I have an idea in mind for a general system that could be adapted to multiple settings, much like GURPS. The basic idea is that every stat or skill has a rating between 1 (Poor) and 5 (Extraordinary) (think Storyteller system), where a rating of 2 is about average. Resolution of actions is pretty straightforward:

1) GM decides difficulty of the action, again a number between 1 and 5. If it's a skill contest between two players, then this number is just the rating of the opposing skill (for example, your target has a defense of 3, so this is just the difficulty).

2) The player rolls an amount of d6's equal to his skill, and the GM rolls an amount of d6's equal to the difficulty.

3) Compare the highest numbers rolled by the player and the GM on any single die (somewhere between 1-6). Highest number wins (player succeeds if he wins or fails if GM wins). If there is a tie, highest rating (the amount of dice rolled) wins. If these match, the player always wins.

What I'm most worried about is that this might be too simplistic. I'm not going for an "epic" feel like D&D where you can have ridiculously high levels and skill mods, but I still worry that only 5 tiers is too little. I guess I could expand it to say 10 tiers, but people I play with complain about systems like Storyteller where you roll a bajillion dice. Does this seem too simplified to be any fun?

Also, I'm rusty on my probability. Anybody have an idea how I would go about finding the chances of success for any given match of two ratings?

Hi keroquack, and welcome

This is very similar to the basic dice mechanism in Ron Edward's Sorcerer and Clinton R. Nixon's Donjon both of which I can highly recommend as excellent games.

It might help you to do a search on odds in Sorcerer in the adept press forum. I know it's been discussed a few times. The difference is that in Sorcerer if the high numbers tie then they are discarded and you look at the next set down for the winner. I'm pretty sure the odds follow the ratio of dice between the two rollers, i.e. if you have an equal amount on both sides that's a 1:1 ratio (50% chance of success either way) where if the player has two dice and the GM has four that's a 1:2 ratio making it a 33% chance of success. I'm not that great with the math though so maybe someone else will speak up for you.

I think one of the things you'll see with d6's (this I know from the Sorcerer discussion) is a LOT of ties. Sorcerer uses d10's by default and Donjon uses d20's both games say you can use whatever size dice you like but the smaller the range, the more ties you will get. 10 siders seem about right for this.

best

Trevis

For the grain issue: that's not something you should bother worrying about. I assure you, your game won't stand or fall because it uses too many or too few dice. To the contrary, it's absolutely crucial to get the probabilities to work for your intent, however many dice it takes (within reason). I'ts even possible that worrying about such a thing is a way to avoid the real questions of design. If you're really bothered about it, consider using rerolls instead of a die pool.

Anyway, the odds. This of it this way: the one player who happens to roll the highest single die wins. Because every die is just as likely to be this one die, the probability of x dice against y dice is x:y, simple as that. Just like Trevis said, it's just a matter of figuring out which pool happens to include the highest die. If you resolve ties in the way you suggest instead of the Sorcerer method, however, the higher pool is favored increasingly at high numbers of dice: when both sides are rolling roughly five dice or so, you can expect to have ties with more or less every roll. We could calculate the exact probabilities for this, but I'm too busy right now.

One way to increase grain without increasing the number of dice in this kind of system is to utilize a "negative" pool: instead of making 1 die the smallest value, continue the number of dice downwards into -2, -3, -4... These numbers are rolled just like positive pools, but you use the lowest value instead of highest. What makes these pools tricky is that the curve of the expected value is strongly bell-like, with 1 die being the middle point. In other words, the greatest changes to your expected result come about in the -2 to 2 range, and the significance of additional dice to either side is comparably small after that. So a negative pool is very unlikely to win against a positive one.