Simulation versus Representation
by Gonzalo Frasca

Page 1 - 2 - 3 - 4

If simulations are not narratives nor "interactive narratives", what are they? The provisionary definition that I am most happy with goes like this:

"Simulation is act of modeling a system A by a less complex system B, which retains some of A's original behavior".

I put together this definition by slighty modifying several definitions of "computer simulation". The reason why I do not here give a more traditional "simulation theory" definition is because, generally, that discipline is more concerned with the prediction of the behavior of the simulated system than with representational issues. If you want more details on this, you should check Chapters III and IV of my Thesis.

For example, Sim City 2000 (system B) simulates a city (system A). As a system, Sim City is less complex than an actual city (i.e. there is not graffitti on the walls in Sim City, nor advertisements on the streets), but it retains some of its behaviors (i.e. buildings need electricty and roads cost money to build).

Probably a better way to understand the difference between simulation and representation is to compare their characteristics. Since I just have used the example of a city, let's compare a representation of a city (London, by Monet) and a simulation (Sim City 3000). Obviously, I just used an image of Sim City 3000 since I can not embed the whole simulation into this text, so use your imagination and pretend that the first one is an image while the second one is a simulation).

Let's start by pointing out their similitudes. Both show the city graphically (even if Sim City also includes sounds). Both include one of the biggest characteristics of cities: buildings -and those buildings follow a certain order. Both can be interpreted or misinterpreted: some people may recognize London on the paiting, while other may recognize NY or any other city on the other example.Both are also partial: they do not show the whole city, but a fragment of it from a certain perspective.

Here is where the differences start to arise. Actually, Sim City could be a 3D model of the city, allowing us to see it from every possible perspective (something that Picasso tried to achieve within paiting) through time. However, we should agree that since the model can not be perfect -otherwise it would be a real city!- the representation is always fragmentary.

One could say that the difference between Monet and Sim City is time: you need time to play the simulation. True, but representation can also be time-based: music and films, for example. Still, we could say that Monet is representing a very particular (foggy, in this case) day in London, while a simulation could show the city under sun, fog, rain, snow, etc. Actually, Monet tried in a different series of pictures to show the Rouen Cathedral under different kinds of light. Each one of these paintings is a particular image. If Monet had a 3D package like Lightwave or Maya, he could have modeled the Rouen Cathedral and then have access to infinite variations on how the light reflects on it. That model would have to include certain rules. In this case, these rules are rules of optics, that can model how light reflects on different surfaces.

Sim City is a model that also includes rules. The user can add and remove buildings, create roads, etc. Those objects also have certain rules: i.e. a factory will produce goods only if you connect it to the power station through power lines. The key concept here is behavioral rules. Sim City is a dynamic system that behaves like a city and also has many characteristics of a city, while the painting only provides the characteristics.

Certainly, instead of using a painting as an example, we could have used a film documentary that showed the mechanics of the city. This film would show narrated events: i.e. a technician connecting the factory to the power station through power lines. Through that story we can infer that factories need electricity, but that rule was interpreted by whoever is watching the film. In other words, the rule can not be dynamically applied to the film: I cannot disconnect the cables on the factory in scene 3 to see if it shuts down. Usually, narrative works in a bottom-up sequence: it describes a particular event from which we can generalize and infer rules (this is why narrative is used so much in education). On the other hand, simulation is usually top-down: it focus on general rules, which then we can apply to particular cases (this is why simulation also works great as a tool for teaching complex rules because, unlike narrative, it allows experimentation).

When systems are not very complex, it is usually better to use representation and narrative to describe its mechanics (remember our first pipe example, the one with the button). But as systems get more complex, simulations become a more attractive tool because they can model the rules that govern the system. If you don't believe me, try to teach somebody to use the gears of a car just showing him a movie or a book. A simulator (as an actual car) can definitively do a better job, because the simulation can portray certain rules like "if you release the shift very quickly, the car will shake and the engine will probably stop" with first-hand experience of the actual relationship between the shift, gear and engine.

Next Page

Go to Page 1 - 2 - 3 - 4

l u d o l o g y . o r g