshots are generated from users’ inputs that are then run through a combinatory algorithm:
save us from desire, tender
fixed, forgotten
save us from fear, wanting
addict & moonlight
They can be quite beautiful, arranging the linguistic jumble of our lives into meaningful configurations, much like latter-day medieval books of hours or baroque sonnets. As Mrs. Ramsay said of the sonnet, so too of the algorithmic haibun: “How satisfying! How restful! All the odds and ends of the day stuck to this magnet.”20 This is how the new magnet, the hard drive, speaks, with the day’s semantic filings arranged into meaningful patterns. It is an ideal textual companion of the future, Virgilian lots for the computational age.
In Darren Wershler and Bill Kennedy’s Apostrophe Engine (2001), clicking on a line of Kennedy’s poem “Apostrophe” submits it to a web-based search engine that collects all phrases that begin with the words, “you are,” and then assembles them in sequential order. “Apostrophe” is the rhetorical term for direct address (Dear reader, hey you, O Death!), and the results produce lines like “You are a dog on a leash like a pig in a pen,” “You are not doing it right, you damn well know you are doing it wrong,” or “You are really depressed and you just want to shout ‘shut up! shut up! shut up!’”21 Every line is itself clickable so that the poem is as potentially large as the web itself. It represents the sum of what we say we are online.
Much of this work—and there is much more of it—goes back to the experiments of John Cayley, one of the forerunners of constructing textual instruments. For Cayley, whose projects depend on what he calls “transliteral morphing,” the algorithm is used less for the purposes of surprise and more to reveal deeper patterns between words and letters. The semantic nonsense that his projects like Riverisland (2007) and Overboard (2004) produce are there to help us see the more elementary relationships between the shapes and sounds of letters.22 Surface noise, intellectual depth, and “going overboard”—these are the core elements of computational reading, what it takes to see form in our massively overburdened textual lives.
Time and again projects like these are pejoratively referred to as “mere games.” Unlike cinema, photography, or books—each with its own aesthetic highs to its more entertainment-laden lows—digital games have not made their way into the expression industry in any canonical sense, this despite the fact that they can constitute huge textual units (one recent game, Dragon Age, consists of over 1.2 million words). However brilliant early text-based adventure games were, we still have no Proust of Zork. I do not mean this ironically. Imagine if you could play In Search of Lost Time. It would be amazing.
When it comes to reading today, we seem to be distrustful of play and the constraints it requires. We tend to privilege virtuosity over failure. For the Dadaist Francis Picabia, whose favorite journal was subtitled Le Raté, or The Failure, the pending arrival of our success-obsessed age was already palpable. He could sense how fearful we were becoming of non-sense, the nonreal, the absence of purpose (or in the words of a fellow traveler: beware of the “literary quacks with a mania for improvement”). In response, a variety of new media artists are taking as their starting point the dysfunctionality of computation, the way electronic technologies, unlike books, are prone to break down.23 It is no surprise that the “outtake,” the genre of deleted scenes, has emerged as an essential new form today. It is success culture’s obscene matter.
And yet playing with texts has always been at the heart of reading. It is a crucial way that we interact with strings of words, the way we make sense of the world’s polyphony. Playing with texts is how we encounter the limits of sense, meaning’s blurry edges. It has been there from the start of reading, and it is at the core of some of the greatest works of the modern age: Faust, Ulysses, The Making of Americans, Gravity’s Rainbow, but also Dada, Oulipo, Fluxus, and the Ramones. The algorithm is the riddle’s heir.
When we read a book we are reading someone else’s modeling of an artificial world, whether real or imagined. We are reading virtuosity, the skill with which someone is able to create ex nihilo. When we read a game, we are reading someone else’s interpretation of how we model the world. We are reading an interpretation of human creativity. In playing with textual games, we learn about our own thought processes, about the nature of
expression, about the formal structures of language-based reason. Through the algorithm we learn about learning, the success of failure.
. . .
That was play, this is work.
I want to conclude this chapter and my book, but hopefully not the book, with a discussion of a new set of computational reading practices. They date back in many ways to the 1970s, a computational high-water mark for many disciplines. But they are only now gradually beginning to alter the landscape of professional reading, what used to be known as philology and now goes by the name of digital humanities.24 It is not without its detractors. But then again, when has this not been the case when there have been changes to how we read?
First, some examples. FeatureLens, which was developed by the Human-Computer Interaction Lab at the University of Maryland, is a program that allows you to view meaningful semantic patterns within large structures of texts (fig. 7.2). For Tanya Clement, who undertook an analysis of Gertrude Stein’s famously difficult and repetitive novel The Making of Americans (1925), the interface revealed a range of structural patterns so far unnoticed by readers.25 Using FeatureLens, Clement was able to see the symmetrical repetition of certain phrases, the way “any such a thing” appears exactly ten times in each of the chapters where it appears. She also noticed that the repetition of the single longest string of words (495 in all) occurs only once at the exact midpoint of the novel, suggesting that the novel has a cyclical rather than random structure. And finally, her research showed that the repetitions of the last chapter are not unique to that chapter, as is the case with the other chapters, but appear scattered throughout the novel. Far from articulating a principle of randomness, this last detail suggests how the final chapter assumes the structure of a refrain (called a “rhapsody” by Stein). Rather than marking an end point, it is the moment when the novel circles back on itself. There is a virtuosity to Stein’s play that takes place at a remarkable level of scale and that only becomes visible when subjected to such quantitative inquiry.
