Game Development Reference
In-Depth Information
might look like Figure 2.5. At first glance, the distribution of the percentages looks
fairly inverse of the spread in Figure 2.4 (the probability of any particular one being
in use). However, is that really what we are trying to accomplish?
FIGURE 2.5
The relative attractiveness of each blade is based
on how long I likely used it.
Making a Decision Based on the Numbers
Since I am aware of my proclivity toward over-using blades 1 and 5, if for some
reason I have to go back and select a blade to reuse, I will make a conscious
choice
of selecting from blades 2, 3, and 4 (Figure 2.6). It is statistically more likely that
those are less-worn, therefore making them more attractive to use. On the other
hand, knowing that I probably used blade five to the breaking point, I'm not going
to be terribly likely to even consider it. That much is obvious. However, despite the
fact that blade 1 is probably in far better shape, I'm still going to be quite certain
that it is worse than any of blades 2, 3, or 4. Why would I even consider blade 1,
even though it may be in reasonably serviceable condition? It is far more likely that
I would choose from that middle trio. The result of this is that the odds of selecting
blades 1 or 5 are pretty much nonexistent. The probabilities of blades 2-4 climb
accordingly.
Notice the subtle difference between the underlying logic of the numbers in
Figure 2.5 and Figure 2.6. The former was simply a relative inverse of the usage sta-
tistics. The latter took into account an
intelligent decision
that I, as the shaver, would
have made because I knew about the scenario that created the data in Figures 2.4
and 2.5. Despite the relatively small amount of difference between the conditions of
the blades, I knew that if I had to choose one that was in better shape, it would have
to be drawn from the three in the middle.
Figure 2.4 is based on observation. Figure 2.5 is based on inferences from the
data in Figure 2.4. Figure 2.6 is designed to model an
intelligent decision
that could
be made from the inferences in Figure 2.5. I will cover ways of modeling these types
of decisions throughout this topic.
