Game Development Reference
In-Depth Information
a significant time commitment involved in execution of the naval attack. While a
portion of the solution was the continual monitoring of the prospective targets for
a change in their status, the weakness in the process was not monitoring what the
resultant decision would entail.
The “attack� that we were deciding to launch only took into account the ex-
pected odds of success or failure and the relative worth of the target. All of those
factors take place between the time the first shot is fired and the time the last defender
(or attacker) gives up and throws down his weapons. There is no accounting for the
entirety
of the action. Put another way, rather than a decision about “what city to
attack,� the decision needed to be “what city to
travel to
and
then
attack.� By
considering the
whole
process as one decision, time is automatically included in the
calculation.
Information Expiry
In addition to the utility of the time we would spend traveling to someplace or
doing something, there is another reason that we need to consider time heavily in
the decision-making process. An
intentional
exploit by a player is not the only rea-
son we should treat the decision to attack a particular city with caution. In the
highly dynamic environment that games tend to be, information has a “shelf life.�
Units are created and moved, defenses are constructed, values change… in short,
things happen
. Even if the decision to attack a city is a legitimate one based on the
information we have, we have to question the likelihood that the information will
still be valid by the time we get there.
This is similar, but not identical, to the “fog of war.� The legendary “fog� en-
compasses how information may change while we are not able to view it.
Information
expiry
deals with the prospect that information probably
will
change over time
whether we are watching it or not.
We touched on this above when we discussed the value of updating informa-
tion. However, by factoring in the possibility that the situation will be different as
part of the initial decision process, we are taking a more proactive approach. Rather
than the “we can always change our mind� method that continual updates provide,
we are saying, “let's not even start.�
While we could conduct complicated analyses of fluctuation frequencies and
ranges, it is usually simpler to build the
idea
that the state of the world
could
change
into the weight given to the time that we will need to spend. We can do this through
biasing the marginal utility in a nonlinear way. Using a simple exponential model,
we could say that our confidence in the validity of information drops away from
100% at an accelerating rate as time passes.
