I’ve been contributing to the conference Wiki and bookmarking stuff in del.icio.us rather than blogging. These activities work differently to blogging though. Right now I need a blog post: Fox Harrell is currently demonstrating his poetry generation system, GRIOT. I tend to like the idea of poetry generation more than its products, but GRIOT generates its poetry line by line in dialogue with a human. So the human types a word (“europe”) and the generator replies, then waits for the human to type another word. When the human is satisfied, she types “end”. The results are rather appealing, and I’d love the opportunity to play with poetry in that way. Unfortunately I don’t think GRIOT is online yet.
I'm Jill Walker Rettberg, author and professor of Digital Culture at the University of Bergen in Norway. I co-direct the Center for Digital Narrative, a 10-year Norwegian Centre of Research Excellence, and I lead the ERC project Machine Vision in Everyday Life. This site started as my blog in October 2000.
Fox Harrell
Dear Jill,
I really enjoyed your talk and the probing, contemplative style of it. I was intrigued by your contention at the end that these various new forms of self representation are often motivated in reaction against commercial culture and commodified ideas of self and beauty. I hope that is the case!
I am glad that you found the output of the “Girl with Skin of Haints and Seraphs” pleasing. I hope you also enjoyed briefly interacting with the same system. I found this post of yours tracing back some of the links from the updated DAC page and looking for photos from the conference (I didn’t take any myself). I had a few responses to it:
(1) The poems output from the system have narrative structures. They don’t just end when the user types “end.” Some are shorter and some are longer but they all have openings, some various types of clauses, and endings.
(2) What you describe is not exactly output of GRIOT itself. GRIOT is the general platform I made to implement and run a particular poetic system like “The Girl with Skin of Haints and Seraphs” or “The Griot Sings Haibun.” These systems have totally different narrative (discourse) structures and different content and themes. So when you interact it is really with a “poetic system” or “polypoem” (polymorphic poem that is different each time it is run). The idea is that the general framework of GRIOT should be useful for other forms of narrative multimedia artwork, not just poetry. Poetry was my first experiment (for a variety of reasons), so it is fair to judge this work in the context of poetry generation though.
(3) The claims being made, and my goals, are not really that the computer generates the poetry. I want help enable a human artist to be able to write text, stories, or multimedia that is meaningfully different on each reading based in interaction with that reader. So I am interested in improvisation in some forms that are not usually thought of as being improvisational (in this case written text). I often find that ideas I want to express deal with shifting metaphors, narrative imagining, and blending and reblending concepts, so this type of medium suits my own artwork.
Thanks so much for you interest and I hope to meet again!
Fox
p.s. If you are interested in some of my motivations (including thoughts about identity), please look here:
http://www.ctheory.net/articles.aspx?id=489