I have an idea for a modern horror rpg. The setting, characters and so forth are easy for me to write. I even have ideas on the rules. However, I have no background in math, game theory, or statistics and I am hopelessly lost when it comes to creating new game mechanics. I don't want to use an existing system.

How do you go about designing new mechanics like conflict resolution? Where do you get the number of dice to roll, what target numbers you want to reach and so forth?I want something fairly simple but not something as easy as Risus.

For example, I want skills to reflect what a person actually learns from education, on the job training and experience. Instead of giving an ex-cop some firearms, driving, investigation and some police contacts and social skills, the character has Law Enforcement Training 3 and the character can roll on any task listed under Law Enforcement 3.However, unlike the standard 'name your power' systems, I want there to be a specific list of skills, each with a list of tasks underneath it.  Skills would go from 1-10 and the numbers would correspond with the amount of experience the characters has with it. A 1 in a skill is a hobby with minimal time spent on it, a 3 is a newly trained professional (rookie), and a 10 is a world class master who has sacrificed many things to achieve that level of skill.

One thing I would suggest is to start with the flavor of the game and try to translate that into game mechanics. For example, I'm working on a game called Satanic Panic, so the game uses three six sided dice for everything (666). The point of the game is to make fun of all those misinformed people who think that D&D and Yahtzee are satanic, so I'm trying to incorporate Yahtzee rules into my game to convey that idea.

This is a modern horror game, so how do you convey a sense of fear and anxiety in the game's mechanics?

Joe:  Relax!  All the games you like?  They were created by people making a really awful first draft, and then tinkering about with it ... playing it with their poor abused friends and noticing where it terribly broke down, then (armed with that knowledge of how things work) remaking it better.  You sound like you're unwilling to begin until you see the path to the finish.  I'm pretty sure, though, that it's not supposed to work that way.  You begin on faith, and then the path either opens or doesn't as you move forward.

Now, on number of dice, and target numbers:  Some popular-mechanics level statistics ...

• A single die, compared against a number, will provide the least predictable results.
• More dice, added together, provide progressively more predictable results compared with the scale of the total:  1d10 can provide a 10 as easily as a 5.  100d10 will very, very, very rarely total to anything outside the 400-600 band.
• So if you want a sense of grim inevitability then give people lots of dice.  If you want wildly unpredictable successes and heart-breaking failures give them just a few.

I hope this helps!

John Kim's system design has some useful links and articles about mechanics. In particular, you may want to check out the comparisons of dice mechanics, and Torben Morgensen's article about dice probabilities.

Those two articles should help you decide where to start with the mechanics. The rest, as Tony said, is about playtesting. Just pick a starting point and go for it!

Thanks for the advice. I now have a starting point to begin designing. I'm very intrigued by the fact that the number of dice rolled sets the predictability.

My new idea: Set the conflict resolution mechanic on rolling two types of dice. For example, normal skill checks use a 3d6 roll, like GURPS but supernatural/desperate characters can spend resources or make a sacrifice of some sort to roll a d20 instead of 3d6. Rolling a 1-3 or 18-20 gives vastly disproportional results when compared to 4-17 and a d20 roll intensifies the result regardless of the number rolled.

Skill and attributes don't affect the roll, but cap the max level of success and minimize the effect of failure on a normal 3d6 skill check. D20 rolls are more dangerous for the PC as skills and attributes don't minimize failure on those rolls.

How does that sound?

clockworkjoe wrote:
My new idea: Set the conflict resolution mechanic on rolling two types of dice. For example, normal skill checks use a 3d6 roll, like GURPS but supernatural/desperate characters can spend resources or make a sacrifice of some sort to roll a d20 instead of 3d6. Rolling a 1-3 or 18-20 gives vastly disproportional results when compared to 4-17 and a d20 roll intensifies the result regardless of the number rolled.

Skill and attributes don't affect the roll, but cap the max level of success and minimize the effect of failure on a normal 3d6 skill check. D20 rolls are more dangerous for the PC as skills and attributes don't minimize failure on those rolls.

How does that sound?

Alarmingly, no one has yet chimed in with this tactic: have you thought about taking the setting ideas you've got, grabbing a pre-existing system and playing with it for a few sessions, then noting where that system supported your setting's desired feel and where it didn't? As much as I fear advising people to start designing from someone else's end point (that is, from someone else's finished system), having problems with an existing system could give you some helpful signposts as to where to take your new system. Worst case scenario, of course, is that you have no problems with another system and you save yourself a lot of work. :-)

And as an aside: the number of dice rolled doesn't always set the probability, certainly not always in the same way, and (most importantly) not always in a way that really means something. The hardest thing about mechanics (for me, anyway) is that none of them ever exist in a vacuum - everything that surrounds them is going to affect them and alter their effectiveness.

The best advice I've ever seen for how to think about mechanics is in this old blog post of Vincent Baker's. That post got infected with a horrible and irrelevant debate in the comments, so I'll summarize: don't focus on the fictional stuff you're trying to simulate; focus on the experiences you want the PLAYERS to have.

I guess I am on the first step towards game design enlightenment. This has given me a lot to think about. I will be back as soon as I get something written.