Using the Rules Better: Fixing Contested Checks

Created with Sketch.

Using the Rules Better: Fixing Contested Checks

I started out this article planning on analyzing the possibilities for Passive Stealth, which is something that a couple of my players asked about. I’ve already written an article about using passive checks for more than just Perception, but when I started thinking about something like Passive Stealth, I came up hard against the concept of contested checks in D&D 5th Edition. And suddenly, the article stopped being just about Passive Stealth and started being about how contested checks are broken, and why they’re broken, and how we might be able to fix that. So, if you’ve just been reading this intro and were really hoping to hear all about Passive Stealth, you’re going to be disappointed for a little while. Once I figure out how contested checks can be made to work right, I’ll be able to revisit Passive Stealth, as well as a lot of other things that have to do with contested checks. For now, let’s look at those contested checks and figure out where it all went wrong.

Contested Checks: the Problem

The short statement of the problem with contested checks is that they don’t behave quite like anything else in the entire d20 System, and that produces results in unpredictable and undesirable ways. What are the particular ways in which they go wrong? Let’s go back to the basics, and take a careful look at how the d20 System works. More importantly, let’s look at why it works. At that point I’ll be ready to present a couple of example cases, and I think by that point I’ll be able to articulate the problem with contested checks, and point out why we don’t need them and actually already have something better. I’ll hopefully be able to offer some kind of solution for the problem by the end of the article, and maybe it’ll even be worth having read all of the rest of it to get there.

Going Back to the Basics

The development of the d20 System is probably one of the single greatest innovations in the history of tabletop gaming, and possibly in the history of games in general. It represents a paradigm shift in how games like D&D work in a mechanical sense, the same way that Gygax and Arneson created a paradigm shift in how tabletop games view individuals and groups. The strength of the system – the thing that makes it such a big deal – is that it is simple. Really simple, in a way that previous systems were not. In fact, it is considerably simpler than a lot of the systems currently in use for other tabletop RPG’s: take a look at any system with a dice pool mechanic, and you’ll see what I mean. The d20 System is so simple, in fact, that I can give you a single one-line equation that tells you how it works. The equation varies a little depending on the game, but here’s how it goes for D&D:

1d20 + Ability Modifier + Skill Modifier + Circumstance Modifier = X

That’s it. Once you have your X, you compare it to a Difficulty Class for the attempted task, and then you decide based on the comparison whether the attempt was a success or a failure. But, just for the sake of clarity, let’s break down each of the things that we added together and discuss what they mean and why they matter. If you’ve been around D&D for a while, this should be a pretty straightforward review, but I do like to make sure I define my terms before I start tearing things up.

1d20: the random number, between 1 and 20, generated by the roll of a 20-sided die. This represents the operation of chance or luck in the attempt.

Ability Modifier: a number derived from the character’s relevant ability score. This bonus represents the character’s natural capacity to cope with situations of the type required for the attempt.

Skill Modifier: if the character has any special training or special knowledge that would benefit the character on this specific attempt, an additional bonus is applied.

Circumstance Modifier: a number that accounts for the context of the attempt. If there are factors that are not dependent on the character’s abilities or skills or on the task itself, but on something else, this represents those additional factors. NB: D&D 5th Edition doesn’t actually have numerical circumstance bonuses like 3rd Edition and Pathfinder, but the advantage and disadvantage mechanics perform the same function. If you really wanted to calculate advantage on a particular roll as a numerical circumstance bonus, just subtract the lower d20 roll number from the higher d20 roll number and fill that in this space.

So, the d20 roll value is provided on the spot by rolling the die, the ability and skill modifiers come straight off of the character sheet, and the circumstance modifiers are decided by the DM when the check is made. All that’s left is the Difficulty Class, which is just a numerical expression of how difficult the task is, and there are really only two ways that DC’s are decided.

Sometimes there’s a number on the character sheet that is the DC for a certain check, like the one that tells how much armor the character is wearing, or how potent the character’s spells are, and the rules of the game tell us how to calculate these. Interestingly, a lot of these calculations use the character’s statistics as part of the determination. This will be very important later on, by the way.

The rest of the time the DC is a judgement call on the part of the Dungeon Master, but there are guidelines for what numbers correlate with how hard the task is. A DC of 5 is very easy, so easy that almost anyone can be fairly sure of success. A DC of 30 is almost impossible, such that even very powerful and well-trained characters will need a hefty measure of luck to succeed. All the numbers in between are the shades of difficulty, and they are really very arbitrary. Even when a hardcover module gives a DC for picking a lock or noticing a secret door, that only means that the author chose the DC (as opposed to calculating it in some way) when writing the adventure, instead of the DM choosing the DC at the table.

The Perfect Example: Climbing a Cliff

I think that the best way to talk about all of these factors is to think about a character planning on climbing a cliff. What sorts of things make a difference when someone (in D&D or in real life) is planning on climbing a cliff?

Strength is important. People who are in better shape physically will do better on tasks that require a lot of muscle, and that’s the Strength ability modifier for a D&D character. High scores in Strength mean larger modifiers added to the d20 roll, and that makes sense. Strong characters will benefit from their strength when climbing cliffs, and a character’s strength is represented numerically right there on the character sheet, as calculated by the rules of the game. When we’re calculating the success of the climb, we use the calculated bonus as the ability modifier in that equation above.

Skill is also important. If someone has experience in climbing, or even in doing the kind of physical tasks that are relevant to climbing (like doing pull-ups, for example), they’ll have an easier time making the climb. 5th Edition expresses this as a proficiency bonus, which means that characters who are considered proficient in a skill get an extra modifier to represent training or aptitude. Again, it’s on the character sheet, and the rules of the game give us the number to use. If a character has the right skills, we use the calculated bonus as the skill modifier in the equation.

Circumstances are also important, but there’s a difference between circumstances that increase or decrease the DC and circumstances that just modify the d20 roll. For the purposes of the cliff climb, the DC changes based upon the features of the cliff: is it a sheer rock face, or does it have lots of roots and rocks for handholds? Those are the type of factors that are considered when the DC is set, and the thing that they have in common is that they are integral to the cliff itself.

But, as the DM, I might consider non-cliff factors to be relevant. I could grant advantage if the character has a set of climbing gear. I could impose disadvantage if there’s a really strong wind blowing while the climb is taking place. Those sorts of things definitely make the climb easier or more difficult, but they are incidental. The cliff would still be there without them, and it would still have the same features that would justify a particular DC. You can tell these kinds of factors apart because you can imagine how things would change without them: what if the character waited for the wind to die down? The cliff would still be there just like always, but the disadvantage would no longer apply. If there are context factors of that sort, then they form the circumstance modifier from the equation. If there aren’t any of those kinds of complications, we just leave that part out.

And, finally, there’s the d20 roll. Unexpected things happen when you attempt a task, and the die roll represents that by providing a random number to work into the calculation. For the cliff climb, the character might accidentally miss a handhold, or put a foot on a loose tree root that was supposed to be solid, or develop a cramp halfway up. It doesn’t really matter what the unfortunate circumstances are, although if you’re the DM you might want to narrate them if they cause the climb to fail.

But, if the beneficial modifiers are substantial enough, even lousy rolls aren’t sufficient to make the attempt fail. Players have a certain justifiable sense of security about this sort of thing. “My character is strong and fit and has a set of climbing gear, so this climb should succeed even if I roll badly.” Decisions are based on this kind of thinking, and making the decisions the character would make is a good working definition of roleplaying.

So that’s our cliff climbing example, and it illustrates exactly how the system works. And the system works really well: all of the factors that contribute to success or failure are accounted for and expressed as numbers that we can add, subtract, and compare. But then the rules take a left turn, and suddenly there’s a double standard going on. When we’re evaluating character success against the environment, things work just like our cliff climbing example. Evaluating character success against other characters, though, is sometimes treated differently. In many (but not all) situations where living beings (who are generally called “characters” in this article, but also include what you might call “creatures”) are contending with one another, there are no more DC’s to compare against. Now success is determined by comparing one character’s check against another character’s corresponding check. That’s what a contested check is, and that’s where a lot of trouble begins.

Contested Checks: Throw Out the Basics

The key notion about how contested checks work is that when two characters are trying to accomplish different and incompatible goals, the thing to do is to pit their abilities and skills against each other. Let’s look back at the equation from before:

1d20 + Ability Modifier + Skill Modifier + Circumstance Modifier = X

When we talk about contested checks, we take that X from the right side of the equation and replace it with all the things that are on the left side. Now we have the results for one character expressed numerically on the left side of the equals sign, and the results for the other character expressed in just the same way on the right side. Bear in mind that the circumstance modifier is expressed in 5E as advantage or disadvantage, and that they cancel each other out; if both characters have advantage, neither one gets to use it, and the same goes for disadvantage. That really doesn’t matter for this discussion, but it’s just as well to mention it and get it out of the way.

Contested checks tend to work one of two ways. Either each contender will be using the same ability (and possibly the same skill) against each other, or else they will be using different abilities and skills which have been declared (by someone) as properly opposing one another. Let’s have some examples.

Example One: Fighting Over a Door

Let’s consider two characters with a closed door between them. One of the characters wants to force the door open, and the other one wants to keep it closed. According to the rules, this is a time when we would use a contested check. Whichever party wins the check gets their way with the door. The rules even tell us that if the contested check is a tie, then we maintain the status quo. This entire notion is problematic.

Remember that with a contested check we are comparing one character’s modifiers to the other character’s modifiers. This seems completely fine, because we are actually pitting the contenders against one another. Fighting over the door is a test of strength, so comparing one character’s strength to the other’s makes good sense. If they were fighting over a trapdoor, the DM might even consider letting the character pulling from the top add in an Athletics proficiency, because pulling up on something that offers resistance is something that an athlete might do when lifting weights. However we decide to adjudicate the individual numbers, the situation is pretty clear-cut: the stronger contender will get his way.

Then we put a d20 roll into the situation, and things get a little odd. And then, we put a second d20 roll into the situation, and things get really odd. Remember, using a random number provided by a d20 roll is meant to represent the uncertainty inherent in a realistic world. When we were climbing cliffs, we thought of that uncertainty in terms of missed handholds and treacherous tree roots. Now that we’re fighting over doors, what uncertainties are we facing that we need a die roll to represent?

The fact of the matter is that I can’t really think of any, and that’s a problem, because when we’re deliberately making sure to integrate random chance into a situation (and we’re doing it twice, once for each character), we had better have some idea as to what random chance looks like in that situation. If we don’t have some conception about what things are uncertain, and therefore merit the inclusion of an uncertain element in the way the situation is resolved, then we’re essentially creating ungoverned randomness.

Ungoverned randomness is a problem, because it doesn’t represent a particular set of uncertainties, and therefore just represents the wild whims of fate (if it represents anything at all). If we have a troll trying to open up that door in order to eat the gnome wizard trying to keep it closed from the other side, giving each of them a d20 roll doesn’t actually contribute to a meaningful resolution of the conflict. Comparing their strength, endurance, and athleticism is meaningful, and in any such comparison the troll wins and the gnome loses. Throw in some d20 rolls, and you have the possibility of the gnome winning and the troll losing. How could that happen?

Easily. All it takes is a good roll for the gnome, and a bad roll for the troll. Remember, because there are two d20 rolls being made, it’s possible for the difference between the results to be very large indeed. And, because no unpredictable elements were represented by either one of those d20 rolls, they were never meaningful, and meaningless die rolls lead to results that make no sense. Like a gnome overpowering a troll. You can justify this nonsense by talking about how maybe that gnome had a sudden heroic burst of strength, but that’s faulty reasoning and you know it. A 30-pound gnome can’t out-pull a 500-pound troll, and the only reason it can happen in a contested check situation is because the contested check mechanic is broken. It includes randomness without actually considering if randomness is needed and what random events might occur. And when you include randomness that way, you make it possible for just any old thing to happen. You let randomness run wild, because you didn’t bother to give it a purpose.

I have to add at this point that no DM should have ever called for a contested check in the circumstance we just discussed, because there’s clearly no reasonable way for the gnome to succeed. If there’s no chance for success, there should be no check called for and no dice rolling involved. Granted, that’s not part of the rules of the game as written, even if I think it should be. Even so, following the rules as written should not result in bizarre and impossible outcomes, and the rules as written call for a contested check between the gnome and the troll, in which bizarre and impossible outcomes are uncomfortably likely.

Oh, and one more thing about the door: apparently if the contested check is a tie, then we maintain the status quo. In our example, this would mean that because the door was closed to begin with, if the gnome and the troll get the same number for their individual checks, then the door stays closed. Well, that’s great, but what happens next? I suppose they both just shrug their shoulders and walk away from that door, instead of continuing to struggle over it. Maybe they’ll come back sometime later and fight over it again, if it works out for their schedules, and maybe then the numbers will work out so that someone actually wins. Oh well.

Example Two: Lying to a Guard

I’ve already expended a lot of words and a fair bit of sarcasm on the last example, and most of that also applies to this one. However, when the gnome and the troll were fighting over the door, they were pitting strength against strength. This example is different, because we’ll actually end up using different skills against one another. We will, in fact, be using different abilities against one another as well. And, as before, it will make a certain degree of sense, up to a point.

Now we’re going to consider a city guard who has been told not to let anyone through a certain door except for the king’s concubines, and a bard who is going to try to sell the story that he is in fact one of those concubines. This is also supposed to be a contested check, but in this one we won’t be pitting the same abilities and skills against one another. We will actually be using the guard’s Wisdom ability with the Insight skill. This will be contested by the bard’s Charisma ability with the Deception skill.

Now, this mash-up of abilities and skills is defined arbitrarily by the game rules, which is mildly questionable, but it’s something I’m willing to let slide, as long as it makes decent sense, and in this case it does. If the guard is a level-headed and skeptical fellow (high Wisdom) with a keen nose for bullshit (Insight proficiency), then he’ll be a tough guy to trick with clever lies. If the bard is a charming smooth-talker (high Charisma) with a convincing line of patter (Deception proficiency), then he’ll come across as plausible even when he’s playing fast and loose with the truth. When these two gentlemen end up at cross-purposes, either the bard will fool the guard, or else the guard will see right through the bard’s lies.

Of course, maybe there will be a tie, in which case we’ll have to come up with some way to maintain the status quo, which will not be anywhere near as clear-cut as deciding that a door is going to still be closed. I’m not even going to get into how a DM might go about resolving such a tie, because coming up with little patch-jobs for a broken mechanic is pointless, especially when I’ll be proposing a solution fairly soon that will make little patch-jobs irrelevant. Moving on, then.

We run up against the same problem with this contested check as we did with the fight over the door, except there’s an additional level of subtlety, because we’re not just figuring out whether a gnome is stronger than a troll. Insight and deception are less clear-cut, and it seems like using d20 rolls to figure out whether the lies work or not might be a good idea. And, to a certain extent, it is rather a good idea. The issue pivots on whether the guard’s skepticism and the bard’s lies are evenly matched. If they are, there might be a call for the dice to come out. If they aren’t, then the dice will just be meaningless random elements, like the ones in the gnome vs. troll strength contest.

So, if the guard is a simpleton with just enough mental capacity to hold a spear and avoid drooling at the same time, the smooth-talking bard shouldn’t have any trouble deceiving him, and rolling d20’s on both sides is just interjecting randomness for the sake of randomness. It might seem not quite as bad as the previous example, but it really is the same thing. If you can’t account for what the randomness actually represents, then including it doesn’t enhance the realism of the game situation: it destroys that realism, because real things don’t just happen at random. That last sentence pretty much sums up the sentiment that I spent a few paragraphs on in the last example, so I’ll just move on instead of belaboring the point. If you need to look back a bit to refresh yourself on how I feel about ungoverned randomness, go ahead. Otherwise, we go forward.

How to Avoid Contested Checks

This is actually going to be a really simple answer, but it’ll require a few caveats. The really simple solution to the problem of contested checks is just not to use them. Just leave them out of your game. You don’t really need them, and there are reasonable ways to get results that are at least as good. Some of those ways will even get you results that are better. We’ll start at the beginning, with how you can eliminate contested checks from certain situations by just eliminating all checks from those situations. And, for situations that do need checks, we’ll discuss why contested checks are just not the best option. There’s even a well-established precedent for doing it without them and continuing to use the system that works for everything else in the entire game.

Don’t Roll Pointless Dice

The problem with the gnome-and-troll door situation wasn’t really a problem with contested checks. It was a problem with any checks, and it just so happened that the rules called for a contested check. It’s entirely possible for contested checks to get a bad reputation for that reason: there are times when you need a check and times when you don’t. If you call for a check (including a contested check) when you don’t need one, that’s wasteful DMing, and it leads to ridiculous results. You should know that the gnome has no chance to keep that door closed if the troll wants it open, and based on that knowledge you should just tell the players the way it’s going to go down. Sorry, Gnome’s Player, your gnome wizard has absolutely no chance of holding the door closed against that troll. End of story, no dice rolls needed. To roll them would only be inviting foolishness to enter the game.

The situation with the bard and the mentally-absent guard is really the same thing, although it seems more complicated. You just need to be asking yourself whether there’s a meaningful chance for the bard to fail to sell his story. If there isn’t one, don’t roll any dice. It’s the same thing as the gnome and the troll, although the foolishness isn’t as glaringly obvious. Imagining the gnome holding the door closed while the troll tries impotently to tear it open is laughable, but it’s no worse than imagining that guard suddenly sucking his saliva back into his slack jaw and then suddenly seeing through the bard’s fabrications with eyes that have suddenly taken on a steely gaze worthy of Joe Friday.

Meaningful Competition

The only time you should ever even consider rolling a contested check is when there’s uncertainty about who is going to win. That statement is actually true of all checks, but this article is about contested checks, so I’m specifying those today. Don’t fool yourself, though, because the contested checks that you actually do roll aren’t really working the way you think they are, and the way that they actually function doesn’t make any sense even when there’s a genuine question about success and failure.

Let’s go back to the gnome and troll fighting over the door, and tweak the situation. Instead of a gnome wizard trying to hold the door closed, it’s a half-orc barbarian, and now there’s some uncertainty as to who might win the fight over the door. We could probably just compare the troll’s statistics to the barbarian’s and decide who’s stronger, and then award the victory to the one with the higher numbers. That would probably be the most reasonable way to adjudicate the situation, but there’s a major problem with it: it’s really boring. Uncertainty creates excitement, and random numbers from dice create uncertainty, and we want our games to be exciting, so we roll a lot of dice. Makes sense, right?

Actually, it does make sense, but only up to a point. The problem with contested checks is that they have too many dice being rolled, because each contender is getting a d20’s worth of randomness introduced on their side of the equation. In point of fact, we can get the level of uncertainty we need to keep things interesting with just one d20 roll. We know this because it’s the way that the d20 System handles every other situation. Remember the equation?

1d20 + Ability Modifier + Skill Modifier + Circumstance Modifier = X

That’s right. A single d20 roll, a set of modifiers, and a comparison to a DC. It’s the core mechanic of the game, and it works for pretty much everything. And by pretty much everything, I really mean absolutely everything. With some quick thinking and creativity, a DM can use that blue-boxed equation to figure out success or failure for any action a player can think to attempt. What are the circumstances that make contested checks so different that we need to start tinkering around with something that works so comprehensively and so well?

The obvious answer is that contested checks deal with living beings pitting their abilities and skills against one another. We can set a DC for a cliff, because a cliff isn’t alive: it doesn’t have attributes like strength or wisdom, or skills like acrobatics and persuasion. Cliffs don’t have those things, but creatures and characters do have them. That means that when creatures and characters are in competition, we should be taking those abilities and skills into account, right? Absolutely… and we already do.

Combat Is a Type of Competition

Yes. That’s right. We take abilities and skills into account all the time in combat situations, but we don’t use contested checks. We express the abilities and skills as a Difficulty Class. Think about Armor Class, for example. Armor Class is actually no different from any other DC, except we give it a different name because it has to do with armor. Look at how we decide whether an attack hits:

1d20 + Strength Modifier + Melee Weapon Proficiency


1d20 + Dexterity Modifier + Ranged Weapon Proficiency


Advantage or Disadvantage or Neither

Familiar, right? It’s just the same as the basic equation that we’ve been looking back at all along, but we fill in some of the specifics when we know we’re comparing the result to an Armor Class.

It also works this way with saving throws against spells:

1d20 + Saving Ability Modifier + Saving Throw Proficiency


Advantage or Disadvantage or Neither

In this case, we know what to fill in based on what kind of save the spell requires, and whether a proficiency bonus is added to the roll because of a class feature or a stat block item. There’s also the possibility of advantage or disadvantage, either because of a feature of the character or creature, or because of the context of the situation. In this situation, the DC that we’re comparing it with is the Spell Save DC of the character casting the spell.

Armor Class, Spell Save DC, and Everything Else

So now we realize that AC and Spell Save DC are just two very specific types of Difficulty Class, and that they require us to fill in the parts of that equation in a particular way. Remember, we’re not actually changing the equation. The components are exactly the same no matter what kind of DC we’ll be comparing the results with, but when we know that what we’re trying to do is hit an armored enemy with a crossbow, we know which numbers to choose when we fill in the equation.

Let’s take a different look at an attack roll in combat. You have one combatant trying to use a weapon to damage another combatant. The other combatant is trying to either dodge the attack or absorb it with armor in order to avoid being damaged. That’s two living beings competing with one another based upon directly opposing goals, so why aren’t we using a contested check? The short answer is that we don’t need one, because all of the factors that a contested check is supposed to be accounting for are already being managed by the rules of combat.

The combatant trying to deal damage is filling in all of the numbers that go on the left side of that equation, including the d20 roll. The combatant trying to avoid being damaged is contributing as well, by providing the Difficulty Class that will determine the attacker’s success or failure. Furthermore, that DC is based directly on the abilities (like Dexterity for dodging) and skills (like proficiency with armor and shields) of the defending combatant. Put it all together, and we have everything we need to figure out whether the attack succeeds in doing damage or fails to do damage.

What we don’t have is a d20 roll being made by the combatant who is trying to avoid taking damage, and that’s where the good-old standard attack action is different from a contested check. If we were going to use a contested check, then both combatants would make a d20 roll. The fact is that we don’t need two d20 rolls, because one of them is sufficient to account for the uncertainties of combat, and the attacker is responsible for making that roll. The defender’s abilities and skills are being sufficiently represented by Armor Class, which we’ve already established is just a particular type of Difficulty Class, so there’s no need for another d20 roll. Essentially, the element of luck for both combatants is being handled by a single die roll, even though there are two combatants, and there doesn’t seem to be any problem with that. We do it all the time, and it works. All the time.

You can see how the Spell Save DC works almost the same way; the only difference is in which combatant rolls the die. The Spell Save DC is calculated using the abilities (the spellcasting modifier) and skills (the proficiency bonus) of the spellcaster. The d20 roll and any modifiers are provided by whomever is attempting the save against the spell. We don’t need the spellcaster to make a d20 roll as well, because the Spell Save DC is already representing the spellcaster’s magical potency. As with an attack, only a single d20 roll is needed to account for uncertainty when deciding how a spell affects a particular target.

That’s how it works in combat, so why should the system work any differently outside of combat? Attacks succeed or fail based on the comparison of a d20 roll with modifiers to a DC (which we just so happen to call AC). Spell effects are determined by the comparison of a d20 roll with modifiers to a DC (which is the Spell Save DC). There is no reason why fighting over doors and lying to guards needs to be handled any differently than swinging swords and dodging fireballs.

If contention in combat doesn’t require a contested check, nothing else really needs to either. The only reason I can think of that we keep this double standard is that there’s a significant difference between Armor Classes, Spell Save DC’s, and all other Difficulty Classes. The difference doesn’t have anything to do with how they work, though. The only difference is how we get the number in the first place.

Calculated vs. Arbitrary DC’s

Armor Class is a calculated number. You get your character’s Dexterity modifier from your character sheet, and then you look up the section of the Player’s Handbook on armor, which tells you what kind of armor you can wear and how it combines with your Dexterity to get your total AC. Then you work in any class features or spells that might apply, which are also described in the Player’s Handbook, and you get the number that is your AC. You write it down in the space on your character sheet that looks like a little shield, and now you’re ready for battle.

Spell Save DC is also a calculated number. Your character’s class has an ability used to cast spells, which is specified in the Player’s Handbook. Your character also has a Proficiency Bonus based on character level, which is also in the Player’s Handbook. When you add the appropriate ability modifier and your character’s proficiency bonus together, and then add eight (and, no, I don’t know why eight, and it also doesn’t matter), then you have your character’s Spell Save DC. Flip your character sheet over and write it in the little block at the top, and now you can start casting spells and finding out whether they work the way you intend.

All other Difficulty Classes are arbitrary. Yes, all of them. The DC for climbing the cliff, or picking the lock, or disarming the trap without setting it off, or noticing the secret door, are all totally made up. Maybe they were made up by the author of an adventure module. Maybe they were made up by whomever invented the stats for alchemist’s fire and manacles when they were writing Chapter 5 of the Player’s Handbook. Maybe your DM made them up on the fly because nobody else had done it and it suddenly became important because of some wackball thing that your character decided to try without a word of warning. Point is, there’s no specified method for calculating the DC’s of things. Basically you just start with “10 is medium, 15 is hard, 20 is really hard,” and then you wiggle it around from there. I have actually come across some decent ways of calculating DC’s so that you can have traps and locks that get harder to handle as the characters become more powerful, but the fact is that those are arbitrary as well: they’re just a way that someone has invented to wiggle DC’s.

But wait, there’s more, and it will all begin to come clear when you think of a certain magical number. This is really quite literally a magical number, and the magical number is… eight. What’s the significance of eight? As we established in the paragraph before last, eight is the number that you add to your spellcasting ability modifier and your proficiency bonus in order to get your Spell Save DC. I put in parentheses that I have no idea why they chose eight, but I was lying. They chose eight because seven wasn’t quite enough, but nine was a little too much. That’s right, your Spell Save DC was wiggled by the D&D 5th Edition design team, just the same as the DC to jump on the hill giant’s back and pull a sack over his head was wiggled by your DM when you tried that nonsense last week. And, while we’re smashing illusions, your Armor Class was wiggled by whomever was in charge of filling out the armor chart in the PHB, because someone had to decide that carrying a shield was worth +2 instead of +1 or +3.

When I said that all DC’s are arbitrary except for AC and Spell Save DC, that wasn’t really true. What is true is that all DC’s are made up. All of them, even the ones that the PHB tells you how to calculate. But they aren’t truly arbitrary, because they are based on a system that has been designed and tested to determine success and failure in a way that is fair and realistic. That makes them discretionary, which is different from arbitrary, but still means made-up. When your DM sets a DC of 15 for balancing your way along the handrail of that bridge, it’s a judgement call, but it’s not arbitrary. Arbitrary would be setting the DC at 15 because it happens to be the 15th day of the month, and thankfully you won’t find a lot of DM’s, or adventure module authors, or game designers, who make decisions that way.

So, Yeah, Contested Checks…

The point of all of this, other than to be fascinating and insightful, is to convey the message that contested checks (because this is an article about contested checks, after all) are broken because they include too much randomness, or the wrong kind of randomness, or both. And, because I truly do try to offer at least some kind of solution to problems that I spend thousands of words identifying, I’ll finish up with how to get rid of contested checks altogether.

Two d20 rolls for a single attempt at any action is too many. It doesn’t matter what your character is trying to accomplish, and it doesn’t matter who or what the opposition is. We need uncertainty to keep the game exciting, but a single d20 roll is sufficient to provide as much uncertainty as we’re going to need. If you don’t believe me, just look at the combat system. Combat is full of living beings trying to hurt each other and avoid being hurt, and they can do it all without any contested checks.

When You Might Actually Use Two Rolls on a Single Action

There is actually a type of situation where you might want to use two d20 rolls for a single action, but in this type of situation both of the rolls are being made by the same player for the same character. Occasionally, I’ll require two checks when there are two skills that are required to achieve a certain outcome.

For example, the wizard might need to make both an Arcana check and a Perception check in order to get information about a particular mysterious rune on a doorframe. The Perception check is there to decide whether the wizard notices the rune, and the Arcana check is there to decide whether the wizard actually realizes that she’s noticed something important. Knowing about the meaning of the rune doesn’t guarantee noticing it, and seeing the rune without realizing it’s significant doesn’t do any good either.

This is not something to overuse, but it is a situation in which two d20 rolls can be appropriate in relation to what would otherwise be a normal check. The salient feature is that both rolls are being made by the same player for the same character, so it’s not a contested check, but just one check with multiple interdependent parts.

If you’re rolling a d20 to add uncertainty, you need to be sure of what uncertain things are part of the situation. Going back to the cliff climb, I can give a long list of things that could go unexpectedly wrong (or unexpectedly well) when someone is climbing a cliff, and that means that I can include that d20 roll with confidence and purpose. If the only reason you’re rolling dice is to evaluate “luck”, then maybe you should make sure that you’ve considered what “lucky” and “unlucky” look like. They don’t have to be complicated answers, but be sure that you aren’t interjecting gratuitous randomness into situations that would be fine without any.

One more thing to get out of the way is how to handle ties, because even if you don’t use contested checks it’s possible for there to be a tie between the DC and the d20 roll with modifiers. For all you DM’s out there, here’s how to deal with that: just don’t let it happen. Someone is going to win the fight over the door. If you’re thinking about that situation where there’s a tie, and how something different should happen to account for that special circumstance, just get over it. It’s a ridiculous outcome, and allowing ridiculous and unworkable outcomes is something that DM’s shouldn’t do, because they don’t have to. Be the boss and don’t allow any ties. There will be a winner and a loser, and if you declare a tie then you are letting your dice kick you around. I personally resolve all ties in the favor of the players, because it’s a simple way to deal with it that doesn’t take up any extra brain-space when I’m running a complicated game already. Go ahead and figure out how to resolve ties any way you want, but the upshot of it all ought to be that there are no ties in D&D. They are almost always more trouble than they are worth.

Fixing Contested Checks

This is an easy one, DM’s. Every time the rules call for a contested check, don’t use one. Instead, decide which of the contenders will be rolling a d20 and adding modifiers, and which contender’s powers and advantages will be represented by a DC. Most of the time the players will be the ones rolling, and that’s good because players like rolling dice. Now all you have to do is figure out the DC that they’re rolling against.

All of the tools to figure out the DC are already there, and a really good place to start is using passive checks. If you need a starting point for the DC, add all of the relevant modifiers to 10, and see what you get. But don’t stop there, because now it’s time to wiggle. Maybe the guards in this town use a particular stat block that specifies that they have +1 to Wisdom, but that doesn’t need to apply to this particular guard. Maybe he’s a little smarter than the average guard, and that’s why he’s on this particular important assignment. Maybe he’s a really dim guard, and they put him on this door because they had to give him somewhere to stand that didn’t matter much. You can even give proficiency to anything with a stat block on the fly; just pretend that the CR is actually a character level and figure out the bonus that way. And there’s always advantage and disadvantage, so maybe that male bard is going to have a harder time convincing the guard that he’s a concubine because every other concubine is actually female.

The point is that you can calculate a preliminary DC for anything, and you can adjust those DC’s to fit the situation. Don’t try to change how you calculate AC or Spell Save DC’s, because you’ll just get into trouble. But, for everything else, all that needs to happen is for a reasonable DC to be set, and it doesn’t really matter how it gets done. Maybe you just use Passive Whatever-Applies-Best. Maybe you decide that 10 is medium and 15 is hard and that convincing this particular guard should be about a 12, and throw all of the stats and calculations out the nearest window. No matter how you do it, it’ll be better than a contested check, because you’ll be consistently using the system that makes the entire game work, instead of creating special cases and exceptions that only make things more complicated and less reasonable.

Ending by Waxing Philosophical

We use dice in D&D because we want to create unpredictable outcomes. Straight-up comparison of character sheets and stat blocks is boring, and running a game that way will just create a series of foregone conclusions. If you’re not planning on having a level of chance, then everyone can just sit at home alone and compare numbers, because without dice we might as well just save the time and travel, because you don’t need anyone else’s input if you’re just going to compare stat blocks. And you won’t actually have any fun, for the same reason that reading the box scores in the newspaper is not entertaining, but watching the baseball game is. Also, dice are pretty colors and shapes and make exciting sounds when you roll them, and aesthetics are never truly irrelevant.

The thing to realize about rolling dice is that the random numbers that come up don’t always represent the same type of predictability. When we were climbing that cliff a few thousand words ago, the d20 roll stood for all the little things that can go wrong with attempting a complex and dangerous activity. When we were pitting a gnome against a troll in a test of strength, the d20 roll stood for meaningless uncertainty grafted shamelessly onto a situation which had no room for any uncertainty. When we were deciding whether the fast-talking bard would sell his unlikely story to the canny city guard, the d20 roll stood for creating a surprise as to which of two worthy adversaries would triumph.

So, go forth and use contested checks no more. You don’t need them in order to make the game work, and they actually make the game work less well than if you didn’t use them in the first place. We already have a versatile and robust system for figuring out success and failure, and we should just use it consistently. Making little exceptions for certain situations in which we don’t use the same rules is foolishness. And, in the case of contested checks, the situations that supposedly need special accommodations are actually no different than every other situation. Competition between living creatures with relevant attributes and skills happens all the time in combat. You don’t need contested checks in combat, and you don’t need them anywhere else, either.

What DM’s do need is a way to come up with Difficulty Classes for all sorts of situations, and fortunately this is not difficult. If you’re dead set on factoring in the competitors’ stats and numbers, just pretend you’re making one side’s contribution into a passive check and work from there. If you’d rather just set a DC based on that wobbly scale of easy to difficult, choose something reasonable. Whatever method you choose (and you can mix it up, of course), just realize that you have to have a DC on one side of that equation. If you can do that, then you’re using the core mechanic of the game just as it’s intended to be used. If you have a DM screen, nobody even needs to know how you’re doing it on any particular occasion. I even roll some dice behind the screen sometimes just to make the “sound of uncertainty” (if any of my players are reading this, I guess that jig is up, although you still won’t know when my dice sounds are genuine uncertainty or not, so I win anyway).

As for the rest of it, I hope that there are at least a few of you out there who got some benefit from the analysis of how the d20 System works. That wasn’t really the point of the article, but it turned out to be a useful part of it. Theory isn’t everything, but you have to understand how something works before you can start tearing pieces out and replacing them with whatever you think will work better.

So, class dismissed. And now, I think I might actually be able to figure out that Passive Stealth problem… so stay tuned.

3 Responses

  1. erichthegreen says:

    I believe I know why the number 8 is added to ability modifier and proficiency modifier to get spell save DC. It is the same reason that base AC before adding ability modifier and armor bonus is 10, and why Passive Perception (a form of DC) is determined by adding ability modifier and skill (if any) to 10. In all cases, the real base number is 10, which is also the average or in-lieu-of-rolling value of a d20 roll. The reason spell save DC uses an 8 instead of a 10 is because, at first level, the proficiency bonus is +2, which means the 1st-level DC (before ability adjustments) starts at 10. Basically, 2 is subtracted from 10 to zero-out the initial proficiency, but still allows the DC to rise as proficiency rises with level.

    For some reason, probably to slant play more toward taking action and succeeding, attack bonus and skill checks start by adding the full skill and ability modifiers, rather than zeroing-out the initial +2 proficiency bonus at level 1.

    As a result of these various starting bases, all the initial values are 10 (unless you have a negative ability modifier).

    – AC is 10 if you have no armor and no DEX bonus.
    – Passive Perception is 10 if you don’t have the Perception skill and no WIS bonus.
    – Spell Save DC is 10 at level 1 if you don’t have an ability bonus. (It is assumed that you don’t have a spell save if you aren’t proficient with casting spells, so your initial proficiency bonus will always be 2; adding this to 8 thus creates an initial value of 10 plus ability modifier)

    The magic number 10 stands in for one of the two rolls of a d20 in a contested situatoin. So with attack vs. defense (represented by AC), the attacker rolls a d20 while the defender is assumed to have rolled a 10. With a spell attack, the victim rolls a d20 saving throw (or in some cases, ability check) vs the spell save DC, which is calculated on the assumption that the caster rolled a 10. (Or an 8, depending on how you want to see it).

    The only time BOTH characters roll is in a contested check, but to incorporate this idea into your revision, just assume that the more passive character has rolled a 10, and use that to create the DC. So for your lie-to-the-guard example, the DC is 10 plus the guard’s skill and ability modifiers, if any (thus accounting for whether he’s a smarter or dimmer than average guard). One could, of course, roll a die to determine if the guard is smarter or dimmer than average, but a d20 would be wildly too random for that.

    Hence I like your suggestion for fixing contested checks except instead of determining a DC arbitrarily or situationally, you should act like that character/monster has rolled a 10 on their d20 and then add skill, ability, and situational modifiers as usual.

    (Yes, I know that the actual average roll of a d20 is 10.5, or a 10 or 11 result, but trying to account for that in this context would be unnecessarily complicating. Although one could use that as justification for having to BEAT, rather than just TIE, the DC of a contested check, resulting in a DC of, say, 12.5 which you beat on a 13 but lose to on a 12. With DCs values always .5, there could be no such thing as a tie roll. So use my suggestion but always assume the character rolled a 10.5 on the d20.)

  2. erichthegreen says:

    “When you add the appropriate ability modifier and your character’s proficiency bonus together, and then add eight (and, no, I don’t know why eight, and it also doesn’t matter), then you have your character’s Spell Save DC.”

    Well, I believe I actually do know why the number is eight, and why it (sort of) matters, and how it suggests the solution to the contested check problem.

    “Under the hood”, as it were, 5e is based on a d20 system but with an underlying assumption of contested checks, the kind you are (with good reason) trying to write out of the game. Of course for the reasons you elaborate, having everything determined by a contested check would mean a LOT more rolling and a LOT less predictability or, conversely, a lot more wild or counter-intuitive results. So instead, in most cases, the 2 actors contesting the check (characters, NPCs, or monsters) are assigned to “active” and “passive” roles (somewhat arbitrarily, as we will see), and the one in the “passive” role, instead of rolling a d20, is assumed to have rolled a 10. It is notable that this mechanic, assuming a d20 roll of 10 instead of rolling, explicitly appears in some class features.

    This is also the case for passive checks, such as Passive Perception, which is calculated as 10 plus any proficiency or ability (WIS) modifiers. Armor class also functions like this. In combat, each combatant is “active” – one tries to hit the other, while the other tries to avoid being hit. However, the defender is considered in 5e to be more “passive”, so their roll of a contested d20 is assumed to be 10. That is why basic AC starts at 10 plus bonuses for armor/shield, ability (usually DEX), and feature modifiers. (In contrast, the attacker rolls a d20 and adds their proficiency, ability, and feature modifiers). If combat were handled as a contested roll, AC would start at 0 plus (minus) modifiers, and then the attacker and defender would EACH roll a d20 and compare results, with the attack hitting if the total modified roll of the attacker was higher. Instead, an assumed d20 roll of 10 is modified to see what the attackers’ modified d20 roll must meet.

    In the case of a saving throw against a spell, the actor trying to save is considered “active” (hence rolling a d20), while the spell effect is “passive”. Except instead of using a deemed roll of 10 plus proficiency & ability (& situational) modifiers, the number is 8, as you noted. I believe this is because a spellcaster essentially MUST be proficient with casting spells in order to generate the saved-against effect. (In contrast, one can attack with a weapon one is NOT proficient with, or roll against a skill in which one does not have proficiency. An attack or skill roll can be made with no proficiency, but a spell cannot). Therefore, there is always a minimum +2 proficiency bonus in the creation of any spell effect. I think using 8 instead of 10 is a way to zero-out this minimum +2 proficiency. A caster with a higher proficiency based on level advancement or higher high CR will add their additional proficiency bonus (beyond the basic +2), raising the basic spell save above the starting value of 10 (before ability or other modifiers).

    Your suggested way for getting around contested checks can be simplified by using a “deemed 10” for the more passive of the two parties, then modifying that by proficiency, ability, skill, features, situation, etc.

    I would note that non-actor DCs are set using similar thinking. By this I mean checks against a passive “opponent”, such as to climb a cliff or wall or rope, cross a slippery bridge, jump over a bonfire, force open a stuck door, etc. No trivial task should require a DC, hence an “easy” task (DC 10, an assumed d20 roll of 10) is one that you should be able to accomplish at least half the time but have a reasonable chance of failure. A very easy task to beat assumes a d20 roll of 5, a medium or hard task a roll of 15 or 20, etc. “Nearly impossible” gets a deemed d20 roll of 30, which is nearly impossible to beat by rolling a d20.

    As another note, using an assume 10 (rather than the real d20 average of10.5), and allowing the “active” party to win ties, shows a bias toward the active party in each contested check. Which makes perfect sense for a game based on high fantasy and cinematic storytelling – it rewards taking action and taking risks.

    • Thanks for the insight on the reason why 8 is used. I’ll tag on here that I have started using something in my games since writing this article, which I prefer to call “roll for mishap”, but which is often actually called “roll for stupid”. It basically turns the DC system upside down a bit, and the players roll against a low DC (generally about 5) to decide whether just bad luck will result in failing a task. It works fairly well for things like skirting a pit along a narrow ledge: it should be fine, as long as no one does anything stupid.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.