How Difficulty Class and the D20 engine ruined roleplaying

It seemed revolutionary at the time. 3e came out and made a unified mechanic. Roll 1d20 against a target number to see if you succeed. In combat, AC is the target number. For everything else, it's a Difficulty Class [DC]. That's it. 

Some of those DCs are calculated as a function of level, opposing ability score etc. But what we also got from this system was a way of ad hoc determining the difficulty of something and then simply saying "roll against that target number to succeed". 

In its core form, this is wonderfully simple and intuitive. All you need to internalise is the size of the numbers on a d20 in relation to overall difficulty and then you can resolve basically anything with it.
The part about size of numbers has proven to be a bit of an achilles heel for d20 over the years, but that is a different point I will address further below.

No, the real point here is that there's an unintended side effect to DCs as a unified mechanic. There are other downsides to it as well that I will go into below too, but this first one is the big one. 

It turns everything in the whole gaming world into a mechanical widget. At first glance, one might think "marvellous, innit? It resolves everything." A quantum gravity theory of RPG in one simple equation. But, subtle as it may be, as it turns out this is actually something horribly insidious that tilts the game in a direction from being a roleplaying game to being a board game without the board. 

How Rolling Against DC ruined Roleplaying

Let me try to illustrate how DC turns the game from this

A world of fantasy and wonder restrained only by the imagination

into this:

Scanning for Difficulty Class

In short - the d20 mechanic enables you to resolve everything. If everything you encounter becomes something you can interact with mechanically and assign a DC to, a widget, then you are no longer actually roleplaying in a fictious world. You are just interacting with the mechanics of a game with a thin veneer of fiction layered on top. 

Now, this fiction may be fun and appreciated by the players who may very well get their roleplaying needs satisfied by adding this on top. But it is also incidental to the game. You don't actually need it to interact with the game, even if it is always there to validate the mechanical interactions.

BadWrongFun

This is a problem when it comes to keeping our elfgames free of BadWrongFun. 

Now, the D20 player may argue that this is only true when taken to its logical extreme, most tables don't go to that far off to the end of the scale. And this is perhaps true. Let me argue for why they still go a lot further than they need or ought to:

Here is a radical proposition:

For the vast majority of cases, 

it is unnecessary to determine the difficulty of an attempt in order to resolve it.

One of the issues with DC is that it applies to stuff that it should not apply to. I submit as evidence a table from the 3.5 DMG.

If it's not obvious, let me say - this is terrible. Ridiculous even. Let's ignore the fact that according to this table a DEX10 commoner will fail 45% of the time when attempting to tie a firm knot. There should be no roll for it in the first place. 

I am using a silly (albeit RAW) example to exaggerate make my point, but the point holds true - When you have a mechanic that turns everything into a button and places that button right in front of you, the tendency becomes to push that button. 

Tying a knot is a trivial example, but where this becomes a real issue is when pushing the button placed in front of you starts replacing what should be the resolution approach for most things - Roleplaying. 

Let's imagine an example and game it out according to the old school and the D20 school. The party enters a tavern and the fighter wishes to make a good impression on the barkeep whose starting attitude as they enter is indifferent.

Old School: "I flash the barkeep my best smile, order a cup of ale and pay with a handsome tip and try to get him talking about the local rumours in a chatty friendly manner."
DM considers the scene and factors in the fighter's 14 charisma and decides that a good impression is made.

Note that this example is hardly expert roleplaying. It's just a short descriptive interaction. The player is not being asked to do 1st person improvised NLP to influence the outcome with roleplaying.

D20: "I flash the barkeep my best smile, order a cup of ale and pay with a handsome tip and try to get him talking about the local rumours in a chatty friendly manner. Actually a Persuasion roll. I roll 12, +2 from Charisma and +2 from Proficiency, so 16."
The DM gives another +2 for the handsome tip and decides 18 is good enough to make a good impression.

In the first draft of this article, I had the d20 player just roll at the start. I changed that on reflection because plenty of tables do after all do the roleplaying bit and I don't want to caricature too much here if I don't need to. And including the roleplaying bit actually illustrates the point even better - In the D20 example, the same dialogue happens but the roleplaying doesn't actually affect the outcome. 

The player end up interacting with a world of game mechanics instead of interacting with a world of fiction.

Now, some tables might demand that the player attempts to roleplay the dialogue before rolling, which is one way of keeping the roleplay alive in a world full of DC buttons to roll against, but this effectively turns it into a gatekeeping activity ("roleplay to be allowed to roll"). Others may argue they give bonuses or minuses dependent on the roleplaying (as seen above where the tip gave +2).
Both of which is still missing the fundamental point made above: 

Roleplaying is the basic resolution "mechanic" of roleplaying games. Dice rolls are there to to resolve outcomes not readily resolved through roleplaying.  

We don't actually need a game engine to determine the difficulty of most outcomes in the world. The needs are much more narrow than that. The game engine should fit those needs and get out of the way when it isn't needed. 

The D20 unified mechanic does the opposite. It's designed to resolve everything and as such inadvertently invites resolution opportunities for resolving everything (as demonstrated by the ridiculous table above). 

The net result is encouraging a push-button mentality where the players more and more begin to inhabit a world of game-mechanics with roleplaying on top, rather than a roleplaying world where mechanics briefly enter and leave in certain windows to resolve certain outcomes in the fiction. 

This is, in a nutshell, how D20 ruined roleplaying. 

Later, it would further compound this ruination by creating all kinds of mechanical widgets that are clearly designed to interact with its own world of mechanics rather than the world of fiction, but this is a topic for a follow-up rant.

Back to the topic of task resolution and how D20 seems really elegant but actually sucks massive balls. 

How to resolve tasks without ruining roleplaying

Here is another proposition:

For the cases where one does need to determine difficulty,

a resolution factor of 3-4 degrees will suffice.

Another issue with the D20 unified mechanic is the d20 itself. With a granularity ranging from 1 to 20, it doesn't really lend itself that well to gauging the numbers in relation to difficulty. Let's look at a table from the 5e DMG:

This is obviously much simplified from the 3.5 Dmg and for good reason. What is a 13 really and how much harder is it than 10 and how much easier is it than 15? Let's just round it. This also makes it tricky to work with modifiers in an intuitive way. How much weight does a +1 really have, or +3, in a system that habitually rounds to the nearest 5 anyway just because the smaller numbers don't say much for the roll anyway.

Below this table, we're even told "If the only DCs you ever use are 10, 15, and 20, your game will run just fine". 

Now, I actually very much agree with the KISS intent of this paragraph. My issue with it is that it demonstrates why d20 is a bad resolution for this kind of mechanic in the first place. If the game runs just fine with 3 increments of "Easy, Moderate and Hard", why are we rolling a die with a granularity of 20 increments?

The answer seems to be because the derived stats from ability scores and proficiency are scaled to a d20. Well, just derive some new ones from those numbers then. No, because we want a unified mechanic. Well, here's another argument for not having a unified mechanic then. It's clearly not the best tool for the job.

With all these things in mind, what's the better alternative then? Here's my radical proposition:

A "roll under" ability check + assorted other mechanics

Yes, the old school ability check.

"But Anders", the frustrated reader remonstrates, "That is still a d20 roll!"

Yes, but the d20 die here doesn't have the flaws it does in the D20 unified mechanic, since we are not working with increments on the die to establish difficulty. There is a fundamental difference in approach however, that gives us a baseline for a resolution mechanic that we actually want to have: 

• With the D20 unified mechanic, you roll against the situation and apply ability modifiers
• With the "Roll Under" Ability check, you roll against your own ability and apply situational modifiers

This avoids the inadvertent effect of DCs creating a world of mechanical widgets to interact with, because you aren't rolling against those. On the contrary, you are rolling against the one constant mechanic component always present in the game - Your own character sheet. 

This simply preserves the a priori state of affairs true of any roleplaying game from character creation - The character has stats already. The world doesn't. 

"Situational modifiers" can be resolved very simply, with three simple categories (much like the 5e DMG suggests the D20 mechanic should be pidgeon-holed into):

• Challenging
• Challenging but with favourable circumstances
• Very Challenging

This baseline is different to what the 5e DMG suggests, since I hold that anything less than "challenging" shouldn't need rolls - This is where roleplaying should be able to determine the outcomes just fine (if it doesn't, let's treat it as the player making it challenging for himself with his roleplaying and end up rolling). And it is very easy to resolve mechanically, thanks to the most elegant of 5e mechanics: 

• Challenging - a standard roll under ability check
• Challenging but with favourable circumstances - same, but with advantage on the roll.
• Very Challenging - same, but with disadvantage on the roll.

This I find to have a satisfactory level of granularity for just about any scenario. It avoids having to assign mechanic modifiers to each component influencing the scenario and assess it in a way very similar to how you would in the fiction: "Given all this, is this challenging or very challenging?"

One may consider adding a "nearly impossible" category for those rare occasions, to which I would give a -4 modifier on top of disadvantage.

Probabilitywise, this makes each step more or less an increment of 25% (incidentally, again, what the 5e DMG would like to pidgeonhole the D20 DCs into).

For more on the elegance of thebability check and why it actually charts really well with B/X' near-universal 2in6 roll and old-school ability score modifiers, see Using Ability Checks in B/X.

Of course, this creates problems for systems that employ Ability score increases, as D&D 3e+ does. For those, I would use some form of Action Throw mechanic, somewhat adjusted for the fact that ability scores will go higher than in TSR D&D. Or for 5e, maybe 8+proficiency+ability score modifier. Someone else will need to do the math on that.

Riding check DC 25 to appease a horse clearly fed up with it all

Assorted Mechanics

There are other issues with the D20 mechanic as a universal task resolution. This article is getting long, so I am just going to list them here and then briefly go through each before explaining why "assorted other mechanics" is the better choice over a unified mechanic.

• Gradated vs binary outcomes
• Linear vs bell curve distribution of outcomes
• Turning situations that aren't really contests of skill into contests of skill

Gradated vs Binary outcomes

Another thing the d20 mechanic handles poorly is determining gradated outcomes. Rolling against a target number tends to encourage binary outcomes, best reflected of course by attacks and saves. And d20 rulesets don't seem to do much to discourage this mentality. In a unified mechanic, why should ability rolls behave differently. To its credit, the 5e DMG does discuss the option of Success at a Cost (the 3.5 DMG simply discusses degrees of failure). But it is all in all a tertiary aspect and one the mechanic does not suggest towards.

Linear vs bell curve distribution of outcomes

There are also times where you don't actually want a linear distribution of outcomes. Modifiers, as well as the advantage mechanic, can tilt that towards high or low results, but how do you resolve something that leans towards the middle and makes very high or very low results a lot less likely?

The OSR has come up with some excellent solutions for both issues. Adapting the reaction table to an all-purpose resolution track.

Alternatively, I would consider using a d8 with 1 being terrible, 2-3 poor, 4-5 qualified success, 6-7 unqualified success and 8 fantastic. 5 outcomes distributed over 8 increments are close enough to each other that any modifier applied should be easy to grog and we have a normal distribution of the three middle outcomes that make up 75% of outcomes and the two outliers taking up 25% of outcomes.

Turning situations that aren't contests of skill into contests of skill

Another issue with d20 is that it tends to turn events that aren't actually contests of skill into contests of skill. The most notorious example is social ones that somehow turn into "social combat" where the player is trying to "overcome" his targets will or some such, rather than swaying him. I've written before on why this is a bad idea and how such situations should be resolved.

A note in all this concerning unified vs assorted mechanics: As far as the player is concerned, they are not exposed to all these assorted other mechanics. With the suggestions given (and there are many other ways of doing this), their complexity level has only changed from "d20 unified mechanic" to "d20 roll high for combat/saves and d20 roll under for ability checks." Overall, I'd call that simpler, since they don't need to fidget with modifiers on those ability checks anymore.

If you really want that player-facing unified mechanic, you could always use Target20 instead of roll-under for ability checks, but I wouldn't like to, since it invites the question "why not target25 for this situation?" etc. and I'd really want the mechanic to emphasise that you are simply rolling against your own ability score and the situation only determines how you roll against it. 

Read Magic would reveal invisible glyphs on the archway spelling "Spot Check DC 15"

Conclusion

Over the course of this article, I've tried to show 

• why the D20 mechanic ruins roleplaying by tilting play towards interacting with mechanics over fiction
• why it is a poor fit for task resolution in general due to the increments of the d20
• why using it as a universal mechanic is a poor fit for a lot of situations where you want non-binary or non-linear outcomes
• and how it distorts situations that aren't really contests of skill into being so, in the name of unified mechanics.

And I think any sane person who has read to the end would now agree that the d20 mechanic should die in a fire. It was an interesting experiment. Maybe we are all better off for having tried it. But we are not better off for persisting with it.