Chapter 3
Computation as Fragmentation

Pre-modern concepts of computation typically relied on an equivalence of macrocosm and microcosm and so conceived of algorithms as demiurgic creation, metaphysically. The game by contast is a central model of 20th century computation, both in the arts and in technology. The main ontological difference between theosophic and game computing is that games can do without a reference to higher powers—which is why religions declared many games, dice and card games for example, sinful—and may impose arbitrary restrictions that do not logically follow from a higher natural order. A game, in other words, can be its own autonomous, self-contained world.

Any game is a process based on rules, a formal source code that can be expressed in logical language. There exist multiple models of computing as games. Best known might be Conway’s Game of Life (figure 1 ) with its particular implementation of so-called cellular automata. Cellular automata consist of simple elements in a matrix that have a finite amount of states. Most typical is a binary state of either zero or one or, in a graphical matrix, black and white. Obeying very simple transformation rules, the automata alter their states in dependency of the states of neighboring cells. Cellular automata have existed as a computational model since the 1940s. Conway’s Game of Life, first published by the British mathematician John Horton Conway in 1970, is based only on the following rules: “If a black cell has 2 or 3 black neighbors, it stays black. If a white cell has 3 black neighbors, it becomes black. In all other cases, the cell becomes white.” 1 Remarkably, this simple game allows Turing-complete computation. That means, any calculation can be made, and computer programs can be written on the basis of cellular automata. There also exist computer programming languages like Logo which are based on a game logic. Designed for children, Logo allows arbitrary computations with the help of the screen graphic of a turtle. Users have to program the turtle to make certain movements on the two-dimensional screen and this way perform calculations. In the 17th century, Georg Philipp Harsdörffer conceived of his linguistic and poetic combinatorics as games, worked under the name of “The Playing One” in a language research society and described most of his combinatory devices in dialogical fiction books that instructed poetry as a social game. 2 But the playful nature of these poetic and artistic language computations made them subject to parody and ridicule, later oblivion, from 1700 onwards until roughly 1900.

Gulliver’s Travels

In 1705, German universal scientist Daniel Georg Morhof published a treatise De Arguta Dictione, which for the first time combined the Jesuit acumen rhetoric (see p. 38 ) of witty points with Lullism in a rhetoric based on combinatory principles. 3 Next to Harsdörffer’s mid-17th century works of scientific and poetic instruction, Morhof’s book is one of the few systematic computational poetics, and theories of language and composition, based on algorithms. It stands, next to Quirinus Kuhlmann’s poetry, as another culmination and end point of the 17th century boom of Lullist thought. In many respects, Lullism was a continuation of the medieval scholastic thinking in Aristotelian categories. Through Newton and scientific empiricism, and the shift towards individual genius and against poetic rules in literature, Lullism became outmoded. As an obsolete and seemingly bizarre scientific and poetic method it ended up being subject to rationalist parody.

In Swift’s Gulliver’s Travels from 1726, the first-person-narrator visits the flying academic island of Lagado and witnesses the mechanics of a combinatorial machine (figure 2 ):

The first professor I saw was in a very large room, with forty pupils about him. After salutation, observing me to look earnestly upon a frame, which took up the greatest part of both the length and breadth of the room, he said perhaps I might wonder to see him employed in a project for improving speculative knowledge by practical and mechanical operations. But the world would soon be sensible of its usefulness, and he flattered himself that a more noble exalted thought never sprang in any other man’s head. Everyone knew how laborious the usual method is of attaining to arts and sciences; whereas by his contrivance the most ignorant person at a reasonable charge, and with a little bodily labor, may write books in philosophy, poetry, politics, law, mathematics, and theology, without the least assistance from genius or study. He then led me to the frame, about the sides whereof all his pupils stood in ranks. It was twenty feet square, placed in the middle of the room. The superficies was composed of several bits of wood, about the bigness of a die, but some larger than others. They were all linked together by slender wires. These bits of wood were covered on every square with paper pasted on them, and on these papers were written all the words of their language, in their several moods, tenses, and declensions, but without any order. The professor then desired me to observe, for he was going to set his engine at work. 4

A shift in culture manifests in the fact that Swift’s description of the text writing machine could work as a parody at all. In 1674, Quirinus Kuhlmann had envisioned, without a bit of irony, an Ars magna librum scribendi, a Lullist ars of writing books, which would mechanically generate all existing and all possible books. 6 Unlike Swift whose fictitious machine computes foreign, unreadable letters, Kuhlmann thought, just as Harsdörffer, of human language as something inherently computable. therefore suffices as a potentiality and thought experiment on language and writing, and needs either an actual machine, nor its output to make its point.

The Library of Babel

The speculative premise of Jorge Luis Borges’ short story The Library of Babel from 1941 is very similar to Kuhlmanns ars magna librum scibendi. 7 It envisions a nearly total library whose books are generated from a combinatorial mechanism and arranged into a virtually infinite array of hexagons. Yet the story is told from the point of view of an inmate of the library. He has no access to its source code, but tells how this source code, or generative principle, became reverse-engineered. In the first sentence of the story, he refers to the library as “The universe (which others call the Library),” expressing a personal opinion that echoes cosmological Lullism. Like in other stories of Borges, the subjective account is a simulatenous device of uncertainty and irony. Fictitious editorial footnotes and remarks make the text a product of fake philology. Because of the subjective narrative perspective, all information about the library has to be taken with a grain of salt. The first-person narrator appears to be a librarian since he speaks of “the hexagons under my administration.” But his reliability as an information source is no better than that of an inmate of Plato’s cave. So it is merely his theory that “the Library includes all verbal structures, all variations permitted by the twenty-five orthographical symbols.” In fact, the opposite can be concluded from his statement. He also says that “each book is of four hundred and ten pages; each page, of forty lines, each line, of some eighty letters which are black in color,” so there are 80 * 40 * 410 = 1344000characters per book, consequently 25 ^1344000books in total, then, from other data he gives, 32 books *5 shelves *5 walls = 575 books per hexagon, and thus hexagons in the entire library.

But it is just the speculation of the first person narrator that the library can be explained by “combinative analysis,” as a computed whole. Twelve years before the story appeared, Gödel had found the logical paradox that formal systems cannot fully describe themselves, or, if they are consistent, cannot prove their own consistency. In the Library of Babel, the book that describes the computation and hence source code of the library, is only one arbitrary book of 25 ^1344000at least one of which contains a refutation of this theory. The cataloguing systems are part of the dilemma rather than its solution. There exist “thousands and thousands of false catalogues,” and a “catalogue of catalogues.” Through the letter combinatorics, differentiation between data (books) and metadata (catalogues) becomes arbitrary. The library ends up being a self-referential linguistic universe in which subject can no longer be told from object, and it is not clear what represents what. While the library may contain all knowledge—as envisioned by Kuhlmann earlier—it is incapable of classifying it. In computer science terms, the namespace of its classifications is either flat or infinite. And while Kuhlmann believes that the Ars magna librum scribendi expands knowledge and wisdom, the Library of Babel marks a melancholy and sceptical view of the same intellectual experiment: Through its totality, the library is contingent. Everything that can be thought has been thought in it already. At the same time, this melancholy subverts itself. After all, there is a story, written by Jorge Luis Borges, called The Library of Babel. This story is one clear-cut and solid block of reference and metadata. The Library of Babel might refute itself in any possible sense, but it does not refute itself being the Library of Babel. According to its inmate’s speculation, it is objective, mechanical, without history, existing “ ab aeterno.” Yet the library functions only through its human inmates who read the books. Without them, it would be just storage of blotted paper. The books are meaningless without human interpretation. While all human acts might be anticipated in the book, those acts don’t exist without the acts of reading. The melancholic contingency of writing, it follows, is the motor of its interpretations. Only the interpretations, i.e. the cultural appropriation of what only appears to be a natural, not human-made artefact, make the library what it is. Its code is cultural even if no culture, subjectivity and interpretation were involved in its initial creation.

Finally, the account of the first person narrator has to be seen as a product of the library. So it implies its own refutation—a refutation that such a library does exist at all. In the end, as the narrative points out, even the combinatorics of the library itself is subjective, a phantasm of its inmates. On the one hand, the systems of order, structure and control inherent in the library are non- semantic. On the other hand they get continuously destroyed through semantics being read into them—theological semantics, for example since there are religious sects in the library holding certain beliefs about its sense and inner workings.

Borges’ story renders combinatorics a purely speculative figure of thought. It is a speculation about a speculation: a speculation created around the speculative, subjective account of the first-person narrator. This self-destructive, paradoxical moment sets the story apart from a plainer, non-ironic speculation like Kuhlmann’s. For Kuhlmann, there is still an identity of art, science, technology, philosophy, religion and cosmology. This leads to an integralist model of computation as totality. Computation does not yet imply, as in Borges, the negative of fragmentation and disintegration. Swift’s fiction of the Grand Academy of Lagado stands precisely between Kuhlmann and Borges because it cuts into the totality of Lullist epistemology and, with its rationalist agenda, separates science from fancy, and observation from speculation. As a result, Lullist language computation is no longer part of serious science, but ends up in the realm of the obscure, occult, para-philosophical thought experiments, and from there in the realm of fiction and the arts. Only as arcane and speculative knowledge, is Lullism able to enter Borges’ fantastic fiction.

Romanticist combinatorics

After Swift, combinatorial encyclopedism survives only in the niches of literature and speculative poetic science. In his 1798/99 encyclopedic fragment Das Allgemeine Brouillon, 8 German romanticist poet and essayist Novalis (Friedrich von Hardenberg) sketches a new comprehensive integration of all science and knowledge on the grounds of an—itself sketchy and fragmentary—“calculus.” It is a new attempt of a Lullist encyclopedia after Alsted, a speculative, essayistic encyclopedia which consists of fragmentary entries on knowledge in the light of a transcendental poetic philosophy. Its systematics takes inspiration from Leibniz’ mathesis universalis, the project of a thorough mathematical language and description of thought. However, Novalis’ “sketches” do not technically pursue the concept, but remain an idealist experiment. They shift combinatorics and computation from concrete, formal-technical manipulation of symbols—as before in 17th century proetic poetry—to a purely intellectual-reflexive, rather vague figure of thought: a meta-computational reflection of philosophical computability.

Computation as a romantic figure of reflection exists also in the Livre, the last project of French symbolist poet Stéphane Mallarmé. 9 The book, which was never finished, should consist of ten volumes which could be shuffled at will. Like 17th century Lullism, and like the proto- Kabbalah of the Sefer Yetzirah, Mallarmé calculates the permutations of the Livre, as 3628800. The volume which did get published and is best known today is a visual-typographic poem, Un coup de dès (A Throw of the Dice). 10 It juxtaposes, so-to-speak, the mathematical permutation of the Livre with the random computation of a die that gets superimposed on its protagonist. A sailor, the poet’s alter ego as it seems, navigates a ship in a storm. Having forgotten to make nautical calculations, he still refuses to throw the dice and give up to fate: “ Un coup de dès jamais n’abolira le hasard,” a throw of the dice will never abolish chance, is a key sentence of the poem. The last page imitates, by its typographic arrangement of words on the page, a starry sky showing the polar star with the little bear, culminating in the line “ UNE CONSTELLATION,” “A CONSTELLATION.” It is, just as the mention of a book on combinatorial analysis in Borges’ Library of Babel, a point where the text turns, recursively, into an index (or meta-data) of itself. After all, “constellation” references both the spatial arrangement of elements on the page and literally means the arrangement of stars in the sky.

With its double meaning, “constellation” rehashes the correspondence of macrocosm and microcosm known from Pythagorean, Neoplatonist and Kabbalist thought. It no longer does so in the context of a scientific worldview, but on the contrary, in self-chosen resistance to the modern scientific paradigm. Its formal experiment of integrating poetry, visual arts and musical composition into a Wagnerian total artwork is close to pre-modern art and thought such as Quirinus Kuhlmann’s integration of combinatorics, cosmology, metaphysics and letter permutations on rotary dials.

Concrete poetry

Mallarmé’s poem inspired and instigated a whole experimental literary genre, the “constellations” of concrete poetry. Developed in the 1950s, they likewise stood for spatial arrangements of letters on pages, sometimes in conjunction with permutational poetic forms. Yet the philosophy and metaphysics of concrete poetry was quite contrary to Mallarmé. Its name was derived from Bauhaus artist and designer Max Bill and his coinage of “concrete art” in 1936. In the spirit of high modernism and functionalist architecture, most of concrete poetry sought to systematically reduce, rationalize and functionalize literature. A constellation of concrete poet Eugen Gomringer from 1969 reads:

no error in the system
no reror in the system
no rreor in the system
no rroer in the system
no rrore in the system
no rror ein the system
no rror ien the system
no rror ine the system
no rror in ethe system
no rror in tehe system
no rror in thee system
no rror in the esystem
no rror in the seystem
no rror in the syestem
no rror in the sysetem
no rror in the systeem
no rror in the systeem
no rror in the systeme
eno rror in the system
neo rror in the system
noe rror in the system
no error in the system 11

Max Bense and “information aesthetics”

The poetics of concrete poetry and its constellations was first written down by Gomringer himself in his 1954 essay From Line to Constellation: 12

Our languages are on the road to formal simplification, abbreviated, restricted forms of language are emerging. The content of a sentence is often conveyed in a single word. Longer statements are often represented by small groups of letters. Moreover, there is a tendency among languages for the many to be replaced by a few which are generally valid. Does this restricted and simplified use of language and writing mean the end of poetry? Certainly not. Restriction in the best sense—concentration and simplification—is the very essence of poetry. [. . . ] The aim of the new poetry is to give poetry an organic function in society again, and in doing so to restate the position of poet in society. Bearing in mind, then, the simplification both of language and its written form, it is only possible to speak of an organic function for poetry in terms of the given linguistic situation. So the new poem is simple and can be perceived visually as a whole as well as in its parts. It becomes an object to be both seen and used: an object containing thought but made concrete through play-activity (denkgegenstanddenkspiel), its concern is with brevity and conciseness. It is memorable and imprints itself upon the mind as a picture. Its objective element of play is useful to modern man, whom the poet helps through his special gift for this kind of play-activity. Being an expert both in language and the rules of the game, the poet invents new formulations. By its exemplary use of the rules of the game the new poem can have an effect on ordinary language.

Bense wasn’t interested in Shannon’s theory as a pragmatic engineering perspective, but as a philosophical tool with which to reinvent aesthetics and poetics, stripping it bare, as he hoped, from any concept of meaning. His combination of semiotics with technical information theory was an artifice for turning the humanities and art criticism upside down. It resulted in a purist modernism which was ideological just in its ostensible refusal of ideology, and metaphysical in its radical refusal of metaphysics. After all, Bense’s conflation of information theory with semiotics stripped computational forms of art and writing from their historical, intellectual implications. Mallarmé and other late romanticist poets like the German Stefan George had been the main sources of inspiration for the formal innovation of concrete poetry. Bense’s theory however sought to look only at the form and cut all philosophical-metaphysical ties.

Pythagorean and occult metaphysics of macrocosmic and microcosmic equivalence had been pushed outside rational science since the 18th century. But not only romanticism remained heavily indebted to this tradition, but also, for example, the avant-garde wordplay poetry of the Russian Futurist Velimir Khlebnikov. On the contrary, Bense and the post-war formalist avant-garde wanted to do away with metaphysics entirely and refound aesthetics on the grounds of modern science and technology. This move obscured the romanticist roots not only of concrete poetry, 15 but also of Bense’s philosophy itself. Like Novalis in his encyclopedic project, Bense took heavy inspiration from Leibniz’ mathesis universalis, the post- Lullist attempt of unifying all human knowledge on mathematical grounds. Bense’s move to clear the project from romanticist connotations and indeed any kind of overt metaphysics reflects, obviously, his experience with totalitarian politics of The Third Reich and Stalinist post-war East-Germany.

Bense’s idea of formal programmation as a model of an ‘objective’ art and philosophy ran in parallel to French linguistic structuralism of the 1950s and 1960s. It anticipated the so-called “linguistic turn” in the humanities of the late 1960s. The linguistic turn did not only happen in theory, but also in art in what critic Lucy Lippard described as the “dematerialization of art” around 1970. Pop art and Fluxus turned into concept art that, in the works of Joseph Kosuth, Art and Language, Lawrence Weiner and On Kawara for example, consisted only of written instructions. The first concept art show, curated by critic Jack Burnham in New York in 1970, accordingly had the title Software. It juxtaposed concept art works with experimental computer software development projects such as Ted Nelson’s first prototype of a hypertext system. Burnham, a close collaborator of artist Hans Haacke before the latter turned from rigorous formalism to politically activist art, took his inspiration from cybernetics and general systems theory. The theoretical base was similar, but not identical to that of the continental European discourse of Bense and literary semiotics and structuralism. 16 Burnham’s ideas, too, imposed a rigorous scientific formalism onto art. They were more concerned with the visual arts than poetry though, and with the detachment of art from material objects more than with computational algorithmics. Software was, as Edward A. Shanken puts it in an essay on Burnham’s exhibition, a “metaphorical premise,” or device, that emphasized software as being different from hardware, not software as executable instruction code.

The underground art and activist journal Radical Software appeared under that metaphorical premise as well. Founded in the same year as the Software exhibition, it propagated an “Alternate Television Movement,” juxtaposing aesthetic reflection with political debates about free media and publicly accessible radio spectrum, much like the contemporary free wireless network movement. In addition, it contained hands-on technical instruction for building and manipulating video equipment. Otherwise, the journal conceived of “software” purely as dematerialized art, and did not cover computer programming.

Modelled after the Bauhaus, Bense’s older program turned poetry, concrete poetry in particular, into language design and art into visual design. These efforts were systematically pursued at the Hochschule für Gestaltung (School for Design) in Ulm, where Bense taught next to his regular professorship in nearby Stuttgart. Hochschule für Gestaltung was founded as a “second Bauhaus.” Its director was Max Bill and faculty included former Bauhaus professors Johannes Itten and Josef Albers. Later in the 1990s, the ZKM media arts center nearby in Karlsruhe was established as yet another attempt of a new Bauhaus, this time as a “Bauhaus of Second Modernism” according to its initiators.

Situationism, Surrealism and psychogeography

In a counter-reaction to Bill, Bense and their functionalist thinking, Danish painter Asger Jorn, previously a member of a Surrealist splinter group in 1940s Paris, founded a International Movement for an Imaginist Bauhaus in 1956. In 1958, it became part of the Situationist International. In January 1959, the German section of the S.I. continued the opposition to Bense and Bill with a prankish attack on Bense in Munich: A public lecture of Bense was announced and once the audience had gathered, a tape recorder was switched on and the voice on the tape declared that Bense was unable to come and would instead give his talk in “cybernetic form.” The talk was a deliberately nonsensical cut-up of German, Latin and French phrases with garbled quotations from Marx and Hegel. Yet the audience stayed through the lecture and applauded in the end. In the prank, the Situationists took Bense’s cybernetic poetics and turned it as a tactical device against himself. The stunt displayed that his attempt to do away with semantics has its blind spot precisely in the semantics of his own statements that negated semantics. Secondly, it debunked the concept of technologically produced information as objective which the Situationists countered with a post-romantic and post- surrealist concept of aesthetic subjectivity. Like Jorn, the German Situationists attacked Bense, according to the S.I.’s report of its 3rd conference, for his “perfect continuation of constructivism.” In his essay Open Creation and its Enemies, published in 1960 in the fifth issue of the journal of the Situationist International, 17 Jorn likened Bense to the concrete poetry-like French Lettrist poets, declaring him “the German equivalent of this anecdote of systematic, paradialectic, and deadly boring ‘Lettrist thought’.”

The flip-side of this critique was hostility of the Situationists to both artistic experimentation with new technology and philosophical reflection on computation. This hostility manifested itself particularly in the repeated Situationist attacks on communication theorist Abraham Moles (see p. 151 ). With their polemics against formalism and for “imagination,” the S.I. clearly continued the ideas of the French Surrealists who in turn were heavily indebted to romanticism. In his 1924 Manifesto of Surrealism, André Breton wrote that “We are still living under the reign of logic: this, of course, is what I have been driving at. But in this day and age logical methods are applicable only to solving problems of secondary interest.” 18 His surrealism expresses laconic indifference to new technology: “Radios? Fine. Syphilis? If you like. Photography? I don’t see any reason why not. The cinema? Three cheers for darkened rooms. War? Gave us a good laugh. The telephone? Hello.” 19 Along the lines of this technological scepticism, Surrealist “automatic writing” for example was not computational, but a psychic automatism that took the unconscious as its source code, not a calculus. It was still a foreign idea to Surrealism that computational formalisms could themselves be highly subjective and culturally coded, as the Pythagorean and Kabbalist tradition and the “semantics” of, for example, Llull’s “alphabetum” suggest. The Situationist concept of “psychogeography” had its roots in the aimless Surrealist drifts through Paris described in Breton’s 1928 novel Nadja and in Louis Aragon’s 1926 novel Le Paysan de Paris, and meant a purely subjective, para-scientific exploration of (chiefly) urban spaces through aimless drift. The surrealist drifts in turn were indebted to the romanticist “flâneur,” a wanderer “botanising the asphalt” as cultural theorist Walter Benjamin put it in his essay on 19th century poet Charles Baudelaire. 20

Computation and romanticist urban drift did not converge until the invention of “generative psychogeography” in the late 1990s through the Dutch artistic project http://www.socialfiction.org . Its .walk is a “psychogeographic computer,” operated by pedestrians who walk through street grids like electrons flow through the gates of computer chips. The .walk computer can execute simple program code like the following:

// Classic .walk  
Repeat  
{  
1 st street left  
2 nd street right  
2 nd street left  
}

• pataphysical; pataphysics is a “science of imaginary solutions” according to its inventor, late 19th century pre- Surrealist novelist and dramatist Alfred Jarry. It could also be called a poetic, absurdist para- and anti-science. A pataphysical appropriation of computing factually exists in the work of the Oulipo poetry workshop (see p. 145 );
• “casting spells on the OS,” reflecting magic as a forerunner of code execution;
• “social engineering,” especially in the form of artificial intelligence chat robots; which reminds of the S.I.’s tactical use of a manipulated tape recording as an analog Max Bense chat roboter. Since Alan Turing, chat robots are the fundamental to artificial intelligence research (see p. 174 )
• questioning the traditional computer; thus, by implication also reflecting on the cultural history of computing aside from its materialization in electronic hardware.

“Speculative programming” reads as an attempt to sum up all philosophies and the complete cultural-imaginative history of computation, including everything that is described in this booklet, too. Computing becomes a figure of thought and reflection not only in theory, but also in artistic practice. While the same could be said about Bense’s philosophy, the implications are contrary. Instead of acknowledging subjectivity and imagination put into computations, computation becomes a token for a culture of scientific and engineering objectivity. Where Bense models art, criticism and aesthetics after computing, superimposing the latter on the former, speculative programming does the opposite. It models computation after the arts and and speculative imagination.

Markov chains

In Bense’s “information aesthetics,” art criticism turns, literally, into computation of data. Interpretation of meaning is substituted with formal analysis according to objectively quantifiable parameters, for example through word statistical analyses of writing. In Italo Calvino’s 1979 novel “If on a Winter’s Night a Traveller,” the first person narrator, a writer, encounters a woman who refuses to read his novels, but instead feeds them as data into a statistical program:

She explained to me that a suitably programmed computer can read a novel in a few minutes and record the list of all the words contained in the text, in order of frequency. “That way I can have an already completed reading at hand,” Lotaria says, “with an incalculable saving of time. What is the reading of a text, in fact, except the recording of certain thematic recurrences, certain insistences of forms and meanings?. . . ” 21

“To construct [order-1 letter-level text] for example, one opens a book at random and selects a letter at random on the page. This letter is recorded. The book is then opened to another page and one reads until this letter is encountered. The succeeding letter is then recorded. Turning to another page this second letter is searched for and the succeeding letter recorded, etc. A similar process was used for [order-1 and order-2 letter-level text, and order-0 and order-1 word-level text]. It would be interesting if further approximations could be constructed, but the labor involved becomes enormous at the next stage.” 23

Independently from Bense, Lutz and Oulipo, critic and Joyce expert Hugh Kenner wrote, in collaboration with programmer Joseph O’Rourke, a text recombination program based on Markov chains. Dubbed Travesty, its source code was published in a 1984 issue of the popular computer magazine BYTE. For the algorithm, Kenner credited the “long-ago idea from the Father of Information Theory, Claude Shannon.” The code was adapted in 1990 by Larry Wall, creator of the Perl programming language, and published as a programming example in the first edition of the book Programming Perl.—The second edition of the same book featured examples of “ Perl poetry” 25 (see p. 153 )—With poet Charles O. Hartman, Kenner co-authored Sentences, a volume of poems generated with the help of travesty. 26 In 1994, experimental poet and former Fluxus member Jackson MacLow generated a number of his 42 Merzgedichte in memoriam Kurt Schwitters with the Kenner’s travesty program and, one year later, Austrian contemporary composer Karlheinz Essl reassembled a Bach violin sonata through Markov chain computation, calling it “Bach sausage.” Likewise, the DOS-based poetic language manipulation toolkit POE designed by Austrian experimental poets Ferdinand Schmatz and Franz Josef Czernin (see p. 184 ) includes a Markov chain function. Around the same time, numerous popular Markov chain-based text manipulation programs were written, such as Dissociated Press, a standard function of the GNU Emacs text editor, TextMangler and Deconstructor for MacOS, dadadodo by Jamie Zawinsky, the former project leader of the Mozilla web browser, or Mark V. Chaney for DOS.

Since Shannon’s Mathematical Theory of Communication, Markov chains have arguably become the third most popular algorithm for computational text generation, next to permutation and recursion. Markov chains are a stricly analytical, not synthetic method, being agnostic to the data they process. They work on arbitrary input, contrary to older syntheticly combinatory methods such as Llull’s Ars and Renaissance proteus poems which always processed fixed, pre-inscribed words. While Llull’s ars abstracts data from algorithms through its separation of the tabula from the figurae, the figurae are designed to process only the particular, immutable elements of the static tabula. Even Swift’s and Borges’ dystopias of language computing, with their ontological randomness and contingency, still rely on a fixed data set, the alphabet.

Tristan Tzara and cut-ups

One could believe that poetic computations of mobile data sets could not be imagined before modern computers were invented. However, the first modern art work based on a computational process and arbitrary input data dates back to 1923. Tristan Tzara’s advised To Make a Dadaist Poem as follows:

Take a newspaper.
Take some scissors.
Choose from this paper an article the length you want to make your poem.
Cut out the article.
Next carefully cut out each of the words that make up this article and put them all in a bag.
Shake gently.
Next take out each cutting one after the other.
Copy conscientiously in the order in which they left the bag.
The poem will resemble you.
And there you are—an infinitely original author of charming sensibility, even though unappreciated by the vulgar herd. 27

While the Dadaist poetic algorithm is the same as the word permutations in the Sefer Yetzirah and in proteic poetry since Optatianus and Scaliger, its imagination and ideology is reversed: There is chaos instead of order, fragmentation instead of totality. It also lacks, on the first glance at least, any metaphysics and cosmology of macro- and microcosm. If one juxtaposes the Dadaist poem to Mallarmé’s Livre, then it reads as the latter’s radical other; it uses a similar algorithm, by implication also a similarly free form typography, but contradicts both the romanticist aesthetics of the genius and the representation of a Pythagorean macrocosmic order in the microcosm of the artwork.

Still, the ostentative nihilism of Tzara’s random computation creates another metaphysics. Gysin’s and Burroughs’ “Cut-ups” for example are a straight-forward adaption of Tzara’s method. Burroughs’ essay The Cut-Up Method of Brion Gysin acknowledges this in its first sentence:

At a surrealist rally in the 1920s Tristan Tzara the man from nowhere proposed to create a poem on the spot by pulling words out of a hat. A riot ensued wrecked the theater. [. . . ] In the summer of 1959 Brion Gysin painter and writer cut newspaper articles into sections and rearranged the sections at random. 28

John Cage’s indeterminism

The same can be observed in the adaption of the Dadaist random montage music in the “indeterminist” music of John Cage his contemporaries and students from Earle Brown, Morton Feldman, Christian Wolff to the Fluxus movement. In John Cage’s piece Atlas Eclipticalis, an astronomical map serves as a superficially random score, and establishes, once again, a macrocosmic- microcosmic correspondence which justifies a randomness of art through a higher-order randomness of nature. Cage also uses the Chinese I Ching, the Book of Changes, as a compositional algorithm for a number of his works, among them Music of Changes and Roaratorio. 29 For these purposes, he later used a computer software adaption of the called IC, written in C for DOS by Andrew Culver. 30 Cage replaces the cosmological-mathematical order of Pythagorean Western music with a metaphysical anarchism. This anarchism nominally draws from Eastern philosophy, but in fact relied only on superficial studies—through the attendance of a few evening lectures—of the non-traditional, highly Americanized Zen of Daisetz T. Suzuki.

Just like Tzara’s Dadaist poem manifests an anti- romanticism that nevertheless shares many formal traits with late- romanticist literary experimentation, Cage’s approach to musical composition could be seen as anti-romantic and ultimately anti-Western while owing more to Western tradition than it acknowledged. Cage’s anti-tradition is most pronounced in Credo in Us from 1942, a piece that pokes fun at late romanticist symphony music by playing it randomly from gramophone records. It could be called the first musical piece using “scratching” and “Plunderphonics.” Still, Cagean indeterminism remained reciprocal to the mathematical determinism of Western music. It only replaced the Pythagorean formula of symmetry through a formula of chance. It seems only logical that Cage formalized his composition methods and implemented them into computer software tools. He used 24 custom-written computer programs for his composition. 31 Cage wrote computational poetry, too. With his method of “reading through,” 32 he compressed literary works like Joyce’s Finnegans Wake to a fraction of their original size. The algorithm he used were mesostichs, the retrieval of words in a text whose middle letters match the letters of a name; electronic poet Jim Rosenberg wrote him a custom computer program Mesolist for this purpose. 33

It seems paradoxical that Cage sought to create indeterminacy through algorithms. While this contradiction superficially resembles the link of algorithms and the imaginary in psychogeographical computing, it in fact is more naive since it relies on a dubious understanding of “chance.” Algorithms can be used for chance operations as Tzara’s Dadaist poem shows, and—with much older origins probably in ancient India—the dice as a random computing device. But such random operations create stochastic chance, not philosophical- ontological chance. Throwing a die is a stochastic chance operation with the possible outcome of 1, 2, 3, 4, 5, 6. Since these results are foreseeable as the set of potential results, they represent not an ontological, but a deterministic chance. Mallarmé describes it precisely in his sentence: “throwing the dice never abolishes chance.” Ontological chance, and therefore true indeterminacy, would occur if the die would crack, vanish, or show the number seven. With pieces like Atlas Eclipticalis, Variations I/II and all I Ching-based compositions, it seems as if Cage misreads stochastic chance for ontological chance. The mere fact that a musical piece sounds “like John Cage” or “Cagean” because of its atonality and lack of development for example, disproves Cage’s claim of “indeterminacy.” If his music were really, ontologically indeterminate, it should—for example—be able to sound like a Britney Spears song, too.

This example illustrates a general limit of computation and software. There is, first of all, no way of turning them into anti-formal, anti-determinist technology, either through stochastic chance, disruption of semantics (as in the cut-ups) or a higher-order complexity of programming (as in artificial intelligence, see p. 174 ). Secondly, the formalisms are always meaningful. In the case of Cage, they seem to subvert the intended anarchist ontology of chance—as it is grounded in 20th century aesthetics, philosophy and religion—into a stochastic determinism. This unveils the blind spots of Cage’s compositional approach and the reason why an “indeterministic” chance composition of Cage can sound strikingly similar to totally deterministic serial compositions of Stockhausen or Boulez. If one compares Cage and Stockhausen and their seemingly contrarian understanding of musical composition, one can practically observe how deterministic over-complexity turns into chaos and conversely, chaos ends up as a formalism and cliché.

Tzara’s poem struggles with this problem only to a lesser degree. As a parodistic and cynical device, it does not have to match up to an ontological indeterminism. The relative lack of cynicism is what distinguishes Cagean aesthetics from Dadaist provocation. Neither do Burroughs’ cut-ups run into the trap of deterministic chaos because they are tools for reaching another, magical, hallucinatory and ecstatic order, and use language to overcome it.

In addition to permutational shuffling, the cut-ups frequently employ recursion, the processing of something through itself, as their poetic method. In Cut-Ups Self-Explained, the last paragraph of Burroughs’ The Cut-Up Method of Brion Gysin applies the method—or algorithm—to its own instruction:

ALL WRITING IS IN FACT CUT-UPS OF GAMES AND ECONOMIC BE HAVIOR OVERHEARD? WHAT ELSE? ASSUME THAT THE WORST HAS HAPPENED EXPLICIT AND SUBJECT TO sTRATEGY lS AT SOME POINT CLASSICAL PROSE. CUTTING AND REARRANGING FACTOR YOUR OPPONENT WILL GAIN INTRODUCES A NEW DIMENSION YOUR STRATEGY. [. . . ] 34

Italo Calvino and machine-generated literature

In his 1967 lecture Cybernetics and Ghosts, 35 Italian novelist Italo Calvino concludes that computer-generated poetry tends to be “classicist;” classicist in the understanding of Italian, French and Spanish literary history, as a poetry governed by strict normative poetics and rules of form. In spite of this restriction, he sketches a general model of language and narration as computations. Calvino speculates that early humans had a limited repertory of sounds and words and therefore needed combinatorics to expand a scarcity of vocabulary into a richness of communicative means. It follows for him that literature is a “combination game.” He, the poet, can be replaced by a “mechanical device.” In his reasoning, Calvino cites the linguistic and semiotic structuralism of ethnologist Claude Lévi-Strauss and Russian philologist Vladimir Propp. This reference reflects Calvino’s exposure to the structuralism of the Tel Quel group around Julia Kristeva and Roland Barthes whose meetings he attended in Paris.

In his 1927 book Morphology of the Folktale, Vladmir Propp concluded from a structural analysis of Russian fairy tales that their plots could be reduced to one universal formula. 36 This formula denotes both possible combinations and variations. It is, in fact, an algorithm which could be adapted as a computer program for generating arbitrary fairy tale plots. Propp’s formalist philology appears like a straightforward continuation of Lullist encyclopedism, the exhaustion of a field of knowledge through combinatorial means. Vice versa a Lullist language generator like Harsdörffer’s Denckring manifests proto-structuralist linguistics. In both Harsdörffer and Propp, synthetic totalism and analytical fragmentation are two sides of the same coin, and their algorithms can be used for either.

As a device of analysis, Propp’s morphology first of all gives insight into the formulaic, constructed nature of folk tales. It thoroughly debunked the 19th century romanticist view of the folk tale as an irregular and chaotic manifestation of popular fantastic imagination (which was superior to highbrow art). The same romanticist idea continued in surrealism, with its fondness of popular culture and trash—like, for example, the Grand Guignol gore theater shows—and to some degree in Situationism with its detournement of comic strips and martial arts movies (like René Vienet’s Can Dialectics Break Bricks?). With its discovery of a popular culture which is so formulaic that it is computable, Propp’s analysis could on the other hand be considered a forerunner of psychogeographical computing with its populist low-tech approach to programming. With their emphasis on popular imagination and popular cultural practices, both Propp and psychogeographical computing themselves owe to romanticist thinking. Italo Calvino’s Invisible Cities, a novel which describes imaginary cities and generates them through a combinatorics of fantastic attributes, manifests another intersection of computation and romantic urbanism. 37

Software as industrialization of art

Movie plot generators

In a newspaper essay Make Your Own Movie which appeared in the book Misreadings, Umberto Eco turns Propp’s combinatory plot generation and, by implication, structuralist morphologies into a parody. He proposes a number of algorithms for generating movie plots, each of which sums up the stereotypical mannerisms of the prototypical angry young film director and the filmmakers Michelangelo Antonioni, Jean-Luc Godard, Ermanno Olmi and Luchino Visconti. 38 The Antonioni algorithm works as follows:

An ^xempty ^ylot. ^zShe ^kwalks away. ⁿ

Variants Key

x:

Two, three an infinity of. An enclosure of. A maze of.

y:

Empty. As far as the eye can see. With visibility limited due to the sun’s glare. Foggy. Blocked by wire-mesh fence. Radioactive. Distorted by wide-angle lens.

z:

An island. City. Superhighway cloverleaf. McDonald’s.

[. . . ]

Authorship and subjectivity

Cornelia Sollfrank’s Net.art Generators

Perhaps the most consequent, if ironic, project of industrializing art was Andy Warhol’s “factory” art. It industrialized its production, aesthetics and even the identity of the artist: Warhol’s art was collectively and industrially produced, used industrial mass media images as its material, and factory members with silver wigs appeared as Andy Warhol for public lectures. Through its seriality, his work resembles machine-generated art and later proved to be ideal source material of artistic computations. In 1997, artist Cornelia Sollfrank developed, in collaboration with a number of programmers— Ryan Johnston, Luka Frelih, Barbara Thoens, Ralf Prehn, Richard Leopold—, a series of Net.art Generators, web-based software programs that retrieve websites according to user-entered search terms, reassembling them as digital collages. Sollfrank was not interested in an autonomous generative art, but in the political and philosophical issues the net.art generators create. She and her programmers, for example, did not anticipate that the generators would create endless variations of Andy Warhol’s flower pictures (figure 3 )—which in turn are based on a botanical photograph by American photographer Patricia Caulfield from 1962. Next to the many unauthorized variations of the Warhol flowers circulating as postcards and poster prints, the computer-modified flowers created unforeseen questions of authorship and originality. No dubious “artificial intelligence” (see p. 174 ) was necessary to create these issues. It was sufficient that the artwork was transformed from a solid entity to an automatic process. Moreover, it was designed by multiple people and working on arbitrary input data. As a result, an exhibition of Sollfrank’s net.art generators and their Warhol variations was cancelled by the organizers out of fear of being sued for copyright violation.

What Tzara’s Dadaist poem puts into ironic terms, the “charming sensibility” or subjectivity of the artist who creates a work according to an algorithm, is at the center of Sollfrank’s philosophical and legal reflections. Who exactly is the creator of a Warhol flower variation computed by the net.art generators? Caulfield as their original photographer, Warhol as their first artistic adopter, Sollfrank as the artist who created the concept of the net.art generators, the programmers who technically designed and implemented them, the users of the net.art generator, or the running program itself?

Conventional software for artists still sells the idea of the artist as an autonomous creator who works with the aid of, but isn’t replaced by, algorithms. Artistic generators like those of Sollfrank reverse the model. They redefine authorship as the artistic design of an algorithmic process and, once this process is set into motion, the observation and reflection of its effects. The industrial nature of its results puts into question traditional categories of authorship, originality and genius. This approach builds and expands on the questions and provocations Tzara’s Dada poem instigated. Programmer Richard Leopold who wrote one of Sollfrank’s generators credits Dada as its main inspiration and calls the generated pages “Dada content” (but employs Markov chains instead of permutation as the generative algorithm).

What sets apart Sollfrank’s generator from Tzara’s poem however is that it explores its philosophical, aesthetic, legal and political implications in a more rigorous and systematic way. The idea that artists turn into designers and philosophical explorers of computations, paying only secondary attention to the output of the process per se, contradicts and subverts the industrial logic of the artistic software tool. Even Plots Unlimited, a rare example of computational-generative poetics within a commercial software package, still clings to the myth of being a transparent “aid” to the artist. Its advertising literally says that “with Plots Unlimited you’ll develop your own original material [. . . ]—stories that sell [emphasis as in the original text].” 41 Other computer programs for artists disguise themselves as mere tools in the hands of artists in more subtle ways. Word processors, graphics editors, desktop publishing and musical composition software are all based on two cultural premises:

1. To emulate, in look and feel, analog tools—typewriters, paint brushes, layout desks, scores and pianola rolls;
2. To pass themselves off as “transparent” tools, i.e. technology that obeys the user.

The latter is a fiction to the degree that those tools create their own aesthetics and have their own implied politics.

Software involves interface paradigms with encoded cultural preconceptions of what, for example, a “document,” “writing,” “designing” is. It has embedded concepts of the order of things, of communication and workflows. To this extent, software controls its users. Yet it sells the illusion that the user is fully in charge. Since the early 1990s, pop cultural graphic design has largely been driven by short-lived fashionable gimmicks and plug-ins in programs like . The whole musical genre of bootleg pop remixes would not exist without the programs Acid by Sonic Foundry and Traktor by Native Instruments. Even these programs maintain the fiction of transparency and user control, excluding algorithms that do not just aid, but actually generate work. Among the few exceptions are Plots Unlimited and Band-in-a-Box, a program that automatically creates musical arrangements.

Signwave Auto-Illustrator

The taboo of “transparent” software—to not openly interfere with the artist—is systematically addressed and subverted in Adrian Ward’s computer program Signwave Auto-Illustrator. 42 It transforms vector graphics software into a generative program with an agenda of its own, or rather, of its programmer who codes his subjectivity into algorithms. With a user interface that precisely mimicks the commercial graphics program Adobe Illustrator, Auto-Illustrator implements, for example, a text tool that writes its own, randomly generated texts. Other functions turn artwork into “instant Bauhaus,” leave “bugs” that wander around in the illustration, or render circles as smilies. However, the program is functional. It generates proper graphics files and has been practically employed for the graphic design of flyers and record covers. Auto-Illustrator combines encyclopedism and fragmentation as anticipated by Lullism and Dadaism respectively. They do not contradict, but complement each other in the program. In the many years of its development since 2000, Auto-Illustrator has acquired an encyclopedic wealth of features, similar to commercial software that, incorporating more and more functions, strives to be the ultimate tool for its purpose. Autor-Illustrator openly renders the same totality absurd by accumulating eccentricities and personal fancies.

Algorithms as subjective expression were not conceived of in classical computational poetics. Both occult and scientific computations, be it magic, Kabbalah, encyclopedism or algorithmic calculus, rely on the idea that computation expresses a higher objective order; be it divine law, or the laws of logic and mathematics. It’s hardly surprising that computation often oscillated between the occult and scientific poles, blurring its boundaries: from Pythagorean mathematics to eventually metaphysical computations such as those of Karlheinz Stockhausen and John Cage. Nevertheless, as the “semantics” of formal languages, i.e. the choice of their cultural denominators like Llull’s alphabetum indicates, computations were never able to do without superimposed meaning, inscribed subjectivity, embedded metaphors.

Since the 1970s, the field of “generative art” relies on the same illusion of objectivity through computation. The most comprehensive theoretical foundation for this field is provided by Max Bense’s writings. Even today, media theory remains centered around and preoccupied with the idea of an anti-subjective, post-human autonomy of machine processes. The roots of this media theory prominently lie in McLuhan’s statement that the medium is the message—i.e. the technology itself bears the meaning instead of being its neutral purveyor—and Heidegger’s philosophy of matter and technology. These theories neglect that the machine processes were designed by humans in the first place and can be seen on the contrary as embedding subjectivity into formulas, processes and hardware. When Cornelia Sollfrank states that “a clever artist makes the machine do the work,” it still implies that the artist makes it work in the first place. 43 Poiesis, making, becomes a second-order poeisis of making something that makes something else. So poetry, making, turns into poetics, the making of making. When making turns into meta-making, subjectivity simply shifts to a second order position, residing in the formula instead of the product. This fact is being repeatedly ignored by critical observers whose perspective remains fixated on the product and who wrongly conclude, in a fallacy reminiscent of Plato’s cave, that technology has done away with the subject behind the work.

Calvino’s lecture on Cybernetics and Ghosts ends its theory about language and poetry as a “combination game” with the speculation that one day, the machine might be able to defy its own rules. This utopian hope has been, throughout the history of arts and computing, frustrated again and again (see chapter 4 ). It is an ultimately magical hope: That computing may one day transcend formalisms, and thereby its own technical grounds and limitations, in a moment of ontological chance similar to that of a die displaying the number 7. The same utopian hope drives artificial intelligence research, artificial life research and “virtual reality” computing. It also underlies the expectation of more “humane,” “fuzzy” computer interfaces which began with the invention of the graphical user interface at Xerox PARC and its popularization with the Apple Macintosh.—The “Humane Interface” was the last, unfinished software development project of Macintosh co-creator Jef Raskin before his death in 2005.—The same hopes and expectations drove the mainstream of digital “media art” and media theory in the 1990s. More critical artistic works, notably those of Jodi and I/O/D (see p. 193 and 155 ) next to Sollfrank’s and Ward’s, have questioned these utopias and contrasted them with the criticial insight that

• computation remains, technically, a formalism;
• contrary to the assumptions of “ artificial intelligence,” “virtual reality” and company, semantics can be had, and already exists, even without transcending formalisms or making them complex.
• formalisms, on the contrary, have a cultural semantics of their own, even on the most primitive and basic level. With a cultural semantics, there inevitably is an aesthetics, subjectivity and politics in computing.

Pataphysics and Oulipo

Alfred Jarry and the Collège de Pataphysique

The insight that even the most simple formalism has a cultural impact resolves the old conflict between computational poetics as higher-order, metaphysical or scientific objectivity on the one side and subjectivist aesthetics on the other. Since Jonathan Swift, empirical and aesthetic thinkers opposed computation precisely on those premises. Romantic technology, such as Surrealist games, Situationist and computational psychogeography, is another model of resolving the opposites. The contradiction between “objective” science and eccentric subjectivism however is most comprehensively subverted, through a simple clashing or “discordia concors” of the two poles, in “mad” or “poetic science.” Pataphysics was coined by its inventor, 19th century absurdist French dramatist and novelist Alfred Jarry as a “science of imaginary solutions.” His 1898 novel Dr. Faustroll defines pataphysics as the “science that added to metaphysics, either in itself or outside itself, and extend as far beyond metaphysics as the latter extends beyond physics.” 44 With this humorous foundation, pataphysics is, factually, a poetic science. Unlike previous poetic sciences and epistemologies such those of Novalis and Mallarmé, it does not take refuge in older metaphysical and theosophical paradigms of macrocosmic and microcosmic analogy. It is a poetic science that may, as Jarry’s novel shows, be founded on empirical absurdism, but in general exists outside the categories of rationalist physics and theosophical metaphysics.

Founded in 1949 by among others Raymond Queneau, Michel Leiris and Boris Vian, the Collège de Pataphysique continued a Parisian-French avant-garde tradition at whose center previously had been André Breton’s surrealist group. Queneau had been a member of surrealism for a short period, and satirized it in his novel Odile. His chief fields of interests were mathematics and linguistics, and his day job was to be the editor of encyclopedias in the Editions Pléiade. Perhaps under his influence, the Collége de Pataphysique abandoned the romanticist, communist-political and psychoanalytical pretext of surrealism. This legacy was continued instead by the Situationist International which consequently criticized pataphysics as a “new religion in the making.” 45

Raymond Queneau’s 100,000 Billion Poems

In 1960, the Collège de Pataphysique founded its own literary chapter, the Oulipo (Ouvroir de littérature potentielle, “Workshop of Potential Literature~). Its core members were Raymond Queneau and mathematician François Le Lionnais. Earlier in 1947, Queneau had spelled out his own obsessions with mathematics, encyclopedism and street slang in the Exercices de Style, a narrative of one short everyday scene in 99 different stylistic variations. 46 In 1961, he extended this concept into a computational poem, the 100,000 Billion Poems, a combinatory sonnet in ten variations. 47 It was printed in a book whose lines were individually sliced so that each line of poem could be turned like a page and picked from ten alternatives. From ten alternatives for the twelve sonnet lines, 10 ¹²possible poem combinations result. Queneau subverts the rigorous classicism of the sonnet form and its Alexandrine meter through slang colloquialisms and through stereotypes sarcastically perpetuated in the poems. His preface credits the playful form of the book to children’s books and disclaims any influence from surrealist games. The encyclopedism of his 100,000 billion poems is one of a perpetuated proverbial, trivial wisdom. As a pataphysical work, they contradict modern science through their combination of a Lullist synthetic scholasticism with an empirism of proverbial clichés. Queneau remarks that

Dans un ordre plus abstrait, [. . . ] l’Absolu, selon Jarry, “les clichés sont l’armature.” (In an abstract order, [. . . ] clichés are, according to Jarry, “the armature of the absolute.”)

Algorithmics as constraint and anti-formalism

Oulipo’s combination of poetry and mathematics was neither strictly rationalist, nor formalist. In an essay Potential Literature that appeared in 1964, Queneau clarifies that

We are not are not concerned with experimental or aleatory literature (as it is practiced, for example, by Max Bense’s group in Stuttgart). 49

Oulipo conceived of formalist poetics as a game-like artificial restriction on writing. As in a game where players have to find clever solutions around the hurdles superimposed by the rules, the algorithmic restrictions had to be compensated by imagination. A whole set of formalisms was created, such as lipogrammatic constraints which prohibit the use of certain letters. Poetic imagination was challenged to find a creative circumvention around the constraint. Oulipo member Georges Perec, for example, wrote a bulky novel La Disparation (English title: A Void) without a single occurence of the letter “e.” 50 Oulipo’s alleged naivité was its sophistication. It neither bought into the fallacy of scientific objectivity and machinic intelligence like Bense’s group, nor did it regress into magical and theosophic metaphysics like Burroughs and Gysin.

Whilst the culture and pompous rites of the Collège de Pataphysique with its “satraps” and “magnificences” parodically reworked the association of science and metaphysics, the parody of Swift’s academy of Lagado was practically made a live performance in the Oulipo. It operates on the grounds of the Swiftian assumption that poetic combinatorics are a fancy, but therefore have an artistic-fantastic potential to be exploited and explored. At the same time, the formalist constraints create a program for channeling artistic subjectivity, as opposed to—for example—the boundless romantic subjectivism of the anti-formal, but pointless Lettrist sound poetry of Isidore Isou and Maurice Lemaître from which the Situationist group emerged, too.

As opposed to Bense’s school with its rigorous constructivist modernism, Oulipo considered formalisms and computations neither a poetic end in themselves, nor a philosophical-ideological base. Oulipo created a computational poetics as anti-computational poetics, using computational formalisms for the sole end of circumventing them artistically. While Bense fought against semantics and imagination, Oulipo made it their game to let imagination fight against self-imposed formalism and have it triumph in the end.

Still, Oulipo’s poetics did not really reflect computations as cultural, and formalisms as loaded with meaning. Like others, it conceives of culture, imagination and meaning as something foreign to and struggling against formal rules. The slogan of 1990s digital art collective I/O/D, “software is mind control, get some” is anticipated in Oulipo only in its first half. Mind control is not embraced in Oulipian poetics, but overcome in a factually romanticist move. It comes as little surprise that Oulipo largely gave up on formal- algorithmic methods after spinning off a computer programming division ALAMO (“ Atelier de Littérature Assisté par la Mathématique et l’ordinateur,” “Laboratory of Mathematically and Computer-Assisted Literature”) in 1973 and the deaths of Raymond Queneau in 1977 and Georges Perec in 1982. Nowadays, Oulipo focuses on improvisational, non-computational games like the writing of poems in between two subway stops.

Italo Calvino’s speculation on literature as computation also reflects upon Oulipo in which he had been a regular member. Cybernetics and Ghosts documents his exposure both to the structuralism of Tel Quel and Oulipian poetics, synthesizing the former’s theoretical, analytic understanding of language as a combinatorial system with the latter’s practical, synthetic poetics. 51 In a later contribution to Oulipo proceedings, Calvino refers to his combinatorial prose composition as “anticombinatorics.” 52 He distances himself from his earlier theory that the poet could be replaced by a machine, and adopts the Oulipo angle of the algorithm as a self-imposed constraint. With his own personal background as a neorealist and fantastic novelist who heavily drew from folk tales, his late prose works like Invisible Cities, The Castle of Crossed Destinies and “If on a Winter’s Night a Traveller” combine rigorous Oulipian composition—comparable to the novels of Georges Perec—with a fantastic imagination of space rendering them (and in particular the novel Invisible Cities) another para-manifestation of psychogeographical computing.

Abraham M. Moles’ computational aesthetics

Just as Oulipo’s anti-formalist thrust was less clear when the group started off as a “seminar of experimental literature,” its literary work was initially perceived as computational experimental literature. In 1962, physicist and philosopher Abraham M. Moles wrote a first manifesto of permutational art (erstes manifest der permutationellen kunst) 53 which was published by Max Bense in Stuttgart. Later, in 1971, he expanded it into a book Art et Ordinateur (Art and Computer). 54 The “manifesto” combines structuralist and cybernetic theory with examples of concrete poetry, Oulipo works and pre-modern Lullist and proteic literature. In Moles’ definition, permutational art is—unlike the Oulipian concepts of naive artisanships and self-imposed constraints—“experimental to the highest degree,” striving to “narrow down and exhaust the field of possibilities accessible through a set of rules.” Along with Bense and his information aesthetics, he even conceives of perception as something that can be formally described through technical information theory and functions as a reverse combinatory process.

In a seeming rehash of 17th century encyclopedic Lullism, Moles proposes to refound the arts on the basis of permutational combinatorics. The disciplines his manifesto covers in brief individual sections are mathematics, music, literature and poetry, mysticism, erotica and painting. These disciplines are, as in Leibniz’ and Novalis’ mathesis, reunited on the grounds of mathematical combinatorics. In tune with Bense, Moles writes both a poetics and an aesthetics in which a near-infinity of generative output likewise corresponds to a computational decoding of aesthetic phenomena. It is a totalism that includes both synthetic creation and analytical perception. In the attempt to map all arts and thinking onto algorithmic processes, Moles’ theory resembles artificial intelligence research and its project to formally describe semantics as a higher complexity syntax. Moles’ program appears not as rigorously totalist, however, because of its sketchy and highly speculative elaboration in this brief manifesto. In tune with Bense’s philosophy and its grounding of aesthetics on technical information theory, it conceives of “permutational art” as a “fundamentally anti- semantic activity.” The manifesto is a historical document of cybernetics and its attempt at a universal science of technology that investigates human-machine interaction. Cybernetics largely faded out and was often considered obsolete in the 1970s until many of its ideas—most importantly the description of cultural processes in terms of technical processes—resurfaced in the 1990s, in the guise of technical media theory.

Source code poetry

One year after Moles’ manifesto, the 9th issue of the Situationist International journal ran a polemic against Abraham Moles. Guy Debord calls him upon to “médite sur la valeur anti-combinatoire du mot,” “ponder the anti-combinatory value of a word.” 55 The phrase is reminiscent of Oulipo poetics and Calvino’s “anti-combinatorics.” Such overlaps actually existed. Noël Arnaud, member of the Collège de Pataphysique since 1952 and co-founder of the Oulipo, was involved also in the pre- situationist Cobra painters group and later, after the schism between Northern European Situationists and Debord’s circle, became co-editor with Jacqueline de Jong of the Situationist Times. In 1968, Arnaud published a book Algol. The poems it contains are based on the vocabulary of the Algol programming language translated into French. The idea of Algol poetry came from Le Lionnais’ first Oulipo manifesto from 1962. Still in the spirit of “experimental literature,” it proposed “forays [. . . ] notably into the area of special vocabulary (crows, foxes, dolphins; Algol computer language, etc.).” 56

For the first time, and unlike in classical computer-generated poetry like that of Theo Lutz and Brion Gysin, the programming language source code was acknowledged as having a poetic quality of its own. Arnaud’s Algol source code no longer generated poetry, but was the poem itself. Program code ceased to be seen merely as a “transparent” technical tool detached from the perceived art work, but as an aesthetic object itself. For the first time, artistic programming was not a means to another end. Computer programming languages appealed to the Oulipo as yet another formal constraint. In comparison to standard human languages, all programming languages are drastically limited in their vocabulary and syntax, Algol even more than others. A “poor” language that limited a writer’s freedom, Algol was yet another playground for Oulipo poets to overcome artificial constraints through imagination and cleverness.

This poetics highly resembles a “hacker” approach. Hacker culture, which—according to its mainstream historifications and legends—originated at MIT in practically the same years as the Oulipo, 57 had the very similar goal of appropriating technology in creative ways which could, if necessary, include clever circumvention of superimposed barriers and limitations. So it is perhaps not surprising that hacker culture re-invented the genre of programming language poetry, but without knowing of its Oulipo precursors. In an internet newsgroup posting of 1991, Larry Wall, creator of the Perl programming language (itself a “hack” in its syntactical mixture of older Unix scripting languages), left the message

Print STDOUT q  
Just another Perl hacker  
Unless $pring

Jodi

Source code poetry was reinvented for a third time in the mid-1990s net.art of Jodi. Jodi stands for Joan Heemskerk and Dirk Paesmans, a Dutch-Belgian artist couple. They were part of a network of artists who radically redefined digital art from so-called “interactive” high-tech graphical simulations to ironic low-tech works that played with bugs, incompatibilities and disruptions in software. Contrary to a slick, visually immersive digital art which treated the computer as a black box, Jodi aestheticize computers as self-immersed, often absurd generators of contingent data streams. The work OSS, for example, makes small browser windows which evade manual control pop up and fly about, forcing the user to take down the computer in order to regain control. The OSS desktop expands the misbehaving, uncontrollable user interface onto the entire Windows desktop. Yet Jodi’s art does not tell an imaginary truth underneath the surfaces of software user interfaces. Instead, it exposes the surrealism of formalisms. A newer Jodi work for example simply uses a commercial car driving computer game, and lets the car make infinite, lawnmower-like circles, with high speed and squeaking tires, in the front garden of an American suburban family home. On http://www.jodi.org visitors are confronted with code in the form of cryptic error messages, blinking machine symbols and contigent tabular listings of numbers. It is code which often simply refers to other code—an error message for example linking to another error message linking to yet another error message. But moreover it is code which is not what it seems to be. The web site features fake software which in reality is just animated graphics and blinking browser text. ‘Surgery/havoc” http://wwwwwwwww.jodi.org/100cc/havoc/ looks like a jump’n’run computer game with several zoom functions, but is actually a clever arrangement of animated graphics files. Employing the same means, BinHe X http://wwwwwwwww.jodi.org/100cc/hqx/i900.html simulates system crashes and computer virus infection.

Next to their web site, Jodi began to use net cultural mailing lists as their medium, bombarding them likewise with cryptic, repetitive code messages such as the following:

Date: Tue, 25 Dec 2001 23:35:19 +0100  
From: 7061 <404@jodi.org>  
Subject: [P1] : [P1] Auto-Su(MMi6 2001 (MMi6 2001 21:48:24 +0100 (MMi6 200  
              1 21:48:24(MMi6 2001 (MMi6 2001 21:48:24 +0100 (MMi6 200  
          1 21:48:24mitted (MMi6 2001 (MMi6 2001 21:48:24 +0100 (MMi6 200  
             1 21:48:24: auto-g(MMi6 2001 (MMi6 2001 21:48:24 +0100 (MMi6  
 200              1 21:48:24enerated (failure)  Mime-(MMi6 2001 (MMi6 2001  
 21:48:24 +0100 (MMi6 200              1 21:48:24sion:   1.0 Subject:  
 Returned   Resent-From: rspi  Sta >> iBo f  Ap k. >> FAQ
istFAQ.shtml> >>  
 >> To u gu(MMi6 2001 (MMi6 2001 21:48:24 +0100 (MMi6 200              1 

 21:48:24m> >> To sw   ibook- >> dige  shguy.com> >>
Nee(MMi6 2001 (MMi6  
 2001 21:48:24 +0100 (MMi6 200              1 21:48:24d help   o(MMi6 2001  
 (MMi6 2001 21:48:24 +0100 (MMi6 200              1 21:48:24(MMi6 2001  
 (MMi6 2001 >> ---------- >> Sm o >> ---------- >> Sm o
>> ---------- >>  
 Sm o >> ---------- >> Sm o >> ---------- >> Sm o
>> ---------- >> Sm o >>  
 ---------- >> Sm o >> ---------- >> Sm o >> ----------
>> Sm o >>  
 ---------- >> Sm o >> ---------- >> Sm o >> ----------
>> Sm o >>  
 ---------- >> Sm o >> ---------- >> Sm o 21:48:24 +0100
(MMi6 200  
      1 21:48:24k- >> requ com> >> ---------- >> Sm o To
Win A  | >>  ---  
 Cano iBook!   | >> >> iBookPl ore. >> >>    Road es,
$20  >>  Traveler >>  
   Cool tools.com> 01 21:48:24 +0100 (M Mi6 2001 21:48:24 +0100 (M Mi6  
 2001 21:48:24 +0100 (M Mi6 2001 21:48:24 +0100 (M Mi6 2001 21:48:24 +0100  
 (M Mi6 2001 2(MMi6 2001 21:48:24 +0100 (MMi6 200              1 21:48:24  
 +0100 (MMi6 2001 (MMi6 2001 21:48:24 +0100 (MMi6 200              1  
 21:48:24 +0100 (MMi6 2001 (MMi6 2001 21:48:24 +0100 (MMi6 200  
  1 21:48:24 +0100 (MMi6 2001 (MMi6 2001 21:48:24 +0100 (MMi6 200  
     1 21:48:24 +0100 (MMi6 2001 (MMi6 2001 21:48:24 +0100 (MMi6 200  
        1 21:48:24 +0100 (MMi6 2001 (MMi6 2001 21:48:24 +0100 (MMi6 200  
           1 21:48:24 +0100 (MMi6 2001 (MMi6 2001 21:48:24 +0100 (MMi6 200  
              1 21:48:24 +0100 (MMi6 2001 (MMi6 2001 21:48:24 +0100 (MMi6  
 200              1 21:48:24 +0100 (MMi6 2001 (MMi6 2001 21:48:24 +0100  
 (MMi6 200              1 21:48:24 +0100 (MMi6 2001 (MMi6 2001 21:48:24  
 +0100 (MMi6 200              1 21:48:24 +0100 (MMi6 2001 (MMi6 2001  
 21:48:24 +0100 (MMi6 200              1 21:48:24 +0100 (MMi6 2001 (MMi6  
 2001 21:48:24 +0100 (MMi6 200              1 21:48:24 +0100 (MMi6 2001  
 (MMi6 2001 21:48:24 +0100 (MMi6 200              1 21:48:24 +0100 (MMi6  
 2001 (MMi6 2001 21:48:24 +0100 (MMi6 200              1 21:48:24 +0100  
 (MMi6 2001 (MMi6 2001 21:48:24 +0100 (MMi6 200              1 21:48:24  
 +0100 (MMi6 2001 (MMi6 2001 21:48:24 +0100 (MMi6 200              1  
 21:48:24 +0100 (MMi6 2001 (MMi6 2001 21:48:24 +0100 (MMi6 200  
  1 21:48:24 +0100 (MMi6 2001 (MMi6 2001 21:48:24 +0100 (MMi6 200  
     1 21:48:24 +0100 (MMi6 2001 (MMi6 2001 21:48:24 +0100 (MMi6 200  
        1 21:48:24 +0100 (MMi6 2001 (MMi6 2001 21:48:24 +0100 (MMi6 200  
           1 21:48:24 +0100 (MMi6 2001 (Mon the discussions between  
 Mrinmoy / Shubhodip (bluwregraseqr werMi6 2001 21:48:24 +0100 (MMi6 200  
            1 21:48:24 +0100 (MMi6 2001 (MMi6 2001 21:48:24 +0100 (MMi6  
 200              1 21:48:24 +0100 (MMi6 2001 (MMi6 2001 21:48:24 +0100  
 (MMi6 200              1 21:48:24 +0100 (MMi6 2001 (MMi6 2001 21:48:24  
 
[...]

The above message appeared not in the E-Mail message body, but entirely in the subject line, trying to provoke errors, crashes and confusion in the distribution and display of the message. With such interventions, Jodi tried to refashion net cultural discussion forums from theoretical discussion forums about art and culture into practical artistic playgrounds. At the same time, their E-Mail art once more reduced the technical complexity of digital art, in sharp contrast to high tech museum installations. With their conflation of English everyday language, code fragments from programming languages, character encodings, markup languages and network protocol code, these messages furthermore created uncertainty and paranoia among naive readers who might think that the signs were malicious and contained viral code. Consequently, the Jodi website was temporarily shut down because its network provider believed that OSS was a computer virus. This was the practical proof that Jodi’s artistic simulations of code in the medium of code were efficacious. (More on on p. 185 .)

As imagined viruses, the codes raised doubt about their origins. All the more when a growing number of pseudonymous entities began sending out similar messages to mailing lists. Media theoretician McKenzie Wark characterized those message as follows:

This might be mangled machine English, or perhaps an English written by a machine programmed by someone who speaks English as a second language, or someone producing a simulation of some such. 59

1337 speech

Codeworks reflect the uncanny underbelly of network communication in an age where the Internet is accessed largely by graphical browser and client programs, but with the constant awareness that non-graphical codes are running underneath the system. In mainstream computing, these codes revealed themselves only in error messages or as “blue screens of death” when the operating system crashes. Computer hackers had written source code poems and typograms with ASCII characters out of the aesthetic limitation of non-graphical terminal and command line computing between the 1960s and early 1990s. For the net.artists who worked in the age of the web browser, these non-graphical constraints were voluntary and self-imposed, very much like Oulipo’s constraints. Since ASCII typograms were hacker circumventions of technical limitations, they had an aura of subversion, and were hybridized with slang.

A particular slang developed in the subculture of “crackers,” hackers who break into network computers or crack the copy-protecting schemes of computer games. It replaces letters with numbers, for example “e” with “3” and “t” with “7” (because of their typographic similarity), so that a “1337 [leet] hax0r” becomes code speech for “elite hacker.” These substitutions originated in about the same time as similar codes in hip hop culture where, for example, “2pac” stood for rapper Tupac Shakur. Writing in 1337 slang, among others on websites defaced by crackers, became digital graffiti. Since the letter substitutions can be automated, there also exists simple text filter software—like the Unix filter “b1ff”—which transforms standard English writing into 733t speech:

SINCE THE LETTUR SUBSTITUSHUNZ CAN BE AU2M8D. THURE ALSO EXISTZ SIMPLE TEXT F1LTUR SOFTWARE—LIKE THE THE MANEFRA1M O/Z F1LTUR B1FF—WH1CH TRANS4MZ STANDARD ENGLISH WR1T1NG IN2 733T SPEECH.

Codework

In the early 1970s, Alan Sondheim began to work with programming code as material in the larger context of conceptual art. He experimented with Unix command line code in experimental writing as early as in the 1980s, and collaborated with younger net.artists since the 1990s. One of his codeworks reads as follows:

From: Alan Sondheim <sondheim@panix.com>  
To: _arc.hive_@lm.va.com.au  
Date: Thu, 9 Jan 2003 17:17:20 -0500 (EST)  
 
 
sleeping and running zombies through bodies  
 
 
CPU states:  4.7% user, 5.8% system, 0.0% nice, 89.4% idle:36 processes:  
35 sleeping, 1 running, 0 zombie, 0 stopped:1m 4:20pm up 8 min, 1 user,  
load average: 0.54, 0.26, 0.11: :Mem:  38664K av, 35084K used, 3580K free,  
14956K shrd, 15080K buff Write vaginas through my CPU states:  4.7% user,  
5.8% system, 0.0% nice, 89.4% idle! CPU states:  4.7% user, 5.8% system,  
0.0% nice, 89.4% idle:36 processes: 35 sleeping, 1 running, 0 zombie, 0  
stopped:1m 4:20pm up 8 min, 1 user, load average: 0.54, 0.26, 0.11: :Mem:  
38664K av, 35084K used, 3580K free, 14956K shrd, 15080K buff Write vaginas  
through my CPU states:  4.7% user, 5.8% system, 0.0% nice, 89.4% idle!  
load average: 0.54, 0.26, 0.11: :Mem:  38664K av, 35084K used, 3580K  
free,:35 sleeping, 1 running, 0 zombie, 0 stopped:1m 4:20pm up 8 min, 1  
user,:CPU states:  4.7% user, 5.8% system, 0.0% nice, 89.4% idle:36  
processes::Write vaginas through my CPU states:  4.7% user, 5.8% system,  
0.0% nice,:89.4% idle! CPU states:  4.7% user, 5.8% system, 0.0% nice,  
89.4% idle:36 processes: is sufficiently well-inscribed. - I consider the  
following again, your CPU states:  4.7% user, 5.8% system, 0.0% nice,  
89.4% idle:36 processes: ... enunciation inscribes me upon your token! CPU  
states:  4.7% user, 5.8% system, 0.0% nice, 89.4% idle:36 processes:, load  
average: 0.54, 0.26, 0.11: :Mem:  38664K av, 35084K used, 3580K free,  
remembers my chisel My load average: 0.54, 0.26, 0.11: :Mem:  38664K av,  
35084K used, 3580K is your language... load average: 0.54, 0.26, 0.11:  
:Mem:  38664K av, 35084K used, 3580K free, calls forth births inscription,  
hungered, making things. upon the time, load average: 0.54, 0.26, 0.11:  
:Mem:  38664K av, 35084K used, 3580K free, is here, 00], 35 sleeping, 1  
running, 0 zombie, 0 stopped:1m 4:20pm up 8 min, 1 user,? ... inscription  
is stopped:1m 4:20pm up 8 min, 1 user, load average: 0.54, 0.26, 0.11:  
:Mem:  on black stone, it’s inscription? Are you satisfied with your load
 
average: 0.54, 0.26, 0.11: :Mem:  38664K av, 35084K used, 3580K free,?  
load average: 0.54, 0.26, 0.11: :Mem:  38664K av, 35084K used, 3580K free,  
1086 is the perfect proclamation. CPU states:  4.7% user, 5.8% system,  
0.0% nice, 89.4% idle:36 processes: 35 sleeping, 1 running, 0 zombie, 0  
stopped:1m 4:20pm up 8 min, 1 user, load average: 0.54, 0.26, 0.11: :Mem:  
38664K av, 35084K used, 3580K free, 14956K shrd, 15080K buff Write vaginas  
through my CPU states:  4.7% user, 5.8% system, 0.0% nice, 89.4% idle!  
load average: 0.54, 0.26, 0.11: :Mem:  38664K av, 35084K used, 3580K  
free,:35 sleeping, 1 running, 0 zombie, 0 stopped:1m 4:20pm up 8 min, 1  
user,:CPU states:  4.7% user, 5.8% system, 0.0% nice, 89.4% idle:36  
processes::Write vaginas through my CPU states:  4.7% user, 5.8% system,  
0.0% nice,:89.4% idle! load average: 0.54, 0.26, 0.11: :Mem:  38664K av,  
35084K used, 3580K free,:35 sleeping, 1 running, 0 zombie, 0 stopped:1m  
4:20pm up 8 min, 1 user,:CPU states:  4.7% user, 5.8% system, 0.0% nice,  
89.4% idle:36 processes::free,:35 sleeping, 1 running, 0 zombie, 0  
stopped:1m 4:20pm up 8 min, 1:38664K av, 35084K used, 3580K free, 14956K  
shrd, 15080K buff Write vaginas Your enunciation names my stopped:1m  
4:20pm up 8 min, 1 user, load average: 0.54, 0.26, 0.11: :Mem:  !  

The work is based on the output of the Unix system command “top” which displays a list of running processes, memory and central processor load. “Zombie” is a technical Unix term for a program process that can no longer be terminated with the “kill” command. Sondheim’s text takes these descriptors—or “semantics,” as computer science would call it—literally. He reads the output of the program as a physical inscription of bodies, as performance art and a subjective utterance in the medium of computer software. It gets reformatted and partially rewritten, in an operation that blurs the boundaries of machine and human, syntax and semantics, with words and phrases mapping bodies and sexuality. Subject and object, syntax and semantics, formalism and culture become inseparably entangled, crisscrossing and writing over each other.

This is not simply a poetic metaphorization because the technical apparatus of writing becomes a part of the text. There is a feedback of textual input, output and processing inside the text and within the medium of code. This conflation of source code, data and processing could be called recursive, since recursion means that a process processes itself. Yet it is not a formal-mathematically “clean” recursion. The text is not a result of a pure computation, but involves human editing, rendering the recursion a simulation, rhetorical, reflexive.

The “codeworks” by Jodi, mez, Alan Sondheim and other artists manifest a most radical understanding of formalisms as meaningful. That their codeworks rarely execute properly, being imaginary code and conflations of machine code with human conversational language, reflects this understanding. The codeworks appropriate languages that were designed to be asemantic—programming languages, protocol code, shell commands—to unveil and elaborate their metaphorical and physical inscriptions, implications, and engendered meanings lurking between the lines. In other words, they do not just reflect words made flesh, but words and codes being flesh.