When I first started playing roleplaying games more than 30 years ago, I noticed a problem. The games I played would talk about what the average person could do, but then their mechanics were built around average dice rolls. That’s two different sorts of average, and they don’t line up.
When we talk about an average person, we mean a typical person, the sort you expect to find most commonly. That’s the everyday idea of average; it’s a typical thing, it’s what usually happens, it’s what’s most likely, or it’s the default.
When we talk about an average die roll, we’re referring to what’s called the arithmetic mean. This is the sort of average we all learned in math class: add together a group of numbers then divide the total by however many numbers were in the group. We’ll call this average in a technical sense.
An average die roll then (in the technical sense) is the arithmetic mean of all the possible results. It’s usually equal to half the highest possible value, plus one-half, so the average die roll for a d20 for example is 10.5. Many games—the most common games—use this technical average because well-made dice don’t have most-likely results. (That’s the point actually: each outcome is equally probable.)
Unfortunately, the real world does have usual results, so the math used in those game designs clashes with our intuitive understanding of how things are supposed to work. Imagine a situation like driving a car. When I get into the car, I expect a typical outcome: I’ll drive with my usual level of skill and will get the same result that I most commonly get.
Something like the common d20 mechanic would create a very different experience. I would only know a general likelihood of success. I could guess how often I would succeed if I attempted several trips, but the results for my current trip would vary across a wide range, with each outcome being as likely as the next. I couldn’t expect a typical outcome because there wouldn’t actually be one. Would any of us get into a car if the world worked like that?
And it gets even more counter-intuitive when we remember that driving requires training. A game might give a player a bonus if they have training in a skill. Bonuses adjust the probability, but don’t really capture the different expectations for “I can do this comfortably” versus “I’ve never done this before.” They don’t scale the way that we expect.
Think about driving again. I don’t just have a higher probability of success than a new driver: I’m almost certainly going to succeed at my normal level, while they’re almost certainly going to fail in a way that scares everyone in the vehicle. Unless the bonus for training is absurdly large, a game based on technical averages can’t represent what the experience is really like.
If the bonus is absurdly large though, that only creates other problems. How can the game represent driving tasks that someone like me would find challenging? How does it represent someone who’s a professional or expert driver? If the game isn’t minimizing real distinctions, it becomes wildly inflated.
There are games that don’t use the sort of model I’m critiquing—it’s called a flat resolution system, if you’re curious—but the most common and popular ones do. If you’ve played a game that has you roll a d20 to resolve actions, you’ve played a game built around this sort of model.
Someone might also object that the goal of game design may not be to accurately reflect reality. Considering how many tabletop roleplaying games involve fantasy elements, it might be fair to say that the goal is often specifically not to reflect reality. I think that misses the point though.
The content of a game can be as fantastic as anyone pleases, but the rules are there to help us interact with that fantastical content, and the real world is where we develop our intuition about how people interact with any world. Game characters can do amazing things, but they also do ordinary and predictable things, which are only possible if the world responds in predictable ways.
If a character can’t guess what the results of any action might be, the character can’t meaningfully make choices. I can choose to open a door, because I know what opening a door means. If one time opening a door meant moving the door, and one time it meant turning my ears into tomatoes, and one time it meant changing the meaning of the word “yellow,” and one time it meant nothing, then I couldn’t really choose to open a door at all, and the storyteller would have less freedom to tell a story, because suddenly even the basic building blocks are quicksand.
I care about telling good stories, so I care about basic building blocks and expectations. I want characters to make choices, so I want a game system that reflects as much as possible what we understand about making choices. I’ve come to think of this as mechanical verisimilitude. It’s why I started making Bardworthy all those decades ago.
I’ll have more to say about mechanical verisimilitude. At heart I’m a storyteller, but I’ll probably talk a lot about math. Rules are like bones, and stories are like muscles. All the athletic and artistic movement in the world is only possible because bones are stubbornly inflexible. A storytelling game will inspire better stories if its rules are sound.
Until next time,
Daniel