Chris Crawford on Interactive Storytelling (38 page)

True, but the fact is there are no ideal words for what I’m seeking. Here’s a long-winded example of the problem: Consider the terminology for defining colors. That terminology is completely arbitrary; in fact, different colors slice up the color spectrum in different ways. Some cultures merge what we call blue and green into a single color. Some cultures define just a few colors, and other cultures recognize many colors. The words and the concepts are all completely arbitrary. However, in terms of the human eye, there are three basic colors. That is, any color distinguishable by the human eye can be defined in terms of three numbers. A common system for defining color is RGB, for red, green, and blue. Any color can then be defined as some combination of red, green, and blue values. Or there’s HSB: hue, saturation, and brightness. Again, any color distinguishable by the human eye can be defined in terms of those three numbers. I suppose hundreds of different three-dimensional systems could define color, but here’s the rub: No color terms correspond precisely to the mathematical elements of these color systems. The term “red” covers many colors, but “red” in the RGB system is a precisely defined color. For conversational use, the sloppy term “red” works fine, but for mathematical use, the term “red” in the RGB system has a much narrower meaning.

In the same fashion, my use of terms such as
Virtue
or
Power
is mathematical, not conversational. I capitalize terms that have a special meaning in my models to differentiate them from common usage. These terms in my system don’t correspond precisely to their meanings in normal language. They are storybuilding-specific versions of common terms; they suggest the concept in use but don’t properly define it. Accordingly, I must supply precise behavioral definitions of
how these terms work. Rather than argue about the semantics of the terms, in the following sections I show how they actually operate in a personality model. If you don’t like my usages, you’re welcome to create your own. Good luck; English words never quite fit into mathematically neat slots.

One of the more confusing problems in personality modeling concerns the polarity of variables.
Unipolar
variables have a maximum value and a minimum value of zero. In other words, Actors can have lots of
Integrity
or zero
Integrity
, but they can’t have negative
Integrity
. Similarly, Actors can have positive
Wisdom
or zero
Wisdom
, but they can’t have negative
Wisdom
. That’s the unipolar model.

In the
bipolar
model, variables have a maximum that’s positive and a minimum that’s negative. The value of zero indicates a middling or neutral quantity of that variable. Thus, in the bipolar model, an Actor with zero
Integrity
is average, a dishonest actor has negative
Integrity
, and an honest actor has positive
Integrity
.

The choice between unipolar and bipolar variables brings up a variety of tricky considerations. In some cases, a variable that makes perfect sense as a unipolar variable is nonsensical when treated as a bipolar variable. For example, it’s easy to understand a high value of greed, but what would a negative value of greed imply: charity or generosity? If you wanted to build an inclination formula representing the intensity of an Actor’s desire to obtain some trinket, you would naturally multiply the value of the trinket by the Actor’s greed. In this case, you would expect that a saint would simply have zero desire for the trinket. But using a bipolar value for the greed variable would make the saint’s desire for the trinket negative—is that reasonable?

My own experience has led me to the conclusion that bipolar variables are, in general, more practical to use than unipolar variables. However, I am certain that a clever designer could create a perfectly workable personality model using unipolar variables. And nothing prevents mixing unipolar variables with bipolar variables, so long as you keep them straight.

The problem that drove me to using bipolar relationships was rather messy. I wanted to determine the reaction of one Actor to another’s behavior toward a
third. To make this easier to follow, here’s an example: How is Mary to react if she observes Fred do something to Tom? Will her overall reaction be positive or negative? That depends on two factors: how nice or nasty Fred’s action was, and how much she likes or dislikes Tom. In other words, there are four possible variations:

Fred does something nice for Tom, and Mary likes Tom: Mary is pleased.

Fred does something nice for Tom, and Mary hates Tom: Mary is displeased.

Fred does something nasty to Tom, and Mary likes Tom: Mary is displeased.

Fred does something nasty to Tom, and May hates Tom: Mary is pleased.

Now, to model this behavior in arithmetic relationships, you need to multiply niceness/nastiness with affection/hatred, but you
also
need to ensure that both niceness/nastiness and affection/hatred are bipolar values. In other words, both nastiness and hatred have to be negative numbers. Here’s the same set of statements using arithmetic values:

Fred does something +2 for Tom, and Mary feels +4 for Tom: Mary is pleased +8.

Fred does something +2 for Tom, and Mary feels -3 for Tom: Mary is pleased -6.

Fred does something -4 for Tom, and Mary feels +4 for Tom: Mary is pleased -16.

Fred does something -4 for Tom, and Mary feels -3 for Tom: Mary is pleased +12.

This example shows why relationships need to be bipolar. And because a relationship is nothing more than a perceived intrinsic variable, these variables must be bipolar, too.