May 10, 2016

Understanding 5e D&D Skills

Today's topic is 5th edition D&D skill checks.  I'm going to examine 5e skill DCs.  And I'll give you a few pointers along the way.
The design philosophy for 5e D&D skill checks was to make DC 10 feel Easy, 15 feel Moderate, and 20 feel Hard. Then it gradually raises the PCs' competency in their best skills by about 6 points over 20 levels.  (You get 4 from Proficiency and 2 from Attribute increases).

Mathematical Probabilities and Subjective Feelings
This isn't a math article, but we're going to go over the math a little at the beginning.  Here's the refresher:  Each number on a d20 is one in twenty, which means 5%.  The probability of failure for any DC is (DC-Skill Bonus-1)*5%.  The probability of success is (Skill+21-DC)*5%.  Got it?
DC 10 is a 55% chance of success for a person with +0 in the skill.  Does that feel "Easy" to you?  Not at all!  Failing almost half the time is not easy!  But who has a +0?
Well, you do.  Consider any of the sample characters in the Starter Set or the Pregens on the DM's Guild.  Each one has many skills in the -1 to +1 range -- in fact, most of their skills fall in that range!  So how is DC 10 Easy?
That was a rhetorical question, of course.  Here's how:  The party picks the best person for every task.  Someone in the party has a decent modifier in just about every skill.  There are 18 skills.  A party of four PCs has enough proficiencies among them to have a good modifier in almost every skill, between them.  Players naturally collaborate to make their characters so there's not too much overlap.  This protects their niche ("I'm the sneaky one") and ensures the party can take on most challenges.

The core assumptions
The core three DCs (10/15/20) are designed for situations that meet the following criteria:
  1. There are four or five PCs.  As a result, there are enough skill proficiencies in the party that every important skill is represnted. 
  2. The characters have minimal skill overlap.  As a result, when the party is together, there's a PC with a decent rating in almost every skill.
  3. The party is together so they can select the most skilled character for the task.
  4. The DM calls for a single roll for a single task.
  5. The party is first level

There's a PC with a decent rating in almost every skill when the party is together.
A level 1 Rogue probably has +7 Stealth.  (That's +3 from Dexterity and +2 Proficiency, doubled for Expertise).  DC 10 Stealth is Easy for that character.  The rogue has a 90% chance to succeed.  That's easy both from a subjective point of view and an objective, mathematical point of view.  DC 15 to pick a lock is Moderate for that character.  The rogue has a +5 in Thieves' Tools Proficiency, using Dexterity; so that's a 55% chance to succeed.  Moderate.  DC 20 Acrobatics is Hard for that character.  They have +5 there, too, so that's just a 30% chance to succeed - and that's pretty Hard!
Can the party Cleric pick the lock?  Assuming the cleric even has Thieves' Tools, the task is DC 15, and the cleric has 10 Dexterity.  There's a 30% chance the Cleric will succeed.  That's not Moderate.  That's Hard.  The Fighter isn't much better off, with a +2 (40%).  The elf Wizard is fumbling about with a +1 (35%).  It's Hard for three-fourths of the party but Moderate for the Rogue, so DC 15 is Moderate for the party.
So if the whole party is together when they get to the locked door, and the party contains a good mix of skills, the locked door task is Moderate difficulty.  If the party doesn't have someone proficient in Thieves' Tools for some reason, or if the Thief is off doing something else, the task becomes Hard.

Single task rolls work; multiple rolls for a single task don't
When the Rogue is picking the lock, it's just one roll.  Succeed, and the door opens.  Fail, and the patrols find you before you can open the lock.  If the DM requires multiple rolls to succeed,where just one failed check spells doom, the probabilities change.
Here's an example.  Let's take that Rogue with +7 Stealth again.  The Rogue has to sneak past six sentries.  The Sentries have 10 Wisdom and are not Proficient in Perception, making each roll DC 10 - Easy.  The task truly is Easy if the DM has the Rogue roll one time for the whole task.  The task is not easy if the rogue has to roll for each sentry.  If just one sentry spots the Rogue, they will raise the alarm, and the scouting mission has failed.
The chance to succeed at a "get 6 successes before 1 failure" task is [Chance to Succeed once]^6
90%^6 is 53%.  The challenge is not Easy; it's Moderate at best!
If the DM had set the task at Moderate (DC 15) for each sentry, the Rogue would have a 65% chance to make each check.  When we do the math (0.65^6) the rogue has just an 8% chance to succeed them all without failing even once.  That's not just Hard.  That's Very Hard.

Aside: Group Checks
Group checks are another example of multiple rolls, but they actually work, mathematically.  Using a Group Check, the DM calls for the whole party to attempt a skill check.  If at least half succeed, the whole party succeeds.  As long as the party has a diverse mix of skills, at level 1, most Group Checks work out using the standard difficulty spectrum. A DC 10 Group Check is Easy; a DC 15 Group Check is Moderate; and a DC 20 Group Check is Hard.

Tasks get easier as you go up in level
The last criterion is important.  The 5e designers didn't want a "treadmill" where the skill difficulties moved up along with the PCs' skill levels.  The skill DCs stay put.  The PCs get better.  
Your attributes improve every handful of levels, and your Proficiency bonus improves, too.  So that Rogue, at level 20, has +17 Stealth.  Very Hard checks aren't Very Hard anymore.  They're pretty easy.
Now the level 20 Rogue has an Easy time at Elaborate Stealth challenges, and a Moderate to Hard time with Impenetrable Stealth challenges.  That makes more sense.
This also means that if you obey the other three criteria, you need to use more and more complex tasks.  Don't just raise the DCs you use -- make the tasks harder first, then give them higher DCs to reflect the greater challenge.
  • At levels 1-7:  Most tasks you set before the PCs should be Easy and Medium challenges.  You can sprinkle a few Hard challenges in when there are good stakes for a "you'll probably lose" wager.  
  • At levels 8-12:  At level 8 or 9,  the party will be able to succeed almost any Easy task even if they roll a 1.  Practically speaking, the PCs will still encounter Easy tasks, but you won't be able to build exciting encounters around them.  You'll need to use mostly Medium DCs with occasional Hard and even a few rare Very Hard tasks where the stakes are good for a "you'll probably lose" wager.
  • At levels 13+:  When the party is all together, use a mix of Medium and Hard tasks, with occasional Very Hard tasks and even a few Near Impossible tasks that play to the party's strengths, when the stakes are right for them.


If there is no Wizard, go easy on Intelligence skills
It's generally acknowledged that 5th edition D&D is the "dumbest edition" because the Intelligence attribute has so little value to every class except Wizards.  Most characters take Intelligence 8 or 10.  You rarely see a non-Wizard character with Intelligence higher than 12.  Consequently, a level 20 Cleric will likely have Religion at +5 or +6.  
This means that you can continue to use easier tasks for Intelligence skills for all 20 levels, unless the party has a Wizard.   (Note that Knowledge Domain clerics and College of Lore Bards may also have good Intelligence skills, even if they don't have high Intelligence attributes.)

Everyone should have a social skill
Every class in 5th edition D&D gets at least one useful social skill.  Persuasion, Deception and Intimidation are the most commonly used social skills, but you can use History or Insight as well.  Between those five skills, any character should have one, if not two, skills that are useful in a social scene.  
In this case, it's good for the PCs to overlap.  It would be terribly dull if only one character took Persuasion, so only one player ever attempted to persuade NPCs of things! 
Because the skill you use depends on the approach you take, social skills are often interchangeable.  Intimidation might be the most obvious approach for dealing with goblins, but you can use Deception if you try to trick them or Persuasion if you try to bribe them.  
In social scenes, it's tempting to call for individual task rolls.  But that encourages the Sorcerer, Warlock, Paladin, or Bard to do all the talking.  Instead, encourage everyont to participate by using group checks (see above).  Everyone who participates gets to make a check using the skill relevant to the approach they used against a DC based on the NPC's social savvy and how much leverage the PCs have (or how much leverage the NPCs have over them!).  

When they split the party, go back to easier tasks
You've heard the cry:  "Don't split the party!"  But it happens all the time.  When the party is split, use simpler tasks to challenge the PCs.  
For instance, if the level 16 Rogue is scouting ahead in the forest, and has to find the hidden goblin lair, an Easy Survival check might be enough of a challenge for the poor city kid.  The Rogue probably has only a +1 in Survival, even at level 16.

Justify your Skill DCs!
Remember, your task DC choice should always be justified by the game-world conditions.  This isn't just a D&D thing - this is a GM core competency, even in more narrative story-focused games!  If climbing a stone wall was Easy at level 1, it should still be Easy at level 14.  Yes, I know there's no way the Proficient characters can fail an Easy skill check at level 14.  But either call for a roll and let the PCs feel powerful for being able to succeed even on a 1; don't call for a roll at all; split the party so the non-proficient characters have to climb the wall unaided; or change the conditions:  If there's a rain pouring down and the PCs have to rush up the wall, maybe it's not Easy anymore.
If you justify higher DCs, the players feel like they're growing in ability.  If you just raise the DCs arbitrarily, the players feel like they're slipping backwards on a treadmill.  You don't want that.

Skill monkeys
Bards and Rogues will dominate your skill scenes if you let them.  They get all kinds of bonuses to skill checks.  They're often called "Skill Monkey" classes.  There are two ways to avoid that.
  1. When you want to design a longer exploration scene, choose multiple opportunities and threats within the scene that relate to a variety of skills, then get everyone to participate.  If the PCs find themselves in a corridor with strange runes along the walls, odd tracks in the dust, and a rumbling sound ahead around a corner.  The Rogue goes to scout the rumbling sound, the Wizard reads the runes, the Cleric helps interpret the religious prophecy the runes spell out, and the Ranger tries to learn something from the tracks.
  2. Typically, the Skill Monkey is going to leap in on every skill scene with a useful contribution.  That's OK - their character is designed for that.  Once the Skill Monkey gets to work on their part of the scene, turn to another player and say "While Darkblade VanRogue scouts the hallway, what is Wizard Wandsworth doing?" 
A few details about high level 5e skill monkeys...
  • Reliable Talent:  At level 11, Rogues treat all rolls below 10 on a d20 for the skills in which they have proficiency as a 10.  Bards of the College of Lore get Peerless Skill which also dramatically improves skill rolls.
  • Expertise:  Expertise (which Bards and Rogues get) doubles Proficiency for a limited set of skills.  This balloons fast.  


Spells solve skill scenes

Many utility spells effectively negate skill checks.  Invisibility and Dimension Door make sneaking past sentries a snap.  Clairvoyance makes scouting easy.  Charm Person can handle Persuasion for a brief scene.  Spider Climb and Fly can make it easy to get over and past obstacles.  If the spell obviously resolves the challenge, it resolves the challenge.  But if there's any question, use this guideline:
  • A spell-caster's spells can be classified as follows:  Their best spells are the spells of the highest level they can cast.  Their second-best spells are the spell level below the highest level they can cast.  Their cheap spells are their spells of two or more levels below that.  
  • Spellcasters' cheap spells should provide bonuses to skill checks.  
    • For instance, a 9th level wizard may use Feather Fall to make it easier for her to walk across a narrow ledge, reasoning that if she's as light as a feather, it's easier to balance.  The GM might let her claim Advantage on the check in exchange for using the spell in a creative manner.
  • Spellcasters' best and second-best spells may resolve the challenge entirely.  
    • For instance, the 9th level wizard could use the Fly spell, cast at 5th level, to let three characters fly.  That would get herself, the cleric, and the fighter across the narrow ledge.  The rogue can easily make the Acrobatics check to cross it.  The spell basically solved the challenge for the whole party.  That deed cost her the only 5th level spell she can cast today, and one of just four spells of 4th-5th level she can cast today.
    • Level 6-9 spells always count as a spellcaster's best or second best spells.  Even a 20th level Wizard doesn't get a lot of them.

Summary

  • The 5e task DCs work best when the group is all together.  At low levels (1-7) use a mix of Easy and Moderate tasks, mostly.  Use slightly harder tasks at mid levels (8-12).  Phase out Easy tasks and mostly use Moderate tasks.  Use harder still at high levels (13+).  Mostly use an even mix of Moderate and Hard tasks.
  • Group checks work!  Use them in social scenes to encourage everyone to participate.
  • If you call for multiple rolls to achieve a single task, where failing once is catastrophic, use easier DCs.
  • If they split the party, use easier tasks.
  • Use more complicated scenes to let the "skill monkey" shine while giving other characters something useful to do.
  • Justify your skill DCs!
  • Spells can "solve" exploration scenes.  If it's not obvious how to handle it, the rule of thumb is the better the spell slot sacrificed, the more effective the spell should be.