In Raymond Queneaus combinatoric sonnet, Cent milles milliards de poèmes, 10 sonnets of 14 lines each are cut in strips. By selecting strips the reader can create 10^14 recombinations, which, as the title says, is 100,000,000,000,000 possible poems. But one can arrive at 10^14 poems by many paths, both in print and online. How do hypertext editions of Cent milles milliardes de poèmes compare with print editions?

Im not sure which edition of the work is pictured in this beautiful photo, however it looks like the strips were part of the original binding, with a slight gap to prevent rubbing (rather than being cut later). This is how Ive heard the 1961 Gallimard edition described. With even damaged copies selling for between $350 and $400, I havent yet laid hands on one. Instead, Ive read uncut reprints (for example, One Hundred Thousand Billion Poems is reprinted in The New Media Reader) and interacted with web editions that try to adapt the paper-strip aspects of the original through hypertext. Im still looking for an affordable pre-cut edition - and considering making my own - however Im also wondering how the online alternatives work in creating a similar or different experience.

One example, Magnus Bodins edition of Cent milles milliardes de poèmes offers the original French text and two translations into English and Swedish. Bodin offers no selection or navigation - instead, clicking a New Poem button simply refreshes the page with a random selection. Like a Haiku generator, each new poem appears out of the ether, and the interface emphasizes the vast number of possibilities by streamlining the process of generating outputs. This fully automatic method is quite unlike the print edition, which allowed readers to work with the text, manually building each poem sequence through a series of selections. Still, there is some rhetorical insight to be gained from a simple button labeled New Poem: with so many possible poems, it is quite likely that any given poem has never been read by anyone before… and may never be read by anyone again.

Jacob Smullyans edition of Cent milles milliardes de poèmes provides a command line interface to the text of Stanley Chapmans English translation 100,000,000,000,000 Poems. While Bodins edition demonstrates the scale of the possibility space through random uncontrollable motion, Smullyans edition demonstrates scale through a blind precision. Entering a number between 1 and 100,000,000,000,000 will specify exactly one of the total possible combinations, which is then displayed. Smullyan calls these Sonnet Numbers, which is an interesting idea. While referring to Cent milles milliardes sonnet no. 7,225,081,256,601″ seems reminiscent of other numbering schemes, such as Shakespearean sonnet numbers or classical music opus numbers, it seems preposterous to imagine the numbers actually being used in a similar way: Ah, yes, I remember good old 7,225,081,256,601, the opening line matches the final couplet so well, etc…. on the other hand, the Smullyan number is a compact way to indicate which textual configuration is being indicated - a set of generation instructions for any example. Is such a thing useful? If it isnt, it might be worth asking why we dont commonly discuss specific arrangements when discussing combinatoric art. Is it because working with examples is too difficult without a good reference system? Or because we find any given specific arrangements beside the point?

Perhaps the most technically impressive implementation is Beverley Charles Rowes edition of Cent milles milliardes de poèmes. Rowes own original English translation is characterized by following Queneaus rhyme scheme and by the use of intentional archaisms. The interface is a 14×10 grid whose squares mimic the array of choices afforded by the strips in the print codex. After clicking the random button, highlighted squares in the grid create a sense of orientation within the text space - a sense not present in the interfaces of either Bodin or Smullyan. In addition, Rowe makes intentional choice possible. Clicking on any square selects that line, while clicking on any column number selects that entire sonnet. Rowes interface goes further by aiding parallel line selection - clicking any row number displays a list of all 10 available choices for that row. Finally, every line of output text can be switched between English and French with a touch of the mouse, and some lines are further annotated with translators notes on word choice, meaning, and etymology.

[Some of these features are expanded in the new edition (v2) which came out on January 19, 2006, but Ive experienced serious javascript problems in Safari, Omniweb, and Firefox - try Opera.]

Florian Cramers edition, on the other hand, allows manual recombinations using drop-down menu selection. After using Rowes intreface, Cramers interface might seem simple and slightly awkward, however the disadvantage of using listboxes might also be the advantage: incrementally browsing the lists interface is somewhat analogous to the experience of flipping through packets of paper strips. While Rowe preserves the sense of the total text space, Cramer preserves the sense of difficulty and the sensation of using something.

Cramers edition only contains four of the original ten French sonnets, along with a fascinating disclaimer. Cramer says that, while the work is still under copyright, this sampling represents fair use, as the sample represents a mere 268435456 of the 100000000000000 total poems. Sampling 40% of the extent text only shows you 0.0000027% of the total work? This might seem counterintuitive, but it is true - 4^14 is only the tiniest fraction of 10^14. Im curious if anyone has ever attempted to make this argument about combinatoric art in a copyright claims lawsuit, and if so how the argument was received. A decision might depend on whether Cent milles milliardes is defined as the sum of its parts, at 140 lines, or as more than the sum of its parts, at 100,000,000,000,000 possibilities. If we take Queneaus title seriously, fair use would handily cover presenting a mere nine of the ten material sonnets. That might be 90% of the source text, however those would render only a modest 23% of the total poèmes.

My modest proposal that we sample 90% of Queneaus work may seem deliberately provocative, but the underlying question is serious. Which is the artwork we might fairly use - the parts, or their combinations? When experienced through Rowes grid, the answer the artwork is 140 lines feels intuitive, as the parts are all laid out before us. On the other hand, a few minutes of using Smullyans command line, and the argument for the artwork is 100,000,000,000,000 possibilities is much more clear, as possibilities are what the interface presents us with. While Rowe and Cramers interfaces are navigators which focus on manipulating the parts of the work, Bodin and Smullyans interfaces treat part-manipulation as the domain of the machine. Instead, they present us with the raw output, the poems themselves, each one distinct and all equal.

The fair use issue is inseperable from the interface method. Both are different ways of considering what constitutes the artwork in a combinatoric piece. As a personal experiment or a class project, it might be interesting to learn about Queneaus work by creating your own - beginning with a simple print version, and letting some vision of what the text really is guide the adaptation.

10 Responses to “Cent milles millardes de poemes”

  1. 1 Jill

    I bought my reprint of the original, which I think is the same as in the photograph, for ???32 at a few years ago. They still have it there - I couldn’t find it with the simple search form on the front page, but get to the advanced search and put “cent” in the title field and “queneau” in the author field and it shows up.

    I’m very happy with it - it’s big, has a lovely hard cloth-bound cover with a shiny removable transparent plastic dustcover, as you say the lines are pre-cut with a little space between them, and it’s really a very useful prop for explaining what the heck these things are about to students, in lectures and presentations.

    I’ve only used that once, but there were no problems with the order or with their shipping internationally.

  2. 2 Jeremy Douglass

    Thanks Jill - sure enough, I found the chapitre reprint of cent milles milliards and I’ve ordered one - can’t wait. I hope the survey of web versions was useful - I’m still thinking about using all these version together in a class at some point.

  3. 3 Jill

    Great, Jeremy! And yes, looking at the different ways it’s been implemented digitally is excellent - really helps in thinking about the differences of paper and digital, and of different kinds of digital - would make a great class, I think.

  4. 4 Magnus Bodin

    Most interesting write-up!

    Thanks for a good read about a phenomenon I like.

  5. 5 Beverley Charles Rowe

    You might like to look at the new version of my Quenau website.

    I have changed some dozen lines of the text, I hope for the better. The look of the page and the way it operates are completely different.

    In response to your comments I have introduced two methods of numbering the sonnets. One is the same as Jacob Smullyan’s. The other was developed in conjunction with Charles Clunies-Ross and has advantages and disadvantages.

    I have also introduced a feature that some people will object to but which is entirely optional. Since any lines that rhyme can be interchanged, I have allowed for random shuffling of rhyming lines. This increases the number of variations so that it would now take 8 million times the age of the universe to read them all.

    As usual, comments gratefully received.

  6. 6 Beverley Charles Rowe

    Thinking over what Douglass says in his original posting, he is wrong in describing Queneau’s book as a series. It is much more like a matrix.

  7. 7 Jeremy Douglass

    Beverley - the text as a 10×14 matrix? An interesting idea. I can understand imagining it as an array - although I’m not sure how the mathematical idea of a matrix helps to understand the way the book read. Which matrix operations would correspond to reading? Matrix addition, multiplication…?

    Certainly I’d agree that the text itself is not a series. When I said “the print edition, which allowed readers to work with the text, manually building each poem sequence through a series of selections” I simply meant that the process of building a poem could be described as a process of 10 choices - reading is not a selection, but a series of selections (matrix selections, perhaps).

    There are other many other ways to interact with the book, of course, making selections in any order, and fewer or more than 10, and reading not whole poems but local snippets as one wanders through the space. Still, I find that the paper edition tends to fight back a bit against browsing - the packets of lines are fussy about being parted, and intractable when held back in groups, and this physical quality encourages me to make a series of rapid choices, then hold back the page and view the completed poem in full. Perhaps this impression is only palpable in contrast to a digital grid interface like your edition of the poem, which encourages true wandering.

  8. 8 Amanda

    I have been trying to locate the English translations of Cent Milles millardes de poemes with no luck. I know Stanley Chapman and Beverley Rowe and John Crombie all published English versions and I am really really trying to find one in print. Does anyone know?

  1. 1 WRT: Writer Response Theory » Blog Archive » Curious George Landow
  2. 2 The Output-Oriented Game (OOG) — The Movies at WRT: Writer Response Theory

Leave a Reply

thesis writing service