Game Development Reference
In-Depth Information
What's more, if we added a fifth homework assignment, we would not have to
consider the
weight
of that assignment beyond the homework category. We don't
have to consider if the points for the new assignment are in balance with the quiz
tomorrow or the test this Friday. As long as it blends in well with the existing four
assignments, we are finished. The
result
would cascade out and down toward the
final grade just like any other changes, but only based on the already established re-
lationship that
all
homework be taken together and (in this case) figure in as 30%
of the final grade.
Noting a couple of more items of interest in the example in Figure 3.7, the 91%
grade on quizzes seems good until you realize that, at 20%, the quiz weighting isn't
enough to bring the grade up that much. Even if the teacher gave 20 more quizzes
and nothing else, if you maintain that 91% grade on them, the final grade would
not change. Even if the quizzes were worth 100 points each and you received a
grade of 91% on them, the final grade would not change. That is another one of the
features of building decision algorithms in a modular fashion such as this.
That simple act of grouping like factors together to determine an aggregate
also cuts much of the complexity down. If all items, whether test, quiz, or home-
work, were to be tallied up without weighting, every time we added a new one of
any
of those, we would have to compare its weight to
all
of the others that have
come before. In actuality, in the school environment, if you wanted to maintain a
certain spread of the weights, you would even have to plan into the future and
preweight everything you would do for the whole term. So by simply constructing
a logical, mathematical algorithm, we not only make things simpler to use, but
maintain more control over what we want to see as an end result.
I
NTHE
G
AME
Expanding the Engagement Decision
In the “engagement decision� example in the previous section, we used only one
criterion, “agent's health,� to decide whether we should attack, hide, or flee.
Unfortunately, the data that we use to construct our decisions is rarely as simple as
that. Something as simple as “agent's health� is usually not the
only
consideration
in making a decision. Often, we need to account for other factors. For example, in
the engagement decision above, we may want to take into account such things as:
Agent's health
Enemy's health
Agent's weapon
Enemy's weapon
Number of enemies
