Pages in this blog

Wednesday, August 30, 2017

Computer-generated stories in the 1960s [#DH]

James Ryan, Grimes' Fairy Tales: A 1960s Story Generator, Expressive Intelligence Studio, University of California, Santa Cruz, jor@soe.ucsc.edu
Abstract. We provide the first extensive account of an unknown story generator that was developed by linguist Joseph E. Grimes in the early 1960s. A pioneering system, it was the first to take a grammar-based approach and the first to operationalize Propp’s famous model. This is the opening paper in a series that will aim to reformulate the prevailing history of story generation in light of new findings we have made pertain- ing to several forgotten early projects. Our study here has been made possible by personal communication with the system’s creator, Grimes, and excavation of three obscure contemporaneous sources. While the ac- cepted knowledge in our field is that the earliest story generator was Sheldon Klein’s automatic novel writer, first reported in 1971, we show that Grimes’s system and two others preceded it. In doing this, we reveal a new earliest known system. With this paper, and follow-ups to it that are in progress, we aim to provide a new account of the area of story generation that lends our community insight as to where it came from and where it should go next. We hope others will join us in this mission.
From the article itself: 
But why should we care about old, forgotten work? If we view story gener- ation as a vast design space, we can think of each implemented system as an exploratory vessel that ventures into a previously uncharted sector. If these ex- ploratory missions are successful, they signal directions that future systems may move further into to find greater success. When success is not had, the failed projects tell us which areas to avoid. In this way, we learn about spaces that incrementalist research may push further into, dead sectors that we should not return to, and all the other still uncharted areas that we do not know much about at all. Thus, both good and bad systems generate new knowledge that is useful to contemporary and future practitioners. But when we forget about past systems—novel explorations in design space—we lose the knowledge that was generated by those systems: we forget what has been explored and what has not, and which areas are worth exploring further. As we discuss below, more than fifty years ago, the system we profile here anticipated, and then abandoned, an approach to story generation that is currently in vogue.

Beyond these fundamental practical reasons lies one of principle: as a field and as a community, we owe it to ourselves—and our forebears and our successors—to record an accurate historical record. How would you like your work to be forgot- ten? We, moreover, owe it to ourselves to maintain a record that encompasses not just a series of names and dates, not just a series of system architectures, but also the intellectual through lines that trace our history. Story generation is an applied technical area, but all human endeavor, especially in the area of research, has intellectual underpinnings and emerges out of intellectual contexts. Even in technical areas, there is a history of ideas that undergirds the evolution of systems over time. Returning to practical concerns, good ideas for systems can lead to bad implementations of them, and so we should track ideas too so that we might have another stab at carrying them out well.

No comments:

Post a Comment