FIVE WAYS TO APPROACH POETRY GENERATION (AS A NATURAL LANGUAGE RESEARCHER)
Hi folks!!! eabigelow’s recent post “TEN WAYS TO MAKE IT AS A DIGITAL WRITER (AND THEN FADE AWAY)” was pretty interesting to me because while it seemed like it might be useful to a lot of folks, it didn’t really reflect what I learned generating poetry on Gnoetry Daily…
After a bit of thought I realized it was probably because I was coming to the field as a natural language researcher doing poetry generation, rather than a poet doing computers. So I thought I’d do a post like eabigelow’s aimed at any other researchers in Natural Language Processing / Computational Linguistics / Artificial Intelligence who might think about doing poetry generation.
Also, I recently bought my kid a bunch of Tsai Chih Chung’s comic book adaptions of Daoist and Zen classics, so just for kicks I wrote it partly in that style! Anyways, see below for the full thing.
FIVE WAYS TO APPROACH POETRY GENERATION (AS A NATURAL LANGUAGE RESEARCHER)
1) See poetry in all research.
Every time you encounter a research artifact (algorithm, toolkit, corpus, result, … ), ask yourself how it might be used to generate poetry. If you do only this, you will benefit.
2) Integrate the human, and instantiate.
Consider how a human could interact with a research artifact to generate poetry.
You are the human. Master the artifact. Implement a generator swiftly and minimally; do not be distracted by irrelevant details, but attend to what parameters you frequently change. Make a graphic user interface and upload it, if feasible.
3) Know the Ways of all Practices.
There are four Practices in poetry generation.
Research Practice investigates issues in language, meaning, and computation.
This is the Way of the Scientist.
Procedural Practice creates new methods of generating poetry.
This is the Way of Oulipo.
Resource Development Practice develops tools for generating poetry.
This is the Way of the Hacker.
Aesthetic Practice produces poems.
This is the Way of the Digital Poet.
4) Understand the true nature of poetry generation.
When you develop a generator, it does not matter if even a single poem is output or read; you have created an infinite number of possible poems and audiences. When you generate poetry, you are sampling from that infinite space. When you interact with a generator you are a heuristic, guiding its path through state space.
Some of your output will have the beauty you see when surveying data or alpha-testing a prototype. This is related to the way of the Language poets. Some of your output will have the beauty you see in incongruous or unexpected results. This is related to the way of the Flarf poets. Some of your output will have a beauty you would not have otherwise imagined.
Output is subjective and software becomes obsolete, but output sets are infinite and methods and algorithms are eternal. All past beings offer their texts as inputs. Your peers scattered over future decades find you through searches.
5) Write explanations for those you might want to know.
Someday you may want your child or a friend to know what you do. Write brief guides and explanations that any intelligent youth could understand. This is related to the hacker ethos of giving back.
Someday your peers’ search programs will locate you. Write appropriate answers for their queries.
When you generate poetry as described above, you have no funders to report to, no program managers to satisfy, no auditors to review your code, and no audience to concern you. Your knowledge and abilities are constrained only by your will. This in itself is poetry.
edde addad, March 12-14 2011
You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback from your own site.
3 Responses to “FIVE WAYS TO APPROACH POETRY GENERATION (AS A NATURAL LANGUAGE RESEARCHER)”.
wow thanks for this, it’s crisp. i am afflicted by #1. I can’t write science anymore because i want it to be poetic, but then no one will understand me. I wish cartooning was my art-affliction. everyone loves cartoons. poetry is so misunderstood. i have no qualms, it’s more magical if we leave it ineffable, methinks. thus i will eschew #5. instead i write gushy staccato commentpoems to semi-strangers, cross my fingers, hope they write back with an essay about the many different grades of hungarian paprika.
This is an interesting read. Thanks!
I enjoyed that, Edde.
1. When I learn new programming stuff, such as a new language, I find that if I consider the sorts of art software I’d like to write in the language, I learn better. Not surprising, I suppose, since that’s really why I’m learning the language.
2. “…attend to the parameters you frequently change. Make a graphic user interface and upload it, if feasible.” Very good. It doesn’t usually take too much to turn all parameters even remotely of interest, first, into variables that can be easily changed in the code itself, and then, later, make those parameters–or perhaps just the ones you’ve found to be most relevant–configurable in a graphical user interface.
3. “Research Practice investigates issues in language, meaning, and computation. This is the Way of the Scientist.” Yes, but also the philosopher. Because philosophy and logic are so strongly related, these days. The theory of computation–given not only its relation with computation but also with, say, Godel’s and Turing’s work, which is fascinating, philosophically–is a kind of a philosophical and theoretical foundation for the poet-programmer. Of course, it is even more widely misunderstood (or not understood at all) than, say, French theory in literature.
Thanks for the thoughtful article, Edde.
Leave a comment.