Here Come the Academics

Here Come the Academics!

By the mid-90s, many computer scientists had decided that storytelling was an exciting research area, and some interesting work was being done. The games community continued to pursue its cut-scene approach, but the academics were trying a completely different tack. Actually, they were trying a number of completely different tacks.

Agent Technology

The idea here is to anthropomorphize a hunk of software. When attempting to design a big, complicated software system, let's think of that system as a little man who single-mindedly pursues our desired task. What would that little man think? What would be his motivations? How would we implement his desires? Computer users first saw agents as little "help wizards" who would pop up on the screen and offer assistance. The concept was sound, but the implementation succeeded only in annoying millions of people.

While software agents as "intelligent help systems" or "proactive user manuals" weren't much of a hit, a number of people realized that they might be useful in a narrative context. Joe Bates at Carnegie-Mellon University developed this concept with the Oz project; that project was followed by more ambitious efforts by Barbara Hayes-Roth at Stanford and Joe in his new startup. While the results so far are interesting, nobody using this technique has generated anything that can be called interactive storytelling. Agent technology has utility in a variety of fields, but it's still not a technology for interactive storytelling.

Story Generation Systems

The concept here is quite old: Can we write software that generates stories? The first such effort, TaleSpin, was built in the 1970s, and it was very impressive at that time. Then came Brutus, a system narrowly focusing on betrayal as a narrative schema. Brutus was an impressive demonstration of AI techniques, but as a technology for interactive storytelling, it suffered from two problems: First, it could handle only betrayal, with no indication that other themes could be addressed, and second, it offered no handle on the problem of achieving interactivity. Brutus generated a story using parameters supplied by the user, but the system was intrinsically inimical to interactivity; it plotted out the narrative arc in a single pass.

A variety of other systems are also under development, but have received little attention. Certainly a great deal of talent is attacking this problem, and I expect to see a great many interesting experiments in coming years.



Chris Crawford on Game Design
Chris Crawford on Game Design
ISBN: 0131460994
EAN: 2147483647
Year: 2006
Pages: 248

flylib.com © 2008-2017.
If you may any questions please contact us: flylib@qtcs.net