I have just returned from the ACM conference in Milwaukee, excellently run and as interesting as ever, but smaller than it has been in a number of years. I also co-chaired the Hypertext and Narrative workshop with my old student Charlie Hargood (now a postdoc at Southampton).
The Narrative workshop had its inception seven years ago as ‘Hyperstructure 2005‘, and I remember really struggling at the time to find a term that captured the breadth of what we were after, the CFP makes the problem clear: “This workshop [is about]: narrative, music, rhetoric, cinema and games (structures that are effected by rules about the syntactic and semantic arrangement of content at a relatively high level of abstraction).” Oh dear.
These days we just say narrative and hope that readers will have the same broad interpretation that we do. The workshop accepted some great papers, and Mark Bernstein has written a good overview of them on his blog, but there were also some interesting unconference sessions where we had some nice discussion. Thanks to everyone that contributed, it was fun to meet you all!
In the morning Stacey Mason, inspired by a recent trip to the ELO Conference, challenged us as to why we would want a machine to tell stories at all. It’s one these seductive questions that I think is actually a non-question. Dijkstra, when asked about the sense of artificial intelligence, said that asking whether we should teach a machine to think is a bit like asking whether we should teach a submarine to swim, and I believe the same principle applies.
I remember once getting an extraordinarily cold response when pitching a narrative project to a large group of scholars that included many people from the humanities. I think the problem was that they saw narratives as sense making and interpretive, and hated the idea that they could be generated by a machine. They saw machine narratives as dangerously subjective perspectives from a machine that otherwise pretended to be objective.
Machines can’t tell stories in the same way we do, but they can simulate and approximate and support, and to do that we need computational models of narrative that may well be simplistic, but that should at least be grounded in theories from linguistics and narratology. It may not be real swimming, but who cares if it helps us to move through the water.
In the afternoon we asked ourselves what we wanted from narrative tools, and how best we might evaluate them. Our conclusions were that the demands were very contextual, but that with narrative tools we had the ability to add higher level structural support, rather than forcing the user to build their own higher level structures and processes from low level ones. One challenge is that if we give authors a structural toolbox them it limits their imagination, but a possible response would be to use structural parsing (just like a spatial parser in a spatial hypertext system) to identify and highlight high level structure when it appears – and perhaps in this way we can reinforce and encourage more sophisticated narrative structure without railroading the author.
There were also some great narrative papers in the main conference. Annika Wolff presented the decipher project, and work done at KMI to use narrative in museum contexts; Alex Mitchell gave a presentation on his thesis work on rereading and the invariant story (asking the question, when does reading a hypertext become rereading and what constitutes closure?); and David Kolb gave a characteristically enthralling presentation of his hypertext paper Story/Story on the inevitability and character of meta-narrative. Efstratios Gallopoulos also gave a wonderfully energetic presentation on analysing children’s ergodic literature, which was just as memorable for its account of his discovery of the form with his daughter, as it was for highlighting a much-neglected area of hypertext narrative.
Our own paper, on the Poetics of Strange and Fractal Narratives, bemused an audience that had mainly gathered for a session on social media analytics. It’s an account of a simple hypertext engine we developed for building hypertexts where clicking on a division in the text reveals more text in that place, with further divisions to click – producing a deep, unfolding narrative. Our aim was to explore the poetics that emerged from such a structure: if node/link hypertexts with their cycles and tangles inspire stories that play with reinterpretation and/or confusion, then what would this deep unfolding symmetrical structure inspire? Instead we learnt that the poetics of a new form must be nurtured, supported by the technology, and emphasised in teaching. To many it will be no surprise that in art, as in life, it is all much more complicated than we first think.
My last thought on this years conference is a reflection on myself, and what a cantankerous old hypertext curmudgeon I am becoming. I’m somewhat of a Hypertext veteran these days, and am developing the dangerous habit of bemoaning each new paper with the cry of ‘we did that twenty years ago!’ Rarely is the work in question my own work, so the irritation is not so much the lack of acknowledgement as it is the fact that we seem to reinvent rather too much, treading water without moving forward. I made the observation a few years ago that as a community we seemed rather satisfied with our lot, and that the conference had turned into a grand safari into the work of others. I can’t help think that to make progress we need more new tools, more experiments and more imaginative interventions.
In short, we need more submarines.