Jurgen, I think your list of possibilities is absolutely on the mark.
Except for the fact that we have no real RPG combat in our system,
each of your suggestions has been implemented to a degree in our
system. We have a physical model with all your properties, as well as
a very general implementation of the sensory properties David Baggett
suggested in his earlier post of the 24th. I agree with him
completely that different sensory modalities must be handled
differently. For example, you would like speech to be muffled when
you are eavesdropping on two characters through a closed door. But
perhaps if the door had a glass window, you could use your lip-reading
talents to help you understand.
One nice property that makes everything "feel like `real'" is that all
of our descriptions are generated by the computer. There is very
little canned text, except at the level that a book is denoted by the
text "book." Each object in the world must have properties that allow
a reasonable description to be generated. The down side of this
stance is that our text is boring to read, and completely devoid of
any style. One of our students is pursuing how to stylistically
generate text to avoid this problem. I'm afraid, however, that this
is a monumental task.
Our characters (NPC's) are weaker than we would like. Currently,
characters we have built have integrated their actions and emotions,
but have weak language ability. (One of our students is pursuing his
thesis work in the integration of all three.) As a consequence, our
most sophisticated characters are animals, including a cat named
Lyotard who is scarily realistic. If you feed him, he may become your
friend. But, if you chase him around, he won't come to you for
feeding unless his hunger overcomes his fear.
We also pursue one idea strongly that is rarely mentioned here. We
want our interactive fiction to be more like movies or plays, and less
like puzzle-fests. For this reason, we study how to best track and
react to the emotional response of the user in his or her experience
with a piece of interactive fiction. Our idea is to gently guide the
users experience so it conforms in some way to an artistic destiny,
while at the same time allowing the user complete freedom of action.
In this way, for example, stories about difficult choices can be go
from being voyeuristic (where the character in a movie is making this
choice) to personal (where YOU the user is making the choice.)
So, Jurgen, we need to talk timetables. The physical world model took
about 1.5 years for me (with my advisor) to design and implement. I
think that if you've thought about these issues you could come up a
preliminary design for your system. I would choose a couple of the
properties you mentioned and try to design a system consistent with
them. Perhaps your feelings on treating NPC's and players the same,
which entails multiple players, and not making "the mistake of
dividing the world in characters, objects and area's" while keeping in
mind David's sensory thoughts would make a good initial step. I would
say the character work is quite ambitious, and if you designed your
physical system to be receptive to later adding the characters you
want, you have gone far to creating a good adventure design tool.
Oz Project Member
... about the Oz system ...
>One nice property that makes everything "feel like `real'" is that all
>of our descriptions are generated by the computer. There is very
>little canned text, except at the level that a book is denoted by the
>text "book." Each object in the world must have properties that allow
>a reasonable description to be generated. The down side of this
>stance is that our text is boring to read, and completely devoid of
Isn't this problem inherent in the term fiction, doesn't it imply that
we want someone to author the story rather than just tell us the
events that results from the simulation. In short does it not put the
finger on exactly the difference between interactive fiction and a
>We also pursue one idea strongly that is rarely mentioned here. We
>want our interactive fiction to be more like movies or plays, and less
>like puzzle-fests. For this reason, we study how to best track and
>react to the emotional response of the user in his or her experience
>with a piece of interactive fiction. Our idea is to gently guide the
>users experience so it conforms in some way to an artistic destiny,
>while at the same time allowing the user complete freedom of action.
>In this way, for example, stories about difficult choices can be go
>from being voyeuristic (where the character in a movie is making this
>choice) to personal (where YOU the user is making the choice.)
Now you are starting to talk about literature and the *art* of
interactive fiction, this sounds very interesting. Although I do not
really see that you need an extensive simulation for this. I think
this is possible in any of the simpler tools also, since it is mostly
on the part of the author, and would if considered in 'games' today
make for much, much better playing/reading. In fact I think these
ideas would be even simpler to implement in a not so advanced system
as you wouldn't need to try to 'gently guide the users experience',
but could just tell it straight and then set things up to allow for
the experience (of course, not at all such an interesting research
project, but would really allow for better games/literature!).
>Oz Project Member
Little languages go a long way...
Thomas Nilsson Phone Int.: (+46) 13 12 11 67
Stenbrotsgatan 57 Phone Nat.: 013 - 12 11 67
S-582 47 LINKOPING Email: th...@softlab.se
That is: assuming that the project is ever at a `stable' state, will it be
released to the net, and then presumably subsequently upgraded?
FWIW: I used to work at ICOM Simulations, Inc. and was the project leader on
Deja Vu II: Lost in Las Vegas. My teammates and I used to spend a lot of time
talking about a next-generation FRP system (surprise, surprise).
I'm a fluent Common Lisp and Scheme hacker, and can wander around Natural
Language Processing issues pretty darned well, thanks. I'd _really_ love to
take a shot at integrating Oz characters' language skills with the world/
emotional models that Oz apparently supports.
Oh well. I'm not at CMU (sniff), so I suppose any chance of that dream
coming true is quite limited. ;-)
Clicking his heels together three times...
I have been getting a number of requests for more information
about the Oz project. A very good first reference for our
work appears in the journal PRESENCE. Here is an exact reference:
Bates, Joseph, 1992. Virtual Reality, Art, and Entertainment.
PRESENCE: Teleoperators and Virtual Environments 1:1, 133-138.
MIT Press. Cambridge, MA.
Better than I can in real-time, this article presents our views
thoughfully. Its bibliography contains references to our more
technical papers, if you are interested in the details.
Concerning Thomas's comments I agree. A pure simulation is
at best as amusing as real life. Although some may argue that life
can be very amusing, it usually does not capture what we want out
of interactive fiction. If it did, would we be talking? However,
the role of an author is not gone in a simulation. The objects
must be created, and the spaces layed out. If this is done
carefully, there may be style in the design.
It is also one of our objects to do stylistic generation. One of
our students is studying how to generate emotional speech from our
characters. This basic technology might also be useful for designing
systems to generate interesting descriptions of either events or
A key difference between a writer of traditional fiction and that
of interactive fiction might be that the actual text written by
the latter may be close to none. I can imagine a system where
the author must build the objects and spaces of the world, create
the characters and personalities, and provide some notion of story.
The art of the "writing" (as opposed to the imagining and creating
the world) could be in programming the stylistic generation of
text. Of course an author could specify changes in this style
for different parts of the story.
Let me close by saying that a realistic simulation may be the
best way to immerse the user in a dramatic experience. If you
"tell it straight" as Thomas suggests, then perhaps you have
lost your interactivity. If what Thomas means is merely a
prelude to the real experience, then that may well be a good
technique for warming up the user.