When I stepped into Jon’s office this afternoon, there was a spectacular view out of his window of a thick haar rolling across the top of Arthur’s Seat. During the course of our conversation the haar dissolved, leaving behind the shadow of a moon and a slowly reddening sky. As I left, it was pitch black, with the moon hanging in place. The moon was particularly apt as we had a very excitable chat about the possibilities of directing gaze around the Puffersphere for my panoramic narratives. These photos of Jon’s whiteboard hold the kernels of the ideas, touching on approach/withdrawal as a metaphor for perspective within writing, perspective of audience, personality dimensions, and left/right hemisphere activity. Using the idea of dualities (light/dark, left/right, past/future, /internal/external, forward/backward, head/tail, true/false) I want to see how one story can be used to insist that the audience must hear the other perspective on it; how meaning can be unlocked on a different level, but made imperative.
Luckily, Jon keeps an inflatable exercise ball in his office, which turns out to be the perfect demonstration tool when attempting to think through an idea at the same time as articulating it (“so, imagine that the happy face is following those three people round the room and shouting at them to come over and listen… and then the sad face is over here, refusing to talk to these people because they’re standing too close.”). Next time I might bring dolls with me to play the audience role.
Today’s subtitle comes from an amusing glitch in Entitlement, the title generating assistant. Over the last three weeks or so it has come on apace in selecting and displaying relevant title words (extracted from abstract and citations for each paper, with different weights being given to different elements to demonstrate possible routes to go down). However, it has reached a sticking point in producing meaningful titles out of the selected words. It’s easy to produce a probable-sounding parody of a title (plenty of those type of generators around- see my last post), but producing something with applicable meaning is surprisingly difficult. After an emergency coffee and ‘why doesn’t it just do what it’s supposed to, automatically?’ chat today, Clare is going to attempt a language model (or ‘sieve’ to laypeople like myself) to rank the improbable titles against meaningful phrases (from the abstract and existing citations) to see if that fixes it. In the meantime it’s coming up with some corkers such as ‘Can The Computer?’ ‘Is The Minimal Closure of Human?’ and (my new personal favourite) ‘An Useful Human: Analyse the text’.
I was delighted to be allowed to attend a meeting of CIRCLE (Creative Interdisciplinary Research in Co llaborative Environments) in Inspace at lunchtime, to hear Joanna Kane talk about her work on the Invisible Archive, and Miranda Anderson talking about a potential literary app based on literature written about Edinburgh. I was amused by the possibilities for a list of properties currently being drawn up for the latter, and particularly hope that a search category based on the Seven Deadly Sins makes it into the final version.
In other news, doodled around with organising principles for Ada and GIL without actually getting anywhere solid (death-by-cut-and-paste), and skimmed through various papers on modelling fiction (summarizers and generators, for the most part) that Micha pointed me towards last week. I also discovered that I have a pigeonhole opposite the office (and post in it! Sadly, not exciting post) and my name is now up on the office door.