I've just finished reading the Kobold Guide to Magic, and it's very good. I've been inspired to try my hand at developing a system of magic for tabletop games. A system of magic has different components, and for the purposes of this exploration I'm going to approach development from the standpoints of mechanics and aesthetics.
My mechanical criteria for a potential system of spellcasting would contain:
- increasing spellcasting abilities with experience
- multiple modes of progression
- the risk of failure
- the risk of over-achieving
The casting of magic spells is a common feature in tabletop role-playing games, and nothing quite beats some good eldritch mayhem. Most systems will tie in the success of performing magic with some sort of magical aptitude attribute or character level, so more experienced characters are better at casting spells. This is a common and good thing. But for me, spellcasting is different than more mundane skills, like sword fighting or lock picking. Spellcasting should feel like the character is tapping into some mysterious and powerful source of energy -- initially very difficult to learn, and always having the risk of disaster. I've found that in common systems the progression of a spellcasting character is fairly linear. If there is some sort of chaotic result, its tied to a critical failure roll -- 1 on a d20, for example. Some systems, like Goodman Games's Dungeon Crawl Classics, have elaborate critical failure tables. Failure should be devastating -- your character, after all, is dealing with powers beyond mortal imagination. I also think being too successful in connecting to those powers might also have effects beyond your character's control -- like connecting your garden hose into the fire hydrant. "Pull out, Betty, you've hit an artery!"
The other day I was thinking about probabilities, and it occurred to me that there might be something in those bell curves that would work for a magic system that satisfies my concepts. With spellcasting success tied to the roll of one die, there are flat odds for any result. But with multiple rolls of one type of die, the results become more of a bell curve. This would seem to fit well with my criteria for failure and over-achievement: there is a "sweet spot", outside of which things bad and good might happen.
This raises two questions: what happens at the edges, and how does one roll into the curve? I poked around at the second question first.
Let's say you have one six-sided dice -- there is no curve, just even odds. With two, your sweet spot is a roll of 7. With three, it's 10 and 11; with four it's 14. The sweet spot is a moving target, which in and of itself isn't a big deal. But it does bother me in an intuitive way; I'd rather the player not have to remember multiple values. So let's add some parameters. If we say that the maximum allowable number of dice per roll is 6 -- chosen arbitrarily based upon convenience and palm size -- our sweet spot is 21. Great! Seems like that should be our target for maximum spell effectiveness.
This seems to take care of a certain amount of advancement -- junior spellcasters might not have six dice, and may use only one or two. That sweet spot is inaccessible for them at the moment. Excellent! Let's make them work for it. If 21 is a perfect execution of a spell, rolling 1d6 isn't going to come close. There might be some effect, but certainly not anything impressive. And that feels good. New spellcasters should be pretty bad at what they do. They should make awful mistakes and die a lot. At least, I think they should.
Experts, on the other hand, rolling 6d6, will hit the sweet spot with some frequency. But they'll also over-roll. They'll be too powerful, or create side effects beyond their control. This works for me, too. 6d6 has a wide median range, meaning they'll be close more than not.
I need to wrap this up for now. I still have ideas on aesthetics which I have yet to address.