Multiplayer - Designing Games - page 189

mediate combined strategies. Each smoothly defined strategy demands a
correspondingly smoothly defined response.
Smoothly defined strategies grow yomi because they allow players to
outdo one another in the nuance of their strategic knowledge. Anybody
can be told that the counter to Marines is Banelings and the counter to
Siege Tanks is Mutalisks. But only the experts know exactly what they
need to defeat arbitrary mixtures of Marines and Siege Tanks. What if he
has 15 Marines and one Siege Tank? What if he has eight Marines and five
Siege Tanks? What is the minimum you need to destroy him? A novice
won't know, but an expert will. This sort of fine-grained interaction is what
pushes the game's skill ceiling into the sky.
Yomi play grows from complex, difficult-to-quantify payoffs.
Different strategies have different potential payoffs. In an artificial
example, we looked at a rock-paper-scissors variant in which you get $5 for
winning with rock and $1 for winning with paper or scissors. Changing
these numbers changes the proportions of strategies players should play.
And, as you'll recall, the correct proportions are those where each indi-
vidual strategy has the same average payoff.
But what if the numbers weren't handed to us? What if strategies had
potential payoffs with qualitative effects, hooking into multiple goals, at
various levels of certainty? These payoffs aren't known ahead of time, and
they can't be described with a single number. Figuring out a good mixed
strategy is no longer just a matter of equalizing the payoffs of the various
strategies. First we have to figure out what the payoffs are. This evaluation
process maintains flow, and heightens skill ceiling.
Uncertain payoffs also mean that we need to guess how the other
player is evaluating his payoffs. If you know your opponent well, you might
find a place where he overvalues or undervalues certain payoffs and exploit
it to predict and defeat him. At even higher skill levels, you might predict
his guess at your evaluations, and so on.
Let 's go back to StarCraft II . Mutalisks and Siege Tanks are expen-
sive, while Marines and Banelings are cheap. This means that for a Terran
player, shooting down Mutalisks with Marines has a better payoff than
shooting down Banelings with Siege Tanks, since the Mutalisks are much
harder to replace and the Marines are easier to make. The same goes in
reverse—the Zerg wants to kill Siege Tanks with Mutalisks more than
they want to kill Marines with Banelings.