I was reading some rules clarifications for Puerto Rico, and I noticed that there were a lot of special notes and exceptions related to combining the abilities of special buildings. It struck me that exceptions were something to avoid when writing rules.
Ideally, every rule in a game is perfectly clear, and applies at all times. In some games, this is easy to achieve. But a lot of games start with a basic set of rules, and then give you ways to start changing them. In Puerto Rico, for example, the violet buildings all give you ways to change what happens when you perform a basic role action. Even the roles act differently depending on whether you are the one who selected it or not. Occupations and Minor Improvements in Agricola achieve a similar effect. All of the races and powers in Small World do, too.
While these are all additional rules that apply in specific circumstances, they are not what I would consider exceptions. Effectively, each player plays by a different, consistent ruleset. But “exceptions” are different. Exceptions are rules that come out of the interaction between all of the supplemental rules. Exceptions tell you how to apply the rule changes. This makes them meta-rules.
Consider a basic case. A basic action lets you take one wood resource. You have an ability that allows you to take one food when you would normally take one wood. You have another ability that lets you take an extra wood when you take one wood. If you try to apply both simultaneously, you run into a problem, because if you take food instead, you’re not taking wood any longer, and the second ability doesn’t apply. Similarly, if you take an extra wood, you’re no longer taking one wood. In this simple case, you might say that replacing wood with food can be done second, but it still introduces an exception. You have to specify what happens when the rules interact.
The reason this is bad is that the rules become too complex. It can become very hard to include all of these cases in the rules, especially if you have multiple interactions. This makes it hard to write the rules, hard to test the game, hard to learn and play since you need to remember more exceptions. There are, of course, ways to develop a system of meta rules, so you always know which ones to apply. This starts to enter the realm of meta-meta-rules.
There are several ways to avoid this. Because it usually occurs when several permanent effects apply simultaneously, you can narrowly scope the behavior to avoid multiple interactions. You can sometimes simply reword the rules to better explain when an ability applies. In Agricola, some abilities trigger when you perform a specific action (such as collecting wood), but others trigger based on using a specific action space (using a space that provides wood).
A different approach is more mathematical, which treats everything as a transformation. Taking wood is +1w. You add the abilities to that equation, +1w + (+1f, -1w) + (…), so you have a systematic method of evaluating the outcome. This sort of a system also reveals a very robust solution, in which effects apply commutatively or “monotonically”, i.e. in any order or in the same direction. San Juan basically follows this route, because all of the building abilities add to what you do, instead of taking things away.
Of course not all of these methods will work. all the time. Sometimes you want these abilities to interact complexly, and you can’t scope them narrowly. The mathematical approach sometimes puts more work on the part of the player to evaluate the outcome, when a short list of exceptions in the rules would be easier to reference. The exceptions in this case should always be easy to find.
There is one type of exception that deserves special mention; that is the ban. Nick Bently discusses it at length in this article from April. Bans are basically exceptions that say you can do everything normally, except this one specific combination. This type of exception should be avoided at all cost, because they completely break the system of interacting rules. A much more favorable way of achieving the same thing is to add a small exception that prohibits small parts from working together. A small and more general exception might lose a small amount of flexibility, but it will make your game feel more robust.
When I designed New Bedford, a lot of the buildings have actions that look like regular actions with an extra ability. It was a struggle to avoid putting in buildings that only granted abilities like in Puerto Rico. This was partly because I wanted everyone to have access to the same abilities, but the end result is that the design avoids a lot of exceptions by forcing everything into discrete actions. You simply have to do what the action space tells you, without wondering if you are missing something. Nevertheless, there is still a strong lure for me to add in a whole variety of abilities to really mix the game up, but I think you’ll find that the base game still gives that feel, while being easy to play.
Exceptions aren’t always bad. They are sometimes necessary to let a designer achieve his or her vision of how the game should play. But too many can feel like the player is mired in details. Exceptions have a special role in keeping the balance between enabling the player and restricting them. Finding that balance will improve your game, which I can say without exception.