Is Uncertainty any good ?

I have to start with a disclaimer: Quantum Physics is just a metaphor for me, a source of inspiration and, in the context of interactive storytelling, is applied to entities (objects, characters, etc…) instead of particles (electrons, etc…). The size of Entities in our case is far bigger than particles. Such big objects are governed by the classical law of physics in our “real” world.

A possible relationship in-between the Quantum world and the Newtonian world in which we live is still subject to debate but a lot of frontier scientists start to prove that such a relation exists and that some of the strange laws of Quantum Physics can also be applied to the world that we experience every day.  I will not try to argue the truthfulness of these statements here but I must say that those Quantum Physics laws applied to full scale objects serve very well my purpose of designing a new kind of Interactive Storytelling paradigm.

The first of the principle that I find extremely useful is the Indeterminacy principle (http://en.wikipedia.org/wiki/Quantum_indeterminacy), that states that particles (entities in our case) car be in multiple “superposed” states at the same time. For instance, being at many different positions at the same time. When observed, those particle’s states “collapse” in a single observable state. The question “why” this happen is still debated in the scientific community. A lot of people think that it’s due to our own consciousness as being an observer. Without having the reply to the “why” question, this phenomenon proves very handy if applied to Entities in our interactive storytelling context. Here is why:

The Indeterminacy principle allows us to split entities into two categories: the one observed (collapsed) and the one unobserved (not yet observed or not any more observed). Let’s name the set of unobserved entities: the backstage, as in theater. The introduction of a very large backstage where nearly the entire world resides (a lot of things are unobserved) gives us a tremendous amount of flexibility. Indeed, our procedural drama manger, all of a sudden, has a large amount of “levers” that it can safely moves to change the story world in backstage in accordance to what the payer is doing on stage (the observed entities).  Those levers, that the drama manger has, are the probabilities of each possible value of each attribute in each unobserved entities (that’s a lot of levers).  

One of the first objections that come in mind is: how we guaranty a coherent story world to the player? This has to do with “temporal causality” and I will try to explain my ideas about this topic in another post. Quantum Physics metaphor can also help a lot there. For now on, we can just acknowledge that we created a flexible backstage to allow our drama manager to safely change the world in the back of the player.

The advantage of storing the different superposed states of entities in the computer memory with their probabilities of being selected at the next collapse time is twofold:

1.       The drama manager can influence the world “analogically” by changing entity states’ probabilities. That is to say, it can store in the backstage multiple dramatic Intentions, for instances: raising the probabilities of having the dramatically needed characters near the player in the backstage, and let the collapse process statistically choose one unique state amongst the possible ones at observation time. This can be seen as a “multiple intentions” and “multiple point of view” statistical reconciliation. One more time, we have to ignore the temporal causality problem for now.

2.       It allows the drama manager to keep track of what the player knows (observed or been observed) and what he doesn’t know (not yet observed or observed since a long time).  This is very important to create a real-time “player model”. That is to say, a dynamically built model that store the player knowledge, values and state of mind.  Tracking what the player knows also allow us to maintain the experience believability, and this while keeping a maximum of flexibility in the story world. 
Here is a simple example: imagine that the story world is full of objects to pickup.  At the beginning, all those objects are simultaneously in all rooms of the story world.  The room where the player stands is the only one observed.  During the “observation” process, all the objects in the story world are considered for collapsing, this work as follow: objects that are statistically not in the observed room are reduced, that is to say : the current observed room is removed from their possible position states. The objects that statistically fall in the observed room are collapsed, that is to say: the player room becomes the only location state that remains for them.
As the drama manager implicitly knows how much object’s position are known by the player (it is intrinsically stored in the world), it can increase or decrease the probabilities of finding other objects in the surrounding rooms (that are still in the backstage). This without breaking the believability of the experience because the player did not yet observed those surrounding rooms. In this example, this is used to regulate the pace of object discoveries without breaking the illusion of coherency.
This can go a bit further if the objects that are not observed since a long time can, again, take multiple positions at the same time, as if they were moved by an undetermined NPC.
This is a very simple example but if we see the drama manager, amongst other things, as a coherent pace manager this can become very handy.

The example I used here is trivial and could be implemented in a simpler way. But imagine that the entire story world works in the same way and that all entities obey to Indeterminacy, for instance: the relationship in-between the characters, the discussion topics, the opinion the character has of other characters, etc..  All those unphysical entities can also be split as being in the stage or in the backstage (observation time become player/NPC dialog time).  This bring to the procedural drama manager a lot of flexibility (it can always change his mind) and due to the linked nature of those non-physical entities, the “simultaneous multiple states” implementation maybe become the simplest one.   This could reconcile reactivity, power of drama and believability.  Off course we will have to tackle the problem of temporal causality first…