Game Development Reference
In-Depth Information
Original games are much harder to plan because the designer has no
foundation of well-tested designs to stand on. An original game composed
of a collection of original mechanics controlled through an original inter-
face set in an original world is a giant web of interlocking uncertainties.
In this sort of situation, the correct planning horizon can be a day or less.
Any plan made a week into the future will be invalidated by tomorrow's
surprises about what does and does not work.
The appropriate planning horizon tends to lengthen over the course of a
project.
At the start of the project, we stand on shifting sands of assumption.
By the end, we're worrying about tiny details within an established struc-
ture. A project might start with a planning horizon of less than a day, as a
small group of developers try out wild ideas. The last few months might be
planned out up front in a spreadsheet listing every art asset and program-
ming task to be completed before it ships.
When the cost of testing is low, we should plan to a shorter horizon.
At the start of my combat design process, I could build and test a
combat idea very quickly. Why spend an hour analyzing an idea when I
can build and test it in 15 minutes, and get a lot more information about
it? It's literally not worth the cost to think hard about it. So I don't think. I
just toss stuff in.
This is the benefit of good tools. It's not just that the tools let you make
the game faster. It's that they change the trade-off between planning and
building, and allow a more experimental development approach by reduc-
ing the cost of failure. Good tools let you take risks. This is how they let
you discover designs that you could not notice if work was so slow that you
had to plan and get everything right the first time.
Plan more deeply when your goal is to make conceptual leaps.
Iteration is what's known as a
hill climbing
algorithm. Imagine every
possible game as points on a landscape. Points on a higher elevation are
better games. Iteration makes the game act like a blind mountaineer who
climbs whatever slope he is standing on. It takes short steps, tests them
