Game Development Reference
In-Depth Information
Wheat and Chaff
Now that we have looked at all the possibilities, however, we can see that, while
“shoot Evil Genius with the rocket launcher� is, indeed, the lowest score, there
were plenty of other options that were very close. How much worse would it have
been for us to attack Boss Man with the rocket launcher instead? In the heat of
battle, would it have looked completely illogical for us to attack Boss Man instead
of Evil Genius? (After all, he's
Boss Man!
And
he's
got a rocket launcher too!) What
about Baddie 3 (who was 30 feet away with his potent shotgun) or Evilmeister
(only 60 feet away with his machine gun)? Those would certainly have been under-
standable and legitimate targets for us—and we see this reflected in the scores.
On the other end of the spectrum, we can see that there were plenty of options
that would have looked just plain odd. With all these Dudes running around near
and far, why would we elect to attack the ones that are far away with a shotgun that
is all but useless at those distances? That explains the bottom five options in the list.
They are simply out of the question.
Drawing a Line
As we theorized, someplace in the middle of this list, the good meets the bad (by
blending into the mediocre). While we would have been comfortable selecting one
of the “reasonable� answers, we want to avoid the “unreasonable� ones. The trick
is deciding where that line is. Everything under 10? Under 50? Under 500?
Pragmatically, we realize that we can't even use the scores in this list as a guide.
In a different situation, we may not have
any
options that score under 10… or even
50. The scores will change all the time relative to whatever arbitrary line in the sand
we draw. However, one thing is certain: No matter what the scores are, we can always
arrange them relative to
each other
. After all, that's how we determined the best
option before. “Attack Evil Genius with our rocket launcher� was the best
relative to
the other scores
. The solution, therefore, lies not in basing the threshold on an absolute
score but on the score relative to other scores.
Instead of limiting ourselves to
only
the single best item, we can select from a
handful of the best scores. How many options we elect to choose from will be very
problem-specific. (Certainly, you are getting tired of hearing that by now.) We
could, for example, choose to select from the top 10 options. However, what if there
are fewer Dudes and fewer weapons available to us? We may not
have
10 options
from which to select.
Usually, it is a better idea to use the top
n
% of the choices rather than the top
n
choices. In the case of the Dudes, for example, we could decide that our cut-off is
25% of the total possible choices. In this case, we would consider the top eight
items. If we had more options, that number would increase. As we removed Dudes
