A New Skill System

Prodigy, like GURPS and Runequest, is a fully skill-based system, using the same skill mechanic for almost every mechanical test undertaken by the players.

There are four skill systems that I've seen used - d20, 3d6, %, and dice pools.  The d20 system works by rolling 1d20, adding modifiers, and then checking the modified result against a target number determined by the task itself.  What's nice about this system is that the difficulty of the task is directly incorporated into the check, as is the character's skill.  However, the d20 roll is hugely variable, and the chance of critical failure and success are the same no matter one's level of skill.

The 3d6 system (and OD&D's ability score checks work the same way) require rolling your dice (3d6 or d20) under a number determined by the character's proficiency with the skill: so a GURPS character with an 11 Merchant skill would need to roll an 11 or less with 3d6 to succeed a given check.    The bell curve provided by the 3d6 means that results are generated much more reliably.  However, the mechanic does not provide a transparent way of incorporating the difficulty of the task into the check itself - you can add modifiers to the character's skill, but it is a very ad-hoc way to do this.

The percentile system combines some of both of the previous two approaches: characters' skills are ranked 1 to 100.  To test a skill, percentile dice are rolled, and the check succeeds if the dice roll under the character's skill level.  There are some cool number games that can be done here, like special effects on doubles rolled under or over a character's skill level and so on, but it combines the wide variability of the d20 with the fixed difficulty of the 3d6, and so has the aspect from each previous system I most dislike.

The dice pool system gives the player a number of dice equal to their level in the skill, and assigns a strict possibility of success (say 8, 9, or 10 on a d10 or 6 on a d6), with more beneficial results occurring when more successes are rolled.  Thus, a player rolling three d10s might think they have a fairly good chance of getting at least one success.  The odds of success, though, are deceptive (rolling three d10s with success on an 8, 9, or 10 conveys only a 65% chance of any successes, with 4 dice bringing the chance only to 75%), and the system has the highest degree of variability of all the systems, since dice that don't succeed are either merely useless or actively bad (if they roll a failure).

Now, I feel that as one becomes more accomplished with a skill, not only can they accomplish more difficult tasks, but they can also accomplish all tasks more consistently.  I also want my system to be able to concretely incorporate the difficulty of whatever task a player attempts - I want challenging tasks to be challenging and easy tasks to be easy, and I want both of those things to happen with a minimum of ad-hoc adjustments.  I also want it to be fairly uncomplicated - no involved table lookups or complicated math.

The way I've accomplished this is by combining many of the approaches together.  As I've said before, I rank a character's level of skill in one of 6 tiers: Novice, Apprentice, Journeyman, Specialist, Master, and Grandmaster.

Skill tier	Dice Rolled	Difficulty Value
Novice	1d6	6
Apprentice	1d6+1d4	9
Journeyman	1d6+2d4	12
Specialist	2d6+2d4	15
Master	2d6+3d4	18
Grandmaster	3d6+3d4	22

This approach borrows from the dice pool system in that as more skill is obtained more dice are rolled.  However, unlike the dice pool system, all dice matter - they are summed together and checked against the task's difficulty value.  Because more dice produce a stronger bell curve effect, the expected value of each die roll is 3 less than the difficulty value of a task of equal tier (because of adding in aptitudes, which are scaled from 1 to 6), which means that increasing skill not only increases the overall total, but it also increases consistency when performing tasks.  Lastly, by assigning difficulty values for targets, the task attempted by the character now absolutely factors into the test.

Moreover, when circumstances change, I can alter the difficulty of the test in one of two ways: I can add or subtract 1 to the DV for minor things or I can change the tier of challenge for larger difficulties.  My options for changing the circumstances of the die roll are restricted because of this tiered system.

While I've yet to run a real game with these rules, I've run several oneshots and combat scenarios, and the mechanic has worked exactly as I like it to, but I'd welcome any thoughts folks have on it.