Game Development Reference
In-Depth Information
FIGURE 15.1
By not taking the travel time into account, the AI fleet does not realize that
it is wasting significant time alternating between two distant goals.
When the fleet approaches the city, the player moves the units back into the
city. Simultaneously, the player removes units from a distant and otherwise safe city
somewhere else. The AI, upon recalculating the situation, views the
second
city as a
prime target—especially when compared to the original target that is now mysteri-
ously bristling with defenders. Using the same, simple utility model, the AI picks
the “best target� and sets off.
The player can repeat this process with a minimum of attention and effort. It
also effectively takes the AI player's navy completely out of the game. None of the
player's cities get attacked because the AI can't ever settle on a decision—at least not
long enough to act on it. If the naval forces are an
invasion
force to the continent,
the benefits are even more significant. The player can live quite comfortably as
long as he needs to in relative peace. He has not built defenses of stone and steel; he
has built a defense of flawed logic and confusion.
Oh yeah… and the AI looks bad. We may as well put him in dark glasses and a
stupid hat.
There are a number of ways that our AI can stumble in the manner exemplified
above. The AI in the above scenario fell prey to one of the classic blunders (the most
famous of which is, never get involved in a land war in Asia): It didn't take into
