Literary texts as

Cognitive assem-
bLages: the Case
of eLeCtroniC
Literature
By N. Katherine Hayles

“The question then is: what kinds of conceptual and artistic frame-
works will help us understand the implications of our participation
in the hybrid human-technical systems that have become essential
to contemporary life in developed countries?”

Suggested citation:
N. Katherine Hayles, Literary Texts as Cognitive Assemblages: The Case of Electronic Literature. Interface Critique Journal 2 (2019),
pp. 173–195.
DOI: 10.11588/ic.2019.2.66991

This article is released under a Creative Commons license (CC BY 4.0).

HAYLES / LITERARY TExTS AS COGNITIVE ASSEMBLAGES

Interactive media include such important millennial moment approaching the two
millennial art works as Simon Penny’s decades mark, computational media are
Traces (1999), Camille Utterback’s Li- increasingly integrated into complex in-
quid Time Series (2001–2010), Utterback frastructural systems, often in ways not
and Romy Achituv’s Text Rain (1999), easily visible but nevertheless crucial for
and other works created around the daily life in developed societies. Airport
turn of the 21st century emphasizing control systems, electrical grids, railroad
interactions between computational and subway controlling and tracking
media and embodied performances by systems, water purification systems, oil
humans. These artworks were accompa- refineries, and a host of other systems
nied by a wide range of books appearing rely on computational media to initiate,
at about the same time arguing for the synchronize, control, and communicate
importance of enactive, embodied, and their activities.
embedded cognition, including Edwin To see how far humans have pro-
Hutchins,1 Andy Clark,2 Francisco Vare- gressed into symbiosis with computatio-
la, Evan Thompson, and Eleanor Rosch,3 nal media, we can engage in a disturbing
Antonio Damasio,4 Gerald Edelman and thought experiment: what would happen
Giulio Tononi,5 and many others, inclu- to the human species if all computatio-
ding my own How We Became Post- nal media were fried tomorrow? All the
human.6 Some twenty years on, these systems that depend on computational
views have become widely accepted media for their activity would instantly
and even pervasive; but we can now also become inoperable, from automobiles
see that they tended to focus on the ac- and trucks to airplanes, railroads, and
tions of an individual person or at most subways, cutting off supply lines; in ad-
a few people interacting with computa- dition, hybrid human-computer systems
tional systems, leaving out of account such as the electrical grid and water pu-
the growing symbiosis between humans rification plants would begin to fail soon
and computational systems. In our post- if not immediately. Communication via
the web, cell phones, radio stations and
1 Edwin Hutchins, Cognition in the Wild (Cambridge, MA 1996).
television would be cut off; markets
2 Andy Clark, Being There: Putting Brain, Body and World would crash; the global economic system
Together Again (Cambridge, MA 1998).
would be plunged into chaos. It is not an
3 Francisco J. Varela, Eleanor Rosch and Evan Thompson, The exaggeration to say that perhaps the
Embodied Mind: Cognitive Science and Human Experience (Cam-
bridge, MA 1992).
majority of humans now living on earth
would perish. Many apocalyptic novels
4 Antonio Damasio, The Feeling of What Happens: Body and
Emotion in the Making of Consciousness (Boston 2000). entertain such imaginaries, but my pur-
pose here is somewhat different. In po-
5 Gerald Edelman and Giulio Tononi, A Universe of Conscious-
ness: How Matter Becomes Imagination (New York 2001). sing this thought experiment, I want
6 N. Katherine Hayles, How We Became Posthuman: Virtual Bod-
to indicate how far into symbiosis with
ies in Cybernetics, Literature, and Informatics (Chicago 1999). computational media we have already

174
 INTERFACE CRITIQUE JOURNAL – VOL. 2 – 2019

come. When two species become sym- bility. Cognizers direct, use, and interpret
bionts, each gains advantages by inter- the material forces on which the assem-
actions with the other, but they also be- blage ultimately depends.9
come increasingly interdependent, thus This schema employs a definition of
making themselves vulnerable should cognition crafted to include humans,
something happen to their symbiont. nonhuman others, and technical de-
This accurately describes our situation vices. In Unthought: The Power of the
today with computational media. The Cognitive Nonconscious (2017), I define
question then is: what kinds of concep- cognition as “the process of interpreting
tual and artistic frameworks will help us information in contexts that connect it
understand the implications of our parti- with meaning”.10 A corollary is that cog-
cipation in the hybrid human-technical nition exists as a spectrum rather than a
systems that have become essential to single capability; plants, for example, are
contemporary life in developed coun- minimally cognitive, whereas humans,
tries? other primates, and some mammals are
To explore this issue, my recent work very sophisticated cognizers. It has been
has focused on what I call cognitive as- traditional since John Searle’s “Chine-
semblages. Following Gilles Deleuze and se Room” thought experiment11 to argue
Félix Guattari7 as well as Bruno Latour,8 that computers only match patterns with
I conceptualize an assemblage as a fle- no comprehension of what that means
xible and constantly shifting collectivity (in his example, no semantic compre-
that includes human and technical ac- hension of Chinese). This view, however,
tors, as well as energy flows and other is increasingly untenable as computatio-
material goods. A cognitive assemblage nal systems become more sophisticated,
is a particular kind of network, charac- learning and experiencing aspects of the
terized by the circulation of information world through diverse sensors and actu-
through human and technical cognizers ators.
that drop in and out of the network in Much depends, of course, on how one
shifting configurations that enable in- defines the central terms “interpretati-
terpretations and meanings to emerge, on” and “meaning.” Searle’s example is
circulate, interact and disseminate. Cog- obviously anthropocentric, since it ima-
nizers are particularly important in this gines a man – a sophisticated cognizer
schema because they make the decisi- – sitting in a room with the rule book
ons and selections that give the assem- and other apparatus he employs, obvi-
blage flexibility, adaptability, and evolva-
9 Hayles, How We Became Posthuman.

7 Gilles Deleuze and Félix Guattari, A Thousand Plateaus: Capital- 10 N. Katherine Hayles, Unthought: The Power of the Cognitive
ism and Schizophrenia, transl. Brian Massumi (Minneapolis 1987). Nonconscious (Chicago 2017).

8 Bruno Latour, Reassembling the Social: An Introduction to 11 John Searle, Minds, Brains, and Programs. Behavioral and Brain
Actor-Network-Theory (Oxford 2007). Sciences 3 (1980), pp. 417–457.

175
HAYLES / LITERARY TExTS AS COGNITIVE ASSEMBLAGES

ously “dumb” affordances. The effect is to argues that writers should strive to
dumb the man down to the level of the maintain complete control over their
affordances he uses, a painful reduction signifying practices. Emphasizing that
of his innate cognitive capacities (hence writing in digital media proceeds via
the anthropocentric bias, which implies multiple layers of code, or as he calls it,
that computers are much dumber than “textual laminates”15, Tenen argues that
humans). We can continue to pat oursel- writers should understand and have
ves on the back for being so much more control over every level of the interlinked
intelligent than computers, but in speci- layers of code that underlie screenic in-
fic domains such as chess, Go and diag- scriptions. Since many commercial
nostic expert systems, computers now software packages such as Adobe em-
perform better than humans. I argue it is ploy hidden code layers that writers are
time to move past thought experiments legally forbidden even to access, much
such as Searle’s and take seriously the less change, Tenen passionately advises
idea that computers are cognizers, mani- his readers to avoid these altogether and
festing a cognitive spectrum that ranges compose with “plain text”16, open-source
from minimal for simple programs up to software that does not demand compro-
much more sophisticated cognitions in mises or sabotage the writer’s intentions
networked systems with complex mul- with hidden capitalistic complicities. We
tilayered programs and high-powered may think of Tenen as the Thoreau of di-
sensors and actuators. gital composition, willing to put up with
The fantasy of being completely auto- the inconvenience of not using pdfs and
nomous has a strong hold on the Ameri- other software packages to maintain his
can imagination, ranging from Thoreau’s independence and compositional integ-
Walden to Kim Stanley Robinson’s Mar-
12
rity.
tian terraforming trilogy13. Yet even Tho- Even if a writer chooses to go this rou-
reau used the planks and nails of a pre- te, however, she is still implicated in my-
vious settler and walked into town more riad ways with other infrastructural de-
often than his notebooks allowed. pendencies on computational networks
In the domain-specific area of writing, that come with living in contemporary
different attitudes toward our symbiosis society. So why single out writing as the
with computational media are manifes- one area where one takes a stand? Tenen
ted. Dennis Tenan, in his excellent book has an answer: compositional practices
Plain Text: The Poetics of Computation,14 are cognitive, and therefore of special
interest and concern for us as cogniti-
12 Henry David Thoreau, Walden; or, Life in the Woods (Boston 1854). ve beings.17 I respect his argument, but I

13 Kim Stanley Robinson, Blue Mars (New York 1997); Green Mars
15 Ibid., p. 5.
(New York 1995); Red Mars (New York 1993).
16 Ibid., p. 3.
14 Dennis Tenen, Plain Text: The Poetics of Computation (Stan-
ford: 2017). 17 Ibid., p. 52.

176
 INTERFACE CRITIQUE JOURNAL – VOL. 2 – 2019

wish to point out that other practices are and a random generator choosing which
also cognitive and so writing in this res- word to slot into a given poetic line. The
pect is not unique. Moreover, resistance method is straightforward but neverthel-
is not the only tactic on the scene. Other ess can generate interesting results. One
writers are of the opposite persuasion, such poem is Nick Montfort’s “Taroko
embracing computational media as in Gorge,” written after he had visited the
a symbiotic relation to human authors; picturesque Japanese site.19
they are interested in exploring the pos-
sibility space of what can be done when
the computer is viewed as a collaborator.
For writers like these, new kinds of ques-
tions arise. How is creativity distributed
between author and computer? Where
does the nexus of control lie, and who
(or what) is in control at different points?
What kinds of selections/choices does
the computational system make, and
what selections/choices do the human
authors encode? What role does ran-
domness play in the composition? Is the
main interest of the artistic project ma-
nifested at the screen, or does it lie with
the code? These are the questions that I
will explore below.

Slot algorithms:
Nick Montfort’s Fig. 1: Screen shot of Nick Montfort‘s Taroko Gorge.

“Taroko Gorge”
The vocabulary evokes the beauty of a
One way of enlisting the computer as natural landscape, including nouns such
co-author is to create what Christopher as “slopes,” “coves,” “crags,” and “rocks,”
Funkhouser calls “slot” algorithms, 18
and verbs like “dream,” “dwell,” “sweep”
with databanks parsed into grammatical and “stamp.” Montfort posted the source
functions (nouns, verbs, adjectives, etc.) code at his site, and in a playful gesture,

18 Christopher T. Funkhouser, Prehistoric Digital Poetry: An 19 Nick Montfort, Taroko Gorge; https://nickm.com/taroko_
Archaeology of Forms, 1959–1995 (Tuscaloosa, AL 2007), p. 40 gorge/, access: March 26, 2018.

177
HAYLES / LITERARY TExTS AS COGNITIVE ASSEMBLAGES

Scott Rettberg substituted his own voca- the larger identity question for electronic
bulary and sent Montfort the result. literature. Code can do things and have
things done to it that conventional wri-
ting cannot […]. Digital media lend them-
selves to duplication, encapsulation, and
appropriation more readily than did ear-
lier media […]. [D]ifferences of scale may
be […] very large indeed. At some point,
pronounced variations in degree may be-
come effectively essential. At the heart
of this hyperinflation lies the willful use
of databases, algorithms, and other for-
mal structures of computing.”21
Other than the ease with which code
can be remixed, what else can we say
about the computer as co-creator here?
Obviously, the program understands
nothing about the semantic content of
the vocabulary from which it is selecting
words. Nevertheless, it would be a mis-
take to assert, à la Searle, that the com-
puter only knows how to match patterns
Fig. 2: Screen shot of Scott Rettberg’s Tokyo Garage.
in a brute force kind of way. It knows
the data structures and the syntax pro-
In Rettberg’s version, “Tokyo Garage,” cedures that determine which category
the generator produced such lines as of word it selects; it knows the display
“zombies contaminate the processor” parameters specified in the code (in the
and “saxophonists endure the cherry case of “Taroko Gorge,” continuously rol-
blossoms.” 20
Others took up the game, ling text displayed in light grey font on
and Montfort’s site now hosts over a do- green background); it knows the rando-
zen variants by others. Stuart Moulthrop mizer that determines the word choice;
raises important questions about what it knows the categories that parse the
this technique implies: “The extension of words into different grammatical func-
the text into reinstantiation – the reuse tions; and it knows the pace at which
of code structures in subsequent work the words should scroll down the screen.
– raises questions about the identity of This is far more detailed and complex
particular texts. It also brings into focus

21 Stuart Moulthrop and Dene Grigar, Traversals: The Use of

20 Scott Rettberg, Tokyo Garage; http://nickm.com/tokyo_ga- Preservation for Early Electronic Writing (Cambridge 2017), Kindle
rage/, access: March 26, 2018. version, pp. 37–38.

178
 INTERFACE CRITIQUE JOURNAL – VOL. 2 – 2019

than the “rule book” that Searle imagi- nonhuman forces and human deter-
nes his surrogate consulting to construct mination, just as generative poetry is a
answers to questions in Chinese. Using collaboration between the programmer’s
the philosophical touchstones of “beliefs,” choices and the computer’s randomizing
“desires” and “intentions” that philoso- selections along with its procedural ope-
phers like to cite as the necessary pre- rations.
requisites for something to have agency,
we can say that the computer has beliefs
(for example, that the screen will res- Code Comments
pond to the commands it conveys), desi-
res (fulfilling the functions specified by as Essay:
the program), and intentions (it intends
to compile/interpret the code and execu- Sea and Spar
te the commands and routines specified
there). Although the human writes the Between
code (and other humans have construc-
ted the hardware and software essential My next example is Sea and Spar Bet-
to the computer’s operation), he is not ween, a collaborative project between
in control of the lines that scroll across prize-winning poet Stephanie Strickland
the screen, which are determined by the and Nick Montfort.23 They chose passa-
randomizing function and the program’s ges from Melville’s Moby Dick to com-
processes. bine with Emily Dickinson’s poems. This
What is the point of such generative intriguing project highlights a number
programs? I think of John Cage’s aest- of issues: gender contrasts between the
hetic of “chance operations,” which he all-male society of The Pequod versus
saw as a way to escape from the narrow the sequestered life Dickinson led as a
confines of consciousness and open his near-recluse in Amherst; the sprawling
art to the aleatory forces of the cosmos, portmanteau nature of Melville’s work
at once far greater than the human mind versus Dickinson’s tightly constrained
and less predictable in its results.22 As aesthetic; the spatial oxymorons in each
with generative poetry, a paradox lurks work, for example the claustrophobic
in Cage’s practices: although the para- rendering room aboard the ship versus
meters of his art projects were chosen Dickinson’s famous poem “The Brain is
randomly, he would go to any length to Wider Than the Sky,” and so forth. For
carry them out precisely. The end results this project, much more human selection
can be seen as collaborations between was used than for “Taroko Gorge,” neces-

22 N. Katherine Hayles, The Paradoxes of John Cage: Chaos,

Time, and Irreversible Art in: Permission Granted: Composed in 23 Nick Montfort and Stephanie Strickland, Sea and Spar Between;
America, ed. Marjorie Perloff and Charles Junkerman (Chicago https://nickm.com/montfort_strickland/sea_and_spar_between/;
1994), pp. 226–241. access: March 26, 2018.

179
HAYLES / LITERARY TExTS AS COGNITIVE ASSEMBLAGES

sitated by the massive size of Melville’s gitude” coordinates, both with 14,992,383
text and the much smaller, but still signi- positions, resulting in about 225 trillion
ficant, corpus of Dickinson’s work. stanzas, roughly the amount, they estima-
te, of fish in the sea. The numbers are stag-
gering and indicate that the words dis-
played on a screen, even when set to the
farthest-out zoom position, are only a tiny
portion of the entire conceptual canvas.
The feeling is indeed of being “lost at
sea,” accentuated by the extreme sensi-
tivity to cursor movements, resulting in
a highly “jittery” feel. It is possible, howe-
Fig. 3: Screen shot of Sea and Spar Between, at closest zoom. ver, to locate oneself in this sea of words
Image courtesy of Stephanie Strickland, used with permission.
by entering a latitude/longitude position
provided in a box at screen bottom. This
move will result in the same set of words
appearing on screen as were previously
displayed at that position; conceptually,
then, the canvas pre-exists in its enti-
rety, even though in practice, the very
small portion displayed on the screen
at a given time is computed “on the fly,”
Fig. 4: Screen shot of Sea and Spar Between, taken at medium zoom. because to keep this enormous canvas in
Image courtesy of Stephanie Strickland, used with permission. memory all at once would be prohibitive.
As Stuart Moulthrop points out, “Stanzas
that fall outside the visible range are not
constructed”24. Quoting Strickland, he
observes that “the essence of the work is
‘compression,’ drawing on computation
to reduce impossibly large numbers to a
humanly accessible scale”25.
The effect is a kind of technological
sublime, as the authors note in one of
Fig. 5: Screen shot of Sea and Spar Between, farthest out zoom. their comments: “at these terms they
Image courtesy of Stephanie Strickland, used with permission. signal, we believe, an abundance excee-
ding normal, human scale combined
The project displays as an “ocean” on with a dizzying difficulty of orientation.”
which quatrains appear as couplet pairs.
The authors define locations on the dis- 24 Moulthrop and Grigar, Traversals, p. 35.

play screen through “latitude” and “lon- 25 Ibid.

180
 INTERFACE CRITIQUE JOURNAL – VOL. 2 – 2019

As Moulthrop writes, “‘Sea and Spar Bet- //fragments and generate lines. The remaining
ween’ asks the reader to swim or skim an 5654 bytes (about 50%)
oceanic expanse of language. We do not //deals with the display of the stanzas and with
build, we browse” . Montfort and Strick-
26
interactivity.
land reinforce the idea of a reader lost By contrast, the selection of texts was
at sea in their co-authored essay on this an analog procedure, intuitively guided
work, “Spars of Language Lost at Sea”. by the authors’ aesthetic and literary
They point out that randomness does sensibilities.
not enter into the work until the reader //The human/analog element involved jointly
opens it and begins to read: “It is reader selecting small samples of
of Sea and Spar Between who is deposi- //words from the authors’ lexicons and inven-
ted randomly in an ocean of stanza each ting a few ways of generating
time she returns to the poem. It is you, //lines. We did this not quantitatively, but based
reader, who are random.” 27
on our long acquaintance
An unusual feature is the authors’ es- //with the distinguishing textual rhythms and
say within the source code, marked off as rhetorical gestures of Melville
comments (that is, non-executable state- //and Dickinson.
ments). The essay is entitled “cut to fit the Even so, the template for constructing
toolspun course,” a phrase generated by
28
lines is considerably more complex than
the program itself. The comments make with “Taroko Gorge.” The authors explain,
clear that human judgments played a //We define seven template lines: three first and
large role in text selection, whereas re- four second lines. These
latively more computational power was //line templates and the consequences they in-
expended on creating the screen display volve were designed to evoke
and giving it its characteristic “jerky” //distinctive rhetorical gestures in the source
movements. The authors comment, texts, as judged
//most of the code in Sea and Spar Between is //intuitively by us, and to foreground Dickinson’s
used to manage the strong use of negation.
//interface and to draw the stanzas in the The selections include compound
browser’s canvas region. Only words (“kennings,” as the authors call
//2609 bytes of the code (about 22%) are actu- them) with different rules governing how
ally used to combine text the beginning and ending lines are for-
med:
26 Ibid., p. 36. //butBeginning and butEnding specify words
27 Nick Montfort and Stephanie Strickland, Spars of Language that begin and end
Lost at Sea, p. 8; https://conference.eliterature.org/sites/default/ //one type of line, the butline.
files/papers/Montfort_Strickland__Spars_1.pdf, access: March 26,
To create the compound words, the
2018.
computer draws from two compound ar-
28 Nick Montfort and Stephanie Strickland, cut to fit the toolspun
course; https://elmcip.net/critical-writing/cut-fit-tool-spun-course,
rays and then “joins the two arrays and
access: March 26, 2018. sorts them alphabetically.”

181
HAYLES / LITERARY TExTS AS COGNITIVE ASSEMBLAGES

In this project, what does the com- beliefs, the computer becomes not merely
puter know? It knows the display para- a device to display the results of human
meters, how to draw the canvas, how to creativity but a collaborator in the project.
locate points on this two-dimensional
surface, and how to respond to a user’s re-
quest for a given latitude and longitude. Computers as
It also knows how to count syllables and
what parts of words can combine to form Literary Influen-
compound words. It knows, the authors
comment, how “to generate each type of ces
line, assemble stanzas, draw the lattice of
stanzas in the browser, and handle input One branch of literary criticism, some-
and other events.” That is, it knows when what old-fashioned now, is the “influence
input from the user has been received study,” typically the influence of one wri-
and it knows what to do in response to ter on another. Harold Bloom made much
a given input. What it does not know, of of this dynamic in his classic study, The
course, are the semantic meanings of the Anxiety of Influence: A Theory of Poetry
words and the literary allusions and con- (1973), in which he argued that strong
notations evoked by specific combina- poets struggle against the influence of
tions of phrases and words. Nevertheless, their precursors to secure their place in
the subtlety and scope of the computer’s the literary canon.29 For writers creating
beliefs and intentions far exceed the ste- digital literature, software platforms (and
reotyped “rule book” of Searle’s thought underlying hardware configurations)
experiment. exert a similar insistent pressure, ope-
In reflecting on the larger significance ning some paths and resisting or blo-
of this collaboration, the (human) authors cking others in ways that significantly
outline what they see as the user’s in- shape the final work. To elucidate this
volvement as responder and critic. dynamic, I asked M. D. Coverley (the pen
//Our final claim: the most useful critique name of Marjorie Luesebrink) to descri-
//is a new constitution of elements. On one level, be her process of creating a digital work.
a reconfiguration of a Her account reveals the push-and-pull
//source code file to add comments—by the ori- of software as literary influence (private
ginal creator or by a critic—//accomplishes this email January 20, 2018).
task. But in another, and likely more novel, way, Coverley took as her example a work-
//computational poetics and the code developed in-progress, Pacific Surfliner, inspired
out of its practice by the train that travels to and from San
//produce a widely distributed new constitution. Luis Obispo to Los Angeles to San Diego.
To the extent that the “new constituti-
on” could not be implemented without the 29 Harold Bloom, The Anxiety of Influence: A Theory of Poetry
computer’s knowledge, intentions and (Oxford 1973).

182
 INTERFACE CRITIQUE JOURNAL – VOL. 2 – 2019

She says that she starts “with an idea – sounds, and only then verbal language.
very rough, no text except perhaps a title “Once all the other elements are in place
or a paragraph.” For this work, she wan- – I can see how economical I can be with
ted to include videos of the views from the prose. If something is already evident
the train windows. “In the case of Pacific in the images, sound, videos, etc. then I
Surfliner, I decided to use a simple Roxio need only refer indirectly to that detail in
video-editing program. It outputs mp3 or the actual text.”
mp4 files.” The advantage of simplicity, It is remarkable that Coverley, who
however, is offset by excessive loading began as a novel writer before she tur-
times, a strong negative for digital wri- ned to digital literature, not only places
ters who want to keep users engaged: too words last in her compositional practi-
long a wait, and users are likely to click ce, but also sees them in many ways as
elsewhere. The solution, she writes, “was supplements to the non-verbal digital
to let Vimeo do the compression—but objects already in place. This makes her
then I had to figure out how to get a Vi- practice perhaps more akin to film and
meo file to play on my designated HTML video production than to literary lan-
page.” Note that these negotiations with guage, although of much smaller scope
the software packages precede actual since it can be accomplished by a single
composition practices and definitively creator working alone or perhaps with
shape how it will evolve. “These decisi- one or two collaborators. She remarks, “I
ons,” Coverley acknowledges, “have al- have always been surprised (and deligh-
ready constrained many elements of the ted) at how much descriptive text can be
message.” dispensed with in hypermedia narrative.
Once she has decided on the software, This way of writing is one of the chief
then comes a period of composition, re- joys of the medium for me.” Here is influ-
vision, and exploration of the moving ence at the most profound level, transfor-
images she wants to use. “For a long ming her vision of how narrative works
stretch, I will arrange and rearrange, crop and offering new kinds of rewards that
and edit, expand some ideas, junk others, lead to further creativity and exploration.
maybe start over several times […]. I did The point is not so much the influence of
about 36 versions of the video [for Pacific specific software packages and operating
Surfliner] and that is about standard.” It is platforms, although these are still very
significant that the images and sounds, significant, but rather the larger context
rather than words, come first in her com- in which she sees her work evolving and
positional practices, perhaps because the reaching audiences. To find a comparab-
archive of images and sounds is cons- le context in conventional influence stu-
trained compared to the possibility space dies, we may refer to something as loo-
of verbal expression, which is essentially ming as literary canonization in Bloom’s
infinite. So software first to make sure theory, a driving motivating force that
the project is feasible, then images and in his view propels poets onward in the

183
HAYLES / LITERARY TExTS AS COGNITIVE ASSEMBLAGES

hope of achieving some kind of literary and visual artist Håkan Jonson,30 takes
immortality. (Of course, “immortality” in the computer’s role one step further,
the digital realm is another matter enti- from collaborator to co-creator, or bet-
rely, beset as the field of electronic lite- ter perhaps poetic rival, programmed
rature is with problems of platform ob- to erase and overwrite the words of the
solescence and media inaccessibility.) Heldén’s original. Heldén is a true poly-
Nevertheless, the tantalizing prospect of math, not only writing poetry but also
reaching a large audience without going creating visual art, sculpture, and sound
through conventional publishing gate- art. His books of poetry often contain
ways and the opportunity to experiment images, and his exhibitions showcase
with multimodal compositional practi- his work in all these different media.
ces function in parallel ways to literary Jonson, a computer programmer by day,
canonization, the golden promises that also creates visual and sound art, and
make it all seem worthwhile. And this their collaboration on Evolution reflects
is only possible because of networked the multiple talents of both authors.
and programmable machines. This is The authors write in a preface that the
the large sense in which computers are “ultimate goal” is to pass “‘The Imitation
our symbionts, facilitating and enabling Game’ as proposed by Alan Turing in 1951
creative practices that could not exist in […]; when new poetry that resembles the
their contemporary forms without them. work of the original author is created
or presented through an algorithm, is it
possible to make the distinction between
Computer as ‘author’ and ‘programmer’?”31
These questions, ontological as well as
Co-Author conceptual, are better understood when
framed by the actual workings of the
Sea and Spar Between does not invoke program. In the 2013 version, the authors
any form of artificial intelligence, and input into a database all ten of the then-
differs in this respect from Evolution, extant print books of Hélden’s poetry. A
which does make such an invocation. stochastic model of this textual corpus
Montfort and Strickland make this expli- was created using a statistical model
cit in their comments: known as a Markov Chain (and the cor-
//These rules [governing how the stanzas are responding Markov Decision Process),
created] are simple; there is no elaborate AI ar- a discrete state process that moves ran-
chitecture domly step-wise through the data, with
//or learned statistical process at work here.
By contrast Evolution, a collaborative
30 Johannes Heldén and Håkan Jonson, Evolution (2013); https://
work by Swedish poet Johannes Heldén www.johanneshelden.com/evolution/, access: March 26, 2018;
Johannes Heldén and Håkan Jonson, Evolution (Stockholm 2014).

31 Heldén and Jonson, Evolution (2013).

184
 INTERFACE CRITIQUE JOURNAL – VOL. 2 – 2019

Fig. 6: Screen shot of Evolution, generation 18.

each next step depending only on the pre- functions, for example putting a word
sent state and not on any previous ones. where there was white space originally
This was coupled with genetic algo- (all the white spaces, coded as individual
rithms that work on an evolutionary “letters” through their spatial coordinates
model. At each generation, a family of on the page, are represented in the data-
algorithms (“children” of the previous base as signifying elements).
generation) is created by introducing va- The interface presents as an opened
riations through a random “seed.” These book, with light grey background and
are then evaluated according to some fit- black font. On the left side is a choice
ness criteria, and one is selected as the between English and Swedish and a
most “fit.” In this case, the fitness criteria slider controlling how fast the text will
are based on elements of Heldén’s style; evolve. On the right side is the text, with
the idea is to select the “child” algorithm words and white spaces arranged as if
whose output most closely matches on a print page. As the user watches, the
Heldén’s own poetic practices. Then this text changes and evolves; a small white
algorithm’s output is used to modify the rectangle flashes to indicate spaces un-
text, either replacing a word (or words) dergoing mutation (which might other-
or changing how a block of white space wise be invisible if replaced by another

185
HAYLES / LITERARY TExTS AS COGNITIVE ASSEMBLAGES

Fig. 7: Screen shot of the same run-through of Evolution, generation 40.

space). Each time the program is opened, coffee per episode of Twin Peaks”32. These
one of Heldén’s poems is randomly cho- playful selections mix cultural artifacts
sen as a starting point, and the display be- with terrestrial environmental parame-
gins after a few hundred iterations have ters with astronomical data, suggesting
already happened (the authors thought that the evolutionary process can be lo-
this would be more interesting than st- cated within widely varying contexts.
arting at the beginning). At the bottom of The work’s audio, experienced as a con-
the “page” the number of the generation tinuous and slightly varying drone, is ge-
is displayed (starting from zero, disregar- nerated in real time from sound pieces
ding the previous iterations). that Heldén previously composed. From
Also displayed is the dataset used to this dataset, one-minute audio chunks
construct the random seed. The dataset are randomly selected and mixed in
changes with each generation, and a total using cross-fade, which creates an am-
of eighteen different datasets are used, bient soundtrack unique for each view.33
ranging from “mass of exoplanetary sys-
tems detected by imaging,” to “GISS sur- 32 Heldén and Jonson, Evolution (2014), n.p.
face temperature” for a specific latitude/ 33 Johannes Heldén and Håkan Jonson, The Algorithm, in: Heldén
longitude and range of dates, to “cups of and Jonson, Evolution (Stockholm 2014), n.p.

186
 INTERFACE CRITIQUE JOURNAL – VOL. 2 – 2019

The text will continue to evolve as long tually be – or never be able to be […] read
as the user keeps the screen open, with by any human having the mind and body
no necessary end point or teleology, only to read them.” “Mind and body” repeats, as
the continuing replacement of Heldén’s do “composed/composition,” “troubled,”
words with those of the algorithm. One and “never actually be/never be able,” but
could theoretically reach a point where each time in a new context that slightly
all of Heldén’s original words have been alters the meaning. When Cayley speaks
replaced, in which case the program of being “troubled,” he refers to one of
would continue to evolve its own cons- the crucial differences in embodiment
tructions in exactly the same way as it between human and machine: whereas
had operated on Heldén’s words/spaces. the human needs to sleep, eat, visit the
In addition to being available online, bathroom, the machine can continue
the work is also represented by a limi- indefinitely, not having the same kind
ted edition print book, in which all the of “mind and body” as human writers or
code is printed out. The book also has readers. The sense of excess, of expo-
appendices containing brief commen- nentially larger processes than human
taries by well-known critics, including minds and bodies can contain, recalls
John Cayley, Maria Engberg, and Jesper the excess of Sea and Spar Between and
Olsson. Cayley seems (consciously or gestures toward the new scales possible
unconsciously) to be influenced by the when computational media become co-
ever-evolving work, adopting a style that creators.
evolves through restatements with slight Maria Engberg, in “Appendix 3: Chan-
variations. 34
For example, he suggests ce Operations,” parallels Evolution to the
the work is “an extension of his [Heldén’s] Cageian aesthetic mentioned earlier. She
field of poetic life, his articulated breath, quotes Cayley’s emphasis process over
manifest as graphically represented lin- object. “‘What if we shift our attention,’”
guistic idealities, fragments from poetic Cayley writes, “‘decidedly to practices,
compositions, I assume, that were pre- processes, procedures—towards ways of
viously composed by privileged human writing and reading rather than dwel-
processes proceeding through the mind ling on either textual artifacts themsel-
and body of Heldén and passing out of ves (even time-based literary objects) or
him in a practice of writing. […] I might be the concept underpinning objects-as-
concerned, troubled because I am trou- artifacts?’”35 In this instance, the concept
bled philosophically by the ontology […], underpinning the object is itself a series
the problematic being […], of linguistic ar- of endless processes, displacing, muta-
tifacts that are generated by compositio- ting, evolving, so the distinction between
nal process such that they may never ac- concept and process becomes blurred, if

34 Cf. John Cayley, Appendix 2: Breath, in: Heldén and Jonson, 35 Maria Engberg, Appendix 3: Chance Operations, in: Heldén and
Evolution (2014), n.p. Jonson, Evolution (2014), n.p.

187
HAYLES / LITERARY TExTS AS COGNITIVE ASSEMBLAGES

not altogether deconstructed. artificial and biological consciousness.”39

Jesper Olsson, in “Appendix 4: We How does Heldén feel about his dis/
Have to Trust the Machine,” also sees an re/placement by machinic intelligence?
analogy in Cage’s work, commenting: I had an opportunity to ask him when
“It was not the poet expressing himself. Danuta Fjellestad and I met Heldén, Jon-
He was at best a medium for something son, and Jesper Olsson at a Stockholm
else.” 36
What is this “something else” if restaurant for dinner and a demonstrati-
not machinic intelligence struggling to on of Evolution (private communication,
enact evolutionary processes so that it March 16, 2018). In a comment reminis-
can write like Heldén, albeit without the cent of Cage, he remarked that he felt
“mind and body” that produced the poe- “relieved,” as if a burden of subjectivity
try in the first place? A disturbing analo- had been lifted from his shoulders. He
gy comes to mind: H. G. Wells’ The Island recounted starting Evolution that after-
of Doctor Moreau and the half-human be- noon and watching it for a long time. At
asts who keep asking, “Are we not men?” first, he amused himself by thinking “me”
In the contemporary world, the porous or “not me” as new words appeared on
borderline is not between human/ani- screen. Soon, however, he came to feel
mal but human/machine. Olsson sees that this was not the most interesting
“this way of setting up rules, coding wri- question he could ask; rather, he began
ting programs” as “an attempt to align to see that when the program was “wor-
the subject with the world, to negotiate king at its best,” its processes created
the differences and similarities between new ideas, conjunctions, and insights
ourselves and the objects with which we that would not have occurred to him
co-exist”.37 Machine intelligence has so (this is, of course, from a human point
completely penetrated complex human of view, since the machine has no way
systems that it has become our “nature- to assess its productions as insights or
culture,” as Jonas Ingvarsson calls it.38 He ideas, only as more or less fit according
points to this conclusion when he writes: to criteria based on Heldén’s style). That
“The signs are all over Heldén’s poetic this fusion of human and machine intel-
and artistic output. Computer supported ligence could produce something better
lyrics about nature and environments, than either operating alone, he commen-
graphics and audio paint urbannatural ted, made him feel “joyous,” as if he had
land-and soundscapes […]. We witness helped to bring something new into the
the (always already ongoing) merge of world based on his own artistic and poe-
tic creations but also at times exceeding
36 Jesper Olsson, Appendix 4: We Have to Trust the Machine, in: them. In this sense Evolution reveals the
Heldén and Jonson, Evolution (2014), n.p. power of literature conceived as a cogni-
37 Ibid. tive assemblage, in which cognitions are
38 Jonas Ingvarsson, Appendix 5: The Within of Things, in: Heldén
and Jonson, Evolution (2014), n.p. 39 Ibid.

188
 INTERFACE CRITIQUE JOURNAL – VOL. 2 – 2019

distributed between human and techni- as a programmer, the important part was
cal actors, with information, interpreta- the more accurate description of genetic
tions and meanings circulating throug- algorithms as “population-based meta-
hout the assemblage in all directions, heuristic optimization algorithms”40.
outward from humans into machines, Whether this counts as “artificial intelli-
then outward from machines back to hu- gence” he regarded as a trivial point.
mans. Nevertheless, to answer the skeptic,
we can consider stronger forms of arti-
ficial intelligent such as recurrent neu-
Super(human) ral nets. After what has been described
as the “long winter” of AI when the ear-
intelligence: ly promise and enthusiasm of the 1950s
and 60s seemed to fizzle out, a leap for-
The Potential of ward occurred with the development of
neural networks, which use a system of
Neural Nets nodes communicating with each other
to mimic synaptic networks in human
In several places, Heldén and Jonson de- and animal brains. Unlike earlier versi-
scribe Evolution as powered by artificial ons of artificial intelligence, neural net-
intelligence. A skeptic might respond works are engineered to use recursive
that genetic algorithms are not intel- dynamics in processes that not only use
ligent at all; they know nothing about the output of a previous trial as input for
the semantics of the work and operate the next (that is, feedback), but in addi-
through procedures that are in princip- tion change the various “weights” of the
le relatively simple (acknowledging that nodes, resulting to changes in the struc-
the ways random “seeds” are used and ture of the network itself. This amounts
fitness criteria are developed and applied to a form of learning that, unlike genetic
in this work are far from simple, not to algorithms which use random variation
mention the presentation layers of code). undirected by previous results (because
The power of genetic algorithms derives they rely on Markov chains), use the re-
from finding ways to incorporate evo- sults of previous iterations to change
lutionary dynamics within an artificial how the net functions. Neural nets are
medium, but like many evolutionary pro- now employed in many artificial intel-
cesses, they are not smart in themselves, ligence systems, including machine
any more than are the evolutionary stra- translations, speech recognition, com-
tegies that animals with tiny brains like puter vision, and social networks. Recur-
fruit flies, or no brain at all like nematode rent neural networks (RNN) are a special
worms, have developed through natural class of neural nets where connections
selection. When I asked Jonson about
this objection, he indicated that for him 40 Heldén and Jonson, The Algorithm, sp.

189
HAYLES / LITERARY TExTS AS COGNITIVE ASSEMBLAGES

between units form a directed graph tionally tractable. Whereas AlphaGo was
along a sequence. This allows them to trained on many human-played games as
exhibit dynamic temporal behavior for a examples, its successor uses no human
time sequence. Unlike feedforward neu- input at all, starting only with the basic
ral nets, RNNs have internal memory rules of the game. Then it plays against
and can use it to process inputs, which itself and learns strategies through trial
is particularly useful for tasks where the and error. At three hours, AlphaGoZero
input may be unsegmented (that is, not was at the level of a beginning player,
broken into discrete units) such as face focusing on immediate advances rather
recognition and handwriting. than long-term strategies; at 19 hours
A stunning example of the potenti- it had advanced to an intermediate le-
al of neural net architecture is Alpha- vel, able to evolve and pursue long-term
Go, which recently beat the human Go goals; and at 70 hours, it was playing at a
champions, Lee Sedol in 2016 and Ke Jie superhuman level, able to beat AlphaGo
in 2017. Go is considered more “intuitive” 100 games to 0, and arguably becoming
than chess, having exponentially more the best Go player on the planet.
possible moves, with a possibility space Of course, programs like this succeed
vastly greater than the number of atoms because they are specific to a narrow
in the universe (10 240
moves vs. 10 74
knowledge domain, in this case, the
atoms). With numbers this unimagina- game of Go. All such programs, inclu-
bly large, brute computational methods ding AlphaGoZero, lack the flexibility of
simply will not work—but neural nets, human cognition, able to range across
working iteratively through successive multiple domains, making connections,
rounds of inputs and outputs with a hid- drawing inferences, and reaching con-
den layer that adjusts how the connec- clusions that no existing artificial intel-
tions are weighted, can learn in ways ligence program can match. The race is
that are flexible and adaptive, much as on, however, to develop General Artificial
biological brains learn. Intelligence (GAI), programs that have
Now DeepMind, the company that de- this kind of flexibility and adaptability.
veloped AlphaGo (recently acquired by Many experts in AI expect this goal to be
Google), has developed a new version reached around mid-century, with a 90%
that “learns from scratch,” AlphaGoZe- confidence level.42 In this case, the AI
ro.41 AlphaGoZero combines neural net would combine the best of human intel-
architecture with a powerful search al- ligence with the powers of machine cog-
gorithm designed to explore the Go pos- nition, including vastly faster processing
sibility space in ways that are computa- speeds, much greater memory storage,
and the ability to operate 24/7. There is

41 Deep Mind, AlphaGoZero: “Learning from Scratch“ (2017); htt-

ps://deepmind.com/blog/alphago-zero-learning-scratch/, access: 42 Nick Bostrom, Superintelligence: Paths, Dangers, Strategies
March 26, 2018. (Oxford 2016), S. 23.

190
 INTERFACE CRITIQUE JOURNAL – VOL. 2 – 2019

no guarantee that humans would suc- tes, however, when a fly in the simulated
ceed in developing constraints to keep ointment causes a glitch, leading not to
such an intelligence confined to fol- great apes but gray drapes. Fixing this
lowing human agendas and not pursuing problem, Trurl succeeds in creating a
its own desires for its own ends. 43
multi-story robot that can only produce
It is easy to see how this could be a doggerel. Lem, ever the satirist, recounts
scary prospect indeed, including, as how he finally solves the problem: “Trurl
Stephen Hawking has warned, the end was struck by an inspiration; tossing
of humanity. 44
However, since this is an out all the logic circuits, he replaced
essay on literature and computational them with self-regulating egocentripetal
media, I want to conclude by referring to narcissistors.”46
Stanislaw Lem’s playful fable about what Demonstrating the Electronic Bard for
would happen if such a superintelligence his friend (and sometimes rival const-
took to writing verse. In “The First Sally ructor) Klaupacious, Trurl invites him to
(A), or Trurl’s Electronic Bard”, 45
Trurl (a devise a challenge for the robot versifier.
robot constructor who has no mean in- Klaupacious, wishing his friend to fail,
telligence himself, although with very invents a nearly impossible task: “a love
human flaws) builds a robot versifier se- poem, lyrical, pastoral, and expressed in
veral stories high. Rather than working the language of pure mathematics. Ten-
on a previously written (human) poem, sor algebra mainly, with a little topology
as Evolution does, Trurl re-creates the and higher calculus.”47 Although Trurl
evolutionary process itself. Reasoning objects, the versifier has already begun:
that the average poet carries in his head “Come, let us hasten to a higher plane
the evolutionary history of his civilizati- / Where dyads treat the fairy fields of
on, which carries the previous civiliza- Venn / Their indices bedecked from one
tion and so on, he simulates the entire to n, / commingled in an endless Markov
history of intelligent life on earth from chain!”48 So the Markov chain surfaces
unicellular organisms up to his own again, although to be fair, it is far, far ea-
culture, descendants of the preceding sier to imagine such a versifier in words
human civilization. Something goes than to create it through algorithms that
wrong with the emergence of the prima- actually run as computational processes!
There follow scenarios reminiscent of
43 Ibid; David Roden, Posthuman Life: Philosophy at the Edge of the predictions of those worried about
the Human (New York and London 2014). superintelligence, although in a fanciful
44 Rory Cellan-Jones, Stephen Hawking Warns Artificial Intelli- vein. The Electronic Bard crosses “lyrical
gence Could End Mankind. BBC News, December 2, 2014; http:// swords” with all the best poets: “The ma-
www.bbc.co.uk/news/technology-30290540, access: March 26,
2018.
46 Ibid., p. 46.
45 Stanislaw Lem, The First Sally (A), or Trurl’s Electronic Bard,
47 Ibid., p. 52.
in: Stanislaw Lem, The Cyberiad: Fables for a Cybernetic Age (New
York 2014), pp. 43–57. 48 Ibid.

191
HAYLES / LITERARY TExTS AS COGNITIVE ASSEMBLAGES

chine would let each challenger recite, We may suppose that this fanciful
instantly grasp the algorithm of his ver- extrapolation of Evolution is “all too far
se, and use it to compose an answer in away” to bother us, so we plunge back
exactly the same style, only two hundred into our present reality when compu-
and twenty to three hundred and forty- tational media are struggling merely to
seven times better.” The Electronic Bard
49
come close to simulating human achie-
enacts the same kind of procedure ani- vements. Lem’s fable does not quite va-
mating Evolution, but vastly accelerated, nish altogether, however, suggesting
the faux precision underscoring its ab- that even the most vaulted preserve of
surdity. human consciousness, sensitivity, and
Just as critics warn that a superin- creativity – that is, lyrical poetry – is not
telligence could outsmart any human necessarily exempt from machine col-
constraints on its operation, so the Elect- laboration, and yes, even competition.53
ronic Bard disarms every attempt to dis- By convention, symbionts are regarded
mantle it with verses so compelling they as junior partners in the relationship,
overwhelm its attackers, including Trurl. like the bacteria that live in the human
The authorities are just about to bomb it gut. We are now on the verge of develop-
into submission when “some ruler from ments that promote our computational
a neighboring star system came, bought symbionts to full partnership in our liter-
the machine and hauled it off” . When
50
ary endeavors. The trajectory traced here
supernovae begin “exploding on the sou- through electronic literature demonstra-
thern horizon,” rumors report that the ru- tes that the dread with which some an-
ler, “moved by some strange whim, had ticipate this future has a counterforce in
ordered his astroengineers to connect the creative artists and writers who see
the electronic bard to a constellation of in this prospect occasions for joy and re-
white supergiants, thereby transforming lief.
each line of verse into a stupendous so- Whatever one makes of this posthu-
lar prominence; thus the Greatest Poet man future, it signals the end of the era
in the Universe was able to transmit its when humans could regard themselves
thermonuclear creations to all the illi- as the privileged rational beings whose
mitable reaches of space at once” . The 51
divine inheritance was dominion over
scale now so far exceeds the boundaries the earth. The complex human-techni-
of (human and robot) life, however, that it cal systems that now permeate the inf-
paradoxically fades into insignificance: rastructure of developed societies point
“it was all too far away to bother Trurl.” 52
toward a humbler, more accurate picture
of humans as only one kind of cognizers
49 Ibid., p. 54.

50 Ibid., p. 56. 53 For an analysis of a posthuman strain within the lyric, see:
Sumita Chakraborty, Signs of Feeling Everywhere: Lyric Poetics,
51 Ibid., p. 57.
Posthuman Ecologies, and Ethics in the Anthropocene. Dissertation
52 Ibid. (Atlanta 2018).

192
 INTERFACE CRITIQUE JOURNAL – VOL. 2 – 2019

among many. In our planetary ecology,

co-constituted by humans, nonhumans References
and technical devices, we are charged
with the responsibility to preserve and Bloom, Harold, The Anxiety of Influence:
protect the cognitive capabilities that all A Theory of Poetry (Oxford: Oxford Uni-
biological lifeforms exhibit, and to res- versity Press, 1973).
pect the material forces from which they Bostrom, Nick, Superintelligence: Paths,
spring. If we are to survive, so must the Dangers, Strategies (Oxford: Oxford Uni-
environments on which all cognition ul- versity Press, 2016),
timately depends. Cayley, John, Appendix 2: Breath, in: Jo-
hannes Heldén and Håkan Jonson, Evo-
lution (Stockholm: OEI Editör, 2014), n.p.
Cellan-Jones, Rory, Stephen Hawking
Warns Artificial Intelligence Could End
Mankind. BBC News, December 2, 2014;
http://www.bbc.co.uk/news/technolo-
gy-30290540, access: March 26, 2018.
Chakraborty, Sumita, Signs of Feeling Eve-
rywhere: Lyric Poetics, Posthuman Eco-
logies, and Ethics in the Anthropocene.
Dissertation (Atlanta: Emory University,
2018).
Clark, Andy, Being There: Putting Brain,
Body and World Together Again (Cam-
bridge, MA: Bradford Books, 1998).
Coverley, M. D, private email communi-
cation, January 20, 2018.
Damasio, Antonio, The Feeling of What
Happens: Body and Emotion in the Ma-
king of Consciousness (Boston: Mariner
Books, 2000).
Deep Mind, AlphaGoZero: “Learning from
Scratch“ (2017); https://deepmind.com/
blog/alphago-zero-learning-scratch/, ac-
cess: March 26, 2018.
Deleuze, Gilles and Félix Guattari, A
Thousand Plateaus: Capitalism and Schi-
zophrenia, transl. by Brian Massumi
(Minneapolis: University of Minnesota
Press, 1987).

193
HAYLES / LITERARY TExTS AS COGNITIVE ASSEMBLAGES

Edelman, Gerald and Giulio Tononi, A Uni- holm: OEI Editör, 2014), n.p.
verse of Consciousness: How Matter Latour, Bruno, Reassembling the Soci-
Becomes Imagination (New York: Basic al: An Introduction to Actor-Network-
Books, 2001). Theory (Oxford: Oxford University Press,
Engberg, Maria, Appendix 3: Chance Ope- 2007).
rations, in: Johannes Heldén and Håkan Lem, Stanislaw, The First Sally (A), or
Jonson, Evolution (Stockholm: OEI Editör, Trurl’s Electronic Bard, in: Stanislaw
2014), n.p. Lem, The Cyberiad: Fables for a Cyber-
Funkhouser, Christopher T., Prehistoric Di- netic Age (New York: Penguin, 2014), pp.
gital Poetry: An Archaeology of Forms, 43–57.
1959–1995 (Tuscaloosa, AL: University of Montfort, Nick, Taroko Gorge; https://
Alabama Press, 2007). nickm.com/taroko_gorge/, access: March
Hayles, N. Katherine, How We Became 26, 2018.
Posthuman: Virtual Bodies in Cyberne- Montfort, Nick and Stephanie Strickland,
tics, Literature, and Informatics (Chicago: cut to fit the toolspun course; https://
University of Chicago Press, 1999). elmcip.net/critical-writing/cut-fit-tool-
Hayles, N. Katherine, The Paradoxes of spun-course, access: March 26, 2018.
John Cage: Chaos, Time, and Irreversible Montfort, Nick and Stephanie Strickland,
Art in: Permission Granted: Composed in Sea and Spar Between; https://nickm.
America, ed. Marjorie Perloff and Charles com/montfort_strickland/sea_and_
Junkerman (Chicago: University of Chi- spar_between/; access: March 26, 2018.
cago Press, 1994), pp. 226–241. Montfort, Nick and Stephanie Strickland,
Hayles, N. Katherine, Unthought: The Po- Spars of Language Lost at Sea; https://
wer of the Cognitive Nonconscious (Chi- conference.eliterature.org/sites/default/
cago: University of Chicago Press, 2017). files/papers/Montfort_Strickland__
Heldén, Johannes and Håkan Jonson, Evo- Spars_1.pdf, access: March 26, 2018.
lution (2013); https://www.johanneshel- Moulthrop, Stuart and Dene Grigar, Traver-
den.com/evolution/, access: March 26, sals: The Use of Preservation for Early
2018. Electronic Writing (Cambridge, MA: MIT
Heldén, Johannes and Håkan Jonson, Evo- Press, 2017), Kindle version.
lution (Stockholm: OEI Editör, 2014). Olsson, Jesper, Appendix 4: We Have to
Heldén, Johannes and Håkan Jonson, The Trust the Machine, in: Johannes Heldén
Algorithm, in: Johannes Heldén and Hå- and Håkan Jonson, Evolution (Stock-
kan Jonson, Evolution (Stockholm: OEI holm: OEI Editör, 2014), n.p.
Editör, 2014), n.p. Rettberg, Scott, Tokyo Garage; http://
Hutchins, Edwin, Cognition in the Wild nickm.com/tokyo_garage/, access:
(Cambridge, MA: MIT Press, 1996). March 26, 2018.
Ingvarsson, Jonas, Appendix 5: The Robinson, Kim Stanley, Blue Mars (New
Within of Things, in: Johannes Heldén York: Spectra, 1997).
and Håkan Jonson, Evolution (Stock- Robinson, Kim Stanley, Green Mars (New

194
 INTERFACE CRITIQUE JOURNAL – VOL. 2 – 2019

York: Spectra, 1995).

Robinson, Kim Stanley, Red Mars (New
York: Spectra, 1993).
Roden, David, Posthuman Life: Philoso-
phy at the Edge of the Human (New York
and London: Routledge, 2014).
Searle, John, Minds, Brains, and Pro-
grams. Behavioral and Brain Sciences 3
(1980), pp. 417–457.
Tenen, Dennis, Plain Text: The Poetics of
Computation (Stanford: Stanford Univer-
sity Press, 2017).
Thoreau, Henry David, Walden; or, Life in
the Woods (Boston: Ticknor and Fields,
1854).
Varela, Francisco J., Eleanor Rosch, and
Evan Thompson, The Embodied Mind:
Cognitive Science and Human Experi-
ence (Cambridge, MA: MIT Press, 1992).

195