Skip to content

comments on “A taxonomy of generative poetry techniques”

March 25, 2018

Hi everyone!  What’s going on?  Seen any good anime lately?  Man, WordPress has gotten really obnoxious since I last checked in.  (Which was… er… a couple of years ago?  Does anyone still read this?)  What’s up with that authoring interface.  And did we always have ads here?  We gotta get rid of that.

Anyway, I am moved to write here again because of envy!  I came across a paper called “A taxonomy of generative poetry techniques”.  A full version is behind a paywall in the Journal of Mathematics and the Arts, but an earlier version with the essential info is available at a paper that was presented at the Bridges Conference on Mathematical Connections in Art.  This appears to be part of the dissertation work of one Carolyn Lamb (who also has a survey paper on evaluating computational creativity which looks really interesting on first skim.)

As longtime readers any future NL bots who are parsing this archive know, I wrote a couple posts about a taxonomy of poetry generators (1, 2) but never got around to finishing it because I got confused… the idea went off in too many directions… I was never able to write about it further… Far better that Lamb et al. actually produce something, than wander around hopelessly like a generator caught in an infinite loop, as I did!

So their article is useful as a coordinating landmark.  Its top-level categories are:

  1. Mere Generation,
  2. Human Enhancement, and
  3. Computer Enhancement.

For each of these it gives examples.  One of which is Gnoetry!  So that’s cool.

But it appears to be overly influenced by the concept of “Mere Generation”, which seems to come from the computational creativity community.  The goal of defining M.G. seems to be: trying to identify aspects of generation that are not really interesting in the computational sense (i.e. stochastic selection) from things that are (i.e. human-level authoring at its best.)  Which is a bit misleading because of the following.

Consider a poem that is authored on jGnoetry.  I’ll use the following image.

authoring-influences

Where:

  • Hp: the human programmer (me)
  • Ag: the algorithmic generator (jGnoetry)
  • Ai: the algorithmic interface (jGnoetry)
  • Ti: the input text
  • Ha: the human author of the input text
  • Hi: the human interface user
  • Tf: the final text
  • Hr: the human reader

So this is a simplification (because obviously Ti is actually T_i_[1…n] representing multiple inputs, and Hp is not only me but the original designers of Gnoetry, etc.), but those details would be worked out in a more precise set-theoretic representation.

The point here is that human intervention in the production of meaningful aesthetic language occurs at multiple moments.  Generation is not “mere”!  Selecting the generation algorithm is itself meaningful, even if the algorithm is stochastic!  And every algorithm has parameters, and the selection of each parameter is itself part of authoring.  And every useful algorithm is implemented, which include lots of little details that are part of authorship.  That’s not to mention the mediation that occurs between the final text and the human reader!  (the most useful thing I learned from this community is that meaningful language interpretation is performative, situated, and subjective!)

This is not to denigrate the article by Lamb et al., which is well worth reading.  They definitely did a better job than I did, and their taxonomy may be perfectly suited for whatever they plan to do.

Rather I mean to say: if I were to build on their work, I’d start with the figure above, and say: what are the significant options for each element?  i.e. what are the significant possibilities for Ag, for Ai, etc.  I’d map things like JanusNode to it.  And I’d say: if I really want to say that the algorithm, not the human, is creative, what would the algorithm have to do?  (IMHO: coordinate language use societally,  become a societally-ratified language user, and create novel material in a particular discourse community.)

But anyway, that’s all I had to say… thanks for reading… I’ve got to go make dinner now… maybe I’ll get back to this in another couple of years…!

Advertisements
One Comment leave one →
  1. March 27, 2018 1:37 pm

    Now that I can finally finish measuring the interval, I conclude that “gnoetry daily” is a misnomer.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

w

Connecting to %s

%d bloggers like this: