such an amazing review -- and i actually agree with you mostly about Derrida, his stuff on writing and machines is about as good as theory gets about writing and machines (so far). i don't think he's lacking responsibility for the abandonment of the restricted economy that the next generation undertook -- and i'm not sure i'd go as far as to hold up Ulysses Gramophone as an example of the countertrend, i want a fictional JD who connected the general and restricted economies by way of ideology, cultural laws, disciplines -- but it's not so important. very grateful for this, and would just add that "Weatherby's Saussure" was Pourciau's Saussure first
Thank you for such a generous reply! re the Pourciau correction, yes my bad and will amend. I think I see what you mean by wanting the fictional JD—the bridge is still missing even in the work I’m citing. Thanks again for the book
right, so the media-theory/local analysis axis is pretty good! depending on which deconstructionist you get ahold of; what bugs me is the parole-langue or close-reading-to-linguistics axis got totally lost. DH btw promised to fix this and hasn't, at all.
I've been grappling with these questions for a while--aided also by Weatherby's incredible book--and this review is great food for thought. In particular, I appreciate the point about countersigning: "To countersign is to let a sentence obligate you, to put your name next to it and be on the hook for what it does in the world." I think the question of who takes responsibility for AI writing will be increasingly important. In teaching, I've built my AI policy around accountability, following the journal Nature's policy from Jan 2023. Regardless of origin, our words will need to be signed.
Excellent framing of the unseigned yes problem. The seperation between generation and answerability really gets at whats unsettling about LLM outputs, they have all the surface features of authored text but none of the binding structure. I think the Joyce-Derrida connection shows how restricted economy analysis doesnt have to float above technical realities, Derrida was actually counting tokens before tokenization was formalized as an NLP concept.
such an amazing review -- and i actually agree with you mostly about Derrida, his stuff on writing and machines is about as good as theory gets about writing and machines (so far). i don't think he's lacking responsibility for the abandonment of the restricted economy that the next generation undertook -- and i'm not sure i'd go as far as to hold up Ulysses Gramophone as an example of the countertrend, i want a fictional JD who connected the general and restricted economies by way of ideology, cultural laws, disciplines -- but it's not so important. very grateful for this, and would just add that "Weatherby's Saussure" was Pourciau's Saussure first
Thank you for such a generous reply! re the Pourciau correction, yes my bad and will amend. I think I see what you mean by wanting the fictional JD—the bridge is still missing even in the work I’m citing. Thanks again for the book
right, so the media-theory/local analysis axis is pretty good! depending on which deconstructionist you get ahold of; what bugs me is the parole-langue or close-reading-to-linguistics axis got totally lost. DH btw promised to fix this and hasn't, at all.
I think we’re in agreement on this! I’m also disappointed by DH not fulfilling its promise
I've been grappling with these questions for a while--aided also by Weatherby's incredible book--and this review is great food for thought. In particular, I appreciate the point about countersigning: "To countersign is to let a sentence obligate you, to put your name next to it and be on the hook for what it does in the world." I think the question of who takes responsibility for AI writing will be increasingly important. In teaching, I've built my AI policy around accountability, following the journal Nature's policy from Jan 2023. Regardless of origin, our words will need to be signed.
Excellent framing of the unseigned yes problem. The seperation between generation and answerability really gets at whats unsettling about LLM outputs, they have all the surface features of authored text but none of the binding structure. I think the Joyce-Derrida connection shows how restricted economy analysis doesnt have to float above technical realities, Derrida was actually counting tokens before tokenization was formalized as an NLP concept.