Go to Part 1

We’re nearing the end of a campaign, having traced a gaggle of strange incidents in which historical events (or at least replicas of historical events) have erupted into the modern world back to an eery city on the border of the Dreamlands. As we explore the city, we discover that it seems to be somewhere between a palimpsest and a jigsaw puzzle, formed from jagged pieces of different cities around the world and drawn from different eras in history (not all of them apparently our history). The whole place is completely deserted, however, and a strange white mist drifts through the streets.

While we’re checking out the apartment that once belonged to one of the PCs, there’s a car crash outside. Rushing out into the street, we see a girl with stark white hair racing away from the accident. We recognize her: Although she had black hair last time we saw her, she was being kidnapped by some of the strange wraith-cultists who seem to be mixed up in (or maybe causing?) all of this weird stuff.

We give chase and she leads us to the British National Museum (or a copy of the British National Museum?), but then she runs into the room with the Parthenon Marbles and vanishes. Our archaeologist notes that the marble sculptures have been altered and appear to depict a map of the city. We take a rubbing and begin using the map to navigate, visiting a number of strange locations where we experience enigmatic things.

Then, abruptly, a bright white light suffuses everything.

And the world ends.

Huh.

In the post mortem, we discovered what happened: After the car crash, we were supposed to check the trunk of the car. If we’d done that, we would have found the girl — still with black hair — tied up in the back. She would have been able to lead us back to the Home Insurance Building (the world’s first skyscraper) and then… something something something. I don’t remember the details. The cities of the world had all been linked together in a ritual using key skyscrapers and the Girl With White Hair was the black-haired prodigy’s mirror-self from an anti-life dimension.

We didn’t check the trunk, though, and so the world ended.

“It was really exciting to run a sandbox!” the GM said.

THE RAILROADER’S FALLACY

The railroader’s fallacy is surprisingly common:

I ran a sandbox, but the players didn’t follow the one plot that was available!

This often results in the railroader saying things like, “Sandboxes don’t work.”

First, let’s understand the nature of the fallacy here.

A sandbox campaign is one in which the players can either choose or define what the next scenario is going to be. In other words, the experience of a sandbox is more or less defined by a multitude of scenarios. So as soon as you see someone use “sandbox” to describe a campaign in which there was only one scenario — or, even more absurdly, only one plot — it’s immediately obvious that something has gone horribly wrong.

So how does this happen? And why does it seem to happen so often?

Well, we need to start with the railroader. Checking out The Railroading Manifesto might be useful if you’re not familiar with it, but the short is that:

Railroads happen when the GM negates a player’s choice in order to enforce a preconceived outcome.

Railroading can happen for a lot of reasons, but a common one is that the railroader lacks the tools to build RPG scenarios and therefore defaults to the linear plots they see in videos, movies, books, graphic novels, and so forth. This linear sequence of predetermined outcomes is antithetical to the interactivity of an RPG, and so the GM has no choice but to railroad their players into the predetermined outcomes.

At some point, the railroader gets the message that Railroads Are Bad™. The ideal outcome would be that they learn some scenario structures and gain the tools they need to run dynamic, awesome scenarios. Unfortunately, this often doesn’t happen.

One common response is rejection of the premise: “I railroad. Railroading is bad. I don’t want to be bad. Therefore railroading isn’t bad.” (Which is, of course, a completely different fallacy.)

But the other possibility is that they hear about sandbox campaigns. They probably erroneously believe that sandboxes are the opposite of railroads. (They’re not.) But they definitely hear that, “In a sandbox, you can do anything!”

And they think to themselves, “Let the players do anything? I can do that!”

Unfortunately, they still don’t have the tools to prep anything other than a linear plot. So what do they prep?

A linear plot requiring a predetermined sequence of specific choices and outcomes.

The only difference is that the players can now “do anything” (sic), so the GM no longer forces the required choices and outcomes. In the most malignant form of the fallacy, they won’t even signpost the choices.

The end of the world is actually fairly dramatic as an outcome. It’s far more common for the players to miss one of these blind turns and just… discover there’s nothing to do. There is, after all, only the one plot; the one path. Leave the path and there’s simply nothing there: You can try to engage with characters or go to interesting places, but nothing happens. You can “do anything,” but nothing you do results in anything happening because the only thing that matters is still the GM’s plot.

“Sandboxes don’t work.”

THE SOLUTION

The solution, obviously, is: Don’t do that.

If you’re going to move away from railroading (and you absolutely should), then you need to actually abandon that broken structure, not just pretend it’s not there. Check out Game Structures and the Scenario Structure Challenge to start exploring fully functional structures for your adventure design.

For more insight on how the scenario selection/creation dynamic at the heart of a sandbox campaign works, check out Advanced Gamemastery: Running the Sandbox. You might also find the extended practical example given in Icewind Dale: Running the Sandbox enlightening.

ADDENDUM

This post has been live for a couple of days, and I want to clear up a point of confusion:

The scenario described at the beginning of the essay is not a railroad. If it was a railroad, the GM would have enforced a preconceived outcome.

The scenario described at the beginning of the essay is what happens when a railroader preps a scenario that requires railroading to work (because that’s the only thing they know how to prep), but then doesn’t railroad.

This is why I’ve said that railroading is a broken technique attempting to fix a broken scenario.

The fallacy is believing that non-broken scenarios are impossible (or bad or impractical) because your broken scenario doesn’t work.

Go to Part 16: Don’t Write Down Initiative

Go to Part 1

You’ve just finished describing in your Feng Shui game how the evil cyborg clone of Jet Li has leapt through the warehouse window, sending a cascade of glass glittering across the oil-soaked floor and landing in a perfect three-point stance. You pause, but the players don’t immediately respond. So you keep talking: Cyborg Jet Li somersaults forward and raises his arm. The flesh peels back, revealing a machine gun… Oh, god. Still no response. Okay, so the machine gun fires, spraying the room with bullets. Then the cyborg dashes behind a forklift for cover. Then he shouts out, “Your deaths are all part of the program!” Then he summons a couple of his attack drones, which come flying in through the window. And then… and then… and then…

This is something I call fearing the silence. The GM finishes describing something and pauses… but there’s not an immediate response from the players. The silence, however fleeting, is like a vacuum, and the GM feels compelled to fill it. So they start talking again. And what can they possibly talk about? Well, whatever would happen next, right?

Gotta keep talking, gotta keep talking, gotta keep talking.

From the GM’s perspective — either consciously or subconsciously — the players aren’t engaged with the game. If they were, then they’d be declaring an action. But they aren’t. Which means the GM has done something wrong. So the GM has to do something — they have to say something — that will get the players engaged again.

(Oh god. It’s all going horribly wrong. What can I do? Gotta do something. Do something. Do something. Just keep doing things.)

But what’s actually happening is that the players are being boxed out. They can’t make declarations about what they’re doing, so it stops feeling like they’re interacting with the world and starts feeling like they’re just watching it.

This often becomes a cascading problem: Because the players lose “momentum” in interacting with the world, it can take a moment to sort of reconnect and get rolling again… except the moment they need is a moment of silence, and the GM is nervously filling it before they can get going.

WHAT DO YOU DO?

When I find myself fearing the silence and feeling a need to fill it, what I find effective is to consciously choose to fill it by saying, “What do you do?” or “What are you doing?”

In other words, very deliberately and very explicitly pass the ball to the players. Now the silence is their problem!

Or, more accurately, their opportunity.

You may also find it useful to think explicitly in terms of turn-taking, particularly if NPCs are involved in the scene. If all of your NPCs have had a chance to say something or do something, then it’s definitely time to make sure the PCs have a chance to say something or do something. (It’s a challenge to roleplay multiple NPCs talking to each other, but if you get into a rhythm with it, it can be surprisingly easy for the NPCs to just chat amongst themselves endlessly without the players being able to get a word in edgewise.)

It’s perhaps unsurprising that I’ve often seen this sort of problem in the first session of a campaign or the first half of a one-shot, only for things to improve as the game goes on: A lot of this is being driven by nervousness. As the GM settles down and the group finds its groove, the problem goes away. Of course, there’s only one chance to make a first impression, so it won’t hurt to consciously work towards avoiding the initial misstep.

GM DON’T LIST #14.1: BOTTOMLESS LORE

A similar problem occurs with bottomless lore. I commonly find myself falling into this trap when I’m improvising some cool new element of the world. Maybe the players have gone to a location I didn’t anticipate and so I’m creating it on the spot. As I describe the location, I keep getting cool new ideas.

So I Just. Keep. Talking.

There’s this… and this… and you see somebody doing this… and somebody else doing this… and there’s also this other thing… and also… and then the first guy does this other thing… but then…

There’s always more stuff to share about the infinite world! You’re once again stuck in a loop and the players are, once again, boxed out.

This is actually quite similar to the freeze-frame boxed text we discussed in GM Don’t List #13: Boxed Text Pitfalls, where the PCs get stuck while the GM narrates their way through a cutscene. The only difference is that in this case it’s happening spontaneously at the table.

THE REACTION POINT

The solution, however, is quite similar. You need to consciously identify the reaction point — the point at which something happens that the PCs will want to react to — and then you need to stop talking.

Another useful tip here is to set a mental description limit: What’s the maximum number of things you can describe before you need to stop talking. This isn’t hard or fast, but I recommend three to five, and almost never more than seven. (You, of course, don’t need to describe that many things. If you describe three things and that’s everything you need to set the scene, that’s great.)

This limit can feel constraining, but there are a few things to keep in mind that will help:

First, at the beginning of a scene — a new location, etc. — start broad, then specific. You’ve given yourself a “budget” for your description, and you’ll want to make sure the players know that they’re in a forest before you start describing individual trees.

Second, identify the reaction point before you start the description. Ideally, aim your description at the reaction point. At a bare minimum, when you’ve spent all of your “budget” except for the reaction point, arbitrarily stop describing anything else and cut to the reaction point. (“Okay, I’ve said four things about the hotel room. Now I need to mention the ogre with the gun and see what happens.”)

If you identify multiple reaction points in the scene, what you’ll usually want to do is prioritize them. Present the first, let the PCs react to it, and then present the second. Upon occasion there might be two reaction points that are truly happening simultaneously — e.g., the ogre pulls a gun at the very moment that kobold ninjas leap in through the window. That’s fine. It’ll just chew up more of your “budget,” since you’ll need to describe both reaction points.

(For more on reaction points, and why you should almost always put the reaction point at the end of your description, check out Random GM Tips: Reaction Points.)

Third, remember that description can persist through the scene. You don’t need to describe every single detail of the hotel room before the PCs react to the ogre pulling her gun. You may not have had time to describe the lamp, but as Antoine rushes forward to knock the gun out of the ogre’s hand, you can describe how the sickly light flickering out through the cigarette-stained lampshade throws his shadow against the far wall. Maybe you haven’t described what can be seen outside the window, but the kobold ninjas bursting through it gives you a chance to mention the “billboard declaring Mayor Thomas’ re-election campaign” behind them.

And so forth.

This is particularly useful if you find yourself bubbling over with bottomless lore: You don’t need to discard all those cool ideas that are sparking in your imagination; you just have to hit the pause button on them, drop the reaction point, and then look for the opportunity to weave them into the action.

GM DON’T #14.2: SAVING THROW MERRY-GO-ROUND

GM: The Lich-King waves his hand and a cloud of black miasma begins oozing up through the floor. Give me a Fortitude saving throw!

Player: 24.

GM: You quickly gulp in a breath of untainted air. But huge gouts of flame erupt from the ceiling! Give me a Dexterity saving throw!

Player: 15.

GM: You’re scorched by flame! As you grimace in pain, the Lich-King points a finger in your direction. There’s a lance of sickly green energy. Give me a Dodge roll!

Player: 9.

GM: It strikes you for 36 points of damage! The Lich-King pulls a lever on the wall, and a trap door opens in the floor beneath you! You’ll really want to make your Dexterity saving throw!

Presented like this, it’s pretty easy to notice that “the player gets a turn” doesn’t appear in this list of events. But this can often be harder to see at the actual gaming table: There are multiple players reporting results. The GM is weaving evocative descriptions of what’s happening in the scene as a result of the various reactive checks or saving throws. And it can certainly feel as if the PCs are doing a lot of stuff: They’re gulping in air! Heroically diving out of the way! Rolling dice!

And, of course, there’s a similar impulse that we’ve seen in fearing the silence and bottomless lore: You want the game to be exciting and engaging. That means you need to keep talking! And if you’re going to keep talking, then stuff needs to happen for you to describe. And you can make that stuff happen by using the playing pieces you control — e.g., the NPCs.

This is also why this problem most often crops up in action scenes: moments that seem to demand a breakneck pace, and so the GM feels pressured to keep things moving.

The solution to this is largely just Don’t Do It™. Which, of course, is not terribly helpful. Couple things that may help if you find yourself doing this:

First, if appropriate, get into an initiative order sooner rather than later. This doesn’t necessarily require combat or even the presence of NPCs. You can just call for initiative checks and use them to structure the scene.

Second, you don’t have to go for the ultra-formality of an initiative order to think in terms of turn-taking. The Lich-King created the black miasma. Has Sasha had a chance to go since then? If not, ask Sasha what she’s doing before the Lich-King takes another action. If you just make a point of thinking about this, you can probably keep track of it in your head. If not, make a list on the notepad in front of you and use tally marks when people take actions.

(This will also help individual players from dominating an encounter and boxing out their fellow players.)

CONCLUSION: CARE ABOUT THE REACTION

In the broader scope, the big solution to all of these problems is simply caring about what your players are doing.

Minimize the mindset of the Story you’re trying to tell or the World you’re trying to immerse them in. Your NPCs should be awesome, but they are ultimately not the stars of the campaign.

The mindset you want to emphasize, in my opinion, is: I want to see what the PCs do.

I’m having the ogre draw her gun because I want to see how the players react to that. (And I want to see how I react to what they do!) I can have a lot of fun playing around with all the cool lair actions the Lich-King can take, but ultimately the point of the Lich-King encounter is for the PCs to confront him.

It can be a subtle shift in thought, but when your primary focus becomes, “Oooo… I wonder what this will make them do?!” you’ll never forget to give them the opportunity to do it.

Go to Part 15: The Railroader’s Fallacy

Go to Part 1

Boxed text in an RPG scenario is a prewritten narration designed to be read to the players by the GM. It looks like this:

The center of this room is filled with a massive contraption of brass and copper and rotten, worm-eaten wood. Great hoops of metal are suspended about a central sphere, with various lumps, pulleys, cranks, and levers protruding here and there in an apparently chaotic and incomprehensible jumble.

(from The Complex of Zombies)

The advantage of boxed text, of course, is that it can be prepared ahead of time: It can give you a chance to carefully consider and craft your choice of words to best effect. If there’s essential information that needs to be conveyed to the players, putting it in boxed text will virtually guarantee that it’s not accidentally omitted in actual play.

In The Art of the Key, for example, I talk about how these features of boxed text make it ideal for conveying what characters see when first entering a room or location by clearly delineating the information the players should automatically have from the rest of the key. (Even if you don’t use full-fledged boxed text to achieve this effect, you’ll still want some form of not-boxed-text that fulfills the essential function.)

So why wouldn’t you use boxed text?

• Carefully crafting your words is time-consuming. (Which may suggest its elimination by virtue of the principles of smart prep.)
• The result is inherently less flexible. (For example, if a room has multiple entries the boxed text needs to be generic enough to work for any potential entrance. Add to this NPCs, lighting conditions, etc.)
• Reading prepared text to an audience is a very specific performance, and can easily be one that a GM is not comfortable with. (In such cases, the spontaneity and engagement of improvising a description will often be superior to a stilted or rushed reading.)

If you’re running a published adventure with boxed text and you’d rather not use it — for these or any other reasons — you may find it useful to highlight the key facts presented by the boxed text, quickly turning it into not-boxed text:

The center of this room is filled with a massive contraption of brass and copper and rotten, worm-eaten wood. Great hoops of metal are suspended about a central sphere, with various lumps, pulleys, cranks, and levers protruding here and there in an apparently chaotic and incomprehensible jumble.

(As described in The Art of the Key, you can use the same technique to quickly salvage location keys that have failed to differentiate “seen at a glance” information from hidden secrets.)

SINS OF THE BOX

Performance issues and a lack of flexibility, however, are not the only reasons that people dislike boxed text. Often they will have been on the receiving end of bad boxed text, which is all too prevalent in published adventures and, as a result of their poor example, homebrewed adventures, too. Many of these failures are either freeze-frame boxed text or remote-control boxed text

Freeze-frame boxed text is when the GM starts reading and then the PCs are frozen in place while a bunch of stuff happens. These can often get quite elaborate, with entire scenes being played through while the players sit impotently in their seats, boxed out (pun intended) from actually playing the game, but even subtle examples can be incredibly frustrating:

Grasping weeds and vines erupt from the cobblestone street beneath the carriage at the head of the parade. The ox pulling the cart panics, causing the vehicle to careen into a post covered in decorations. The vegetation then wraps around the cart’s wheels and the closest bystanders. A pair of revelers produce weapons, revealing themselves to be guards protecting the Prince of Vice.

(from Journeys Through the Radiant Citadel)

As soon as the players hear, “Grasping weeds and vines erupt from the cobblestone street!” they’ll want to respond to that. Instead, everyone else in the scene – including the ox! – gets to react before they do.

What we’ve identified here is the reaction point. You don’t always need to immediately stop talking when you’ve reached the reaction point (although often you should try to structure you descriptions so that you do), but even if there are other pertinent details of the world to establish, what you should avoid at all costs is having the game world continue to move forward past the reaction point without letting the players react; without letting the players play the game.

This is an easy trap to fall into with boxed text: The author (or GM) wants to establish the key features of the scene – vines appear, ox panics, cart crashes, disguised guards draw weapons – and the boxed format strongly biases you towards pushing all of that together into a single presentation.

When you see freeze-framed boxed text as a GM, though, what you should do is break it up into actionable chunks. And I use the word “actionable” here because you are specifically looking for the actions you can take as GM, allowing the players to have a reaction to each of those actions.

Here, for example, we actually start at the end of the boxed text: There are guards disguised as revelers. Before anything else happens, therefore, you should call for Perception checks to see if any PCs spot them.

(If they are spotted, what do the PCs do with that information? I have no idea. Play to find out.)

The next actionable chunk is: “Grasping weeds and vines erupt from the cobblestone street.”

That signals the start of combat, which means that it should trigger an initiative check. So rather than skipping past that moment, make the initiative check. (Or don’t if you’ve already rolled initiative and are ready to go, go, go! But either way, you’re moving into tracked combat time.)

The other actionable chunks are:

• the ox panicking and crashing the cart
• the guards drawing their weapons and moving to attack the vines

These can obviously just happen during the first round of combat, with the PCs also taking whatever initial actions they think best, too.

REMOTE-CONTROL BOXED TEXT

Remote-control boxed text suffers from similar problems (preventing the players from participating), but insidiously goes one step further by declaring the thoughts or feelings or (worse yet!) actions of the PCs.

• “You look upon the devastation of the valley and are overwhelmed by sadness.”
• “You step forward and return the king’s greeting with a deep bow.”
• “As you return to Waterdeep, you smile, thinking fondly of the ale at Trollskull Manor.”
• “You see a strange creature crouching upon the boulder. As you step into the room, it looks up with wide, yellow eyes, gives a deafening call of alarm, and then scurries away.”

There are two major problems with this sort of thing.

First, a player controls exactly one thing: their character. When you take the one thing they control away from them — even for a little bit — you have effectively removed from the game. They are, in fact, no longer a player, but merely a spectator.

Second, for many players, the damage that you do in those brief moments of seizing control can extend far beyond the moment itself. If their character does something that isn’t what they would have chosen to do, it can often feel as if there’s something “wrong” with the character. Do it enough — or do it at just the wrong moment — and the player may dissociate entirely from the character. When that happens, you may have easily just ruined the entire campaign for them.

So… don’t do this. As the GM you literally have control over the entire game world. Be content with literally the entire universe of toys you have to play with.

Focus on showing the players the scene and letting them react to it. Don’t tell them how they’re reacting to it.

Those reactions, it should be noted might be:

• physical actions
• emotional reactions
• reflective thought
• dialogue

And so forth. There’s a wide panoply of possible experiences, and some of them may be entirely internal to the player. You may never know, for example, how their character truly felt about something. That’s okay. The important part is that they know, and it will shape their actions and the course of the entire campaign.

Go to Part 14: Fearing the Silence

Go to Part 1

GM: One of the villagers approaches you. “Thank you, brave hero, for slaying the dragon! You have saved all of our lives!”

Hero: You are most welcome.

GM: “I have a task that I believe only you could possibly accomplish!”

Hero: If it is within my power, I will do it.

GM: “You are most kind, mighty hero!”

Hero: And what is the task?

GM: “Please deliver this letter to my niece in Watertown.”

Scenario hooks like this — where the PCs are tasked to perform as mundane messengers — are surprisingly common. I think of these as mail carrier hooks. They don’t always involve a literal letter or message; sometimes it’s an object that needs to be delivered, or maybe the PCs need to go and retrieve something instead of delivering it.

The root of the problem, I believe, is that mail carrier hooks are extremely common in (a) published adventures and (b) video games.

They’re common in published scenarios because the hooks in a published scenario are, by necessity, generic: The writer doesn’t know who the characters are or what’s been happening in your campaign, so they can’t tie the hook to any of those elements. It’s unfortunately really easy for “generic” to trip over into bland.

Such scenarios are also often set in specific locations that the writer feels obligated to bring the PCs to. Having an NPC literally say, “You need to go to there,” is the easiest possible way to make that happen. And the most generic possible reason for an NPC to say that is a sealed envelope that needs to be delivered.

They’re even MORE common in video games, where “I have turned on a switch state and put an item in your inventory, go to Point B to turn off the switch state and remove the item from your inventory” is nearly the most simplistic programming possible. Want to implement a lot of content quickly and/or signal the players that it’s time to move onto the next zone? Mail carrier hooks are super-easy to implement.

And because published scenarios and video games are probably the most significant exemplars for new GMs, these boring, generic scenario hooks infect their scenarios, too. Eventually many people — players and GMs alike — come to accept them as a rote expectation of the game.

GM DON’T #12.1: MAIL CAR ON THE RAILROAD

The problem with mail carrier hooks is that they tend to reduce the PCs to mere errand boys. A particularly insidious implementation of these hooks, therefore, is to string them out in a linear sequence and then railroad the players through them.

An NPC tells the PCs where to go. The PCs go there and meet another NPC. That NPC tells them where to go. Repeat forever.

There’s a certain dark elegance to the scheme’s simplicity. If you’re just going to force the players to do exactly what you want them to do anyway, you might as well just tell them where they’re supposed to go. And, as we’ve discussed, the delivery of a sealed envelope (or it’s equivalent) is the absolute most generic way to do it: Anyone can hand you a letter and they can tell you to take it literally anywhere.

(Obligatory “don’t railroad your players” here.)

Stringing together these arbitrary, generic interactions, however, will often begin to breed a meaningless lack of care in the players: The figurative envelope is, ultimately, empty. The first NPC is not telling them to seek the second NPC because that actually matters; they are doing so in order to move the PCs to the next place they’re supposed to go. And the players are not doing it because they care about what happens; they’re doing it because the GM is telling them to do it.

This, by itself, is a terrible malaise that will sap the strength and vitality of a campaign. But it can become particularly cancerous if this attitude feeds back into the Game Master: Seeing that their players don’t care about the content of the hook (because that content is, of course, a mirage), the GM stops caring, too. The whole structure now becomes a kind of cargo cult: The NPCs, of course, must have a “reason” why the PCs need to go where they tell them to go, but since the reason doesn’t matter, it simply degenerates into a rote recitation divorced from true cause and effect.

The Descent Into Avernus campaign for D&D 5^th Edition is a textbook example of what this looks like in practice. The entire campaign follows this structure of a mail carrier railroad. In one notable example, the PCs are trying to reconstruct the lost memories of an amnesiac friend. They are told that a particular NPC knew their amnesiac friend during the period of their friend’s lost memories. So they seek out the NPC who, of course, recognizes their amnesiac friend and then… nothing. The adventure provides no explanation of what the NPC’s memories of their friend are.

Because, of course, the NPC’s role is not to provide those memories. That would be meaningful and this is a cargo cult which has forgotten meaning. The NPC’s role is to tell the PCs to go and talk to a different NPC, and they do that. And, having done that, the expectation is that the PCs will continue on to the next NPC. It’s assumed that the players are on the same page; that they won’t actually care about why they were sent here, because they have been trained to discount meaning.

If you aren’t part of the cargo cult, the result seems utterly bizarre. The presence of an envelope surely implies the presence of a letter; if you are sent to hear the memories of an NPC, then surely those memories will be shared. But in the cargo cult, of course, the act of delivering an envelope has become entirely separated from the concept of a letter.

The result, of course, is disastrous.

When this structure becomes full-blown, the whole backbone of the campaign is built on nonsense. The inevitable consequence is that this rot will spread into the rest of the campaign. If you’ve learned that the central plot you’re following has no meaning, then it’s a pretty short step towards believing that nothing else in the campaign has meaning.

SCENARIO HOOKS THAT MATTER

The key solution here is to use scenario hooks that matter. If the players care about what they’re doing and/or if what they’re doing is important, that escalates everything else that happens at the table.

Once we understand this, we can see that a mail carrier hook is not fundamentally wrong. The trick is recognizing that the structure of a mail carrier hook is so utterly devoid of purpose that it becomes crucial for the message itself to be of great import.

In Storm King’s Thunder, another 5^th Edition D&D campaign, for example, there are number of mail carrier hooks. Some of them are quite mediocre; things like, “Hey, could you deliver some horse harnesses for me?”

But there’s also mail carrier hooks like, “Giants are invading! The Harpers must be warned!” And that’s clearly meaningful. It matters. The PCs will feel important being asked to do that.

So how do we make scenarios hooks that matter? How do we make the players care about the hook?

Well, one way, as we’ve seen, is to increase the stakes. Put big important stuff — people’s lives — on the line. The entire movie of 1917 is not only one shot; it hangs entirely on the single goal of delivering a message that will save thousands of lives. And it is absolutely compelling.

The most effective thing you can do, though, is simply listen to your players. What do they already care about? People, places, things, goals. Whatever it is, simply tie your hook to that and your work is already done. You just need to make sure that the tie is significant. (Someone the PCs caring about asking them to deliver a generic message isn’t inherently significant. You need to make the message important or, better yet, vital to them.)

If you’re using a published adventure, take the time to identify the generic hooks and make them specific and important to the PCs. The difficulty of this can vary, but it’s generally a lot easier than you might think. The key thing is to identify elements in the published adventure which can be adapted or recast to fit the existing lore of your campaign. (At the beginning of the campaign, this will primarily be drawn from the PCs’ backgrounds. Later it will build on your shared experiences at the table.) This is a topic I discuss at greater length in The Campaign Stitch.

Go to Part 13: Boxed Text Pitfalls

Go to Part 1

This will probably be the most controversial entry I write for the GM Don’t List, because there are a lot of players who absolutely LOVE this. And if the players love it, why wouldn’t you do it?

Well, try to bear with me because we’ll get to that.

The technique we’re talking about is description-on-demand: The GM directs an authorial question at a player, giving them narrative control to define, describe, or determine something beyond the immediate control of their character. Examples include stuff like:

• What is Lord Fauntleroy’s deepest secret?
• You open the door and see Madame DuFerber’s bedroom. What does it look like?
• Okay, so you pull him off to one side, confess your love to him, and demand to know if he feels the same way. What does he say?
• What does Rebecca [your PC] know about the Dachshund Gang? Who’s their leader?
• Robert, tell me what the name of the mountain is.
• Okay, you find some juicy blackmail on Mayor McDonald. What is it that he’s done? What evidence do you find?

If you haven’t encountered this technique before, the key thing to understand is that none of the characters being defined here are PCs: The GM isn’t asking Lord Fauntleroy’s player what their character’s deepest secret is. They’re asking the players to step out of their character and create an element of the game world external to their character (often in direct response to their character taking interest in that element of the game world).

It’s description-on-demand because the GM is demanding that a player provide description.

MANY PLAYERS DON’T LIKE IT…

Description-on-demand tends to be a fad that periodically cycles through the RPG meme-sphere. When it does so, the general perception seems to be that every player thinks this is the greatest thing since chocolate-dipped donuts.

So let’s start there: This is not true. Many players do love it. But many players DO NOT. In fact, a lot of players hate it. There are a significant number of players for whom this is antithetical to the entire reason they want to play an RPG and it will literally ruin the game for them.

I’m one of those players. I’ve quit games because of it and have zero regrets for having done so.

So, at a bare minimum, at least take this lesson away with you: Check with your players before using description-on-demand. Because it can absolutely be a poison pill which will ruin your game for them.

Okay… but why do they hate it?

A brief digression: If you’re not familiar with the distinction between roleplaying games and storytelling games, I recommend checking out Roleplaying Games vs. Storytelling Games. The short version is that roleplaying games feature associated mechanics (where the mechanical choices in the game are directly associated to the choices made by your character, and therefore the act of making mechanical choices in the game – i.e., the act of playing the game – is inherently an act of roleplaying) and storytelling games feature narrative control mechanics (where the mechanics of the game are either about determining who controls a particular chunk of the narrative or about determining the outcome of a particular narrative chunk).

When I’m playing a roleplaying game (as opposed to a storytelling game), I am primarily interested in experiencing the game world from the perspective of my character: I want to experience what they experience, make the decisions that they would make, and vicariously experience their fictional life. The reason I want this experience can be quite varied. Roleplaying can be enjoyable in a lot of different ways (catharsis, escapism, experimentation, sense of wonder, joy of exploration, problem-solving, etc.) and the particular mix for any particular game or moment within a game can vary considerably.

Description-on-demand, however, literally says, “Stop doing that and do this completely different thing instead.”

This is not only distracting and disruptive, it is quite often destructive. There are several reasons for this, but the most significant and easy to explain is that it inverts and negates the process of discovery. You can’t discover something as your character does if you were the one to authorially create it in the first place. This makes the technique particularly egregious in scenarios focused on exploration or mystery (which are at least 90% of all RPG scenarios!) where discovery is the central driving force.

Not all players who dislike description-on-demand hate it as much as I do. Some will be merely bored, annoyed, or frustrated. Others will become stressed, anxious, or confused when being put on the spot. Some will just find their enjoyment of the game lessened and not really be able to put their finger on why. But obviously none of those are good outcomes and you need to be aware that they’re a very real possibility for some or all of the players at your table before leaping into description-on-demand.

…BUT SOME PLAYERS DO

So why do some players love this technique?

And they clearly DO love it. Some enjoy it so much that they’ll just seize this narrative control for themselves without being prompted by the GM. (Which can cause its own problems with mismatched expectations, but that’s probably a discussion for another time.)

So… why?

If we keep our focus on the tension between discovery and creation, it’s fairly easy to see that these are players who don’t value discovery as much. Or, at least, for whom the joys of creation outweigh the joys of discovery.

I’m one of those players. When I’m playing a storytelling game, I love being offered (or taking) narrative control and helping to directly and collectively shape the narrative of the world.

…

… wait a minute.

How can both of these things be true? How can I both hate it and love it?

Well, notice that I shifted from talking about roleplaying games to talking about storytelling games.

Here we get to the crux of why description-on-demand is a poor GMing technique. Because while there are times I prefer to be focused on in-character discovery, there are ALSO times when I’m gung-ho for authorial creation. And when that happens, description-on-demand in a traditional RPG is still terrible.

Remember that this technique gives us the opportunity to experience the joy of creation, but does so only by destroying the joy of discovery. There is an inherent trade-off. But when it comes to description-on-demand, the trade-off sucks. I’m giving up the joy of discovery, but in return I’m not getting true narrative control: Instead, the GM arbitrarily deigns to occasionally ask my input on very specific topics (which may or may not even be something that I care about or feel creatively inspired by in the slightest).

Description-on-demand techniques in an RPG dissociate me from my character while offering only the illusion of control.

In an actual storytelling game, on the other hand, I have true narrative control. The structure and mechanics of the game let me decide (or have significant influence over) when and what I want narrative control over. This is meaningful because I, as a player, know which moments are most important to my joy of discovery and which ones aren’t. (This is often not even a conscious choice; the decision of when to take control and when to lean back is often an entirely subconscious ebb-and-flow.)

Note: This discussion is largely assuming storytelling games in which players strongly identify with a specific character (“their” character, which they usually create). There are many other storytelling games – like Once Upon a Time or Microscope – in which this is not the case. In my experience many of those games still feature a tension between discovery and creation, but the dynamics are very different in the absence of a viewpoint character.

Towards the end of the movie Inception, Eames looks towards the dream vault they’ve been trying to break into for basically the entire movie and says, “It’s a shame. I really wanted to know what was going to happen in there. I swear we had this one.”

Now, imagine the vault door opening. And the GM says: “Okay, Eames, tell me what you see in there!”

For one player, this is great! The importance of this vault has been relentlessly established. The entire narrative has been pushing towards this revelation and now THEY have the opportunity to create what’s inside it!

For another player, this is a disastrous, gut-wrenching disappointment. They’ve spent all this time anticipating this moment; speculating about what the vault might contain, imagining different possibilities, parsing together clues to try to figure it out. And now they’re going to find out! And, instead, the GM announces that there was never any solution to this riddle. There was no plan. No mystery to be solved. Just an empty madlibs puzzle waiting to be filled. “I really want to find out what’s in that vault,” but instead, “Nope, you don’t get what you want. In fact, you have to actively participate in disillusioning yourself.”

For a third player, they don’t really care about having narrative control, but they don’t really have any strong ideas about what should be in the vault and aren’t interested in making a creative decision about that.

And here’s the key thing: You have absolutely no way of knowing which player is which.

In fact, the answer can very easily change from one moment to the next. One player may want an in-character pay-off for the mystery of the vault, but has strong opinions on what Lord Fauntleroy’s deepest secret is and would love to define that. But another may well feel the exact opposite, while a third has no interest in either, and a fourth will be disappointed by any player defining what’s in the vault, because it’ll reveal there wasn’t a canonically true answer.

(And, yes, I have very deliberately chosen a narrative in which the characters do, in fact, have influence — albeit an indirect one — over what the vault will contain. I want you to challenge your preconceptions within the uncertainties of this liminal space. While you’re here, if you’re familiar with the movie, ask yourself whether your opinion on this interaction would be different if the GM was asking Robert Fischer’s player what was inside the vault instead of Eames’ player. Do you see how a different player of Robert Fischer might want the exact opposite answer?)

The cool thing about most narrative control mechanics is that they give you the ability to say, “This is what I care about. This is what I want to create.” And, conversely, “This is not something I care about. This is, in fact, something I DON’T want to be responsible for creating.”

CONCLUSION

Here’s my hot take.

I think description-on-demand is primarily — possibly not exclusively, but primarily – popular with players who have never played an actual storytelling game or who would desperately prefer to be playing one.

Because the thing that description-on-demand does — that little taste of narrative control that many players find incredibly exciting — is, in fact, an incredibly shitty implementation of the idea.

If you’re interested in an RPG, this is like playing Catan and having the host demand that you roleplay scenes explaining your moves in the game. (Just play an actual RPG!)

On the other hand, if you’re craving an STG, then description-on-demand in a traditional RPG is like playing co-op with an alpha quarterback who plays the entire game for you, but then occasionally says, “Justin, why don’t you choose the exact route your meeple takes to Sao Paolo?” and then pats themselves on the back for letting you “play the game.”

(This applies even if you’re playing an RPG and are just interested in adding a little taste of narrative control to it: You would be better off grafting some kind of minimal narrative control mechanic onto the game so that players can, in fact, be in control of their narrative control.)

To sum up, the reason description-on-demand makes the GM Don’t List is because:

• If that’s not what a player wants, it’s absolutely terrible.
• If it is what a player wants, it’s a terrible way of achieving it.

BUT WAIT A MINUTE…

There are several other techniques which are superficially similar to description-on-demand, but (usually) don’t have the same problems. Let’s briefly consider these.

FENG SHUI-STYLE DESCRIPTION OF SETTING. Robin D. Laws’ Feng Shui was a groundbreaking game in several ways. One of these was by encouraging players to assert narrative control over the scenery in fight scenes: If you want to grab a ladder and use it as a shield, you don’t need to ask the GM if there’s a ladder. You can just grab it and go!

Notably this is not on-demand. Instead, the group (via the game in this case) establishes a zone of unilateral narrative control before play begins. It is up to the players (not the GM) when, if, and how they choose to exercise that control. Players are not stressed by being put on the spot, nor are they forced to exert narrative control that would be antithetical to their enjoyment.

EXTENDED CHARACTER CREATION: This is when the GM asks a question like, “What’s Rebecca’s father’s name?” Although it’s happening in the middle of the session, these questions usually interrogate stuff that could have been defined in character creation.

This generally rests on the often unspoken assumption that the player has a zone of narrative control around their character’s background. Although this narrative control is most commonly exercised before play begins, it’s not unusual for it to persist into play. (Conversely, it’s similarly not unusual for players to improvise details from their character’s background.) This can even be mechanically formalized. In Trail of Cthulhu, for example, players are encouraged to put points into Languages without immediately deciding which languages they speak. (Each point can then be spent during play to simply declare, “I speak French,” or the like.)

Because it’s unspoken, however, both the authority and boundaries of this zone can be ill-defined and expectations can be mismatched. (The problems that can result from this are probably yet another discussion for another time.)

There’s also a gray zone here which can easily cross over into description-on-demand. “What’s your father’s name?”, “Describe the village where you grew up,” and “You grew up in the same neighborhood as the Dachshund Gang, so tell me who their leader is,” are qualitatively different, but there’s not necessarily a hard-and-fast line to be drawn.

RESOLUTION OF PLAYER-INITIATED ACTION: So if saying, “You find some juicy blackmail on Mayor McDonald. What is it that you’ve caught him doing?” is description on demand, then what about when the GM says, “You deal 45 hit points of damage. He’s dead. Describe the death blow,” that must also be description-on-demand, right? I mean, the GM even said the word “describe!”

There is some commonality. Most notably, you’re still putting players on the spot and demanding specific creativity, which can stress some players out in ways they won’t enjoy. But this effect is generally not as severe, because the player has already announced their intention (“hit that guy with my sword”) and they probably already have some visualization of what successfully completing that intention looks like.

In terms of narrative control, however, there is a sharp distinction: You are not asking the player to provide a character-unknown outcome. You are not dissociating them from their character.

This is true in the example of the sword blow, but may be clearer in a less bang-bang example. Consider Mayor McDonald and the difference between these two questions:

• “You find some juicy blackmail on Mayor McDonald. What is it that you’ve caught him doing?”
• “You find some juicy blackmail on Mayor McDonald. He’s been cheating on his wife with a woman named Tracy Stanford who works in his office. How did Rebecca find this out?”

In the first example, the GM is asking the player to define an element of the game world outside of their character and their character’s actions. In the second example, the GM has defined that and is instead asking them to describe what their character did. Although it’s become cognitively non-linear (the player knows the outcome, but is describing actions their character took before they knew the outcome), it is not dissociated from the character.

The same is true of the sword blow: The mechanics say the bad guy dies; take a step back and roleplay through how that happened.

(For a longer discussion of closely related stuff, check out Rulings in Practice: Social Skills.)

WORLD DEVELOPMENT BETWEEN SESSIONS: As a form of bluebooking, players may flesh out elements of the campaign world between sessions.

Sometimes this is just a more involved version of extended character creation. (“Pete, it looks like that Order of Knighthood your character’s brother joined is going to be playing a bigger role starting next session. Could you write ‘em up? Ideology, leaders, that kind of thing?”) But it can scale all the way up to troupe-style play, where players might take total control over specific aspects of the world and even take over the role of GM when those parts of the game world come up in play.

The rich options available to this style of play deserve lengthy deliberation in their own right. For our present discussion, it suffices to say that while this is in most ways functionally identical to description-on-demand (the player is taking authorial control beyond the scope of their character), in actual practice there’s a significant difference: Players don’t feel stressed or put on the spot (because they have plenty of time to carefully consider things). And many players don’t feel that inter-session discussions are as disruptive or dissociative as stuff happening in the middle of a session (because they aren’t being yanked in and out of character).

Go to Part 12: Mail Carrier Scenario Hooks