SeRiAlExPeRiMeNtS
Explorer
I like it, maybe they will use the full stat as a static number, like the passive perception in 4e, and stat -10 if you add a d20.
Odd-Numbered Ability Scores
- Thread starter Falling Icicle
- Start date
kitsune9
Adventurer
This works for me.
I think any changes to ability scores bonuses will have to go hand-in-hand with how important the opposed roll becomes in the game. In fact, we don't know if the opposed roll will be like a classic opposed roll, or whether it will simply give a small modifier to the saving throw (although I doubt this will be the case).
In the classic case both roll a d20 and add their bonuses, and the larger result wins. If D is a random variable = random d20 roll and d is the defender's bonus, and A is a random variable = random d20 roll and a is the attacker's bonus, the probability of an attack succeeding (assuming the defender wins ties) is:
A + a > D + d, equivalent to A - D + (a-d) > 0.
A - D is distributed as d20 - d20 (independent rolls), which has a probability distribution that runs from -19 to 19, and is shaped like a triangle with a peak at 0 (a 5% probability). This is a much wider range than a single d20 roll, but it is also more forgiving to large differences in the bonuses because the "average" outcome is more likely than the outside ones. If we assume the 3.5/4e method of ability score bonuses, and assume ability scores will generally range from 2 to 20, the largest usual a-d will be 5 - (-4) = 9. Even in this case, the attack succeeds only 83.5% of the time, and slightly lower than that if a natural 20 on a save always succeeds. In 4e the equivalent (ignoring class adjustments and so on), would be a +5 attack vs. defense 6. In other words, the attack succeeds 95% of the time. In 3e even a 1st level spell would have a DC of at least 16, so a character that starts with no other modifiers to this particular save would need to roll a 20. And in both 3e and 4e the disparity will tend to grow wider over time.
In a more typical case, say an attacker with 18 vs. a defender with 10, the probability of the attack succeeding is d20-d20+4 > 0, which succeeds 66% of the time (again ignoring natural 20s saving). In the case of 4e that would be like a +4 bonus vs defense 10, which has a 75% chance of success. The point is, for any method of calculating bonuses from ability scores, opposed rolls are more tolerant of large differences in ability scores than rolls against static numbers.
Now consider the relative benefit of higher ability scores in 5e if using opposed checks. When a-d is close to 0, each additional +1 is almost the same as in the static case. For example, when a-d is 0 the attack has a 47.5% chance of success, but if it is 1 it has a 52.5% chance of success. If a-d is 2, however, it becomes 57.25%. If the best offenses and the best defenses basically rise together, that means (assuming 3.5/4e ability score bonuses) going from ability 16 to ability 20 would give about a 10% improvement when attacking a target's best defense much like in 3.5 and 4e. However, if one targets an opponent's lowest defense (the more usual strategy), the change is smaller. Suppose two wizards (16 and 20 Int) try to charm the classic dumb fighter (8 Wis/Cha). Then a-d is 4 and 6, respectively, and the first wizard has a 66% chance, and the second a 73.5% chance, for a 7.5% improvement. Whether or not that is still too great a difference I'll leave to you.
The opposed roll has an additional element, in that if the attacker rolls sufficiently high the defender will often need to roll a 20 to save. This is conditional probability, however, so a specialized attacker can't count on it when choosing what to do. One of the more exciting elements of the game is needing to roll the clutch 20, and this will make those moments more frequent without screwing everyone constantly. There is also tension in the game by making the players wonder *if* they'll need to roll a clutch 20 against this next fireball. Even a person with a very good defense will face that on occasions.
I suppose the opposed roll is a little weird in that it will feel more swingy (there will be some pretty wild outliers and they will occur unpredictably) even though it actually tends more to the mean than a single d20 roll using the same ability scores and system of bonuses. For example, fifty percent of all outcomes for a single d20 are in the 6-15 range, while (about) 50% of all outcomes for d20-d20 are in the -5 to +5 range. They have the same absolute size, but different sizes relative to the number of possible outcomes. This is why looking at opposed checks and thinking the bonuses should be twice as large to account for twice as many possible outcomes is a bad idea.
Finally, using opposed rolls probably adds a single extra roll to what we'd expect in either 3e or 4e (that is, the attacker does not roll one for every target). At least in my experience the number of d20 rolls was not the primary cause of slowdowns. If the rest of the game is designed to keep things moving we might not notice a single extra roll most turns.
In summary, I don't think we'll necessarily need to change the way ability score bonuses are calculated, because opposed checks using the 3.5/4e method still tones down some of the strongest disparities due to static checks found in 3.5/4e, and introduces modest diminishing returns for increasing ability scores. Whether that will be enough to discourage the worst min/maxing and prevent the worst relative disparities between characters will depend a great deal on other factors. In either case I think understanding the math of the opposed check is important.
In the classic case both roll a d20 and add their bonuses, and the larger result wins. If D is a random variable = random d20 roll and d is the defender's bonus, and A is a random variable = random d20 roll and a is the attacker's bonus, the probability of an attack succeeding (assuming the defender wins ties) is:
A + a > D + d, equivalent to A - D + (a-d) > 0.
A - D is distributed as d20 - d20 (independent rolls), which has a probability distribution that runs from -19 to 19, and is shaped like a triangle with a peak at 0 (a 5% probability). This is a much wider range than a single d20 roll, but it is also more forgiving to large differences in the bonuses because the "average" outcome is more likely than the outside ones. If we assume the 3.5/4e method of ability score bonuses, and assume ability scores will generally range from 2 to 20, the largest usual a-d will be 5 - (-4) = 9. Even in this case, the attack succeeds only 83.5% of the time, and slightly lower than that if a natural 20 on a save always succeeds. In 4e the equivalent (ignoring class adjustments and so on), would be a +5 attack vs. defense 6. In other words, the attack succeeds 95% of the time. In 3e even a 1st level spell would have a DC of at least 16, so a character that starts with no other modifiers to this particular save would need to roll a 20. And in both 3e and 4e the disparity will tend to grow wider over time.
In a more typical case, say an attacker with 18 vs. a defender with 10, the probability of the attack succeeding is d20-d20+4 > 0, which succeeds 66% of the time (again ignoring natural 20s saving). In the case of 4e that would be like a +4 bonus vs defense 10, which has a 75% chance of success. The point is, for any method of calculating bonuses from ability scores, opposed rolls are more tolerant of large differences in ability scores than rolls against static numbers.
Now consider the relative benefit of higher ability scores in 5e if using opposed checks. When a-d is close to 0, each additional +1 is almost the same as in the static case. For example, when a-d is 0 the attack has a 47.5% chance of success, but if it is 1 it has a 52.5% chance of success. If a-d is 2, however, it becomes 57.25%. If the best offenses and the best defenses basically rise together, that means (assuming 3.5/4e ability score bonuses) going from ability 16 to ability 20 would give about a 10% improvement when attacking a target's best defense much like in 3.5 and 4e. However, if one targets an opponent's lowest defense (the more usual strategy), the change is smaller. Suppose two wizards (16 and 20 Int) try to charm the classic dumb fighter (8 Wis/Cha). Then a-d is 4 and 6, respectively, and the first wizard has a 66% chance, and the second a 73.5% chance, for a 7.5% improvement. Whether or not that is still too great a difference I'll leave to you.
The opposed roll has an additional element, in that if the attacker rolls sufficiently high the defender will often need to roll a 20 to save. This is conditional probability, however, so a specialized attacker can't count on it when choosing what to do. One of the more exciting elements of the game is needing to roll the clutch 20, and this will make those moments more frequent without screwing everyone constantly. There is also tension in the game by making the players wonder *if* they'll need to roll a clutch 20 against this next fireball. Even a person with a very good defense will face that on occasions.
I suppose the opposed roll is a little weird in that it will feel more swingy (there will be some pretty wild outliers and they will occur unpredictably) even though it actually tends more to the mean than a single d20 roll using the same ability scores and system of bonuses. For example, fifty percent of all outcomes for a single d20 are in the 6-15 range, while (about) 50% of all outcomes for d20-d20 are in the -5 to +5 range. They have the same absolute size, but different sizes relative to the number of possible outcomes. This is why looking at opposed checks and thinking the bonuses should be twice as large to account for twice as many possible outcomes is a bad idea.
Finally, using opposed rolls probably adds a single extra roll to what we'd expect in either 3e or 4e (that is, the attacker does not roll one for every target). At least in my experience the number of d20 rolls was not the primary cause of slowdowns. If the rest of the game is designed to keep things moving we might not notice a single extra roll most turns.
In summary, I don't think we'll necessarily need to change the way ability score bonuses are calculated, because opposed checks using the 3.5/4e method still tones down some of the strongest disparities due to static checks found in 3.5/4e, and introduces modest diminishing returns for increasing ability scores. Whether that will be enough to discourage the worst min/maxing and prevent the worst relative disparities between characters will depend a great deal on other factors. In either case I think understanding the math of the opposed check is important.
Last edited:
Another supporter of this. I much prefer the bell curve with hard caps on the maximums. Moving to a linear curve with no limits is IMHO one of the things that contribute to the Swinginess in 3E and the Grind in 4E.
So this also means that no increase in ability scores as you go up in level. Other things improve like saving throws and attack bonuses, so the base scores don't need to.
Tilenas
Explorer
On board. In the end, I even favour the 3d6 roll over 4d6, drop lowest. The only reason to have a 3-18 range and let PCs play in the top division almost everywhere is to give them a sense of superiority vis-a-vis the world's NSCs. Since I don't buy into the PCs-are-heroes-from-level-1-creed, I don't support it. If it must be 4d6, drop lowest, let it be Organic Characters (roll in order, then reroll any one score, then swap any two scores), so that ability creation is more than rolling 24d6, dropping the 6 lowest, and assigning pretty much as it pleases you.
Players can point buy all they want in my campaigns. I'm giving them 15 points. Works perfectly. It's also in the DMG, which maybe fosters acceptance. They may also lower scores under 8 to gain points on a 1-for-1 basis. Last time around, our 3.5 group had one PC who started with an 18 (and that included the racial bonus).
Mark CMG
Creative Mountain Games
This is what I am doing with Griffins & Grottos but G&G is a ten-level system with some advencement features manifesting between the levels rather than only on them. I like that it is more intuitive and makes the math simpler at a glance. I hasten to add that G&G will be released under the OGL long before 5E is finalized, so one of my hopes is that 5E will also be OGL so they have the chance to draw from the OGC released in other OGL systems.
MortonStromgal
First Post
Very close to what I like
3-4: -3
5-6: -2
7-8: -1
9-12: No adjustment
13-14: +1
15-16: +2
17-18: +3
If your rolling stats rolling a 17 is pretty much as cool as rolling an 18. YMMV
Tony Vargas
Legend
'Roll under' (or roll as high as possible without going over) stat checks were a common variant in olden times. I use the latter when I run the mechanically 0D&Dish 1e gamma world.
You still want a high roll, you might fail if you roll very low, but you fail automatically if you roll higher than your stat. Thus an 18 doesn't just succeed more often, it succeeds at things a 9 simply can't.
It's good for 'soft' rolls that need a degree of success rather than a binary pass/fail.
I may debate abandoning point buy if rolling for your stats did not produce very drastically different characters in terms of strength...something like:
03, 04, 05 = -2
06, 07, 08 = -1
09, 10, 11 = +0
12, 13, 14, 15 = +1
16, 17 = +2
18 = +3
(If I remember correctly, classes will add +1 to their main stat. Race will also give a +1 bonus to a stat or two. With this in mind, no stat can go beyond 18 at character creation.)
The majority of folks will be in the +1 range with maybe a stat or two with +2 or even +3.
Most DC checks that can be auto-passed with a high attribute score will be will have DCs of 10, 13 and 15. DC of 20 will be for harder checks.
03, 04, 05 = -2
06, 07, 08 = -1
09, 10, 11 = +0
12, 13, 14, 15 = +1
16, 17 = +2
18 = +3
(If I remember correctly, classes will add +1 to their main stat. Race will also give a +1 bonus to a stat or two. With this in mind, no stat can go beyond 18 at character creation.)
The majority of folks will be in the +1 range with maybe a stat or two with +2 or even +3.
Most DC checks that can be auto-passed with a high attribute score will be will have DCs of 10, 13 and 15. DC of 20 will be for harder checks.
