Sometimes, it isn't enough to help players help themselves (read the last entry!). Sometimes, it's a more fundamental fault with the game mechanic itself. In this entry, I want to look at a way to evaluate and reason about the complexity of a game mechanic, namely through cognitive load theory.
Cognitive load theory, proposed by John Sweller, offers us guidelines that can help encourage and optimize player learning through the way information is presented. It's actually much in the same vein as Nielsen's recovery-from-error heuristic, but this theory will help us examine the intrinsic difficulty of a mechanic rather than error recovery. You can think of it as a way lower the chances player errors will occur in the first place.
The theory is based around a simple framework that posits three types of cognitive load: intrinsic, extraneous, and germane. Let's start with the first one.
Intrinsic cognitive load is the execution difficulty of a mechanic. For a real world example, using chopsticks is harder than using a fork. Therefore, using chopsticks has a larger intrinsic load than using a fork.
For a video game example, in Street Fighter, doing a hadouken, a motion special move, requires a directional motion down-forward and a button press. This is more difficult than executing a normal move which requires only a button press. This equates to a special move having a larger intrinsic load than a normal move, and results in the complexity of special moves going up. This type of cognitive load can be lessened on part of the developers by making the mechanic easier to execute and on part of the players through practice. For the player to overcome intrinsic load is for the mechanic to become muscle memory. The question intrinsic cognitive load asks is how much will the player need to execute the mechanic, so they don't have to think about it to do it?
Extraneous cognitive load is generated by the manner in which information is presented to the players and is under the control of the designers (Wikipedia). For a real world example, I really like the one on Wikipedia: let's say we want to describe a square to someone who has no idea what one is. We could try to describe a square as (courtesy of Google definitions) a plane figure with four equal straight sides and four right angles. Or we can just draw one. The verbal description is far less efficient than the visual and generates extraneous load.
For a video game equivalent, in League of Legends, abilities are described in a variety of ways: text, video, or player execution. Let's take a look at a simple ability: Mystic Shot. We can see it has a range of 1150 and a speed of 2000, and here's its description: "Ezreal fires a bolt of energy in a line, dealing physical damage to the first enemy hit..." (LoL Wiki). We know that Ezreal is the name of the champion, but does "a bolt of energy" mean anything in terms of gameplay? Or what about the width of this bolt? How fast is 2000? How do we aim it? The video fares a bit better at describing the ability. We can see that it will shoot in the direction of the cursor, how fast 2000 is, the width of the Mystic Shot, and that the "bolt of energy" is just a visual description. What's interesting is that we will end up learning just as much as the video as we will by executing the ability. (Of course, Mystic Shot has a larger amount of intrinsic load than say a regular attack, so it will require practice to do what the player wants.) The point is, the tooltip has a larger extraneous load than a video or player execution. What's also interesting to note is that the video and player execution may sometimes fail to give the player the full story, so it's still important to have that text description to fill in the gaps. The question extraneous cognitive load asks is how efficient and comprehensive can the game inform the player how a mechanic works?
Germane cognitive load is the processing, construction, and automation of schemas (a schema, as used here, is a mental framework that organizes and perceives information [both definitions from Wikipedia]). For a real world example, think about the first time you learned algebra. The common refrain is "what are letters doing in my numbers?" The way we've thought about math in our early years makes learning algebra more difficult because it requires a new way of understanding math. Therefore, algebra has a higher germane load than say multiplication or division.
Think of germane load as the mental model that the player has of the game as a whole, and how the mechanic fits in with what they already understand about it. The harder it is for the mechanic to fit in, the higher the germane cognitive load is. Take a look at charge moves in Street Fighter. Charge moves are executed by holding a direction like back for 2 seconds, then moving quickly to a different direction like forward before pressing a button. These special moves can be more difficult than the motion special moves for some players, even though they both have about the same amount of intrinsic load. This can be because charge special moves don't fit in with what players already understand about the game: the motion special moves. The question germane cognitive load asks is how does this mechanic fit within the big picture?
Through these different lenses, we can examine more categorically the complexities of a game mechanic. The main thing to consider is the balancing act inherent in each type. In intrinsic load, we may want to make a mechanic harder to execute to balance out its effectiveness. In extraneous load, we may want to provide a less efficient way of informing the player because they wouldn't otherwise be able to understand it. And in germane load, we may want to introduce new mechanics that vary wildly from previous ones in order to change how the player perceives the game world at hand. Regardless of what design choices we make, we should make them conscientiously and knowingly in how they will affect the complexities within each type.
Recommended Wiki: https://en.wikipedia.org/wiki/Cognitive_load