Quote:Original post by Wavinator
This is another thought experiment for those interested in dynamic plots...
Long description edited out for sake of brevity...
Would This Experience Be A Story?
Assuming this all happens freeform and plays differently each time, would it be considered a story? A story is usually defined as a set of events which are neatly tied together, with everything happening for a reason. But in this scenario, many events (like arguments) could repeat-- nothing is set just so such that it happens at the right time, only once.
Well, I'd consider this to be a story of sorts, but as a lot of people have pointed out it mightn't necessarily be a very good story if the system doesn't try and make it one. If the events feel disjointed from each other, repeat too often, or just feel too 'procedure' based, it won't gel. If the system however managed to pace it like a story arc, then it could work.
One way I think dynamic story could work is if you have different layers of story events that the system to work with, sort of like milestones that the form the backbone of the story arc. For a murder mystery, for example, there needs to be a murder, and the crime needs to be solved. For a murder to make sense, storywise, there needs to a motive, so that need be justfied, and so on. I guess this would work as a backward logic chaining system (not sure if I got the right terminology there).
Now I come to think of it, I'm sure I've read something similar to that example somewhere before, when I was doing some research into representations of storytelling systems. Let me just check my old notes....
Ah yes, there was a system called the Automatic Novel Writer that generated short murder mysteries developed by Sheldon Klein at the University of Wisconsin in the early 1970s. From memory, the system ran a bit like simulation, giving a log of what the characters did in a old mansion type setting, before eventually one character would murder another, a detective would arrive, and one of the other house guests would solve the murder. Here's an extract of what I summarised about this:
Quote:
The plot is created by a simulation of the behaviour of the characters using a series of bejaviour rules written by the researchers for each event in the story. Sample of the types of events are 'Two friends meet by chance, they agree to play tennis, one of the friends flirts with other friend's wife' and 'Pushing your business partner down the stairs to gain control of the business'. The system also keeps track of time, with various events occurring at different times (such as 'Tea is served at 4 p.m. All guests stop their activities when the butler serves tea'). Sove events may provide the triggers for future events (for example, flirting with someone else's spouse provides the trigger for a lover's tryst and adultery, and this tryst provides another trigger, in this case for murder). The choice of rules is done probabilistically using likelihoods determined by past events and character attributes. Each character is given a numerical value for a number of personality traits, such as attractiveness, sexd drive, intelligences and propensity for violence. Eventually, the novel writer will choose to perform one of the six murder events. This starts the chain of events that starts with the selection of a character to play detective and solve the crime.
This is from my notes, but my reference is 'Automatic Novel Writing: A Status Report; Technical Report 186, Computer Sciences Department, University of Wisconsin, July 1973', so I don't know if it can be easily found on the web (I can't remember how I found it.).
I do remember that the tech. report included source code (can't remember the language though). It was basically a bunch of predesigned scenario events that slotted together to make a story.
Not sure if all this makes sense, and I can't even remember if I stayed on topic, as I should have had lunch two hours ago. Oh well, I've written too much to not include this to the discussion now!