Last week, I talked about two common types of numbers used in RPGs, statistics and traits. Statistics are only applicable in certain fictional contexts, traits are rationalized during play.

Of the two types, statistics are much more common. How should we choose the basic statistics for a new game?

Start with the use-cases

This is the most common error. While attributes can describe a character, that’s not what they are made for. They are to be used as input for certain mechanics. So you have to have an idea of what mechanics there are. You can still fiddle with the internals, like what dice are to be used. But you need a good idea of what people will actually roll for during play. What are the regular rolls that are made each session? If you have that, you can also balance your statistics by grouping the less prevalent cases under one statistic.

Make sure you actually need a special statistic for a use case

The way attributes and skills typically work is that different characters have them at different levels and these levels change during play.

Make sure this is what you want. For example if all characters should be able to do this and get better at the same rate, you can just use a general character level.

If characters should have different capabilities in this regard and these relative differences remain fix, you can just give a bonus based on character class or something.

Ignore symmetry

Many games do something like x physical, x social, x mental statistics. If you follow that route, you have to find viable use cases for each number in your roster, which is hard. Of course, if you just follow the previous point, you won’t have this problem.

Use none or all for subsystems

If you have extensive subsystems, say combat, either make all statistics in as category usable there or none. For example, if your main phases of play are courtly intrigue and and hunting monsters, you can either have separate statistics for both areas. Or you can each statistic do double duty in both areas. You can make the pairings however you like, because according the first point, you started with two separate sets of use cases.

Keep layers ortgogonal

Many games use more than statistic in rolls. You do not just roll A, but A and B, combining the two somehow.

The relationship between As and Bs can be of two kinds. Bs could be specializations of As. That means each B is used with exactly one A. On the other hand As and Bs could be separate layers and you combine them case by case.

If you do this, make sure that each A can be used with each B without doing mental gymnastics. It probably won’t work, if As are attributes and Bs are skills. It might work if As are personal traits and Bs are environments, relationships, motivations… Cortex+ Games do a nice job here.

tl;dr: Do not recreate the World of Darkness.