Having discussed my methods of dice-rolling and how they work with the *Fell Gard *story, I want to conclude with a few thoughts on what the whole thing feels like. To get there, it’s worth going through a note or two about probability.

Specifically: random results don’t necessarily look random. I saw a note somewhere on the internet about a high school math professor who gave his class a homework assignment in which each student was to flip a coin a hundred times and write down the sequence of heads and tails. The next day he went around the room, looking at everybody’s sheet of paper with their results listed, and then told them all who’d actually done the assignment and who’d just written down ‘heads’ and ‘tails’ in a pattern that looked random. The teacher was able to do this because random distribution is often counter-intuitive. According to the piece, most of the cheating students knew to put in a run of three or even four results of either heads or tails in a row — but in a series of one hundred coin tosses, one would actually expect a run of six similar results in a row (I think the piece said three runs of six), and probably a run of eight.

Something like that happens with dice rolls. The results cluster together more than intuition would suggest. If you’re rolling percentile dice — two ten-sided dice you roll at the same time to get a number between 1 and 100, with one die representing the tens digit and the other the ones — you don’t get a ’00’ (100) result one time in every hundred rolls. It’s not that even a distribution. What happens is that maybe you don’t get a 00 at all for, say, three or four hundred rolls, then you get it ten times in the next fifty rolls, then not at all for six hundred rolls. Or some such. Unlikely results seem to bunch up together more than one might expect.

And if you roll often enough, repetition becomes a factor. There’s a saw I heard once that if you get twenty people in a room together, there’s good odds at least two people share a birthday. That might sound strange, until you realise that you’re not looking for any one specific person to share a birthday with another specific person. (As I understand it: you have one person in the room, and another enters, so that’s a low chance of a shared birthday between them. A third person enters, who might share a birthday with *either* of the first two people, so another 2 chances. A fourth person enters, and there’s another 3 chances for a shared birthday. And so on. The odds start adding up pretty quickly. The whole thing is laid out here, in the second problem.) So: roll often enough, and at some point you’ll get duplicate rolls closer together than you’d expect. *Something *will duplicate *something*. Earlier today I was rolling percentile dice and came up with 63 twice and I think 76 three times. That’s exceptional, but it gives an idea of what I mean.

It’s all incredibly helpful in building the story. A cluster of improbable results gives me odd things seeming to build off of each other. Unexpected duplicate results suggest a certain structure which might have a reason behind it. Why are trapezoidal rooms entry chambers? They happened to emerge as locations for characters as I was rolling up the first level, duplicate results at key points. So these real statistical effects allow me to build narratives. The narratives don’t ‘explain’ the results, but make use of them.

That’s what’s literally happening when I roll dice. How I perceive the results and think about the whole process is another matter. More on that later.