Imaginary Machines: eSifter
If I had plenty of time on my hands I’d do something like this:
The idea is that you have language resources on the left, which are selected by dragging them to the right. In the example above, the text of “King Lear” is being dragged into the language model being built. The language model is weighed (as in Gnoetry) as suggested by the light blue triangle: the “Ready to Die” text is heavily weighed, while the “Heart of Darkness” text is less weighed.
Similarly, rather than just using n-gram language models, you could use statistical grammar, pos template, and other types of language models. You could even have custom grammars (as in JanusNode) which don’t rely on the text in the previous step. As in the previous step, each resource is weighed.
In the third step, you would drag which items you’d want the poem to be evaluated on. In this case it’d be evaluated based on whether it produced an end rhyme, and whether it matched a set of requested phonemes (as in ePoGeeS).
In the next step you’d be selecting which surface-text transformations to do, like eecummingsification as in JanusNode, or fades and animations as in animated poetry.
Finally, in the last choice, you’d select which of the eOulipesque authoring techniques you’d use. Once the “Run” button is pressed, some kind of authoring interface would pop up to let you execute the authoring technique.
I dunno… it’s kind of a neat idea, but it would require a lot of GUI authoring overhead, and it isn’t really engaging, and I don’t think it’d make a good tool… Probably only a 1 in 10 chance I’ll implement it.. Still, it was fun to think through, I like how you drag resources, maybe that’ll turn into something useful. Anyways, this is classifed as Method 22813e0e-bf00-4362-9a23-e9efa6c58d5f (eSifter).