Dice. As a way of introducing a random element into games, dice have been in use for millennia. For most of that time they were a six-sided cube, although in the last 40 years they've come in a wide range of shapes a sizes. From the 'roll-a-d6-and-halve-it' d3, up to the '3-d10s' d1000, all give a games designer a way to shape their game's mechanics. And the choice of dice can seriously alter the way a game works.

I was torn on which dice to use. The most commonly used by far is the regular six-sided dice. Everyone's got some, and it's what most people think of when they hear the word. So choosing a d6 as the dice for your game is a good way to minimise 'barriers to entry'.

But still, I fancied either using the d10 or the d12. The d10, because it would allow for easy conversion to a percentile system, if Ravensrodd ever developed into an RPG. The d12, because, well, I just really like them. They've got a really nice shape...

In the end, I decided against the d10. This was because they are often used in pairs, as a d100 percentile dice - so instead of a '10' they often have a '0'. And it just felt intuitively wrong to have to keep explaining that '0's were better than '9's...

So I went ahead with my favourite dice - the d12s. I'm using an opposed die roll system as the core game mechanic; each player rolls a d12 and adds the relevant stat. The highest wins, and in some cases the difference between the two scores is known as the Resolution. Here's an example; John has an Attack stat of 7; Geoff has a Defence stat of 5. John rolls a 7, and adds it to his Attack for a total of 14. Geoff rolls a 6, adds this to his Defence of 6, for a total of 12. John wins, with a resolution of 2.

Now, that works perfectly well. But the more I played with it, the more I began to have my doubts...

The problem is probability. A d12 gives an equal 1-in-12 chance of rolling any of its numbers. That means you are just as likely to roll a 1 as you are to roll a 12, or a 7 or a 3. Take our example above again. Instead of rolling a 7 and a 6, this time John rolls a 1, and Geoff rolls a 12. Geoff's now been beaten by a whopping 10. We can be pretty sure, however the damage rules play out, he's dead. Seems extreme?

But that example isn't extreme at all - it's exactly as likely as the first one where they rolled 6s and 7s.

This is a problem, because it makes the rules feel overly random; when results that look and feel extreme occur regularly, the players wind up thinking the game is purely down to luck. And that's not much fun.

So, how did I get around this? Simple - instead of a d12, I used 2d6. The spread of results is pretty much the same; 1 to 12 for the d12, 2 to 12 for the 2d6. But the probability is totally different. With 2d6, the odds of rolling those extreme numbers - 2 and 12 - are 1 in 36. But the odds of rolling a 7 are 1 in 6. In fact, rolling a 6, 7 or 8 will happen about 45% of the time; almost half!

This removes a lot of the extreme-seeming results instantly. Now, if you've got Attack of 9, and your opponent has Defence of 3, you can be pretty confident of a win. Sure, there's still a chance you could roll a 2, and him a 12; but this is far less likely than it would have been using d12s.

So, design decision made - Ravensrodd will use d6s. Time to raid those old Monopoly sets...