Orthogonal Unit Differentiation or “making sure each piece of game material is unique to all others in some way” is tricky to get right. Without it you run the risk of strictly dominanted material, that is some material x which is worse in every way to some other material y . Here’s an illustrative example, let’s design Very Good Racing Game as a simple car-racing game. Let’s limit the game material to cars and tracks. Tracks are represented by simple 2d designs, cars have attributes which define them further, let’s use top speed, acceleration and handling, each with a value from 1 to 5. Here are some initial designs.

green red blue You may call them terrible drawings, I call it rapid prototyping

Now we have 3 cars each suited to different types of a track. Green is best for sharp turns, Red is best on long straights and Blue is best for shallow bends or short straights. If a track is designed with all 3 in mind then each will excel at different times, done correctly this will enable the player to make impactful decisions – something all development should strive for.

track A simple track

In this simple example with only 3 attributes each car can be easily visualised using radars, this helps us to identify the real villain of the piece, strictly dominated material. Strictly dominated material would be any material that is entirely contained within another in a radar plot. So let’s add another well vehicle for people to use, a bike.

radar Basic radar plot

One easy way to balance the cars would be to set a total value for all attributes, say 8, and have one car for every possible combination with each attribute having atleast 1. This gives 18 unique combinations of attributes summing to 8