Forget 'yes and'; 'Yes if' is for DMs

Alexa the Artificer needed a way to contain the immortal shoggoth her party just trapped, but she had just the thing: "I build a rocketship and send the shoggoth to the moon."

"Hang on now," the DM says. "Rocketships don't exist in this world."

"This sucks," Alexa says. "You're not 'yes-and-ing' me. You really need to take some improv lessons if this campaign is ever going to go anywhere."

"Yes and"?

As D&D has expanded from its wargaming roots and picked up more improv storytelling over the years, the question comes up more often: As a DM, is it better to "say yes" to everything, or is it better to tell players that something's impossible when it doesn't fit in your world?

The core mechanic

A balance between the two extremes pivots on the "core mechanic." A core mechanic is gameplay at its most stripped-down, something that forms the foundation of the game and you end up doing over and over. D&D's core mechanic is often described as "roll a d20, add modifiers and compare to a target number to determine success."

I'd like to broaden that, cribbing a bit from the 5th Edition's introduction. So here's D&D's core mechanic, DungeonFuntimes edition:

1. The DM describes a situation.
2. The players describe what they want to do.
3. The DM decides how to resolve that action, citing rules or setting the cost of success if necessary.
4. The players attempt the action.
5. The DM narrates the results of the adventurers' actions.

This should look pretty familiar. But I reiterate it here because I find that steps 2 and 3 are the key to balancing "yes-and-ing" with the limits needed to respect the story you're telling. The players say what they want to do. The DM tells them what it takes to accomplish that.

Yes, if

So it's not the DM's job to say "yes, and." If you were taking part in a fantasy-themed freeform roleplaying exercise, then it would be your job to say "yes, and." But you're playing a roleplaying game, and the rules of that game dictate that it's your job to say "yes, if."

Yes, your character hits that bandit, if your attack roll beats their armor class. Yes, your character persuades the queen to declare tomorrow Wear A Turkey-Shaped Hat Day if you beat this rather high secret difficulty class on a Persuade roll.

But how far does "yes, if" extend? The 5e introduction applies this core mechanic to two cases: 1. It's an easy task that characters can accomplish without rolling the dice. 2. It's a task that includes a risk of meaningful failure, so players need to roll a d20 and add the appropriate modifiers and compare it to a target number (difficulty class, or DC).

In debates about whether a natural 20 should be considered an automatic success, you often see the argument: "Well, if a player wanted to jump to the moon and they rolled a nat 20, would you let them?"
But this highlights a third case in resolving player actions: 3. Actions for which a roll is not required but which are covered by specific rules.
My answer to "would you let a player jump to the moon on a nat 20" is always, "No, because your jump height is determined by your Strength score, not by a roll. Unless there's some complicating factor, you don't need to roll to jump."
Your running high jump is 3+ your Strength modifier in feet. So unless your character is Bruce Lee or has a Strength of 5.2 billion, then no. But I wouldn't ask for a roll to jump, anyway.

But there's a fourth case (see the side note for the third case), and that's what we're interested in: 4. It's an extremely complicated task that can't be resolved in a single action. Instead of telling the player flat-out that something's impossible, the DM can instead turn steps 2 and 3 into a conversation: The player says what they'd like to do, and the DM determines what it would take to accomplish that — but instead of a roll, success might require a longer quest with several objectives. The goal is to let the players do what they want to do, but make the party earn it. If the price for success that the DM sets is too high for the player, both the DM and the player can brainstorm alternatives.

It's not rocket science, it's rocketmancy

Back to the opening example: Alexa says she wants to build a rocketship. Before the DM gets too focused on the rocketship, there's one question that should never be too far from a DM's lips: "What's your ideal outcome?"

Then you have a conversation. If Alexa says "I want to put the shoggoth on the moon," the DM can ask whether the rocketship's a dealbreaker.

If Alexa isn't too hung up on the rocketship, the DM could suggest a consultation with a powerful wizard who's an expert in teleportation magic. Or perhaps there's a Temple of the Moon fabled to open a gateway when the moon's in perfect alignment.

What's your ideal outcome?

Sometimes players say they want to use a certain spell or mechanic, but they're confused about what that mechanic does. For example, a player might say something like, "I want to Thaumaturgy that guard."
Instead of getting too focused on the fact that Thaumaturgy isn't targeted on a creature, you save the lecture and just ask: "What's your ideal outcome?"
Then the player can respond with, "I want to put the guard to sleep," in which case you can refer them to the Sleep spell, or they'll say, "I want to set the guard on fire," and you can recommend a fire-based attack spell, like Fire Bolt or Burning Hands. Or they'll say, "I want to make an ominous whispering sound from the other side of the hall so I can sneak past." Bingo! A by-the-book use for Thaumaturgy that might grant the player advantage on the subsequent Stealth check or even just make the Stealth check possible in the first place. Clarifying questions FTW.

If the rocketship is a dealbreaker, the DM might set the major requirements for success as such:

1. Goal: Understand the basic principles of rocketry. Cost: Decades of research.
2. Goal: Invent rocket fuel. Cost: Decades of research, testing and a source for raw materials.
3. Goal: Invent a rocketship. Cost: Decades of research, testing, and hundreds of thousands of gold pieces to pay for materials.
4. Goal: Understand the mathematics behind orbital mechanics. Cost: Decades of research.

So yes, Alexa can build a rocket, but the shoggoth is probably going to get loose before then. If Alexa can come up with any feasible shortcuts to any of those steps, the DM might allow her to knock off a few decades or even build a rocketship within the timeframe of the campaign.

The key is to identify and address the impossibilities involved. Remember: Your goal isn't realism, it's suspension of disbelief, which is a much lower bar. So if Alexa can have dwarves forge a mithril rocketship hull and a team of magic users casting a torrent of firebolts through a portal that leads to the rocket's combustion chamber and an enchanted space compass that's connected to the controls and keeps the rocket pointed the right way, I'd say let her do it! That sounds fun!

When you need to say no

One last note: I believe there are cases in which you need to exercise DM fiat and flat-out say, "No. Your character does not do that." Those cases should be limited to character actions that violate your group's code of conduct or actions that make other players very uncomfortable. Stuff like sexual assault and child murder. But again, these DM fiats work best if the party has worked out and perhaps voted on a code of conduct before problems arise.