Having committed to d100 games for the bulk of my (occasional) role playing, I have started a process of research, designed to weed out the best ways of handling various games and themes.
A trend has developed, or at least I have re-discovered it, and it comes with some bemusement and mild frustration. Mostly I find it interesting how an idea takes hold.
The term most often associated with d100 mechanics is “whiffy”. Whiffy refers to the even (linear) chance of any single result occurring and the often hard pass or fail effect applied to the roll. This has the effect of making a seemingly more random or unpredictable game than a “bell curve” style game mechanic (one that uses multiple dice added together, or used in a pool, forming a natural seeming average).
This is on one hand interesting and for the most part irrelevant. It seems to me, it is all in how you (a) perceive it and (b) how the mechanics themselves are executed.
Lets first look at D&D, the oldest and most popular of RPG’s. D&D uses 3d6 curve for characteristic generation (as do most d100 games), but a single dice, usually a d20 for tests and combat rolls. The main difference is the nature of the d20 roll. The roll is a “roll high against a target with all mods applied”. This means that you increase your chance of success in direct proportion to the number of positive dice mods you can apply. Another blogger once called this “Optimistic” or open ended rolling and I get that it feels that way. You are trying to overcome a hill resistance and every little “+” helps. D100 games are often a roll under a set value, but sometimes roll as high as you can under it, a bit like that game where you skate a disc as close to an edge as possible without it going over. Brinkmanship vs throw as hard as you can.
The roll is still linear, but weighted heavily towards one end or the other by flat mods. This brings it’s own problems. With some forms of the game effectively making unlikely results plain impossible, there have been soft fixes in late editions, limiting the possible range of the mods, curbing higher level blowout (called “bounded accuracy”).
The very nature of a +/- mod, roll high system makes a key difference to the effect of modifiers. Add +10 to your roll (a huge amount of advantage) and you still only have little over a 50% chance of beating 20+ as a target, but you automatically beat 10+. This is fairly robust and gives the illusion of powerfully leaning one way or the other, but it is still linear and has it’s limits.
The average character with a +3-5 mod on favoured tests has a 15-25% greater chance of passing them than an average +0 mod character (but often an identical chance of a critical pass/fail).
Of greater effect is the new advantage/disadvantage dice system, rolling 2d20 and taking the best or worst by circumstance. This doubles your chance of a better or worse result without changing the range. We will look at this idea some more.
*
The next example is Traveller, the oldest Sci Fi RPG, using 2d6 for a simple mechanic with a mild bell curve. The curve gives a sense of a broad middle (6 out of 36 chances of rolling 7, 16 out of 36 of hitting the middle three numbers) and 1 relatively unlikely chance each of rolling unlikely extremes (2 or 12). Mods have a much stronger effect, literally shifting the average. They are actually moving a curve in a non linear fashion.
The biggest issue comes from these mods, which can blow out the range quite quickly. MT2e has also adopted an advantage/disadvantage dice system, combined with the curve. This gives the player the curve for comfort and the best/worst pair for difficulty. It does not increase the range of possible rolls, only the chance of a favoured/slighted result.Using 2 of 3 dice is less clean than the 2d20 version from above, but it squeezes into the small range of 2d6 well enough.
The reality is, bell curve or dice pool games also have fixed percentages of pass and fail. Needing to roll 7+ with a 2d6 mechanic is a 21/36ths or 58% chance. It looks gentler with it’s middle curve producing the most common results, but it is still a fixed 58% chance. Add a single +1 mod (target now 6+) and that jumps very quickly to 72%, or a -1 mod (8+) and you have roughly 42% of passing. These single value mods have a very strong effect especially at low levels, with less effect as increased, due to the curve shortening. If the above mechanic is applied with the same pass/fail logic then you have a d% system with vastly fewer increments and a “lump” in the middle.
I feel it is not the range or the curve that we rect to, but the application. Again the climb the mountain vs the push to the edge thinking.
*
D100 games usually use roll-under mechanics. The character has a skill/chance level expressed in % points, rolled against with 2d10 (1 as 10’s and the other as 1’s) with a critical pass/fail chance, which differs greatly by game, but may for example be half or 10% of the pass chance, or the lower/higher 5%. The later iterations of these games have been taking advantage of doubles as fumble/critical pass values.
This is where the problems come from. Apart from often weighty maths, a real issue with massed combats etc, the flat pass/fail is seen as too twitchy. Have a 55% skill in driving, that means you fail/crash 45% of the time you are put to the test which is unrealistic, but tends to be the way the system is perceived. Add mods for difficulty and the base shifts proportionately, but depending on system, brings inherent imbalance. Why would a skill 10% character get a 300% increase in skill chance with a +20 mod, when a 80% near master only gets the same +20, which often tops out at an “automatic fail” level of 95%, netting a lower proportionate bonus and a lower flat benefit.
How to fix?
Option 1 (assuming a pass/fail dynamic) is to make most average test rolls an “arrange to taste” mechanic, which is to still roll 2d10, but arrange the two die in the desired order (71 becomes 17 if you want or the opposite if desired). This gives you two things. The first is a heavy weight towards the pass (lower) end with a roughly 70% chance of rolling 50% or lower and the ability, depending on the system used, to push the roll higher for more success “height”.
If this is the base test, then more threatening or volatile tests (like combat without advantage) should still then use the less predictable and most common “take them as you roll them” system. This system works well with a fixed limit of 99% for all skills. World masters always have a 1% chance of failing, combined with an embarrassing, but not life threatening fumble (see below) and 50% skills have a better than 50% chance of passing standard tests.
Option 2 is to soften the pass/fail slug. If the 10-20% points above and below the skill level (or half way points between the skill and the extremes) are used as “soft” pass/fail levels, this gives the GM room to allow a pass with complications or a “fail forward” soft landing. If the half over/under system is used, reference numbers can be kept on a handy chart or just eye balled (math is often sighted as the “problem” with d100 games).
It seems illogical to have 100 possible results and dilute them down to a drastic pass-fail mechanic.
Handling Critical pass/fail levels is also something that seems to change across D100 games. The best I have seen is the “doubles” critical pass/fail version, which elegantly bypasses the messy math thing. D00 lite games, OpenQuest 3, D100 Revolution, Warhammer 4e all use or suggest this system and it is neat and clean. Regardless of other factors, doubles are fixed and offer a 10% chance of something extraordinary happening, be it good or bad. This reduces the 10% broad crit chance to a 1-3%, specific crit result range.
To add more control to this, a simple table of the ten crit pass or fail result levels is a way of curbing the automatic severity of these results, or, with point 2 in mind, crits that fall within the “soft” zone are not too aggressive, but ones outside of that zone are more catastrophic or amazing. A “00” auto fail is a soft crit fail to the 99% master, but a “33” critical is a real stuff-up by the low skilled apprentice. GM pay-in is also allowed of course, but a good guide is invaluable.
Difficulty can also be handled within the 99% maximum range. For difficult tests, take the worst combo, for greater difficulty, add another dice, still taking the worst two, then another etc. For easy tasks, add another dice and choose the best result of two etc. This is equally harsh and fair (less so than a +1 mod on a 2d6 curve!). You always have a chance of extremes, ever increasing/decreasing, but no added math.
We use blue and red dice for easier/harder tests. State your aim and the GM hands you 2+ dice. If they are 1 red and 1 blue, then roll and arrange as pre-determined (red as 10’s). Get two blues and you can take them as you want, three blues, take the best 2, 2 reds are always read the worst way, three reds you still take the worst 2 etc. Imagine the look on a players face when they describe the crazy antics their character is attempting and you nod and hand them 4 reds!
Experience. To improve D% skills, which to the games using this system replace abstract levels, offers another place to add “curve” to the game. the character is usually rewarded with chances to improve after success, or even sometimes failure. If this improvement requires a test, say one related directly to the characters actual skill (roll over?) then improvement becomes slower and less likely. The difficulty (above) can also help determine the chance of an improvement roll.
One system, designed to go with the 99% limit, is to roll an experience test, by rolling over your current skill, each time a critical success or failure (called a learning point) is rolled on anything but a mundane or easier task. This may be for a +1% bonus (maybe a -1 if the crit was a fail and so was the roll) but these could also be automatic +1% gains if you want a more dynamic environment. This can be limited to crits that fall outside of the “soft” zone, meaning you learn from really successful or really unsuccessful results only.
Levels can be used also. 20 levels with 5/3/2/1% gains per level in batches of 5 of each (55% max), added to 2 characteristics (40% max). These diminishing levels are on one hand a realistic representation of a flattening skill curve and on the other, a lowering of player desire to max out a skill.
*
D% games have always been under the shadow of this “twitchy” or “whiffy” cloud, which I feel is unfair. The actual mechanics are linear and granular, both desirable characteristics, but often the application of these is the problem. Look to the core mechanic, the answers are there.
The beauty of most d100 games is, one “fix” can often be applied to all of them or as needed. The core ideas are intact, just the application changes slightly and the reality is, most experienced games have probably applied one or more of these in their games at some time.
My games use most of the above, only requiring a single sheet to apply them to the terms and conditions of the master game. Difficulty levels need wording accurately, advantage mechanics taken into account and some basic choices made. The very versatility and robustness of d% games is your friend here.