October 28, 2003
I somehow managed to pause halfway through Narrative Intelligence, edited by Grand Text Auto‘s own Michael Mateas and Phoebe Sengers, and mentioned on here before. After losing track of the book for a while, I’ve finally finished reading it. Even knowing something of the breadth of inquiry undertaken by those in narrative intelligence, it’s a rich and surprisingly diverse collection. Papers from the AAAI Fall Symposium 1999 are supplemented with other selections important to narrative intelligence research. One effect of the collection is to make me sorry that I missed the NI symposia and the active days of the NI group at MIT. But I’m glad this book is still around as a contribution to the academic discourse, and I hope future work will build on the insights in it.
NI researchers all share a concern with intelligence (human and computer) and with the use of narrative to organize events, but the field (if “field” is the best word for it) encompasses many different concepts and approaches, as Michael and Pheobe explain in their introduction. This means that people are more likely than usual to find a few essays to be gems and to find that others are of no use. In my case – as my interests seem to be pretty in keeping with the “typical” set of NI interests, assuming there is such a thing – I found something to redeem each of the articles, although it wasn’t always what I expected. For instance, an essay that didn’t provide me with any insight into the expected topics of computing and narrative (one about the design of a documentary about a band) offered helpful discussion on topics I didn’t expect to read about, such as the consistent ways that youth culture expresses itself within a mainstream culture.
Some of the high points of the book, in my reading of it, included the theoretical discussions by Jerome Bruner and (in a very different vein!) Phil Agre. R. Raymond Lang’s article on story grammars showed that story generation is still an interesting topic. Many of the articles that took the framework of a project writeup were quite effective; Warren Sack’s discussion of his conversation map was a standout for its thorough description of the technology (including details about the different sorts of linguistic analysis that were done) and the critical motivation for the project. The discussion of Virtual Babyz by (our own) Andrew Stern was also quite well-done, and described well the conditions under which interesting behavior might emerge that could be interpreted in a narrative way. Phoebe Sengers’ discussion of her project was another good case study in critical technical practice, outlining the relevant ideas of the anti-psychiatry movement when discussing the psychopathologies of intelligent computer agents and how to treat them. I’ve already suggested articles from the book to more than one person.
Rather than delving into any specific essay, although I may do that later, I’d like to ask Michael and others where narrative intelligence – as a field, as a movement, as a concept – seems to be headed these days? The Media Lab’s NI list is certainly not hopping with activity. Will there be other get-togethers? Other books? Or has NI done what it needed to do and helped to focus curricular selections and research directions in new ways, so that these trappings aren’t necessary any more?