As I mentioned in the last post after fiddling with the core mechanic for The City I ended up re-assessing the system I was using for Ave Imperator. My original goal for the system was to have to be as “realistic” as possible. Which is somewhat of a silly goal, when people talk about realistic systems they generally mean a system with a decent level of verisimilitude. People want the system to “feel” real. Not necessarily fully simulate reality (because that would be both extremely complex and demanding and also runs the risk of being rather boring). To boil it down even further, all I really wanted was a combat system that felt fairly lethal, as when talking about rpg systems most people equate highly lethal or “gritty” combat with realism. The system I had was decent enough, as far as it went. It was a dice pool system that worked for the most part but had some odd probablity issues (mostly the weird scaling on the probability of success). It was also open to abuse in that attributes were much more useful than skills. Which wasnt really an issue per se as the player would never be manually generating any characters (more on that later). But after playing BRP for the first time in ages last week, and working on dice probability stuff for The City, I just couldnt leave the Ave Imperator system as it was. So I’ve come up with something new, and, I think anyway, better. At least in terms of combat its a bit more involved but also hopefully more fun.
The original system I came up with used a pool of d10’s and modifiers based on skill and attributes. As I said above, it worked, but it had some issues. When I came back to the system a few days ago I was tempted to throw the entire thing out and switch over to a flat percentage system. The benefits of which are a) its easy to understand and b) its easy to see your chance of success. However it also has some disadvantages, a) it scales poorly, b) competency is all over the place, c) in most implementations either stats or skills are clearly more useful. At the end of the day I just cant bring myself to love percentile systems. At least not for a computer game, they work fine in tabletop because its a lot easier for the GM to adjust stuff up and down than it is for me to program the game to adjust things up and down. I also didnt want to go with another dicepool system. Nor am I a big fan of linear probability as seen in d20 systems and derivatives, or straight d10 systems. I know it doesnt make a huge differnce in actual play but I wanted a dice mechanic who’s results when graphed showed something like a normal distribution.
I was originally using a stat+skill system where the human maximum was 5 for each (so the ultimate human would roll five dice and add 5 to the result of each before modifiers). I wanted to expand that range a little, and as 12 was a number used quite a bit by the Romans having 6 be the new maximum for each seemed a good way to go (well possibly 7 for truly legendary characters). I also wanted to use them as a flat modifier, so that gave me a max possible modifier of +12, with the “average” been around +4. Then the fiddling with which randomiser to use began. Single dice solutions, with linear probability scales, were straight out, so goodbye 1d10+2 and the never loved 1d12. I looked at 2d6, which was serviceable, though the probability distribution was more of a pyramid than a bell. I also looked at 3d6, take the two highest. But that skewed a bit too high and compressed the range of successes. I also looked at various Roll xdx and subtract or add Y. I finally pared the ones I liked down to these few http://anydice.com/program/20ca. I suppose I should note here that one of my rules of thumb was that an average character (2 stat, 2 skill) should have a 60-70% base chance of succeeding in a stressful/difficult situation (I mean think about it, do you fuck up your job more than 20-30% no matter how stressful it is?).
In the end I opted to go with 3d6-5 (-6 skewed a bit too low) and a base difficulty of 9 (8 was a little too high at ~75% probability for an average character to pass and 10 at ~50% was too low). Now its not perfect in that completely untrained characters with some form of malaise in the governing attribute (i.e. stat 1, skill 0) cant get higher than -1. Which is an auto fumble. This would be an issue if a) I were planning to use the rules for a generic rpg and b) the chance was higher than ~2%. However, even the shittest slave gladiator is going to have at least a total modifier of 2 (which does mean that 0.46% of the time he may get a result of 0) so its fine.
While I did abandon my dicepool system I do like the idea of successes that a lot of dicepool systems use. So I decided to implement it here. So getting 9 or better counts as one success, for every two full points you get over 9 you gain another success (so thats 2 at 11, 3 at 13 and so on). I am also a fan of crits and fumbles. The only problem is that a normal distribution really cuts down the amount of times they occur (which is one of their selling points as they reward skill over luck). I still havent fully decided what to do, at the moment I’m toying with the idea of a two tiered crit and fumble system. So in essence the results from any roll can be “Fumble (Roll 3 1’s) | Fail (end up with a total result of <9) | Success (End up with a total score of >=9) | Superior Success (Get double 6’s, this adds an extra success onto your total number of successes, may also have extra effects in combat, occurs ~7% of the time) | Critical Success (Get triple 6’s, occurs ~0.5% of the time, allows you to roll 3d6-5 again and add it to your result)” There’s no “Critical Fumble” and fumbles wont occur that often. But thats not really a problem for me.
So thats more or less the core mechanic in it’s entirety. There are a number of sub-systems (i.e. the combat system, the social system, the morale system, etc.) which modify it slightly but the core is always “Roll (3d6-5)+Stat+Skill, base difficulty 9, extra success for every +2 above 9”