Friday, April 17, 2009

Skill Checks, part 2

Neubert had some good questions about my group skill check system, so I thought I would go into it further.

The math: First we ask, what is the average skill bonus of a L1 character? If an L1 character has stats 18, 16, 13, 12, 11, 10 (after racial bonuses), the average stat bonus is +1.5. The average character class has 4 skills, but some characteres may have feats or classes that give more, so assume 4.25 skills, which is ¼ of the total number of skills. This gives an average skill bonus of +1.25. And the average race has +2 in two skills, an average of +0.25 per skill. This gives an average skill bonus of +3.

How this is divided among the party will vary, but let’s assume +8, +5, +2, +0, +0. This means the chance of no one making a DC 20 check is 55% x 70% x 85% x 95% x 95% = 30%, so the chance of success is 70%.

The actual DC chart:
Characters should almost always succeed: DC 16
Characters should usually succeed: DC 18
Characters should succeed more often than not: DC 20
Character success is 50/50: DC 22
Characters should fail more often than not: DC 24
Characters should usually fail: DC 26
Characters should almost always fail: DC 28

The rules I gave for group skill checks are only for the most common type of skill checks, checks in which a success for one person is a success for everyone. Examples:

Almost all Knowledge checks (Arcana, Dungeoneering, History, Nature, Religion, Streetwise)
Noncombat Perception and Insight checks to see if the group notices something

I also like to use it for social skill checks (Bluff, Diplomacy, Intimidate). While you might argue that it is more appropriate for one diplomat to make all the rolls while everyone else stays quiet, I prefer to be a bit more abstract and allow everyone to contribute. I imagine that the charismatic types are contributing more, but the better everyone is at social skills, the more chance the party as a whole has of succeeding. You might imagine, for instance, that the “special effect” of the surly dwarf succeeding in a group diplomacy roll, is that he managed to refrain from saying something rude that would have ruined the negotiations.

Skill checks in which everyone has to succeed – in which only the worst performance counts – are a totally different animal that my rules were not meant to cover. This is a much trickier problem. In practically all traditional skill systems, it is impossible to make a group stealth check, since lots of people are rolling and someone is bound to screw up, especially when that someone is clumsy and wearing plate mail. An idea I’ve been considering is that everyone rolls, and everyone has to succeed, with one change. For every 5 points by which a character makes his initial skill roll, he allows one reroll for one other character. I think this will have the feel I want – the weak link is certainly a weak link, but everyone makes a difference and having some experts in the party helps out a lot. The math for this system is decidedly non-trivial and I haven’t had time to work it out, but I’d estimate DC 10 or 11 would be equivalent to a DC 20 best-roll skill check.

Example: The rogue, ranger, paladin, fighter, and wizard are making a group stealth check at DC 10. The rogue gets a 22, the ranger 17, the paladin 6, the fighter 9, and the wizard 12. The rogue and the ranger have earned 3 rerolls between them, so they give one to the paladin and one to the fighter. Now the paladin gets a 4 and the fighter gets a 20. The fighter does not earn any rerolls (this isn’t the initial check), but there is still one reroll left over from the ranger and the rogue. The paladin rerolls, gets a total of 11, and passes the check. The group is stealthy!

Skill checks in which each person succeeds or fails separately (such as, roll an Endurance check or lose a healing surge) are straightforward and easy to do. The only thing to remember is that such a check is much, much harder than a group check. DC 10 would be about equivalent to a DC 20 group skill check.

The system doesn’t envision “opposed” skill checks, as I’ve been using it for skill checks baked into a scenario. In general, I don’t normally roll the monsters off against the players, I just set a difficulty for the players. This puts more control into the rolls of the player (all that rolling for the monster does is make the difficulty number for the players unpredictable, so sometimes the party fails with a good roll or succeeds with a bad roll). If I’m making the scenario, I don’t really need to look at the stats of the monsters, I just set the difficulty based on the situation. If you need to make an impromptu check, a DC of 12 + monster skill bonus seems reasonable if you want to players to succeed more often than not, the equivalent of a DC 20 group skill check.

Skill challenges are a whole different ball game that needs its own article.

Note that physical skills are slightly harder if the characters are wearing armor, so you may want to make them one point easier. Also note that “trained-only” checks are more difficult, so you would want to make them around DC 16 to be equivalent to a normal DC 20 check.

One thing I forgot to mention last time: I don’t use passive skill checks. Passive skill checks guarantee that nothing matters except the character with the highest skill, and I am trying to avoid that. And rolling dice is more fun for the players.

Neubert asked about converting skill checks from modules. The difficulty with this is that when the module says “DC 21 Religion check”, there is generally no telling how difficult the writer wants the check to actually be, in terms of whether the writer expects group skill checks to be much easier than individual skill checks, and how often the writer wants to party to fail. Even if the writer does nothing but look at the charts in the DMG, everyone can interpret Easy, Moderate, and Hard in different ways. You have to look at the style of the various skill checks in the module.

However, if the difficulty is taken from the errata DMG (which seems to put DC 10 as Moderate), and you assume that “Moderate” is a typical difficulty, then group skill checks will be 10 points too easy and individual skill checks will be right on. I took a look over the adventures in Dungeon 164, and based on those, it appears that if use my system and allow everyone to roll the check, and you want the party to have a decent chance of failing, you should add 5 to the DC of all the group skil checks.

As far as checks in the book with fixed DC’s – I myself don’t often use the listed DC’s, I prefer to rely on my own judgment and just pick a DC based on how hard I think the check should be for the average party. Looking at the skill section of the book, I see that the Arcana checks are pretty harsh already, no need to make them worse. Most of the other group-oriented checks have fixed DC’s, which I don’t normally use. Given the wacky way in which D&D characters become experts in all skills just by gaining levels, I figure that high level monsters should correspondingly be harder to hear, harder to track, and harder to obtain useful knowledge about. I’d need a whole blog article to describe how I’d specifically handle various types of skill checks.

With regards to making combat stealth checks, the new stealth rules are a great improvement over the (lack of) rules in the original players handbook, but I still don't completely understand how the designers intend stealth to work. The rules are quite understandable if there is one player trying to be stealthy against one monster, but what happens when there are multiple monsters is not clear. I've been playing that you make only one stealth roll, but you apply that roll separately against each monster, and you can be hidden against some monsters but not others. In that case, it is not too hard to hide successfully and ambush some monsters. These combat skill rolls have a very different meaning from noncombat skill rolls, and I haven't really played with them enough to have an opinion on whether they are too easy or too difficult.

I thought a little about the progression of skill rolls with levels. You get +1 every 2 levels just from the level bonus. You gain 11 points of statistics every 10 levels, which is about +1 to every skill for every 10 levels. Looking at the book, I see that most of the feats and utility powers than improve skill checks are low-level, more like something to factor into the basic numbers than something you gain with levels. You certainly can gain skill bonus magic items with levels – but you probably only are wearing about 1 even at the highest level. So whether you get a lot of bonuses at high level depends on whether your game encourages your high-level characters to buy hordes of cheap magic items and swap them about before performing each check. If not, you probably won’t gain even as much as +1 in every skill by 30th level.

You will, however, become more polarized – better in your best skills compared to your other skills. This is statistically advantageous. So if your bonuses are +29/+23/+18/+16/+16 at 30th level, your chance of failing against DC 38 is (40% x 70% x 95%) = 27%, pretty close to the same. This works out to increasing the difficulty by 0.6 per level. Not too far from the rate of progression in the DMG, which is 0.66 per level.


  1. Interesting idea regarding the stealth check. It will of course increase complexity slightly, but the amount of stealth rolls the group performs during a session shouldn't be overwhelming either.
    I was thinking there might be a potential issue with only the most stealthy characters sneaking forward, but after writing out a whole example, I think I disproved my own theory.
    I am curious however, would you use this rule for monsters as well? Or would they simply create a set DC for the players to roll against, as you also mentioned in your post?
    (This leads me to a follow-up question regarding your view on using passive perception/insight vs calling for a roll?).

    I think your are running the new stealth rules from the book correctly - at least I also understood the rules as the player making one stealth check and then being hidden from some monsters and seen by others.

  2. With monsters, I would normally just set a DC for the characters. No need to be "fair" to the monsters, I just decide how hard I want it to be for the players to spot them and have them roll the dice. My experience with my players has been that they like rolling dice, so I let them roll the dice in almost all skill situations.

    I rarely use passive skill checks at all. I started writing an answer describing issues involved in passive skill checks, but it seemed involved and should be in a full article. But a simple reason is that it fits my theory that players like to roll dice.