This past week, I’ve been thinking about dice. Specifically, I’ve been thinking about dice-based resolution in role-playing games, i.e. determining success or failure through a dice roll. I’ve recently played a few sessions of Dragon Age, which uses 3d6 as the resolution dice rather than 1d20, and I’ve been led to question the relative merits of different assortments of dice. I even went so far as to create an Excel document to compare probability curves.
The d20 system
I like using a d20; there’s a beautiful simplicity to just rolling a single die and adding a bonus. Another feature of using a single die is that there is an equal chance of getting any individual value. I say feature, rather than benefit, because I’m not really convinced that it is a benefit. I’ve read a lot of criticism about the idea that 5% of the time, you just fail completely. Now this is not a huge problem in a system where the penalty for crit failing is just that you fail; it becomes a more complicated issue if you play in a group that has decided to implement any kind of critical fumble rule.
On the other hand, people don’t seem to mind that they are making a critical hit an equal amount of the time, but it makes you think nonetheless. Should characters who are well trained in combat have an equal chance of complete failure (natural 1) as they have of simply performing well (natural 10 or 11)? And it was this thought that led me to do a survey of different options and compare them.
Other Options
I mentioned Dragon Age above, which uses 3d6 as the resolution dice. Now 3d6 does have more difference than just its probability curve, it also has a higher minimum result and a lower maximum result. While 1d20 can roll anything from 1 to 20, 20 possible results, 3d6 is restricted to 16 possible values from 3 to 18. Those details are not irrelevant, but in the context of this discussion I’m going to push them aside for the moment. As a note, the percentages I give in the rest of this post are not all exact; I’ve rounded them to one decimal place for simplicity’s sake.
The more relevant consequence of using multiple dice is that the probabilities of getting different values form a curve rather than a flat line. With 1d20, each possible result has a 5% chance of being rolled. With 3d6, the extremes of 3 and 18 have a 0.5% chance of being rolled, while the middle results of 10 and 11 each have a 12.5% chance of being rolled. This happens because there are more possible combinations of numbers that add up to those values. All this means that spectacular failure and spectacular success are both less likely to happen.
But 3d6 isn’t the only possible option; approximations can be made with any size of die. 2d12 and 3d8 are perhaps poor options because their maximum result is 24, which might have further ranging effects than are intended for this change. With 2d12, the extremes (2 and 24) each have a 0.7% chance, while the middle value of 13 has an 8.3% chance. With 3d8, the extremes (3 and 24) each have a tiny 0.2% chance of occurring, while 13 and 14 in the middle each have a 9.4% chance each. The two other options have a more useful maximum of 20, with minimums of 2 for 2d10 and 5 for 5d4. 5d4 might also prove to be more extreme than desired, as 5 and 20 each have a miniscule 0.1% chance of appearing, and 12 and 13 in the middle combining for a whopping 30.2% (15.1% each) chance of being rolled.
The 2d10 system
This all leads me to the last option that approximates 1d20 while giving a more middle-weighted probability curve; replacing 1d20 with 2d10. There are quite a few things I like about this option. While there is slightly more adding involved than with a single die, it is still simpler than using 3 (or 5) dice, and so there is still some of that clean simplicity that I like about the d20 system. The maximum (2o) is the same as a d20, and the minimum (2) is only 1 higher.
While the extremes of 2 and 20 are far less likely than they are with 1d20, they still seem like very real possibilities at a 1% chance (down from 5%). On the flip side, the middle result of 11 (and it is actually the middle now rather than 10.5) has a 10% chance of occurring (up from 5%). While this adjustment makes the middle more likely and the ends less likely, the less likely values still seem possible.
If you were to implement this house-rule in 4th edition, a few smaller adjustments would need to be made as well. Saving throws currently succeed on a result of 10 or more, which means an unmodified roll has a 55% chance of success. With 2d10, however, you have a 64% chance of rolling a 10 or higher on an unmodified roll. This is easily compensated for simply by raising the target number from 10 to 11, which would give you the same chance (55%) of success as with 1d20.
The critical hit system might also need adjustment, as rolling a natural 20 (two 10’s) now has only a 1% chance of occurring. This might suit your purposes, but if you want to stay closer to the current chance, you can extend the crit range down to 19 (for 3%) or 18 (for 6%). Another option takes a cue from Green Ronin’s system for Dragon Age, where rolling doubles gives you a boost. Since there is a 10% chance of rolling doubles on 2d10, rolling doubles wouldn’t always be a crit; rather, doubles would only count as a crit if the roll hits. In this system crits do not count as auto-hits but instead just as damage-maximizing hits. If you want double 10’s to be extra special, you can have that result maximize crit dice as well (from high crit and magical weapons).
In Conclusion
I’ll finish this off by saying that none of these suggestions have yet been playtested, but doing so is my next mission. If you choose to use my suggestions, I would love to hear feedback and comments that I can use to further refine the house-rule. I can be contacted at fedosu@gmail.com.