THE MACHINE READING SERIES:
organized by Charles Bernstein and Nick Montfort
hosted by the Kelly Writers House
at Penn University - Philadelphia.
*****MAIN RESSOURCES**********************************************
Kelly Writers House:
http://www.writing.upenn.edu/~wh/
Charles Bernstein:
http://www.writing.upenn.edu/bernstein/
Nick Montfort:
http://nickm.com/
**********************************************************************
Wednesday February 15th: with Loss Pequeno Glazier and Jim Carpenter
Charles Bernstein and Nick Montfort regularly invite poets over at Penn
University for a reading held at the Kelly Writer House. This Wednesday
February 15th, 2006, they invited digital poet Loss Pequeno Glazier and
Jim Carpenter. Also, for the first time this hosting was christened
"Machine Reading Series", signaling a focus on computer form of
experiments in poetry.
Charles Bernstein introduces the event under the auspices of Gertrude
Stein, asking about the difficult relationship between the readership
and experimental text works. That forces the question: What does that
mean? New forms of poetry "revealed" to the public are often received
with a shock (the structure of poetic revolutions?), and digital poetry
more or less faces the same reception problems.
The two invited poets are among the pioneers who have (or still are)
moving poetry digital medium. C. Bernstein here reminded the audience
that experiments with language should be thought about within the
context of databases, methods of classification, archiving,
implementing formats. Writing is as much about creating as about
protecting, keeping tracks of written traces.
The Electronic Poetry Center, directed by Loss Pequeno Glazier (
http://epc.buffalo.edu/),
one of the most extensive resource database about digital writings, was
cited as an example of a long-term involvement with literature but also
with finding new formats of publication, distribution, and access. This
website was built page by page in a very low-fi way, and has always
been up and running since its creation in 1996. A practice of writing
code and links as an alternative to prefabricated corporate formats
Loss Pequeno Glazier was first working in the librarian field. C.
Bernstein emphasized L. P. Glazier's early interest in practices of
publishing: his first book was on the subject of the Mimeo Revolution,
when the poets in the 60s re-invented the DIY (Do It Yourself)
practice by publishing at low cost as little as 2 pages long poetry
reviews (a "Little history of the Mimeo Revolution" can be found here:
http://www.granarybooks.com/books/clay/clay4.html).
Then Glazier developed as a PHD project the Electronic Poetry Center,
and published one of the first book on the subject of computer poetry:
Digital Poetics: Hypertext, Visual-Kinetic Text and Writing in
Programmable Media, University of Alabama Press, 2001.
Jim Carpenter has started as a high-school teacher and is now a
lecturer in computer programming and systems design in the Dept. of
Operations and Informations Management at the Wharton School at Penn
University. He developed the Electronic Text Composition (ETC) project (
http://slought.org/content/11207/).
J. Carpenter specializes in digital and computer applications,
implementing tools for poetry composition. His work might help change
the definition of poetry itself demanding an involved participation
form the public, not in the sense of mere interactivity but by
demanding a knowledge of the tools at stake, avoiding the transaction
intermediary (formats)
More accurate information about these persons can be found on the Kelly Writer House website and here specifically:
http://www.writing.upenn.edu/wh/calendar/0206.html#15
*--*--*--*--*--*--*
*--*--*--*--*--*--*
*--*--*--*--*--*--*
*--*--*--*--*--*--*
The readings
Jim Carpenter
poem 1: "Mystic Writing Pad"
--> a computer reading structured as a sound triptych: three
computers (2 feminine voices, 1 masculine: superimposed) uttering words
and fragments of sentences. These words appear to be thematic keywords
such as performance, evolution, speech, work, write, links, share,
distributed, explore, or aesthetics concern (though without a
reference): good, bad... Some semi-articulated sentences: "Whatever is
supposed public land of number ?", "Write the facts".
While these are being formaluted into computer voices, their symbols as
strings of letters appear in different fonts, sizes and colors on a
minimalistic interface (grey or similar colored plain page)
Jim Carpenter launched the application and put an end to it in a
seemingly arbitrary way. The first thing he said after the multi-vocal
poem stopped was that he was a conceptual artist. Though J.Carpenter
said almost everything in an ironic tone, I will write down what I
remember (almost) as if there was not too much staging in that reading
(it is difficult, because there was much staging).
This text-generation software has been conducted for almost 7 years. It is structured on 2 main supports:
- an extensive language database that incorporates words of the British National Corpus (
http://www.natcorp.ox.ac.uk/
- here I remember that the voices were definitely identifiable as
British), each tagged (semantic relationships) and re-organized in what
Carpenter calls "bigrams". The output is determined by frequency of use.
- a series of software modules structured as trees and based on the
principles of transformational grammar (as opposed to context
free grammar). This allows the use of standard English to be
rationalized. There is no input string but a generation from an initial
semantic ceiling.
The process is repetitive and results in leaving traces: Carpenter
calls this a "geodesic structure", as an analytic analysis of how code
and semantic relation support each other.
It was first a research project that asked: is it possible for a
generated text to compete with a 'normal text' (human words). Here
Carpenter stresses out that there are a lot of text generators out on
the internet but that a lot of them don't work. He also wants the
machines (British accent) to read the poems because he cannot claim
authorship himself (cf. the post on the cyborg author).
Also, he introduces the notion of "competitiveness" opposed to the
notion of "quality". Competitiveness points out to the market of
literary reviews: a competitive poem is one that can compete (can have
the same value) with the standards of these reviews. I infer from this
that a "quality" poem is one that is out of the market standards - I
guess it has to be ahead of these standards. Also, if the competitive
poems are judged in regard to external properties (the correspondence
to standards. i.e. topics, genre, figures, forms), the "quality poem"
should have inherent properties that make it good. These definitions
seem to bring us back to the idea of the poem as a tautologic object:
their quality is what define them its own properties.
Jim Carpenter states that if the poems he produces with his language
engine are not based on quality but on competitiveness, the act of
programming, in contrary, is an aesthetic experience, thus it is based
on quality. I am not sure if this theoretical distinction is really
helpful, because once again this aesthetic experience is based on
values of good and bad that are effective in a competition world. But I
discuss this later.
The poems J. Carpenter is going to read are all implementing an idea of
what is competitive according to J. Carpenter (I would say, he made
good pastiches). But there are also allow multiple readings (a poem can
be feminist & modernist for instance). Carpenter has only edited
lightly the poems: punctuation and particles.
- poem 2: "Nuclear"; a good example to show how the ETC language
machine is good at paratactic structures. This is a good point for
competitiveness as paratax is one of the favorite topics of
postmodernist poetry: it allows the affective description to be
disconnected from the subject (or what is left of it).
- poem 3: "Peripheral"; here the machine performs a tonal capacity
- poem 4: "With a heartfelt unhappy sigh": here, within the range of
romantic modernism (very competitive according to J. Carpenter)
- poem 5 : "Bullshit": an experiment with obscenity words; because they are a minority in the corpus, the program run faster.
Jim Carpenter decided to submit these poems for evaluation, as a
possible test to answer the question: What is the test to prove the
value of research?
In order to verify this, he thought of different procedures in the submission process:
- handing out texts as machine-written to be read and judged by a
fellow poet? --> dismissed because the poet could be too kind too
really evaluate them
- read out loud texts as human-written in a poetry workshop? --> dismissed for the same reason
- send texts to poetry reviews? it worked.
These texts were published under the name Erika T. Carter, an amalgam
of the machine and J.Carpenter, a "bio-mix". J. Carpenter insists on
the fact that they are 2 personalities at work, not conflicting but
influencing each other. For example, he thought that he would be good
that the machine-poet Ericka should be a lesbian precisely because of
the inability of its doppelganger human-poet Jim to conceive a love
relationship with males.
Jim Carpenter ends by talking about programming as an abstraction =
"language spread out into pure signs". Opposed to this, the human
poet's task is to concretize, to inject some flesh into the skeleton
(machine-poet) he created.
~-~-~-~-~-~-~-~-~-~-~-
~-~-~-~-~-~-~-~-~-~-~-
~-~-~-~-~-~-~-~-~-~-~-
~-~-~-~-~-~-~-~-~-~-~-
~-~-~-~-~-~-~-~-~-~-~-
Loss Pequeno Glazier
Glazier admitted as well having always had problems with publishing in
literary reviews and magazines. But he was more frustrated by the fact
that whenever a poem was published, he, as a poet, was urged to change
it, but could not do it. With the digital medium, he could experiment
on that.
Loss Pequeno's homepage on the Electronic Poetry Center website is full of resources and poetry online:
http://epc.buffalo.edu/authors/glazier/
Anatman, Pumpkin Seed, Algorithm (Salt Publishing 2003): a different
book. It shows Glazier's interest with language more than with
computers. And especially language within language: Glazier's poetry
owes a lot to his upbringing within a bilingual family (Mexican mother
& Anglo father) where the Hispanic language could not be spoken
except undercover. Glazier considers Sapnish as a mark up code. His
relationship to computer languages has first to do with this (language
within language), and he first used computer terms in a metaphorical
way within linguistic language. "Getting all mixed up".
You might find this useful for more information:
http://www.saltpublishing.com/books/smp/1844710017.htm
White-faced Bromeliads: (online work :
http://epc.buffalo.edu/authors/glazier/java/costa1/00.html)
This experiment (with text and photos) helped him to deal with the
frustration of having a poem getting published with the impossibility
to change it. Every line has 2 possible lines or variances. You get to
change every line of the poem 1 time by refreshing the page (or waiting
that it refreches by itself). Reading becomes a challenge because lines
change when you're actually reading. It is not a text in the closed
sense but a text in movement (open text - Eco). The question that
arises is : how do you grab a literary analysis (made on correspondence
and structure?) which demand an access to the ensemble. Here the sum of
the part is larger of the ensemble: that's also the variable principle
of the database.
Territorio Libre (online work:
http://epc.buffalo.edu/authors/glazier/e-poetry/Territorio/)
Another possibility for variation, based on word concatenation. There
is a video from the performance at the E-Poetry festival in 2003:
http://epc.buffalo.edu/video/e-poetry2003/day3/glazier02-030425.wmv.
"Touchlexicalby concussion" can be a symbolic concatenated sentence for
the whole work as this was written after a car accident that left him
slightly bewildered and also with a lot of time in his hand.
Io Sono at Swoons (
http://epc.buffalo.edu/authors/glazier/java/iowa/index.html)
Words broken into stems in different ways: a new generation each
time; fully original poems, never reproducible, highly ephemeral (takes
a click on the page or even a simple wait). Again, a video :
http://epc.buffalo.edu/video/le/glazier-021108.wmv
Glazier classifies the evolution and the actual differentiation between digital poetics in 4 categories:
1/ hypertext
2/ visual kinetic text
3/ interactive text: reader can help shape how the poem look (Cog: to play with)
4/ text in variable/programmable media; make things occur unforeseen
The last category is the one in which he wishes to include himself. It
is important that the reader has the reflex of finding and
reading/deciphering the code, however hard it may be (the simple "View
Source" option could be a good start). Himself he often includes jokes
in his working code. Writing code is creative in the sense that it's
interacting with the language on the screen. "Code as potent of
codeine", says he.
His main interest his directed towards the Unix system, as an environment within an environment.
_?_?_?_?_?_?_?_?
_?_?_?_?_?_?_?_?
_?_?_?_?_?_?_?_?
_?_?_?_?_?_?_?_?
_?_?_?_?_?_?_?_?
Discussion/Q&A
- (to J.C.): what are the social consequences of hoaxing?
J.C.: it happens that the literary world is made of isolated islands
that never talk to anybody else. The literary production becomes a
fiction in which you can assume a persona. The process of submitting
things for publishing is highly eye-opening. It is also a harsh world:
what if you're a young writer who wants to gain a career? They probably
won't let you do it. The dynamics of publishing are frustrating. The
hoaxes deal with that.
- (to both): what is your relationship to the machine? a partnership? competitive?
J.C.: as a developer, I can only say that the machine is a means to an
end for me. I often like the most complicate things, not as a challenge
but as a way to make money.
L.P.G: a complex of language (language within language). I like the
metaphor of the 'brick', having to assemble pieces together. I enjoy
working at the code level (non interface) = sorting out signs like you
would sort rocks from beans.
- (to J.C.): how much power of combination and novelty you have at your
disposal? at the same time what would it be like to reverse the
process, to make the machine a reader?
J.C.: There are experiments in speech recognition: you speak into a
microphone and the computer answers; of course it reduces the scale of
human speech, because the computer could not keep up. Also, the Turing
experiments tradition are the holy grail of machine understanding (the
communication language robots Racter or Eliza); but there are far from
what they would like to achieve.
C.Bernstein: You can interpret the machine as a reader in a different
way: a machine reading some traditional literature corpus for instance.
Again we have to talk about database and semantic issues: when you
search for something, what do you consider valuable, significant?
Reading is also determined by metadata...
In a more aesthetical way, you can ask what a poem is in terms of a database that you can read through in several ways.
-----------the-end
-----------the-end
-----------the-end
-----------the-end
-----------the-end
-----------the-end
&~&~&~&~&~&~&~&~&~&~&~&~&~&~&~&~&~&~&~&~&~&~&~&~&~&~&~&~&~&~&~&~&~&~&~&~&~&~&~&~&~&
My perspective on the event:
In my opinion, the readings were focused on what Espen Aarseth calls
the "cyborg" author. In his Cybertext. Perspective on Ergodic
Literature, Aarseth places this notion under the literary authority of
Umberto Eco's Opera Aperta and the "space medicine analyst" Manfred
Claynes.
The latter coined the word "cyborg", from the expression "cybernetic
organism", as a new entity that results from the alliance between
humans and technology.
The former theorized the concept of "open work" that implies aesthetics
that foregrounds the general topic of variable expression in works of
art. A correlate term is "work in movement", i.e. unplanned or
physically incomplete structural units that involve the idea of
randomness. Of course, the context within which Eco is writing is the
experiment of modernist music composition (Stockhausen, Cage). But he
nonetheless tries to define a poetics concept, underlying that these
works should be read as texts and that they convey a new dialectic
between the work and their interpreter.
This last point in particular provides an interesting analogy for text
in the cyborg era to the extent that textuality cannot be envisioned as
a mere author/reader or emitter/receiver relationship, but as a
participatory operation that redefines the practice of authoring,
writing, reading. Aarseth's perspective on the subject is tri-polar:
the text as a kind of machine, a "symbiosis of sign, operator and
medium" that implies automatically the cyborg. He then raised this
question:
"Any cyborg field, as any communicative field, is dominated by the
issue of domination or control. The key question in cyborg aesthetics
is therefore, Who or what controls the text? " (55)
The reactions of the audience during the readings and Q&A, even if
highly ironic and amused, revealed a more or less implicit concern for
this question of control. There was a general uneasiness regarding how
to read these works. The performance of Jim Carpenter most of all did
imply this ironic relationship to the "fake text" (= non human), even
if he was constantly reminding the audience the intimate relationship
based on shared identity he had with his computer double, a certain
Ericka. He was staging his own loss of control over the power of the
machine: not because the machine would suddenly become over-powerful
and get a total control, but because himself as a researcher and a poet
could not help always relying on the machine. The little fiction to
introduce this new relationship was that no poetry review ever wanted
to publish Carpenter's poems, so he thought he could trick them with a
little help from the computer (and a lot of help from this new entity
created: the cyborg).
Here what seemed to be uneasy for the audience was control over the
text not so much in cyborg authorship terms as in reading/evaluating
issues. And this was at the heart of Carpenter's project: quality or
competitiveness? What do you make of a text half-produced by a machine?
How do you judge it? (Should you judge/interpret it at all?)
The binary opposition between 'quality' (value of the poem to 'true'
poets - the term was never defined if I remember well) and
'competitiveness' (value within the literary world of reviews: economic
requisites) was a tricky way to, eventually, not answer this question
of evaluation. It triggered the narcissic conflict between the good
people that think and write, and the bad people that imitate and sell.
As if we didn't know at least since Hume that aesthetic qualitative
values are based on expertise standards that turn them into
quantitative values. There is a more than a continuity between the two:
it is the same process. This competitiveness seems to me a concrete
result of the construction of literary history, an quantitative
indexing (in semiotic terms) of what was first thought as traces of
experiments (what can be thought as quality, with its correlates:
originality, invention, etc.).
The problem over control that I was talking about seemed to me to have
been transfered from the author to the reader within this discussion at
Penn. What do you chose, legitimate, defend what you read? What is
worth reading? In the era of blogs, of course there is no easy answers.
Fed up with post-modernism, we cannot really answer "everything". Also
this is highly dependent on how you read. Etc.
Charles Bernstein tried to orientate the debate over the issue of
database and acess (which is also fondamental to Lev Manevich
informational esthetics in The Language of New Media). He
re-articulated the traditional definition of the poem as an object to
be interpreted in several ways in terms of database and sorting
techniques. This allows the necessary shifting from interpretation to
practise, according to me (= what can you do with a poem?). Thus the
poem is re-inserted in the communication process when it is losing its
status of exception and treated as any text.