[D&D4] Skill DCs Rebuilt From the Ground Up

In game design, role-playing games on September 8th, 2008 at 8:00 am

As I said in my last post, I am less than impressed with the skill check system in D&D 4th edition. Most of these problems are rooted in the idea of resolving checks using d20 roll + skill modifier. The high variability and flat probability curve are a nightmare to work with. Actually fixing the problems would require starting from zero, and pitching the d20 heritage, though.

Instead, I just want to hack the skill system a bit, to get it working well enough to be fun. There will likely be several stages to this, but the first starts at the foundation. The baseline DCs in the Dungeon Master’s Guide and in the first batches of errata released for it just do not match the game’s mathematical structure. They need to be replaced.

To develop my numbers, I’ve used three “test cases” – skill modifiers worked up from more or less typical character build cases. Working from these skill mods, I then worked with the success rates I wanted for easy, moderate and hard tasks, to see what DCs I would have to use. Fortunately, there was a lot of “close enough” results between test cases and the number of DCs you need to cover all the bases is reasonable. While I have expanded the number of benchmark DCs for each level from three to four, I’d feared that five or more would be necessary before I started doing the math.

Please note that these DCs are intended only for use in one-off skill checks. Skill Challenges are their own beast, and need to be built from different math.

The test cases I used to derive DCs are:

  1. Untrained – A character that has 12 (+1) in the relevant stat and no skill training. The character does not acquire magic items relevant to the skill, and does not raise the relevant stat, except at level 11 and level 21 when all stat values increase by 1.
  2. Trained – The trained character begins with a 16 (+3) in the relevant stat, and is trained in the use of the skill (+5). The character acquires a magic item that helps with skill checks once every fifth level (beginning at level five), the item’s bonus equal to the character’s level divided by five. The character raises the relevant stat at every opportunity.
  3. “Min-Max” – The trained character begins with a 20 (+5) in the relevant stat (thanks to a racial modifier) and is trained in the use of the skill (+5) and has taken a Skill Focus feat (+3) for this skill. The character acquires a magic item that helps with skill checks once every fifth level (beginning at level five), the item’s bonus equal to the character’s level divided by five. The character raises the relevant stat at every opportunity. Note that, in spite of the label, this is not a true min-max build. There is still room for characters to develop skill mods higher than this, including acquiring a relevant magic item before level 5* the item’s skill bonus.

This resulted in the following Skill modifiers, with a little inaccuracy here and there due to grouping levels into pairs:

Level Normal Trained Min-max
1-2 2 9 14
3-4 3 10 15
5-6 4 12 18
7-8 5 13 19
9-10 6 15 20
11-12 7 17 22
13-14 8 18 23
15-16 9 20 25
17-18 10 22 27
19-20 11 23 28
21-22 13 25 30
23-24 14 26 31
25-26 15 28 33
27-28 16 30 34
29-30 17 31 36

Next, I decided what what I felt the approximate success percentages should be for easy, moderate and hard tasks. I settled on 85% for easy, 65% for moderate, and 45% for hard. I’m sure that many readers will find these unusually high – I believe WotC aimed for 50% success for moderate difficulties – but these are supposed to be heroes, so I want to decrease the whiff factor. One reason why I am being so transparent about my methods is to make it easy to adjust my results to suit your own tastes.

My assumptions turned out the following DCs for each typical character for easy, moderate and hard tasks.

Level Normal Trained Min-max
Easy (85%) Medium (65%) Hard (45%) Easy (85%) Medium (65%) Hard (45%) Easy (85%) Medium (65%) Hard (45%)
1-2 5 9 13 12 16 20 17 21 25
3-4 6 10 14 13 17 21 18 22 26
5-6 7 11 15 15 19 23 21 25 29
7-8 8 12 16 16 20 24 22 26 30
9-10 9 13 17 18 22 26 23 27 31
11-12 10 14 18 20 24 28 25 29 33
13-14 11 15 19 21 25 29 26 30 34
15-16 12 16 20 23 27 31 28 32 36
17-18 13 17 21 25 29 33 30 34 38
19-20 14 18 22 26 30 34 31 35 39
21-22 16 20 24 28 32 36 33 37 41
23-24 17 21 25 29 33 37 34 38 42
25-26 18 22 26 31 35 39 36 40 44
27-28 19 23 27 33 37 41 37 41 45
29-30 20 24 28 34 38 42 39 43 47

I decided that easy tasks for the untrained should be “say yes” situations – leave the dice on the table. I also conflated similar columns, like hard for untrained characters and easy for trained characters. That led to the following set of benchmark DCs:

Easy Medium Hard Heroic
1-2 13 17 21 25
3-4 14 18 22 26
5-6 15 21 25 29
7-8 16 22 26 30
9-10 17 23 27 31
11-12 18 25 29 33
13-14 19 26 30 34
15-16 20 28 32 36
17-18 21 30 34 38
19-20 22 31 35 39
21-22 24 33 37 41
23-24 25 34 38 42
25-26 26 36 40 44
27-28 27 37 41 45
29-30 28 39 43 47

To settle on an appropriate DC for a Skill check, determine which test case is closest to the best character in the party at that skill. Then assign the DC for a character of that level based on your estimate of the task’s difficulty as follows:

Case Easy Moderate Hard
Untrained Say Yes Easy Moderate
Trained Easy Moderate Hard
Min-Max Moderate Hard Heroic

So far, I’ve only used these DCs in one session with Level 1 characters, so they are not battle-tested. Nevertheless, I already like them much better than what WotC has given us.

Use them in good health.

  1. […] Reconfiguring AD&D: A note on skill DCs Posted by faustusnotes under Uncategorized   Linnaeus takes on the task of defining what a skill DC should be, as part of a revision of the AD&D 4th […]

  2. I have been looking at the skill system a lot as well, and have the odd report about it on my blog. I long since gave up on d20s – I use 2d10, and drop the threat roll for critical hits. Critical hits and fumbles are now so rare that they really do add spice to the game.

  3. I don’t use fumbles, myself. A 3.5 campaign I play in does, and I find them rather frustrating and unheroic.

    Criticals don’t rely on threat rolls in 4e. Instead, a natural 20 that would hit anyway deals maximum damage. It lacks a bit of the pop of the threat system, but it keeps the action moving along at a nice clip.

    Moving away from d20 is very tempting, but it does weird things to probabilities so that bonuses and penalties don’t react in the way you’d expect, so it would require serious thought before I adopted an alternative die roll for skill checks.

  4. […] Rebuilding Skill DCs; well written and persuasive. […]

  5. I like the weird thing it does to probabilities. When you think about dice rolls you think about the “average” roll of 10.5 on a d20, but that is not the most likely roll – on 2d10 it is a number between 9 and 12 (can’t be bothered working out which one, I guess 11 or 12). So when you set DCs you can think of what the most likely outcome would be for a person with a particular skill bonus. And you can think of that 9-12 block of values as the ones you’re aiming the skill DC at.

    I like applying the bell curve to heroic behaviour, I suppose.

    Also I come from a long and tragic history of Rolemaster DM-ing, where fumbles are very definitely heroic, so I quite like them. A particularly heroic Rolemaster fumble would be the one involving the tortoise…

  6. I have played much Rolemaster. They’re definitely part of the fun of RM (along with the criticals), but RM has a much different vibe than D&D4. It’s a vibe much better suited to the long past days of high school, too 😀

    And, the more I think about it, the less enamoured I am of roll+skill mod vs. DC systems. I can live with it in D&D, since it’s a secondary system, but the flaws (for my tastes) are becoming more and more apparent.

  7. very high school indeed! And requiring a little too much attention to detail… which is something I like about a smoothly-run D&D skill system without too much rules-lawyering or fiddly detail.

    I’m not sure I’ve seen many functional alternatives to this skill+roll vs. DC method. The main alternatives seem to be the multiple-dice tests, but they often also seem to be very detail-heavy and equally dependent on DM caprice. Can you think of better alternatives?

  8. Well, the best known option is dice poll systems. Some of those include TN mods, so there is a little overlap, but modifying the number of dice a character gets to roll seems less GM fiat-y than DCs do. I realize the difference isn’t that great in reality, but it feels more correct to me.

    I also play conflict resolution games, and some of them have interesting resolution systems, too.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s

%d bloggers like this: