Working backwards from the end-user application of Jerome Wiesner, we can
look at a demonstration program which offers a compelling visualization of a
content base annotated by keywords. "ConTour,"[8] a visualization program written
by Master's Candidate Mike Murtaugh, provides a mechanism for exploring a media
database, given a keyword approach to annotating media objects. In this
program, the story engine dynamically sequences video clips based on a dynamic
weighting of what has already been played. This mechanism insures the
appearance of continuity. Using the video included in the Jerome B. Wiesner
project, we can begin by selecting the descriptor "Personal Style," for
example. The relevant descriptors grow dynamically with the playout of a
segment. Therefore, if we link "Personal Style" and the "Washington Years,"
the system presents us with a segment from Amar Bose. From here, the system
will look for the combination of descriptors or, failing that, one of the two
descriptors. This might cause the "Washington" descriptor to shrink because
it no longer applies. "Personal Style," on the other hand, remains large. At
any point the viewer can "steer the story" by selecting a keyword. The
dynamics of this interface permits a critical visualization for content; it
invites rather than requires our participation, while emphasizing content-based
continuity.
In addition to exploration, interaction can promote emotional transference
and role-playing by the audience. In a recent piece, "Lurker," the audience is
situated as a supporting character in the story. In cyberspace, a Lurker is to
a hacker, as, in the music world, a groupie is to a star. "Lurker" is run on
our World Wide Web server once 6 people have volunteered to participate in the
story. Shortly thereafter, the Lurkers are given a call to arms: a hacker has
disappeared; they can help find her. The Lurkers work collaboratively as a
"society of audience."
Written in 1995 by a former student, Lee Morgenroth, "Lurker" is an example
of a larger genre which we call "Thinkies;" a "Thinkie" challenges the
audience to think in a style which is appropriate to their circumstance in the
narrative. Through this experience, we believe that the audience can gain new
skills for problem-solving. Normally, this project takes five days to play
out; events and postings of materials are released in a timely way, controlled
by a clock. After the audience/player registers for the game, she is assigned
a pseudonym and invited to explore the Toad Sexer's Pad; one section of the pad
contains all the WWW home pages of the characters. Simultaneously, they
begin to receive e-mail from the hackers and from other Lurkers. Soon they
receive a call to arms from the hackers. One of the hackers, Shira, has
disappeared; the Lurkers are asked to help the hackers find her. Because Bippy
wears a head-mounted camera, the hackers are able to post a video which shows
Shira disappearing down a catwalk. In the process of unraveling the situation,
the audience members perform a series of more or less difficult processing
tasks on the computer. For instance, in the original version, you install and
learn to use a PGP encryption code on your computer. Installing it requires a
"hackerly" mindset. By using this code to unencrypt a picture in the context
of an engaging story, you acquire some thinking and doing skills. I believe
that Thinkies are applicable to many educational contexts. The simulation
running here will provide the attendees with a taste of the complete
experience.
Moving beyond these demonstrations, I would like to consider the future of
interactivity. The astonishing growth of activity on the WWW suggests that a
new medium, which has been trying to birth itself for the past 15-20 years, is
finally at hand. What narrative directions will be most appropriate to future
media? What modes of interactivity will compel us to participate? How will
the society of audience be shaped as the technology moves forward?
This year we began a new project which we call the World Wide Movie Map. In
this project, we hope to engage an international community to participate with
us in creating a repository for personal explorations of place. If the project
succeeds we should be able to hand shake from place to place around the world
by the millennium. Ultimately, we hope that you will all participate with us
to create portraits of Sydney, Melbourne, Cairns and so forth. This idea of
individuals mapping the terrain of the world grew out of an old exploration in
simulated travel, the Aspen Movie Map, which was constructed at the MIT
Architecture Machine Group in 1979-81. The World Wide Movie Map will emphasize
the role of the amateur in building this sort of simulation; as a research
project, our interests are split between understanding community storytelling
in an electronic environment and in understanding the tools which are needed to
increase access to the WWW as a distributed publishing and distribution
enterprise.
In this project, we will provide a navigational substructure, perhaps built
on the Argus Map database, and some tools for submitting material. Our tools
will help you to publish content and attach keyword annotations to your
submissions. Image processing tools might include stabilizing shaky video
shots, translating video into still imagery using "salient still"[9] technology developed at the Lab
and, perhaps, searching for similar pictures. This project introduces a new
level of collaborative and playful construction and should eventually
incorporate all of the modes of interactivity which Jean Piaget observed in
children's game-play.
Another graduate student, Kevin Brooks, is approaching dynamic movie playout
by designing "Agent Stories." The environment contains agents which understand
how to use story parts, such as character introduction, story introduction,
diversion etc. This work has allowed us to think about a new kind of soap
opera. Instead of having a fixed set of characters whose lives move forward
en masse, we might choose to build a distributed character set whose
paths cross infrequently but in profound ways. In a daytime show based on this
premise, a new character could be introduced each week. Characters could be
developed more or less "thickly" and could potentially be written by different
writers. The audience's expectation could be woven into a serendipitous
déjà vú which would occur as the audience happened
to be watching the other side of one of those fleeting interactions.
Working in video, I find myself constantly battling the bandwidth issue. In
order for dynamic, interactive narrative to become a widespread future form, we
need to allay fears about limitations in the bandwidth of the delivery system.
For this reason, we are collaborating with Media Lab researchers who are
working on the problem of model-based or structured video. Using this approach,
the temporal stream is constructed by assembling lower-bandwidth objects
on-the-fly into a 3D representation. This method offers the potential for
extreme compression as it does away with a frame-based representation. Rather
than shipping the whole frame once every 30 seconds, only changes in the frame
data need to be delivered to the presentation device.
All of these projects incorporate strategies for using a two-way channel to
personalize and globalize communication. They provide examples of how the
language of interactivity may affect story structure and, by emphasizing story,
celebrate the diversity of human endeavor. Today, I have the sense that our
culture has finally become deeply democratic; this is reflected more in the
changing nature of our communications systems than it ever was in the
framework of liberal politics. Perhaps we can trace the "home page" to Andy
Warhol, who suggested that everyone in the world had the right to 15 minutes of
media fame. Of course, if I can make something and a million -- or a billion
-- people want to see it, that should also be OK, too.
I think Piaget's taxonomy provides a valuable benchmark which we can use to
describe approaches to the interactive story experience. Both designers and
audience can measure their interest in incorporating some or all of these
interaction types: practice, pretend, rules of the game, and construction.
However, these interactions must be meaningfully mapped into the story
environment. Because the medium is very young, we do not always understand
what we are making. The wonder of being an "interactive multimedia" author
today lies in the discovery, the effective surprise of the creation. The
author only has a limited ability to previsualize a project before it becomes a
functioning system. I encourage designers to throw convention to the wind, not
to dismiss visual intensity of a presentation, but rather to disregard -- in
the beginning -- decisions about buttons and mouse clicks and menus and
branching, and concentrate on finding an underlying structure which can be
driven procedurally and presented with a rich dimensionality. The design,
which grows procedurally out of this underlying structure, can be extensible as
well as memorable.
With that I end my talk and invite questions.