Imagine the scene: Fastfeet the Rogue and Joe Average need to cross a rickety rope bridge before kobolds have time to drop a bolder from the cliffs above. Fastfeet, with dexterity 20, stands as the quickest halfling alive. Joe Average. with dexterity 10, has a hopelessly mundane, non-D&D name. Let’s call him J’oe. Better.
The wobbly bridge has rotting and missing planks, so crossing it without slowing requires a dexterity check. The DM decides that the crossing counts as an EASY check: DC 10. No problem thinks Fastfeet, I’m optimized to have the highest possible dexterity. I just can’t roll a 1…or a 2, or 3, or 4. Hmmm, I may as well try diplomacy.
Fastfeet, the quickest halfling alive, still suffers a 20% chance of missing an EASY check. Despite being the quickest possible character, Fastfeet only gains an extra 25 percentage points in his chance over J’oe average.
The problem stems from the mere +5 that a 20 characteristic adds to the check. The D20 roll swamps it. This leads to two problems:
- Exceptional characters do not noticeably stand out. Whether your character has a poor or a great characteristic, every ability check pretty much feels like a coin flip. This becomes particularly noticeable with checks that encourage everyone at the table to try. That’s when everyone puzzles over an ancient map fragment, the resident sage says she will try a history check, and everyone chimes in, “I’ll try too.” Most times, the expert character gets no chance to shine, because her numerical bonus barely exceeds anyone else’s. The success goes to the person who happened to roll a 19.
- Even when an exceptional character attempts something easy, the outcome remains unpredictable, as in Fastfeet’s case.
I asked Mike Mearls about this issue, and he said that the DM could simply rule that a easy check is an automatic success for characters of advanced ability. The advice patches over bad math with DM fiat. As a DM, I would make that ruling, because the system’s rotten foundation forces it. I would rather see math that works.
You may think that I’m overlooking the skills that address my problem with the math. Forget skills. D&D next has no skill checks or ability checks, only checks. Unlike earlier editions, skills no longer provide a system for determining success, so for example, the skill descriptions no longer include rules for resolution. Skills represent a small number of areas where extraordinary focus and training might help your character make checks. Skills stand as an optional rule for granting a bonus to a limited number of checks. Most checks rely entirely on ability modifiers.
This means that Fastfeet’s +5 won’t get any better. No athletics or balance skill exists to improve the odds. Even if one did, most characters only get 3 skills.
In 3rd and 4th edition, the DM typically asks for skill checks rather than ability checks. Fastfeet probably has acrobatics skill, granting another +4 or +5 to the check. Suddenly that easy check becomes easy.
Third and fourth edition assumed checks would be skill checks, so both the skill and ability contributed bonuses. Next assumes ability checks. Skills add an unusual bonus rather than an inevitable addition.
I think this simplification makes for a better game. In addition to the virtue of simplicity, an over-reliance on skills tends to encourage players to solve problems by looking at their skill list, rather than thinking about other things their character could do in the game world.
I like the new approach, but in D&D next, the system’s numbers still seem to assume characters always get a skill bonus stacked with an ability bonus. In practice, a first level character gets a maximum bonus of +5 to a typical check. Little mathematical difference exists between a character with extraordinary ability and one with average ability. In third and fourth edition, a level 1 character like Fastfeet saw a bonus closer to +9 or +10, big enough to make a practical difference.
The solution seems obvious. For checks, the ability modifier must double, to +1 for each ability score point over 10. Now Fastfeet enjoys a +10 to dex checks, appropriate for the quickest halfling alive and consistent with the bonus typical in earlier editions.
Obviously, Fastfeet cannot also enjoy a +10 on his bow attacks. The original modifier scale must remain as combat modifiers, separate from ability check modifiers. The two scales introduce a small, necessary complexity.
On the other hand, calculating ability modifiers becomes easier. A character with 15 dexterity has a +5 ability modifier. As an added bonus, odd-numbered ability scores gain significance in the game. Suddenly 15 really is better than 14.
I realize this change bucks the history of ability modifiers established in 3rd edition, but I can trump that with an earlier precedent. Check page B60 of the Moldvay basic set from 1981. “To perform a difficult task, the player should roll the ability score or less on 1d20.” The mechanic flips the numbers, asking for a low roll, but your ability score has the same numerical effect as the modifiers I suggest. In the late 70s, I saw this mechanic used frequently. So the change qualifies as old school and it fixes the system. Seems like a win.
Still not convinced? Consider this. Over the course of an adventure, an exceptionally-strong fighter might make a hundred attack rolls. The +5 attack modifier she gains from her 18(00), I mean 20, strength improves them all. She dominates the battlefield. Over the course of the same adventure, the smooth talker with a 20 charisma may get 8 diplomacy checks, tops. Over the course of so few rolls, the 1-20 spread of the die buries the mere +5. The diplomacy skill can help. Still the most charming person you ever meet, in game terms, seems little better than the half orc who picks his nose as he negotiates with the elf king. The player who optimized the smooth talker hardly gets a chance to shine.
Pingback: Two Problems that Provoked Bounded Accuracy | DMDavid
Lets ignore for a moment the two problems of using d20 for all resolution (a hugely swingy mechanic, but unfortunately tradition) and this new skill idea that in practice is actually halfway between reliably there and never there. Given those, I agree with the problem as stated, but not the solution. The most obvious solution is actually to lower the difficulty. If the result of making checks easier across the board is undesired and the decision is made to double the impact of abilities, theres a better way than the one proposed. Rather than having two sets of calculated bonuses (one for combat and one for checks) checks could simply use the raw ability score. DCs would simply be higher by 10. I’m still not crazy about the solution, but at least you don’t need an extra spot on the character sheet.
Goken,
Your idea of using a raw ability score and raising the DC by 10 would fix all the problems I describe in this post. Thanks for sharing your thoughts.
I think your example of Fastfeet and Joe Average is flawed. The d20 is a test of skill levels, not abilities. Skill levels span from 1-20 (assume 20 is max) and hence the d20 is used. The skill test guarantees that a level 20 skill is successful at any roll check (max DC is 10+20).
Your using d20 to test abilities is inappropriate because ability modifier only range from 1 to 5 at level 1. The appropriate dice to use when testing ability is d6, not d20. The test of jumping across a cliff is not something you train for and so there is no Skill Level involved, just raw ability.
I gotta start thinking before I post. The part “The skill test guarantees that a level 20 skill is successful at any roll check (max DC is 10+20).” is obviously wrong. I meant to say success rate is 50% (55% actually) success at max DC.
Also a POSITIVE ability modifier range from 0 to 5 at level 1, not 1 to 5.
Using a d6, Fastfeet has no trouble jumping across the cliff as her DEX modifier of 5 beats the DC of 3, the mid point of d6.
Suppose that instead of walking across bridge, Fastfeet is trying to pick a DC 10 lock. She still has 15% chance of failure at level 1 (assuming 1 rank of pick lock and +5 Dex modifier). That’s because she is a novice rogue.
This discussion leads me to think that there are actually two types of tests to consider, the Skill test and the Ability test. There are several actions that test an ability instead of a skill, like breaking down a door, walking across a rickety bridge, understanding a concept, attracting a guard, surviving poisoning, etc. They aren’t uncommon actions either, so there is a need to use different sided die when making roll checks.
I don’t think it’s *quite* as bad as you make out.
Fastfeet isn’t only 25% better than J’oe, he’s 50% more likely to succeed and, perhaps more importantly, half as likely to fail.
I’d also note that this is exactly the sort of situation that (by my very limited understanding) the Advantage rules were made for. Not only should Fastfeet get +5 on his check, he should get Advantage as well, because this is clearly his bag.
Just came across this article. Thank you for posting. I happen to agree with you, though one piuece of evidence you present is incorrect–Moldvay uses 1d20, not 1d10:
“There’s always a chance. The DM may want to base a character’s chance of doing something on his or her ability scores (Strength, Dexterity, and so forth). To perform a difficult task (such as climbing up a rope or thinking of a forgotten clue), the player should roll the ability score or less on ld20. The DM may give a bonus or penalty to the roll, depending on the difficulty of the action (-4 for a simple task to +4 for a difficult one). A roll of 1 should always succeed, and a roll of 20 should always fail.”
(Moldvay, B60)
Oops. Thanks for the correction!
I’ve wanted to remove the d20 as the major mechanism for a long time, since it really is very swingy. I’ve considered replacing it with 3d6, which narrows the range by two on either end (reducing swinginess) and it has a bell curve, so characters tend to do more towards the average than not (also reducing swinginess). I’d likely expand the crit range in order to stay closer to 5% of the time.
Obviously, they won’t do this with Next, since they’re trying to stay as true to the system that it’s always been while changing it. But I could always implement it as a house rule.
I enjoy your idea, but it seems like the bonus might a little TOO big. That’s more of a general feeling than actually having looked at the number, but I too would enjoy SOMETHING that uses the odd numbers of ability scores. They had an auto-success system early on, where you automatically beat a DC equal to your score – 5 (I think that was the calculation; it’s been a while). But that was found to grow too quickly, and they scrapped the idea completely.
Jacob,
Thanks for the insight.
Many years ago, I remember reading something Steve Jackson wrote on his GURPS design. When he mentioned the game’s 3d6 core mechanic and its bell curve, Steve seemed contemptuous of D&D’s use of a d20. At the time, I failed to see his point, but I get it now. Nonetheless, when I played GURPS, I sorely missed rolling my d20.
The inverse mechanic from Moldvay and the “use ability score, raise DC” from the comments are both ideas I intend to try out in a group. I really really like the idea of making 15 better than 14, because that irks me. Why am I going to spend a point of ability increase on something that gives absolutely no benefit, other than to wait another X levels and have to do it again?
DM David, have your opinions on this changed as of the latest playtest?
Hi Cory,
The final, public playtest adds skills to the core rules, and this makes a tiny improvement. But when compared to 3E and 4E, skill bonuses have a much smaller magnitude and skills cover fewer checks, so the problem still stands. The DM guidelines tell DMs to shunt around the problem by avoiding some checks: “Take into account the ability score associated with the intended action. It’s easy for someone with a Strength score of 18 to flip over a table, though not easy for someone with a Strength score of 9.” I’ll revisit the subject in a post next week.
Hi David,
It’s been a while since you posted this. Have you tried basing these on the ability score-10 rather than modifier? How did it go? What did you do about the proficiency bonus? Did you scale that in some way as well so that taking a skill makes a difference?
Ian
Pingback: Bounded accuracy and matters of taste | DMDavid