Abstract

This paper describes a search for natural ways to make responsive media, investigating the techniques used in oral performances and contrasts these with ways which start from a fixed, recorded form.

It starts with the origins of storytelling and examines the way in which the epic poet constructs stories from themes and formulas as they are being performed. It points to similarities with entity–component–system techniques in games programming.

It continues by looking at a form of improvised theatre from Renaissance Italy that combined oral performance with literary learning where the lack of a script avoided censorship. This was structured around scenarios for scenes that were played out through exaggerated stock characters. The computational challenge of creating this approach to storytelling in a machine involves programming interacting behaviours aiming towards a goal.

Finally, the paper examines the challenge of storytelling in the form of a tour guide, where the narrative plays out in different orders in response to the movement through a museum or town and points towards the development of responsive learning experiences.

Introduction

How can we create media that truly responds to the audience, whilst still telling flowing, meaningful and coherent stories? Since the invention of the computer, much of the focus has been on using the structures of a text adventure or video game and forms of storytelling taken from modernist and post-modernist literature such as the branching narrative, as envisaged by Jorge Luis Borges in his 1941 story, The Garden of Forking Paths.

However, these forms of media can be expensive to produce because branching narratives and alternative versions require the production of many more parts of the story than any one audience member will experience. Also, the audience can often end up disappointed by these experiences because they are left wondering what they have missed out.

The BBC first explored the possibilities of interactive radio drama in 2001 with The Wheel of Fortune and the production of the three synchronised versions, following the three different characters, proved highly challenging and very time consuming. Where projects have created explorable or responsive media forms they have often relied on scheduling collections of video clips that can be viewed in any order or human moderation where the interaction includes linguistic input from the audience.

This piece of work has searched for established examples of responsive storytelling that may enable us to create media that both works for a mainstream audience and can be produced at-scale, on a broadcast budget. Whilst video games are definitely mainstream, their budgets and timescales of production are well beyond the reach of most broadcasters.

What follows is a description of some forms of oral storytelling that provide insights into possible approaches for responsive media. These present challenges in the form of artificial intelligence (AI) problems that will need to be solved, along with the issues for the story-tellers in structuring the stories. Some of these issues were first analysed by Janet Murray over 20 years ago, and her book, Hamlet on the Holodeck, (recently updated) was the starting point for this investigation.

Whilst the video games industry has made great progress over the past 20 years, developments in responsive storytelling have been far more modest. The challenges for responsive storytelling are, in effect, core AI challenges, getting the computer to tell the story, or present elements of a pre-built story structure, in response to the needs of the audience.

As with any computational task we first need to model the problem we are trying to solve and work out what functionality the code should provide. We need to build good models of stories and storytelling.

Download the full tech paper below