The Alexandrian

Posts tagged ‘railroading manifesto’

Big Red - CGP Grey

I’ve just had an interesting discussion regarding the intersection between random generators and railroads.

Hypothetical Situation #1: You’re running a standard hexcrawl campaign. You generate a sequence of six locations. Regardless of where the PCs decide to go, they will encounter those six locations in the order that you prepared them.

This is self-evidently a railroad.

Now, take this hypothetical situation and begin stripping off locations until you’re left with a single location. Regardless of which direction the PCs leave town, they will encounter that specific location.

Still self-evidently a railroad.

Hypothetical Situation #2: You’re running a standard hexcrawl campaign. You create a random encounter table of six locations. When the PCs leave town, you start rolling on the random encounter table. As they encounter locations you cross them off the random encounter table. Since they’re randomly generated, however, the sequence in which they’re encountered may vary.

Is that a railroad?

Does your answer change if I similarly strip off locations until the random encounter table consists of a single location?


Does this mean that random content generators are an example of railroading?

No. But this thought experiment does demonstrate the complexity of these issues and the danger of trying to create some sort of “railroading purity test” for various techniques without considering the motivation, context, and methodology of their use.

The core distinction here is whether or not the players are making a meaningful choice. In this hypothetical hexcrawl scenario, the choice of direction has been rendered meaningless (since you’ll have the same experience regardless of which direction you go). And if the choice is meaningless, why are you having the players make it? Why are you lying to them about the choice being meaningful?

Random anecdote time: Many, many moons ago my players needed to explore the sewers beneath a major fantasy metropolis. I didn’t want to map the sewers out and the only game structure I really understood for that sort of thing at the time was dungeoncrawling. So I came up with a system which randomly generated dungeoncrawling maps for the sewer.

This worked just fine and the players were having a great time… until they realized that the terrain was being randomly generated. Their interest in exploring the sewers instantly evaporated: They knew that their choices were irrelevant. There was nothing that could actually be discovered. They were using a game structure of exploration, but they weren’t actually exploring anything.

This taught me a really important lesson as a GM: In order for an exploration scenario to work, there has to actually be something to explore. If all choices are equally likely to get you to your goal (because your discoveries are being randomly generated or because the GM has predetermined their sequence), then your choices become meaningless. And meaningless choices are boring and frustrating.


I’ve talked frequently in the past about the usefulness of procedural content generators: In a dungeon, random encounters can simulate the activity of a complex. In open campaigns, dungeons can be restocked with random generators. In hexcrawls, random generators can simulate the activity of the wilderness or they can be used to generate new locations on-the-fly.

These tools are incredibly useful. But how do you use them in a way that doesn’t negate player choice? How do you use them without the railroad?

The solution is actually quite obvious: Make sure the players’ choices are still meaningful even with the presence of the random elements.

A really simple example of this is simply allowing the actions taken by the PCs to affect the random generators: In a dungeon, for example, certain activities will create noise and increase the likelihood or frequency of encounters. In a hexcrawl, choosing to go to the Old Woods will cause the GM to roll on a different random encounter table than the Volcanic Peaks. And so forth.

Another straight-forward variation, in the context of an exploration scenario, is to create an environment with enough meaningful detail that it renders choice meaningful while the random content provides an additional patina of variety. For example, for my OD&D open table hexcrawl I keyed specific content to 256 hexes. That pre-existing geography creates a ton of meaningful choice: The random encounters that are being generated on top of it simply provide additional spice. Another example is the dungeon complex where the keyed rooms provide the meaningful choices, while the random encounter table provides a variety of activity throughout the complex.

A more complex variation would be a procedural content generator that creates an environment as the PCs explore it. This only works, however, if the location at which a piece of content is encountered becomes meaningful. This rules out purely ephemeral encounters (you meet eight orcs and you kill them) because the location is meaningless. But if the PCs are heading west and discover that the Salt Flats of Doom are over there and that Castle Vampire is on the far side of that difficult-to-traverse terrain, that geographical placement becomes meaningful if/when the PCs start mounting expeditions to Castle Vampire. (You could also imagine a structure where placing Castle Vampire and the Valley of the Giants next to each other creates a unique alliance that wouldn’t occur if it turns out that the Valley of Giants is on the other side of town.) The ability to revisit and reincorporate content, as is so often the case, is the key factor here.

Let’s consider a non-exploration example: In Technoir, the plot map of the scenario is randomly generated by the GM through play. Both the GM and the players discover the truth of the conspiracy together. However, before the players make any decisions the GM creates the mission seed which forms the core of the conspiracy. The mission seed provides the pre-existing detail which makes choices meaningful as the PCs seek to unravel the conspiracy (and, usually, penetrate to the heart of the mission seed). Even though the outcomes of these decisions are random, the choices remain meaningful because (a) they determine how the random elements connect to the mission seed (even as they also continue to redefine the conspiracy as a whole) and (b) the mission seed itself is a hard truth which can actually be discovered.

There are a lot of other ways in which random generators could be rooted into meaningful choice (or surrounded by it) without losing their utility. (And we would doubtlessly discover even more if we started poking around other game structures.) What you’ll note, however, is that all of these techiques are very different from simply taking the encounter you want the PCs to have and putting it in front of them regardless of which choices they make.

Go to the Railroading Manifesto

Go to Part 1


The Sphinx

The mechanical gate is a specific example of a broader category of chokers in which the potential solutions to a problem are limited. This limitation self-evidently creates an experiential chokepoint.

Limiting solutions merely for the sake of limiting solutions is almost always a railroad-by-design. What can avoid the railroad, however, is only limiting the potential solutions to a problem within a specific paradigm.

For example, the PCs encounter the Sphinx and it demands an answer to its riddle. Within the paradigm of “solving the riddle” there is only one solution (because the riddle only has one answer). Outside of that paradigm, however, the PCs could also kill the Sphinx, teleport past the Sphinx, or hire someone to deal with the Sphinx for them. (If you opt for the last of these, however, I recommend not marrying the handsome young lad who solves the problem for you. It’ll end in tears of blood.)

The example of the Sphinx, however, also reveals why the technique of limiting solutions can be effective: It creates a sense of accomplishment when the answer to a problem is discovered. (Or when you manage to MacGyver your way around it.) An adventure that consists only of Sphinxes who say, “Say any word and you can go ahead.” is a lot less interesting than puzzling out the answer to an actual riddle. (Or, alternatively, slipping on your ring of invisibility and tricking Gollum into leading you to the surface.)


In addition to limiting solutions, you can also limit information. The classic example of this, once again, is the clue in a mystery scenario. (Here the information provided by the clue is also the limited solution to the problem of where you need to go to find the next clue, which also demonstrates how much overlap there is between these different categories of chokers.)

But the limiting (or structuring) of information can also encourage PCs to interact with a situation in a specific way. Or create entirely new ways for them to interact with it.

A simple example: The PCs discover information that they can use to blackmail the programming director of Arcadia Television. Without that information, blackmail is impossible. With that information, blackmail becomes possible. (Limiting information can be used to prevent certain choices from being made by withholding knowledge that they’re possible.)

Another example: Someone is being framed for a crime. If the PCs are aware of the character’s innocence when they begin their investigation, that’s a very different scenario than if the PCs are ignorant of it.

Let’s take a more complicated example. Imagine a simple hexcrawl in which the small village of Laciton is surrounded by six hexes. These hexes are keyed with:

  • The ruins of the Keep of the Dracolich
  • A bleakened fairy ring
  • A mammoth cave system used by goblin marauders
  • A lake that’s home to a prophetic sylph
  • A tower haunted by “ghosts” who are actually dimensionally-displaced wizards
  • The dreamgroves of the ash-scarred ents

In one version of this hexcrawl, the PCs can go to the center of Laciton and see a billboard that lists all six of the interesting locations surrounding the village. There is no choker and the information is freely available: They can simply choose which location they want to investigate first.

In another version of this hexcrawl, the PCs can ask around town. The villagers know that there’s a ruined keep west of town. They know that goblin marauders are attacking the outlying farms, but they don’t know where they’re coming from. They also warn people away from the “haunted tower”. In this campaign, although the default hexcrawl structures still make it possible for the PCs to visit any of the six locations they want, each limited piece of information creates an arrow which points them preferentially towards the things they know about.


In the blackmail example above, the information was a necessary tool and the withholding of that information necessarily removed the ability to make certain choices. Information, however, is not the only resource which can be required in order to make a particular choice possible.

For example, the PCs might want to blow up the demon with a bazooka. But if they don’t have a bazooka, they obviously can’t do that. Similarly, 1st level D&D characters generally don’t have access to a teleport spell, so they won’t be able to teleport into Skull Mountain.

This choker can also appear during play if the PCs are deprived of a resource: Their swords can be sundered. Their hotel rooms plundered. Their evidence confiscated. Their informants killed.

What I generally recommend when it comes to a scarcity of resources, however, is that it is almost always the right course of action to make it possible for the PCs to gain the resources they want. That doesn’t have to be easy, of course: When Buffy the Vampire Slayer and the Scooby Gang wanted a bazooka to blow up a demon, they had to leverage their existing resources (Xander’s vestigial memories of being an enchanted soldier) in order to stage a difficult raid on a military armory.

If the resource they want simply doesn’t exist, is there some way that it can be created? For example, there might not be any blackmail available on the local politician because he’s actually a pretty decent guy. That doesn’t mean they can’t set up a honeytrap and create the blackmail material they need.


When the resource being limited is specifically time, you’ve created a deadline.

The limitation of time has the practical effect of taking certain options off the table. For example, if you know the Priests of Orcus are sacrificing Lady Karna at the stroke of midnight, then the option of performing a tactical retreat and coming back in the morning after you’ve had a chance to recuperate is no longer viable. (Assuming you want to save her, of course.)

However, you can also use deadlines to make choices more meaningful: When the Joker tells Batman that he only has time to save one of the hostages from dying in a gasoline explosion, that creates a crucible for the character which reveals deep truths. (Although sometimes what’s revealed is that you can outsmart the Kobyashi Maru and save everybody. Don’t get so attached to your crucibles that you start negating player choices that would circumvent them.)


The opposite of a limited resource is a unique reward which can be only be gained in a specific way.

A pile of gold coins is not a unique reward; there are a lot of gold coins out there. Even the bazooka from the local armory isn’t a unique reward because, again, there are other bazookas out there.

Having to go to the Lady of the Lake in order to claim Excalibur? That’s a choker.

Like the Sphinx and its riddle, clever PCs may be able to find other methods of obtaining the reward. (And an Arthurian campaign where King Arthur murders the Lady of the Lake and steals her sword is definitely going to be a fascinating iteration of the legend.) But, generally speaking, the unique reward is a giant carrot that says, “Come over and do this really cool thing so that you can get this really awesome reward.”


An external event is one which cannot be anticipated (or prevented) by the PCs because it originates from outside the domain of their experience.

For example, the Red Dragon Gang decides to put a hit out on the PCs. You make a note in your prep documents that on November 18th dragon ninjas are going to track the PCs down and attack them.

If the PCs were previously interacting with the Red Dragon Gang (or, possibly, just aware of them) this is not an external event: The attack could theoretically be prevented if the PCs wipe the gang out before the 18th or negotiate a truce with them or fake their own deaths or just coincidentally kill the dragon ninjas who were supposed to be attacking them on the 18th.

If the PCs were NOT previously interacting with the gang, however, it’s an external event. And it’s a choker because there’s no functional way for the PCs to avoid the attack. (Hypothetically, of course, they might have just left town or something. Which is why this is a choker and not a railroad.)

When taken to an extreme – when the PCs are subjected to an endless sequence of external events over which they can have no influence or control – these chokers can be hideously frustrating. In practice, fortunately, that’s very unlikely: The reaction to being ambushed by dragon ninjas is generally going to be figuring out who sent the dragon ninjas, which immediately gets the PCs involved in events over which they do have control.

And external events can be incredibly useful to the GM because they automatically provide the certainty which simplifies smart prep: Since the PCs basically can’t avoid them, the GM can assume they happen without needing to spend a lot of time on contingency plans. They are bangs that can be carefully incubated and then unleashed at the perfect moment.

External events, properly implemented, can also be very effective in providing a larger structure for the campaign. My Ptolus: In the Shadow of the Spire campaign, for example, started with the PCs waking up after losing two years of their memories. In preparing the campaign, I knew that Act II would be triggered when someone they had hired to find a magical artifact during their period of amnesia returned to tell them that the artifact had been located.

With that external event in my pocket, I could confidently build the interlocking scenarios of Act II without needing to worry about whether or not a specific outcome would emerge from Act I.

It should be noted that this didn’t mean that the events of Act I were irrelevant. Quite the contrary: Ninety sessions into Act II, the decisions they made in Act I continue to resonate daily in the campaign. Intriguingly, this is largely because the decisions they made created a rich tapestry of chokers: They chose alliances and they rejected others. They destroyed some enemies and allowed others to escape. They learned some secrets and gave others away before they knew what they had. All of those decisions limited their resources and the information that they held, which has had a deep impact on how they’ve been able to approach Act II and the problems it presents.


The Czege Principle maintains that, “When one person is the author of both the character’s adversity and its resolution, play isn’t fun.”

This is a principle which applies equally to roleplaying games, storytelling games, and improvisational theater. In the case of roleplaying games, it generally means two things:

First, railroads are boring. The GM creates the adversity and they also create the resolution of the adversity. (Then they force the players into acting out the resolution they’ve already created.)

Second, chokers are necessary. Chokers are either the means by which the GM creates adversity or the result of that adversity in play. A campaign without chokers is a featureless expanse in which the PCs face no adversity of which the players are not the ultimate architect. Here, too, adversity and solution flow from a single spring and the result is lusterless.

Avoiding the railroad does not mean that the GM must abandon their creative agenda. Quite the opposite, in fact. The GM must create richly and they must create deeply, so that when their creations meet the creativity of their players there will be greatness born in the clash of titanic ideas.

Addendum: Random Railroads

Go to Part 1

There’s been a recent memetic trend of attempting to defend the idea of railroading your PCs by, first, claiming that a bunch of stuff that isn’t railroading is a “railroad” and then using that claim to conclude that railroading isn’t bad and we should all just stop worrying about it.

It’s as if I said, “Racism isn’t all bad. After all, apple pies are totally racist. But they taste delicious, right?” It sounds perfectly reasonable until you notice that the bit about apple pies being racist is bullshit.

One thing to keep in mind is that the metaphorical use of the verb “railroad” is not a piece of terminology unique to RPGs. It’s an English word dating to 1884 that means, “To force someone into doing something by rushing or coercing them.” Whenever you see someone trying to defend railroading by claiming that it includes a bunch of techniques that don’t include forcing the players to do something they don’t want to do, they’re abusing the terminology. The term “railroading” in this context has an inherently negative connotation, and it’s simply not useful to attempt to redefine the English language so that you can use the term “railroad” to describe some completely different technique that isn’t railroading.

What would be useful, on the other hand, is a new term that we could use to describe the body of techniques which limit player choice. Fortunately, Zak S. has already created it: Chokers. (So named because they create chokepoints in your scenario design.)

Limiting player choice, you’ll note, is distinctly different from negating their choices. While chokers can certainly be abused in order to constrain choice to the point where the GM is enforcing their preconceived outcome by default, imposing creative restraint upon a given situation creates interest and variety.

For those particularly paranoid about limiting the choices available to a PC, it may be useful to note that these chokers will also appear organically and spontaneously through play. The natural world, after all, often surrounds us by barriers. (For example, I am currently sitting in my office. The only way I can easily leave my office is through the door. I could also climb out the window. I might be able to hack my way through the wall, but that’s unlikely. The nature of the room has limited my choices.)

Being aware of these chokers will not only allow you to take advantage of their strengths, but also avoid their perils. The Three Clue Rule is an example of that: A breadcrumb-style mystery scenario is formed from a series of chokers, with each clue providing a very narrow and very specific path that leads to the next clue. The Three Clue Rule simply adds more clues, removing the choker and allowing the scenario to proceed smoothly.


Map of Faerun - Near Waterdeep

Let’s say that the PCs want to travel from Waterdeep to Neverwinter. Why they want to do this is largely irrelevant: Maybe they just learned that the man who killed their father has fled to Neverwinter. Or they’re traders who’ve heard that their Tethyrian textiles will fetch a good price there. Or maybe they just randomly decided to go there at the end of the last session.

So they look at the map and they say, “Hey. Let’s take the road.”

The road is a natural choker, narrowing their experiences to a sequence of literally linear events. For example, you could look at this journey from Waterdeep to Neverwinter and simply prepare it as:

  • As they enter the Sword Mountains, they’re attacked by the Blood Shield Bandits.
  • Black water kappa will emerge from the Mere of Dead Men as they pass it.
  • North of the Mere of Dead Men they’ll encounter Benedict, an itinerant merchant with a beautiful daughter. The axle of their wagon has broken.

Insofar as the road is the natural choice for traveling from Waterdeep to Neverwinter, the choice of going to Neverwinter triggers this specific sequence of events.

Of course, one can loosen the choker if the PCs can choose from a variety of routes. For example, instead of taking the road from Waterdeep to Neverwinter, the PCs could book travel on a ship. (Of course, the journey by ship merely presents a different set of linear experiences.)

It should be noted, however, that the mere existence of multiple routes is often a false choice unless the choice between routes includes an incomparable difference. For example, if the PCs are concerned with speed and the only difference between the journey by sea and the journey by road is the amount of time they take, then there is no real choice: There’s only a calculation of which route will get them there faster. (See The Importance of Choice for a complete discussion of this.)

What I think is interesting about literal roads, however, is that they usually contain an incomparable choice by their very nature. For example, imagine that the PCs are going from Waterdeep to Triboar. There’s only one road, so that’s the way they have to go, right? Except, of course, they could also choose to journey cross-country. The cross-country journey would be slower, but it would also mean avoiding whatever troubles there might be on the road. (The choice between speed and avoiding the events on the road is the incomparable.)

Of course, not all roads are literal roads. Another metaphor for this sort of choker is a bullet in flight: The PCs load themselves into a gun and then pull the trigger. The choice of pulling the trigger is theirs, but once they have done so their flight will logically and naturally take them through a sequence of experiences. As long as they choose to remain upon the bullet, the bullet will continue to fly. You can either take this opportunity to show them some awesome stuff midflight or you can challenge them in order to make the decision to stay on the bullet a meaningful one.

A third way of understanding this choker is “looking out the train window”. Once they’ve made the decision to get on the train, it’s not necessarily impossible to get off before the next stop, but they’d have to put considerable effort into doing so. The metaphor of the train, however, also highlights the ability to create a rich scenario full of complex choices on the train: There may be a linear sequence of events rushing past the windows in the background, but that doesn’t need to be true of the immediate experiences of the PCs. (Maiden Voyage, a D20 adventure by Mike Mearls, is an excellent example of this type of scenario.)

Where you can get yourself into trouble is assuming that the PCs will take a road when they have no need to do so. (Or, worse yet, strong reasons for not taking the road.) For example, the Horror on the Orient Express campaign for Call of Cthulhu ironically makes it a bad idea for the PCs to take the Orient Express. (And then doubles down later in the campaign by, in fact, making it quite difficult for the PCs to stay on the train.)


A mechanical gate exists whenever the PCs need to succeed on an action check in order to do a thing or go to a place or otherwise have a particular experience.

The simplest example of this is a secret door in a dungeon:

If the PCs fail to detect the secret door in Area B, then they will never discover Area C.

A mechanical gate like this is only a problem, of course, if Area C is of vital importance. For example, if Area C is the only place you can find the cure for the Princess’ disease. If the content in Area C is non-essential, on the other hand, then it simply serves as an optional reward for PCs who successfully leap the mechanical hurdle.

A common example of the mechanical gate is the breadcrumb trail approach to mystery scenario design: Every clue is necessary to reach the next section of the scenario, and thus every skill check to find a clue is a mechanical gate which must be passed through in order for the scenario to succeed. (This is the broken scenario structure which the Three Clue Rule seeks to fix. It’s like adding additional routes by which the secret chamber in the dungeon can be found.)

Go to Part 5: More Chokers

Go to Part 1

Most of the problems that arise from railroading are the direct result of negating the primary strength of the RPG as a medium: Player choice.

Exploring the full scope of how player choice acts as the elemental ingredient of the RPG is beyond the scope of this manifesto, but the short version looks like this: For narratives with predetermined outcomes, literally every other medium is better than the RPG. For interactive narratives that tightly constrain player choice, computer games are clearly superior (including graphics and soundtracks and smoother implementation of more complicated mechanical structures). Non-interactive mediums, on the other hand, benefit from the polish of tight creative control.

What tabletop RPGs have going for them is the alchemy of player agency. Of presenting a situation and seeing what happens when a unique set of players make a unique set of decisions and produce a unique set of outcomes. When you railroad your players, you specifically set yourself at odds with the very thing that makes playing an RPG worthwhile in the first place.

When they’re confronted with a railroad, players will often exclaim, “Gimme a ring when you finish writing your novel! I’ll enjoy it more!” And they’re not wrong.

But that’s not the only problem with railroading. Railroading creates a whole penumbra of problems.

When a GM predetermines what’s going to happen in the game, they become solely responsible for the entire experience. And that’s a ridiculously heavy burden to bear. Are your encounters balanced? Did you include enough “cool stuff” for every player to participate in? Did you incorporate enough elements from each PC’s back story? The list goes on and on.

This is how you end up with GMs stringing together precariously balanced My Precious Encounters™ in a desperate juggling act as they try to keep all of their players happy.

When you allow the players to make their own decisions, all of the pressure and responsibility melts away: They’ll choose the fights they can win. They’ll approach situations in ways that let them do cool stuff. If there’s not enough stuff from their back story seeking them out, then they’ll go looking for it.

There are a couple of potential speed bumps here:

First, the players might choose to do something you, as the GM, don’t enjoy doing. I generally haven’t found this to be a problem. Partly because, even without railroading, I still wield a lot of control over what activities are available in the campaign. Mostly because any RPG campaign inherently includes parameters that everybody explicitly or implicitly agrees to before playing: “Let’s all play starfighter pilots!” simply doesn’t result in the players saying, “Okay. Let’s all buy a moisture farm on Tatooine.”

(If this implausible hypothetical were to occur, I would talk to my players about the mismatch of expectations and try to figure out what sort of game they’re looking for in their new careers as moisture farmers. If that’s a game I want to run, then we’ll go for it. If not, then we’ll play something else. Railroading, you’ll notice, wouldn’t actually solve the problem of them wanting to do one thing and me wanting to do something completely different.)

Second, the players might choose to do something that they don’t enjoy doing. This seems outrageous, but it’s also something that happens frequently with groups who are used to being railroaded. (It’s one of the problems living in the railroad’s penumbra.) They’ve been “trained” to see a plot hook and bite it (because otherwise the plot hook will bite them), so when they see a plot hook rich environment they start biting at everything.

If you let them roam free for a bit, however, they’ll work it out of their system. And then you’re liberated, too: You don’t have to try to read their minds. You can just let them do whatever they want with the expectation that they’ll do what they want to.


Another problem in the penumbra of railroading is the effect that it has on game design.

I’ve talked in the past about the way in which the RPG industry and hobby generally doesn’t appreciate the importance of game structures. RPGs are often designed with the assumption that they’re some kind of pure and unadulterated experience. You’ll hear people say things like, “The players tell me what their characters are doing and then we resolve it! They can do anything!”

Except, of course, that’s nonsense. And, as a result, the RPG hobby has mostly rolled along on the basis of the four major structures that were developed in the first half decade or so of the industry: Railroading, dungeoncrawling, mystery scenarios, and combat. (Recently hexcrawling, as a fifth structure, has been making a resurgence.)

And at most tables things are even more constrained. Many ostensible mystery scenarios are poorly designed and actually fall back into railroading structures. D&D’s communication of dungeoncrawling procedures became anorexic in 1989 and literally disappeared entirely from the 4th Edition rulebooks.

I suspect that the result is that the vast majority of RPG sessions are built around a core structure of railroading and combat.

And this is why I think so many players like jumping into combat. Most combat systems feature robust mechanical systems with clear-cut win/loss conditions, which means that it’s the one place where most railroad GMs finally give their players the freedom to make meaningful choices.

This, in turn, is why the vast majority of RPGs are predominantly based around combat. And that will probably remain true until the industry collectively realizes that a universal mechanic doesn’t meaningfully equate to “You can do anything!” – particularly on the macro-level of scenario design.


… is not a sandbox.

That’s a common mistake, though, so let me unpack it a bit.

An RPG sandbox exists when the players can either choose or define what the next scenario is going to be. In other words, you get sandbox play when the entire world is designed as a situation, allowing the players to decide what their next adventure will be.

The opposite of a sandbox is saying, “I brought Keep on the Borderlands tonight, so that’s what we’re doing.” It’s the prototypical campaign where the GM comes prepared with a specific scenario for the game session and the players are expected to play through that scenario.

And this is where we run into the problem with treating “sandbox” as the opposite of “railroad”: Most people would consider “the GM has a scenario and the players are expected to play it” to be extremely light railroading (if they consider it railroading at all). In other words, I think the severity of a railroad is perceived to increase from the outside in: Predetermining that a particular scenario is going to be played is very light railroading. Predetermining the sequence of encounters is heavier railroading, but not as severe as predetermining the exact outcomes of those encounters ahead of time.

So when we cast “sandbox” and “railroad” as antonyms, we actually end up treating the lightest form of railroading as if it were the most extreme form of railroading. And, in response, the meaning of “sandbox” gets warped towards meaning “anything that isn’t linear”. Neither distortion is useful, with the former radicalizing our understanding of railroads and the latter eradicating the unique utility of the term “sandbox” by turning it into a synonym of “non-linear design”.

The opposite of the railroad is actually “default to yes”, which is a concept I’ll be discussing at more length in the Art of Rulings. Instead of defaulting to a rejection of everything proposed by the players, it’s the state in which players are free to make any choice and for the consequences of those choices to be fully explored.

Go to Part 4: Chokers

Go to Part 1

Railroading can take myriad of forms, but it tends to boil down into a few major methods.

ENFORCING FAILURE: The PCs cast a teleport spell to bypass the frost giants guarding the Necromancer’s inner sanctum (that’s the player choice), but you want them to meet the Frost Giant King (that’s the preconceived outcome), so you decide to add a teleport interdiction field to the inner sanctum (that’s enforcing failure). The wizard casts a finger of death on the Frost Giant King (that’s the player choice), but you need the king to escape so that he can wage war against the human kingdoms in Act III (that’s the preconceived outcome), so you fudge his saving throw to keep him alive (that’s the enforced failure).

In its extreme form, enforcing failure becomes pixelbitching: Absolutely nothing works except for the precise preconceived path that the GM envisioned for the adventure, so the PCs are left to fail and fail and fail again until they finally guess what the GM wants them to do. (The term comes from poorly designed computer adventure games where players had to guess which specific portion King's Quest V - Sierra Gamesof the screen the designer wanted them to click on.)

Enforcing failure can also crop up in adventure design, but it’s a gray area: Bank vaults are supposed to be difficult to steal from. It’s not railroading if you design a bank vault with a big heavy door and a really difficult lock. It’s not even enough to look at the intention behind those design decisions (because, of course, your intention with the big heavy door and the really difficult lock is, in fact, to prevent the PCs from accessing the vault in certain ways). You have to look at the totality of the design here: If your scenario is designed in such a way that the predesigned world is constantly funneling the PCs into a single course of action, you’ve probably got a problem.

Generally speaking, designing your scenario from the perspective of the game world (instead of from the perspective of how to control your PCs and guarantee specific outcomes) generally solves the problem.

Another gray area arises when failure is enforced on a macro-scale: If the PCs successfully take an action that should prevent a specific event from happening, but then the event happens anyway, that’s a form of enforcing failure. For example, let’s say that the PCs are serving as bodyguards for Lord Harrow. Lady Karna sends a squad of orcs to kidnap Lord Harrow and the PCs defeat them. But you really need Lord Harrow kidnapped, so you decide to add another squad of orcs who pop out and grab him. That’s railroading.

But if you’re dealing with smart, active opposition that are trying to make a specific event happen, the fact that they keep trying to make it happen even after you’ve stopped them once isn’t railroading. For example, if Lady Karna responds to the PCs defeating her orcs by sending a different kidnapping team the next day that’s not railroading.

FALSE CHOICE: “We take the corridor on the right.” “You find the Altar of Despair.” [REWIND] “We take the corridor on the left.” “You find the Altar of Despair.”

This is similar to the magician’s choice: The PCs appear to have a free choice between multiple options, but no matter which choice they make the GM forces the outcome he wanted the PCs to choose. (For example, the magician asks the spectator to choose between two cards. If the spectator chooses card A he says, “Flip that card over!” If the spectator chooses card B he says, “Okay, we put that card back in the deck. Flip the other card over.”)

Location-based false choices are quite common: No matter where you go, you end up in the same place.

Event-based false choices are also quite prevalent, although they can run into the gray area of smart opposition. (Lady Karna’s kidnapping teams finding Lord Harrow a second time is railroading if you’ve simply negated whatever security precautions the PCs have made to prevent that from happening. But Lady Karna really wants to find Lord Harrow, so it’s reasonable for her to have invested the cash in a scrying spell if the PCs have exhausted all of her mundane efforts to locate him.)

Another variant of the false choice is the timeline that doesn’t matter. In published scenarios, this seems to often show up in the form of, “No matter how long it takes the PCs to reach the Lost City of Bakkanar, the bad guys have gotten there just before them.” Here, though, you can see the gray area between enforcing failures and false choices: Is this a false choice because it doesn’t matter whether they travel by ship or horse or teleport spell? Or are you enforcing failure by claiming that their effort to use a teleport spell to outpace the bad guys automatically doesn’t work? It doesn’t really matter, of course. It’s railroading either way.

PROMPTING: Prompting is not, strictly speaking, a method of railroading, but it’s a related technique. It basically boils down to the GM making suggestions about what the PCs should be doing, either directly or indirectly. A common variant is the GM NPC who tags along with the group and either tells the PCs what to do or hints at it. GMs can also prompt in the metagame by simply telling the players what their characters are supposed to be doing.

A more complicated form of prompting is simply having negative consequences from failing to take a particular action: Something horrible happens, informing the PCs that they should have taken action to prevent it. If they still have the opportunity to undo it (or stop it from becoming worse), then that’s serving the same function as a nudge. For example, they hear a rumor that there’s a Necromancer in the forest that’s using a bane bone to create skeletons. They ignore it and a week later a group of skeletons attacks the village. They continue to ignore it and the skeletal attacks become larger and more frequent until, finally, they return from the dungeon one day to discover that the Necromancer has taken over the whole village and reared a tower of bone in the town square.

Some prompting has a generally negative effect on the game (because it tends to influence or corrupt the players’ decision-making process). Other prompting has a generally positive effect on the game. But prompting, by itself, isn’t railroading.

Where it intersects railroading, however, is when the GM is using the prompt to tell the players what the GM’s preconceived course of action is. The GM is basically saying, “You should be doing this. If you try anything else, it’s going to fail because this is the thing that you should be doing.”

MIND CONTROL: Mind control is a special case. When a PC has their mind controlled, the GM isn’t negating the outcome of a choice. Instead, they are removing the player’s agency entirely. It’s a complicated issue that deserves a dedicated discussion of its own.

As far as railroading is concerned, my opinion is that mind control is not necessarily railroading in much the same way that failing to hit an ogre with your sword isn’t. Where it becomes railroading is if the GM uses mind control in order to create a preconceived outcome. (Although here, again, there’s a legitimate gray area: It’s not unreasonable for Lady Karna to use a dominate person spell on a PC in order to have them lead her kidnapping squad to Lord Harrow’s hiding place.)

If I was looking for a litmus test here, it would be the GM’s willingness to fudge the PC’s saving throw in order to make sure the mind control happened. That’s a dead giveaway that they’re committed to their preconception.

GUARANTEEING SUCCESS: Is it railroading to guarantee that the PCs succeed at what they want to succeed at?

I don’t necessarily have a satisfactory answer to that question. For the sake of argument, let’s lay aside the issue of game mechanics that guarantee success for the moment (whether that’s universal mechanics like GUMSHOE’s automatic clue-finding or specific instances like a character who’s skill is so high that they’ll automatically succeed on a particular task) and simplify the discussion by simply talking about a GM fudging a die roll to make a PC succeed where they should have failed.

Is that railroading?

My gut says that it isn’t. There is a fundamental difference in kind between negating a player’s choice and enabling that choice.

With that being said, however, the guaranteeing of success does share a number of negative traits in common with railroading: It flattens the gaming experience by making it more predictable. It removes complexity and interest from the scenario. And it can significantly warp the decision-making process of the players.

This becomes particularly true if the GM shows favoritism towards enabling only certain types of decisions. (Which should become really obvious if you consider that one type of decision that can be selectively enabled is “decisions which lead to my preconceived outcome”.)


A categorical distinction is often drawn between visible railroads (where the players can see the tracks) and invisible railroads (where the mechanisms are hidden from them). Invisible railroads are sometimes referred to as illusionism (referring to the fact that the players only have the illusion of free choice), and the simplest example would be fudging a die roll behind the GM’s screen. In any case, the difference between a visible railroad and invisible railroad is more of a spectrum.

Invisible railroads are often invoked as another excuse for railroading:

It doesn’t matter if the players don’t know they’re being railroaded!

In practice, however, I would note a couple of things.

First, GMs tend to overestimate the degree to which their players don’t notice their railroads. Lots of players are polite enough not to pull back the curtain on the Wizard of Oz, but that doesn’t mean that they don’t see his feet poking out from behind the curtain.

Second, the majority of negative effects created by a railroad exist whether the players are aware of the railroad or not.

Go to Part 3: Penumbra of Problems



Recent Posts

Recent Comments