Digitale Code and Literary Text (paper)
Florian Cramer
%% digital texts in Erfurt, Germany - hope it's of interest to some
%% of you. - Florian
\documentclass[oneside,english,12pt]{amsart}
\usepackage{pslatex}
\usepackage[T1]{fontenc}
\usepackage[latin1]{inputenc}
\usepackage{babel}
\usepackage{url}
\setcounter{secnumdepth}{0}
\setlength\parskip{\medskipamount}
\setlength\parindent{0pt}
\sloppy
\begin{document}
\title{Digital Code and Literary Text}
\author{Florian Cramer}
\date{Sept. 27, 2001}
Can notions of text which were developed without electronic texts in
mind be applied to digital code, and how does literature come into play
here?
\maketitle
\section{Code}
In his abstract for this conference, John Cayley takes a position which
to some extent seems to be the opposite to mine. It's exciting
that we will have a debate at this conference, and so I would
like to make my point and clear up my basic assumptions about the term
``code''.
Since computers, the Internet and all digital technologies are based on
zeros and ones, they are based on code. Zeros and ones are an alphabet
which can be translated forth and back between other alphabets without
information loss. The Internet and computers run on alphabetic code,
whereas, for examples, images and sound can only be digitally stored
when translating them into code, which -- unlike the translation of
conventional text into digital bits -- is a lossy, that is, not fully
reversible and symmetric translation. In digital systems, literature is
a privileged symbolic forms for this very reason. We may automatically
search a collection of text files for all occurences of the word
``bird'', but doing the same with birds in a collection of image files
or bird songs in a collection of audio files is incomparably tricky and
error-prone, relying on either artificial intelligence algorithms or
manual indexing, both of which are methods to translate writing (pixel
code) into writing (descriptions).
The reverse is true as well: We can perfectly translate digital data and
algorithms into non-digital media like print books, as long as we
translate them into signs coded according to the logic of an alphabet.
This is what is done, for example, in programming handbooks or in
technical specification manuals for Internet standards. Today there are
two notorious examples of a forth-and-back translation between print and
computers:
\begin{enumerate}
\item The sourcecode of Phil Zimmerman's cryptography program ``Pretty
Good Privacy'' (PGP). The PGP algorithms were legally considered a
weapon and therefore became subject to U.S. export restrictions. To
circumvent this ban, Zimmerman published the PGP sourcecode in a book.
Unlike algorithms, literature is covered by the U.S. First Amendment of
free speech. So the book could be exported outside the United States
and, by scanning and retyping, translated back into an executable
program.
\item The sourcecode of DeCSS, a small program which breaks the
cryptography scheme of DVD movies. Since U.S. jurisdiction declared
DeCSS an ``illegal circumvention device'' according to the new Digital
Millennium Copyright Act (DMCA), the ban equally affected booklets,
flyposters and t-shirts on whom the DeCSS sourcecode was printed.
\end{enumerate}
The fact that code is speech has been stressed by programmers again and
again, and also forms the core argument of Lawrence Lessig's legal
theory of the Internet (\cite{lessig:code}). It is, strictly speaking,
sloppy terminology to speak of ``digital media''. There actually is no
such thing as digital media, but only digital information.
An average contemporary personal computer uses magnetic disks (floppy
and hard disks), optical disks (CD-ROM and DVD-ROM) and chip memory
(RAM) as its storage media, and electricity or fiber optics as its
transmission media. Theoretically, one could build a computer with a
printer and a scanner which uses books and alphabetic text as its
storage media.\footnote{Such a machine would operate slower than with
magnetical or optical media, but on the other hand provide more robust
and durable information storage} Alan Turing showed that no electronics
are needed to build a computer; the Boston Computer Museum even features
a mechanical computer built from wood.
Juxtapositions of ``the book'' and ``the computer'' are quite
misleading, because they confuse the storage media (paper versus a
variety of optical, magnetical and electronical technologies) with the
information (alphabetical text versus binary code). It further ignores,
by the way, the richness of storage and transmission media in
traditional literature which, aside from the book, include oral
transmission and mental storage, audio records and tapes, the radio, to
name only a few.
If there is, strictly speaking, no such thing as digital media, there
also is, strictly speaking, no such thing as digital images or digital
sound. What we refer to as a ``digital image'' actually is a piece of
code containing the machine instructions to produce the flow of
electricity with which an analog screen or an analog printer is made to
display an image.\footnote{Normally, this code is divided into three
pieces, one -- the so-called sound or image file -- containing the
machine-independent and program-independent abstract information, the
second -- the so-called display program -- containing the instructions
to mediate the abstracted information in a machine-independent, yet not
program-independent format to the operating system, the third -- the
so-called operating system --, mediating the program output to the
output machine, whether a screen or a printer. But these three code
layers are nothing but arbitrary conventions. Theoretically, the
``digital image'' file could in itself contain all the code necessary to
make itself display on analog end devices} Of course it is important
whether a sequence of zeros and ones translates, into, say, an image
because that defines its interpretation and semantics. The point of my
formalistic argumentation is not to deny this, but to underline that
\begin{enumerate}
\item when we speak of ``multimedia'' or ``intermedia'' in conjunction
with computers, digital art and literature, we actually don't speak of
digital systems as themselves, but about translations of digital
information into analog output and vice versa;
\item text and literature highly are privileged symbolic systems in
these translation processes because (a) they are coded and
(b) computers run on a code.
\end{enumerate}
Literature and computers first meet where alphabets and code, human
language and machine language intersect, secondly in the interfacing of
analog devices through digital control code. While of course we cannot
think of code without media because we can't read it without them, the
computer does not really extend literary media as such. All those output
media -- electricity, electrical sound and image transmission etc. --
existed before and without computers and digital information.
So I have to correct the position I presented last year at this
conference: If we speak of digital poetry, or of computer network
poetry, we don't have to speak of certain media, and we don't even have
to speak of certain machines. If computers can be built from broomsticks
(and networked via shoestrings); if any digital data, including
executable algorithms, can be printed in books and from them read back
into machines or, alternatively, executed in the mind of the reader, ther=
e
is no reason why computer network poetry couldn't or shouldn't be
printed as well in books.
Perhaps the term of digital ``multimedia'' -- or better: ``intermedia''
-- would be more helpful if we redefine it as the \emph{the possibility
to losslessly translate information from one sign system to the other,
forth and back, so that the visible, audible or tacticle representation
of the information becomes transitory}. Which can't be achieved unless
the information isn't coded in some kind of alphabet, whether
alphanumerical, binary, hexadecimal or, if you like, Morse code.
\section{Literature}
\subsection{Synthesis: putting things together}
When we observe the textual codedness of digital systems, there of
course is the danger of generalizing and projecting one's observations
of digital code onto literature as a whole. Computers operate on machine
language, which is syntactically far less complex than human everyday
language. The alphabet of both machine and human language is
interchangeable, so that ``text'' -- if defined as a conglomeration of
alphabetical signifiers -- remains a valid descriptor for both machine
code sequences and human writing. In syntax and semantics however,
machine code and human writing are not interchangeable. Computer
algorithms are, like logical statements, a formal language and thus only
a restrained subset of language as a whole.
However, I believe it is a common mistake to claim that machine language
would be only readable to machines and hence irrelevant for human art
and literature and, vice versa, literature and art would be unrelated to
formal languages.
It is important to keep in mind that computer code, and computer
programs, are not machine creations and machines talking to themselves,
but written by humans.\footnote{No computer can reprogram itself;
self-programming is only possible within a limited framework of game
rules written by a human programmer. A machine can behave differently
than expected, because the rules didn't foresee all situations they
could create, but no machine can overwrite its own rules by itself.} The
programmer-artist Adrian Ward suggests that we put the assumption of the
machine controlling the language upside down:
\begin{quotation}
``I would rather suggest we should be thinking about embedding our own
creative subjectivity into automated systems, rather than naively trying
to get a robot to have its `own' creative agenda. A lot of us do this
day in, day out. We call it programming.''\footnote{quoted from an
E-Mail message to the ``Rhizome'' mailing list, May 7, 2001}
\end{quotation}
Perhaps one also could call it composing scores, and it does not seem
accidental to me that musical artists have picked up and grasped
computers much more thoroughly than literary writers. Western music is
an outstandig example of an art which relies upon written formal
instruction code. Self-reflexive injokes such as ``B-A-C-H'' in Johann
Sebastian Bach's music, the visual figurations in the score of Erik
Satie's ``Sports et divertissements'' and finally the experimental score
drawings of John Cage shows that, beyond a merely serving the artwork,
formal instruction code has an aesthetic quality and complexity of its
own. In many works, musical composers have shifted instruction code from
classical score notation to natural human language. A seminal piece, in
my opinion, is La Monte Youngs ``Composition No.1 1961'' which simply
consists of the instruction ``Draw a straight line and follow
it.''\footnote{\cite{hundertmark:maciunas}, no page numbering} Most
Fluxus performance pieces were written in the same notation style. Later
in 1969, the American composer Alvin Lucier wrote his famous ``I am
sitting in a room'' as a brief spoken instruction which very precisely
tells to perform the piece by playing itself back and modulating the
speech through the room echoes.
In literature, formal instructions is the necessary prerequisite of all
permutational and combinatory poetry, which I spoke about last year.
Kabbalah and magical spells are important examples as well. But even in
a conventional narrative, there is an implict formal instruction of how
-- i.e. in which sequence -- to read the text (which maybe or followed
or not, as opposed to hypertext which offers alternative sequence on the
one hand, but enforces its implicit instruction on the other). Grammar
itself is an implicit, and very pervasive formal instruction code.
Although formal instruction code is, as I said, only a subset of
language, it is nevertheless at work in all speech and writing.
But what seems remarkable about computing to me is that the namespace of
executable instruction code and nonexecutable code is flat. One cannot
tell from a snippet of digital code whether it is executable or not.
This property does not stand out in the alphabet of zeros and ones, but
is solely dependent on how another piece of code -- a compiler, a
runtime interpreter or the embedded logic of a microprocessor --
processes it. Computer code therefore is highly recursive and highly
architectural, building upon layers of layers of code.
\subsection{Analysis: taking things apart}
The fact that one cannot tell from any piece of code whether it is
machine-executable or not after all is the principle of all E-Mail
viruses on the one hand and of the net poetry of jodi, antiorp/Netochka
Nezvanova, mez, Ted Warnell, Alan Sondheim, Kenji Siratori and others
that pretends to be viral machine code on the other.
I would not attempt to make a theoretical point for the digital poetry
as code poetry here if it wasn't backed up by others' artistic practices
and my own aesthetic preferences in net poetry.
I think the ``codeworks'' (to borrow from Alan Sondheim) of these
writers and programmer-artists are prime examples for a digital poetry
which reflects the intrisic textuality of the computer. But it does so
not by writing, to quote Alan Turing via Raymond Queneau, computer
poetry to be read by computers\footnote{\cite{queneau:poemes}, p.3}, but
by playing with the confusions and thresholds of machine language and
human language, and by reflecting the cultural implications of these
overlaps. The ``mezangelle'' poetry of mez/Mary Ann Breeze, which mixes
programming/network protocol code and non-computer language to a
portmanteau-word hybrid, is an outstanding example of such a poetics.
In comparison to earlier poetics of formal instruction, like in La Monte
Young's \emph{Composition 1961}, in Fluxus pieces and in permutational
poetry, an important difference can be observed: The Internet code poets
do not construct or synthesize code, but they use code or code grammars
they found and take them apart. I agree with Friedrich Block and the
theses he wrote for this conference that digital poetry should be
regarded in context of experimental poetry and its history. A poetics of
synthesis was characteristic of combinatory and instruction-based
poetry, a poetics of analysis characterized Dada and its successors. But
one hardly finds poetry with analytical approach to formal instruction
code in the classical 20th century avant-garde.\footnote{An exception
being the the ALGOL computer programming language poetry written by the
Oulipo poets François le Lionnais and Noël Arnaud in the early 1970s,
see \cite{oulipo-compendium}, p.47} Internet code poetry is being
written in a new -- if you like, postmodern -- condition of machine code
abundance and overload.
When I said that there is no such thing as digital media and that
digital code may be stored in any medium, it comes to no surprise that
the codework poetry excellently verifies this hypothesis. Unlike
hard-coded hypertext and multimedia poetry, most of the artists I
mentioned prefer to write plain ASCII text. This also reveals the
critical implication of its poetics and aesthetics. The poetics of
hyperfiction and multimedia poetry ran more or less parallel to the
construction of the World Wide Web; hyperfiction authors rightfully saw
themselves as its pioneers and, in the course of nineties, continued to
push the technical limits of both the Internet and multimedia computer
technology. But since much digital art and literature became testbed
applications for new browser features and multimedia plugins, it
simultaneously locked itself into non-open, industry-controlled code
formats, thereby playing a partly affirmative role in the proprietary
reformatting of the Internet.
Shifting the focus of the reader back from slick multimedia interfaces
to raw code, code poetry appears to have strong aesthetical and
political affinities to hacker cultures. While hacker cultures are far
more diverse than the singular term ``hacker'' suggests\footnote{Boris
Gröndahl's (German) Telepolis article ``The Script Kiddies Are Not
Alright'' gives an excellent outline of the multiple camps associated
with the term ``hacker'',
\url{http://www.heise.de/tp/deutsch/html/result.xhtml?url=/tp/deutsch/i=
nhalt/te/9266/1.html}},
hackers could as well be distinguished between those who put things
together -- like Free Software and demo programmers -- and those who
take things apart -- like crackers of serial numbers and communication
network hackers from YIPL/TAP, Phrack, 2600 and German Chaos Computer
Club schools. Code poets have factually adopted many poetical forms
that were originally developed by various hacker subcultures from the
1970s to the early 1990s, including ASCII Art, code slang (like ``7331
wAr3z d00d'' for ``leet [=elite] wares dood'') and poetry in programmin=
g
languages (such as Perl poetry), or even belong to both the ``hacker''
and the ``art'' camp, like my fellow conference speaker Walter van der
Cruijsen from \url{http://www.desk.org}desk.org and the ASCII Art
Ensemble.
From its beginning on, conceptualist Net.art engaged in a critical
politics of the Internet and its code, and continues to be closely
affiliated with critical discourse on net politics in such forums as the
``Nettime'' mailing list. In its aesthetics, poetics and politics,
codework poetry departs from Net.art rather than from hyperfiction and
its historical roots in the Brown University literature program.
To resolve the title of my paper ``Digital Code and Literary Text'', I
would like to strongly argue in favor of considering both to be related
and intertwined. Given that literary text, and not digital code, is the
reference measure, one can subscribe to this without, as John Cayley's
abstract suggests, having to subscribe to Friedrich Kittler's
techno-determinist media theory; a theory which, in my opinion,
exemplifies the metaphysical trap Derrida described in ``Écriture et
différence'': By replacing one metaphysicial center (in Kittler's case:
``Geist''/spirit, ``Geistesgeschichte''/intellectual history and
``Geisteswissenschaft''/humanities ) with another one -- technology,
history of technology and technological discourse analysis -- it writes
on metaphysics under a different label, contrary to its own claim to
have rid itself from it entirely.
The subtitle of this text addresses an open question: ``Can notions of
text which were developed without electronic texts in mind be applied to
digital code, and how does literature come into play here?'' For the
time being, I can answer this question provisionally at best: While all
literature should teach us to read and deal with the textuality of
computers and digital poetry, computers and digital poetry might teach
us to pay more attention to codes and control structures coded into all
language. My list of musical compositions and literary forms is
fragmentary and needs to be extended. In more general terms, program
code contaminates in itself two concepts which are traditionally
juxtaposed and unresolved in modern linguistics: the structure, as
conceived of in formalism and structuralism, and the performative, as
developed by speech act theory.
\bibliographystyle{alpha}
\bibliography{diss}
\end{document}
--
http://userpage.fu-berlin.de/~cantsin/homepage/
http://www.complit.fu-berlin.de/institut/lehrpersonal/cramer.html
GnuPG/PGP public key ID 3200C7BA
"c u in he][l][avan" (mez, _Viro.Logic Condition][ing][ 1.1_)