Game Development Reference
In-Depth Information
It is also similar to the gymnastics and skating scoring systems. Without know-
ing how the other contestants did in any given competition, we can glean informa-
tion from a single score. We know that a score of 9.975 on a 10-point scale is pretty
darn good. We can
not
, however, determine if the recipient of that score won that
particular competition without knowing the scores of the other participants.
Therefore, while a weight relative to a maximum allows us to judge the fitness of an
individual score on its merit alone, we can't determine “the best� until we compare
it directly to other scores.
A variation on this method is that the defined point does not need to be a max-
imum. Instead, we can define an
anchor value
to which we compare our values.
The result is similar, but we now have the possibility that our relative weights can
be greater than zero. For example, using the same bars as in our previous example,
we see in Figure 13.2 that setting the anchor to a value of 10 produces different
relative weights.
FIGURE 13.2
The value to which we are comparing our scores does not have to
be a maximum. We can use other anchors. In the right-hand graph, we are comparing
our absolute weights to a value of 10 to arrive at the weighted scores.
Weights Relative to Each Other
The third method, illustrated on the right of Figure 13.1, is another variation of rel-
ative weighting. In this case, the maximum is not a predetermined value. Instead,
we compare all the scores to whichever score is the greatest. Therefore, the fourth bar
(size of 16) is the
de facto
maximum for the moment. We score the other three bars
relative to 16. For example, the first bar (size of 12) is 0.75 the size of the fourth.
One advantage to this method is that we aren't limited by an arbitrary maximum
score. Because of this, the method scales itself as the scores change. For example, as
a situation changes, the idea of what makes a “good score� may vary significantly.
