This paper looks at two recent text generators that are intended to raise questions about the style and substance of academic writing: Andrew Bulhak's "Postmodern Essay Generator" and Ian Bogost's "Latour Litanizer."
Bulhak created his generator in the wake of the Sokal hoax to mock the pretentious vocabularies of aspiring intellectuals who might hope to publish what he considered to be their relativistic and impenetrable nonsense in peer-reviewed scholarly journals in the humanities. The generator creates sentences like this each time the page is refreshed: "If one examines subtextual deconstructivism, one is faced with a choice: either accept dialectic materialism or conclude that the collective is responsible for class divisions, but only if sexuality is interchangeable with consciousness." Bulhak used the Dada Engine to create the verbiage in the essays his program generated. The actual program was written in what he called "inelegant" pb code that used protocol buffers "" a platform-neutral, language-neutral, extensible way of serializing structured data "" and recursive transition networks (or RTNs) to create a diagrammable grammar that Bulhak also retasked in a program for automatically writing adolescent poetry dominated by "i" statements.
In contrast, Bogost devised a much simpler javascript program that uses Wikipedia's random page API to generate lists of things and to provide brief poetic homages to the writing style of philosopher of science Bruno Latour, which frequently includes lists of seemingly disconnected objects. Bogost's program, which ends with the command to "litanize," has also been deployed in his larger critique of writing in the humanities, which is not only bad in his mind because it is obfuscatory but also because it is disconnected from the catalogues of things in everyday experience that he holds dear. A sample list created by the Latour Litanizer reads as follows: "Energy Tax Act, Linux Users of Victoria, Burton House, Fault line, NCAA Women's Division II Basketball Championship, USS Apache, Quinquagesima, Dbrowa, Wieliczka County, Karel Rychlík."
It could be argued that both Bulhak and Bogost use their generators somewhat disingenuously. Bulhak closely engages with the linguistic features of the poststructuralist vocabulary he appropriates and struggles to create grammatical sense by tweaking his program to accommodate such complex, multisyllabic words. Bogost incorporates Latourian litanies into his own diatribes against the digital humanities and the humanities more generally, but each item is carefully chosen in accord with his own writerly ear rather than machine-generated by his online gadget with its procedural poetics of chance.
In the era of highly constrained user-generated content on locked-down content management systems and mass-market social media applications, web generators have become a standard, recognizable Internet genre. These popular web-based forms seem to the general public both to create original online texts and to output data that conforms to fixed conventions. Thus these generators serve to be both generic and generative as they crank out results. As part of a larger realm of what Noah Wardrip-Fruin has called "expressive processing," they are also capable of particular rhetorical effects, particularly when they satirize individual agency in discourse. Because of these properties such generators offer a way to criticize how language itself is used and the rules by which utterances are assembled. Although very different in audience and purpose, Andrew C. Bulhak and Ian Bogost have created text generators that are intended to raise questions about the style and substance of academic writing and institutional expectations about the character of learned prose.
To understand how these web-based generators work and the importance of code-sharing practices among users, it's helpful to look at look at some introductory examples before turning to the generators of Bulhak and Bogost. For instance, the Church Sign Generator at a heavily trafficked site allows users to insert their own phrases into a stock photograph of a roadside church sign. To enable visitors to participate in media-creation activities that otherwise require some knowledge of image-editing software, such as Photoshop, creator Ryland Sanders posted a PHP-based form to his site for personalizing the church signs. By creating the generator, he invited visitors to gratify a form of wish fulfillment that would be prohibited in the offline world, when social norms make manipulation of such private property and public messages of others taboo: "Ever seen those signs in front of churches with the moveable letters? Ever wanted to rearrange the letters to make your own church sign? Well, now you can. Choose a design below, add your text, and a personalized church sign photo will be generated for you! Save it, send it to a friend, put on your website, or use it however you like. Enjoy!"
These websites also support another kind of DIY community: those who share code knowledge by showing how their clever PHP, Perl, JavaScript, or other programs operate. As Sanders explains on an FAQ, his Church Sign Generator program works by "copying each letter from an image with only letters on it (that is, by copying a rectangular region containing each letter from the letters image) and compositing it onto a background image." He justifies this seemingly circuitous approach by explaining that this method unlike "simply drawing text using TrueType fonts" allows him to "apply pixel effects (shading, color variations, beveling, etc.) in Photoshop that couldn't easily be done dynamically." For example, as he explains in his comments, this section of Sanders' code handles the length of each line of text.
# loop through lines array and process each line
for ($l = 1; $l < 5; $l++) {
$x_offset = 0;
$truncate_to = 0;
# loop through each letter in each line and get its width
# and kerning value (if any). if a line is too long to fit
# on the sign, it gets truncated.
for ($i = 0; $i < length($text[$l]); $i++) {
$char = substr($text[$l], $i, 1);
if ($x_offset + ($wx[ord($char)]) <= $max_line_width) {
$x_offset += $wx[ord($char)];
$truncate_to++;
}
}
$text[$l] = substr($text[$l],0,$truncate_to);However, Sanders warns that although he has released his code for others to tinker with and has even provided recommendations for basic primers on programming, he will not suffer fools gladly who expect personalized customer assistance:
In particular, don't send me e-mail that starts out, 'I know you said you don't do support, but"' because these e-mails will be deleted unread. I'm sorry if this seems curmudgeonly, but I don't have the time or the patience these days to deal with people who are trying to learn programming - these are not scripts for beginners.
In answering what is apparently a common question on his FAQ about how to save an image on a computer screen, Sanders responds, "I don't mean to be surly, but I don't have the time or, frankly, the patience to teach you how to use your computer" and suggests that "if your problem isn't covered here, take it up with your internet service provider, your network administrator, or the store where you bought your computer." Despite his rhetoric of reticence when it comes to provide advice to novices, Sanders then provides basic instructions about how to right-click on images to save them to the user's machine. In an appeal to sympathy, he also explains how he has blocked hotlinking to prevent expensive bandwidth costs and bemoans the hassles of maintaining his servers.
Sharing of source code for web generators is also not always welcomed by the authorities. When Indiana University graduate student Christopher Soghoian created a web generator that could print out look-alike versions of Northwest Airlines boarding passes, agents from the federal government forcibly entered his apartment and seized computers and other equipment. Soghoian was soon threatened with arrest. However, Soghoian's protest against potential flaws in the Transportation Security Administration had already drawn supporters, and soon one of them, Matt Watterman, created a humorous "warrant generator" for which "this script has been created for district courts all across the United States with the intent of improving national security by reducing the amount of time it takes for our public guardians to create search warrants." Users can enter names, addresses, and possible legal justifications into the warrant generator and receive a warrant that looks much like Soghoian's original. The warrant generator was also supplied with instructions, such as "Enter desired legal mumbo-jumbo below. Again, make sure to use run-on sentences and try to cite some laws and codes and stuff."
However, soon after his site was taken down at the insistence of cybercrime officials, an imitator who went by then pseudonym "J0hn 4d4m5" made another, more polymorphic version of the Northwest boarding pass generator and shared his source code on the web. Unlike Soghoian's code, the copycat programmer did reverse engineering in order to design a generator that could be "implemented only using HTML and javascript" so that "you do not need a web server where you can run PHP; you need no server at all." All one has to do is "use the open file menu in the browser and open the pass.html to use the generator."
Soghoian never made the source code for his boarding pass generator public. Now, not only could users run the generator in their browsers directly from the downloaded files with more privacy, but also they could use 4d4m5's tool to create generators of their own. According to the explanation provided, this new JavaScript generator was meant "to demonstrate the framework that the NWA generator was written in" to inspire others to create "document generators like this one." As he explained it, the "design goal for the tool was that it be easy to modify any document it generates." 4d4m5 included a "hello.html" file, and he encouraged "people to play with the tool by modifying the hello_content.html and hello_form.html files to produce new generators that make passes for other airlines or whatever you deem appropriate."
Now, not only could users run the generator in their browsers directly from the downloaded files with more privacy, but also they could use 4d4m5's tool to create generators of their own. According to the explanation provided, this new JavaScript generator was meant "to demonstrate the framework that the NWA generator was written in" to inspire others to create "document generators like this one." As he explained it, the "design goal for the tool was that it be easy to modify any document it generates." 4d4m5 included a "hello.html" file, and he encouraged "people to play with the tool by modifying the hello_content.html and hello_form.html files to produce new generators that make passes for other airlines or whatever you deem appropriate."
The way that generators generate other generators in Soghoian's case illustrates the importance of one of the common features of this genre: its power for social as well as computational replication. Often, as content-creators explain how their source code can be deployed, it becomes apparent that they are merely repurposing code from other generators. For example, the code from the Random Compliment Generator uses a syntax based on the Random Insult Generator, as the following explanatory snippet from the code shows:
""" Random Compliment Generator """ # based on the Random Insult Generator but modified for a work environment pronoun = ['you','my','thou'] verb = ['hard working', 'forward thinking', 'fast moving'] # doing words verb.extend(['quick learning', 'compelling','painstakingly', 'encouraging']) verb.extend(['amazingly'] adjective = ['intelligent', 'smart'] # describing word adjective.extend(['vigorous','energetic','resourceful', 'thorough', 'meticulous']) adjective.extend(['reliable','diligent','advanced']) adjective.extend(['avant garde','cutting edge','revolutionary','dynamic']) nounNot surprisingly, academia is a frequently chosen foil for web generators, particularly if its theories and methodologies seem to have been petrified into easily assembled institutional clichés disconnected from everyday speech acts. Three MIT students used SCIgen, a program that generates random computer science papers in Association for Computing Machinery format to fool the organizers of the 2005 World Multiconference on Systemics, Cybernetics, and Informatics, a supposedly peer-reviewed conference in Orlando, Florida, known for its spam-style solicitations of those who work in universities on technology-related issues. The papers that this program generated bore titles such as "Harnessing Byzantine Fault Tolerance Using Classical Theory," "On the Study of the Ethernet," and "Synthesizing Checksums and Lamda Calculus Using Jog," and came complete with graphs, figures, and citations that integrated stock buzzwords. The creators of SCIgen released their source code online with pointers to related resources in Perl, LaTeX/BibTeX, Gnuplot, and Graphviz. They also provided some documentation to help others understand the context-free grammar that drives the program and cautioned that there are "misleading files" in the source code. Automatically generated papers from SCIgen have also been successful at deceiving organizers of mathematics, emerging technology, and new media pseudo-conferences.
Andrew C. Bulhak pointed to the so-called "Sokal hoax" as inspiration for creating the Postmodern Essay Generator, after physicist Alan Sokal cobbled together an essay, "Transgressing the Boundaries: Toward a Transformative Hermeneutics of Quantum Gravity," and successfully submitted it to the peer-reviewed journal Social Text. After the essay was in print, he described in in an "Afterward" as a "mélange of truths, half-truths, quarter-truths, falsehoods, non sequiturs, and syntactically correct sentences that have no meaning whatsoever" (Sokal 338). Bulhak took Sokal's claim that postmodern criticism was randomly assembled nonsense to the next step of creating a generator, so that a machine could produce prose just as authoritative and incoherent as Sokal's essay.
Bulhak applied the Dada Engine to the project to populate web pages with polysyllabic diction and generous footnotes at the push of a refresh browser button. A typical "postmodern" sentence reads as follows: "If one examines subtextual deconstructivism, one is faced with a choice: either accept dialectic materialism or conclude that the collective is responsible for class divisions, but only if sexuality is interchangeable with consciousness." The generator spouts out papers with titles like "Pretextual Theories: Objectivism, Debordist Situation, and Cultural Deappropriation," "Postconstructivist Discourses: Cultural Dematerialism and Capitalism," and "The Failure of Art: Conceptual Neocapitalist Theory, Marxism and Expressionism." Recently Bulhak also created an iPhone application that can also incorporate location information somewhat crudely into the results.
Bulhak makes his criticism of theoretical postmodernism even more explicit in a paper about his generator, in which he compares the system's output of Sokal-like pseudo-postmodern prose to a simulation of "mental debility" like the "ranting of a paranoid schizophrenic street preacher, or perhaps a USENET ranter" and suggests the script could be modified to mimic the writing in "eccentric pseudoscientic/religious pamphlets." Bulhak explains that his other inspiration came from reading Douglas Hofstadter's Gdel, Escher, Bach: An Eternal Golden Braid in which Hofstadter "demonstrated a method of generating meaningless but grammatically correct English text" and "illustrated this method with an example: a selection of 13 fragments of text, ten of which were generated by a computer and three which were taken from a journal titled Art-Language." Bulhak also describes an additional practical application for the source code to explain its origin: "this script was originally written with the intention of generating bogus practice examination papers to be distributed in lectures for the purpose of scaring students." All three explanations have to do with simulating academic discourse: in Social Text, in Art-Language, and in an everyday examination situation.
The actual program was written in what Bulhak called "inelegant" pb code that used protocol buffers "" a platform-neutral, language-neutral, extensible way of serializing structured data "" and recursive transition networks (or RTNs) to create a diagrammable grammar that Bulhak also retasked in a program for automatically writing adolescent poetry dominated by "i" statements. The different morphologies of the preposition/adjective/noun/verb structure in the Dada Engine are unpacked in the engine's documentation. He also gives examples of variables, parametric rules, and operators that allow shortcuts. The source code itself is also generously commented.
Comments in pb are handled as in C++ and Java.
A comment of any length may be enclosed between the sequence `/*' and the sequence `*/'. Alternatively, one-line comments may be preceded by `//'.
The following are valid comments:
/* This is a two-line comment.
See? */
// You are not expected to understand this. Although his generator mocking academic discourse once lived on a server from Monash University in Australia, Bulhak seems to have left academia and now lives in London, where he works in the private sector "in IT, with Linux, Python, C/C++ and web technologies, both professionally and otherwise"; he also periodically DJs, according to his home page. Bulhak had used his outsider status to criticize the insular self-authenticating structures of academic institutions; in the nineties his name was often linked online to "psychoceramics" and others who used the Internet to mock crackpot disciplines that were propagated by the structures of academic print publishing. His generator was even cited admiringly by Richard Dawkins in a 1998 essay in Nature called "Postmodernism Disrobed."
In contrast, Ian Bogost is a career academic with a tenured position at Georgia Tech. His criticism of the jargon of postmodernism, like his use of online generators, serves as a different kind of intervention in academic linguistic practices from Bulhak's satires aimed at broad ridicule. Although Bogost complains that "discourse" has become "the brand name for a device used to manufacture petty snipes—about the etymology of a word, or the truth value of a proposition, or the unexpected exclusion of a favorite French theorist" ("We Think in Public"), he also frequently cites his prior training in critical theory and his deep investments in posthumanist philosophy and literary criticism. Speaking as a "lapsed Derridean," he claims that academics largely fail to reach the public because they express their contempt for the concerns of everyday non-academic life by embracing obscure subjects and expert terminology and refusing to develop simple and straightforward "elevator pitches" and distinctive "brands" to explain their research areas to those outside the academy.
Like the SCIgen team, Bogost has used programs that generate phrases with context-free grammars or "CFCs." As Bogost explains, "in CFCs clauses can be nested inside clauses arbitrarily deeply, but grammatical structures are not allowed to overlap." In A Slow Year, Bogost has used this kind of computer program to create "machined haiku" from programs that require different, more subtle kinds of user interaction than pressing a simple refresh button.
As games, these rely on the procedural representation of an idea that the player manipulates. As poetry, they rely on the condensation of symbols and concepts rather than the clarification of specific experiences. As images, they offer visually evocative yet obscure depictions of real scenes and objects. They are inspired by ideas or experiences I encounter, as attempts to capture something fundamental about how they work.
In his academic rather than poetic pursuits, Bogost has created a much less sophisticated text generator that works with a simple button-push called the "Latour Litanizer," which provides brief poetic homages to the writing style of famed philosopher of science Bruno Latour. Latour was known to frequently include lists of seemingly disconnected objects when discussing how "human, natural, artificial, logical, and inanimate allies" (Harman 18) could form assemblages that generate particular causal effects. Levi Bryant has claimed that this particular rhetorical technique is central to Latour's ontology and epistemology as a philosopher.
What Latour's litanies effectively accomplish is an annulment of prototypes that come to stand for the being and nature of all objects. Through the creation of a litany of heterogeneous objects, the object theorist is forced to think that heterogeneity as such rather than implicitly (and often unconsciously) drawing on one prototypical object that functions as the representative of the nature of all objects.
In his own academic talks and presentations, Bogost has begun to incorporate such litanies, as he does in his concluding lines of "We Think in Public" at a recent conference in honor of the philosopher Richard Rorty.
It's not possible to get through the day without crossing hundreds, thousands, tens of thousands of ignored subjects of philosophy ready to be thought in earnest . . . the whippoorwills, the linen petticoats, the combine harvesters, the paper shredders, the salt water taffy, the magnolia blossoms, the waves that crash to shore, the bits that race from hard disks, the stippling on unseen bookplates
Bogost is careful to explain that his program, which ends with the command to "litanize," creates phrases that "are not just parody." The litanizer has been deployed in his larger critique of writing in the humanities, which is not only bad in his mind because it is obfuscatory but also because it is disconnected from the catalogues of things in everyday experience that he holds dear. A sample list created by the Latour Litanizer reads as follows: "Energy Tax Act, Linux Users of Victoria, Burton House, Fault line, NCAA Women's Division II Basketball Championship, USS Apache, Quinquagesima, Dbrowa, Wieliczka County, Karel Rychlk."
Elements are generated from the MediaWiki platform, which has a "Special:Random" function built into it to show sample articles in the sidebar of Wikipedia. As Bogost explains, "because this is just a web page, it can be queried and its results fetched." The JavaScript code that runs the "litanize" command reads as follows:
function litanize()
{
var xmlhttp=false;
if (!xmlhttp && typeof XMLHttpRequest!='undefined') {
try {
xmlhttp = new XMLHttpRequest();
} catch (e) {
xmlhttp=false;
}
}
if (!xmlhttp && window.createRequest) {
try {
xmlhttp = window.createRequest();
} catch (e) {
xmlhttp=false;
}
}
document.getElementById('litany').innerHTML = "loading litany...
";
document.getElementById('redo').innerHTML = "";
xmlhttp.open("GET", "http://" + location.host + "/speculativerealism/litanizer.aspx");
xmlhttp.onreadystatechange=function() {
if (xmlhttp.readyState==4)
{
//alert(xmlhttp.responseText);
document.getElementById('litany').innerHTML = xmlhttp.responseText; document.getElementById('redo').innerHTML = 'Generate Another';
}
}
xmlhttp.send(null)
return;
}
litanize();In this way Bogost offers up a tool for invention that livens up academic discourse rather than a way to emphasize the meaninglessness of existing templates for scholarly speech.
Both Bulhak's and Bogost's projects respond to the limitations of procedural expression by bringing in their own rhetorical flair. Bulhak closely engages with the linguistic features of the poststructuralist vocabulary he appropriates and struggles to create grammatical sense by tweaking his program to accommodate such complex, multisyllabic words. Bogost incorporates Latourian litanies into his own diatribes against the digital humanities and the humanities more generally, but each item in his litany is carefully chosen in accord with his own writerly ear rather than selected from the machine-generated results of his online gadget with its procedural poetics of chance. By automating content-creation, both projects open up possibilities for a critique of academic speech that incorporates more humor, poetry, logic, and clarity than is typical of conference papers like this one.
Bogost, Ian. "A Slow Year." Ian Bogost: Videogame Theory, Criticism, Design. Web. 10 Oct. 2010.
---. "Latour Litanizer." Ian Bogost: Videogame Theory, Criticism, Design. Web. 10 Oct. 2010.
---. "The Cocktail Party Test." Ian Bogost: Videogame Theory, Criticism, Design. Web. 10 Oct. 2010.
---. "We Think in Public." Ian Bogost: Videogame Theory, Criticism, Design. Web. 10 Oct. 2010.
---. "What is Object-Oriented Ontology?." Ian Bogost: Videogame Theory, Criticism, Design. Web. 10 Oct. 2010.
Bryant, Levi. "Latour Litanizer." Larval Subjects. Web. 10 Oct. 2010.
Bulhak, Andrew C. "acb." Web. 10 Oct. 2010.
---. "Communications from Elsewhere." Web. 10 Oct. 2010.
---. "On the simulation of postmodernism and mental
debility using recursive transition networks." Dept Computer Science Technical Reports 96.254 (1996): n. pag. Web.
---. "The Dada Engine - Table of Contents." Web. 10 Oct. 2010.
Dawkins, Richard. "Postmodernism disrobed." Nature 394.6689 (1998): 141-143. Web.
Harman, Graham. "Latour Litanies and Gibbon." Object-Oriented Philosophy. Web. 10 Oct. 2010.
---. Prince of networks : Bruno Latour and metaphysics. Prahran Vic.: Re.press, 2009. Print.
J0hn 4d4m5. "Document Gennreator." Web. 10 Oct. 2007.
Sanders, Ryland. "Church Sign Maker." Web. 9 Oct. 2010.
---. "Says-It.com - Frequently Asked Questions." Web. 9 Oct. 2010.
Soghoain, Christopher. "Chris's NWA Boarding Pass Generator." Web. 10 Oct. 2010.
Soghoian, Christopher. "Insecure Flight: Broken Boarding Passes and Ineffective Terrorist Watch Lists." SSRN eLibrary (2007): n. pag. Web. 10 Oct. 2010.
Sokal, Alan D. "Transgressing the Boundaries: An Afterword." Philosophy and Literature 20.2 (1996): 338-346. Web.
---. "Transgressing the Boundaries: Toward a Transformative Hermeneutics of Quantum Gravity." Social Text 46/47 (1996): 217-252. Print.
Stribling, Jeremy, Max Krohn, and Dan Aguayo. "SCIgen - An Automatic CS Paper Generator." Web. 10 Oct. 2010.
Wardrip-Fruin, Noah. Expressive processing : digital fictions, computer games, and software studies. Cambridge Mass.: MIT Press, 2009. Print.
Watterman, Matt. "fakewarrant." Web. 9 Oct. 2010.