The Alexandrian

The Railroading Manifesto

March 13th, 2015

Railroad Tracks - Ha Tay

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

Note, however, that both parts of this equation are important: The choice must be negated and the reason it’s being negated is because the GM is trying to create a specific outcome. The players must try to get off the train and the GM has to lock the doors.

A simple failure to achieve a desired outcome is not railroading: If the doors are unlocked, but the players can’t figure out how the door handles work that’s not railroading. For example, a player might want to hit an ogre with his sword. If he fails his attack roll, that’s not railroading. (If the GM secretly changes the ogre’s AC so that the PC misses, that’s railroading.) If the PC tries to break down an adamantine door with a fluffy pillow, that’s still not railroading even if the GM says they have no chance of success.

It’s also not railroading if the GM has a preconceived outcome, but doesn’t negate player choices in order to make it happen. As an extreme example, consider a campaign where the PCs are FBI agents in New York during World War II. On May 2nd, 1945, the newspaper headlines declare that Adolf Hitler died on April 30th. The GM, of course, knew that Hitler was going to die on April 30th long before it happened, but the newspaper headlines are not railroading the PCs.

The same remains true on a more intimate level: The GM might make a note that the beautiful dame Jane Adams is going to contact one of the FBI agents on May 15th with information about a KGB operation targeting Manhattan Project scientists. Unless the PC deliberately goes into hiding for some reason, it’s still not railroading when Jane Adams shows up.

Finally, choices having consequences is also not railroading. If a PC punches somebody in the nose and then they punch the PC back, that’s not railroading. If a player says, “I’m going to hop on I-94 and drive from Minneapolis to Chicago.” Then it’s not railroading when the GM says, “Along the way, you pass through Eau Claire.”

In fact, choices having consequences is the exact opposite of a railroad. Railroading makes a choice meaningless. Consequences make a choice meaningful.

(Of course, not every consequence is a negative one: If the PCs piss off the Red Dragon Gang, the gang might retaliate. But it’s also possible that the PCs might be given a medal by the mayor who also asks them to do a favor for him. Or they might be contacted by the Red Dragon Gang’s rivals who want to hire them as enforcers. And so forth. None of that is railroading.)


Railroading, in the purest sense of the term, is something that happens at the gaming table: It is the precise moment at which the GM negates a player’s choice.

In practice, of course, the term has bled over into scenario prep. We talk about “railroaded adventures” all the time, by which we generally mean linear scenarios which are designed around the assumption that the PCs will make specific choices at specific points in order to reach the next part of the scenario. If the PCs don’t make those choices, then the GM has to railroad them in order to continue using the scenario as it was designed.

However, not all linear design was created equal. And it’s not really accurate to describe all linear scenario design as being a “railroad”.

Linear scenarios are built around a predetermined sequence of events and/or outcomes.

Consider a simple mystery:

Scene 1: The PCs come home and discover that their house has been broken into and an arcane relic stolen from their safe. They need to figure out who did it, which they can do by analyzing fingerprints, looking at their neighbor’s surveillance camera, asking questions around town to see who took the job, or casting a divination spell.

Scene 2: Having discovered that Jimmy “Fast-Fingers” Hall was responsible for the break-in, the PCs track him down. They need to figure out who hired him, which they can do by interrogating him, following him, analyzing his bank statements to figure out who paid him, or hacking his e-mail.

Scene 3: Having discovered that Bobby Churchill, a local mob boss, was the guy who hired Jimmy, the PCs need to get their relic back. They can do that by beating Bobby up, agreeing to do a job for him, or staging a covert heist to get it out of his vault.

That’s a fairly linear scenario: House to Jimmy to Bobby. But because we used the Three Clue Rule to provide a multitude of paths from one event to the next, it’s very unlikely that a GM running this scenario will need to railroad his players. The sequence of events is predetermined, but the outcome of each scene is not.

Non-linear scenarios do not require specific outcomes or events, allowing freedom of player choice.

Linear scenario design and non-linear scenario design exist on a spectrum. Generally speaking, requiring specific events (“you meet an ogre in the woods”) is less restrictive than requiring specific outcomes (“you meet an ogre in the woods and you have to fight him”). And the more specific the outcome required, the more likely it is that the GM will have to railroad the players to make it happen (“you meet an ogre in the woods and you have to fight him and the killing blow has to be delivered by the Rose Spear of Vallundria so that the ogre’s ghost can come back and serve the PC at the Black Gates of Goblin Doom”).

With that being said, it’s often quite trivial for an experienced GM to safely assume that a specific event or outcome is going to happen. For example, if a typical group of heroic PCs are riding along a road and they see a young boy being chased by goblins it’s probably a pretty safe bet that they’ll take action to rescue the boy. The more likely a particular outcome is, the more secure you are in simply assuming that it will happen. That doesn’t mean your scenario is railroaded, it just means you’re engaging in smart prep.

My point here is that you can’t let fear of a potential railroad make you throw away your common sense when it comes to prioritizing your prep. This, by the way, leads to one of the most potent tools in the GM’s arsenal:

What are you planning to do next session?

It’s a simple question, but the answer obviously gives you certainty. It lets you focus your prep with extreme accuracy because you can make very specific predictions about what your players are going to do and those predictions will also be incredibly likely to happen.

Where you get into trouble is when your scenario expects something which is both very specific and also very unlikely.

For example, in the Witchfire Trilogy from Privateer Press, there’s a moment where the PCs have all the information necessary to realize that a specific NPC is the bad guy they’ve been looking for. This makes it incredibly likely that the PCs will simply confront the bad guy. The author doesn’t want that, though: He wants the PCs to put her under surveillance and trail her back to her secret hideout. So he throws up a bunch of painfully contrived roadblocks in an effort to stop the PCs from doing the thing they are nevertheless overwhelmingly likely to do. (So You Want to Write a Railroad? is an almost endless litany of even more egregious design failures from another published scenario.)


Another way of thinking about this is that the more specific and unlikely the necessary outcome, the more fragile your scenario becomes: It will break if the PCs deviate even slightly from your predetermined sequence. Once the scenario breaks, you’ll have to resort to railroading in order to fix it. This is why I often refer to railroading as a broken technique seeking to fix a broken scenario.

It’s fairly typical, for example, to hear someone say, “I only railroad my players if it’s really important.” And when you delve a little deeper, you virtually always discover that “really important” is a synonym for the GM making sure their predetermined outcome happens. These are literally people saying that they need to railroad because they designed a railroad.

Another common rationalization for railroading is that GMs have to use it in order to keep problem players in line. However, if your relationship with your players is that they’re naughty children who are testing their limits and you’re a parental figure that somehow needs to keep them in line, then your relationship with your players is fundamentally broken. More generally, what you’re talking about are issues outside of the game. Attempting to handle those issues with in-game behavior modifications is simply dysfunctional. It’s no different than if a player at your table was cheating or if they poured a drink over the head of another player: These are all problems which require intercession. But none of them are going to be solved through railroading.

One specific example of this is often cited as an exception, however: Behavior which is deliberately disruptive through the agency of the game world. For example, the guy who tries to assassinate the king when the PCs are called in for an audience. Ultimately, however, this example only cycles back to the previous two: Either the guy involved is a jackass (which is a problem that needs to be solved outside of the game) or this is really only a “problem” insofar as it disrupts your preconceived notion of how the royal audience was supposed to play out (which means we’ve arrived back at “I need to railroad them in order to maintain my railroad”).

(Note, too, how often these “problems” can quickly be solved by having the game world respond naturally to the circumstances: Crazy McGee has just assassinated the king. What happens next? Well, the king’s guard is going to try to arrest them. If they escape, there’s going to be a manhunt. Then there’s going to be a power struggle to fill the vacuum. The other PCs need to decide whether to help hunt down their former comrade or help him escape. There may be a rebel group who concludes that the PCs are on their side because of the assassination. And so forth. That all sounds like interesting stuff.)

Nobody minds the railroad if the destination is Awesome Town!

The theory here is that if you offer a big enough carrot, nobody will mind being hit by the stick a few times.

There’s a fair amount of truth to that, but what always strikes me about this popular meme is the extraordinary amount of hubris it demonstrates. See, any time that a player chooses to do something, that implicitly means that it’s something that they want. That doesn’t necessarily mean that they should automatically succeed at everything they attempt, but if you’re artificially negating their choice in order to enforce your preconceived outcome, what you’re saying is, “I know what you want better than you do.”

Which might be true. But I’m willing to bet that 99 times out of 100, it isn’t.

The railroad creates specific situations. The goal is to see how the PCs react to those specific situations.

This is a more nuanced and deliberate application of railroading techniques. The idea is that the choices you’re interested in are those made in specific moments. The methods by which individual moments are reached are of less interest, and, in fact, it’s more important to create specific moments of particular effectiveness than it is to enable choice outside of those moments. You’re basically stripping out the strategic choices of the players in order to create intense tactical experiences.

In practice, however, railroads warp the decision-making process of the players. When you systematically strip meaningful choice from them, they stop making choices and instead start looking for the railroad tracks.

So railroading PCs into a situation to see what choice they’ll make doesn’t actually work: Having robbed them of free agency in order to get them there, you’ve fundamentally altered the dynamic of the situation itself. You’ll no longer see what their reaction is; you’ll only see what they think you want their reaction to be.

I suspect that GMs who habitually railroad have difficulty seeing this warping of the decision-making process because it’s the only thing they’re used to. But it becomes glaringly obvious whenever I get the players they’ve screwed up: Nothing is more incoherent than a player trying to figure out where the railroad is when there’s no railroad to be found.

For example, I had a group who spent all their time trying to figure out which NPC was the GM NPC they were supposed to be following around because that was the method their last GM had used to lead them around by the nose. Since the scenario I was running for them revolved around a conspiracy with multiple factions who were all more than happy to use the PCs to achieve their own agendas the result was… bizarre. (Unfortunately, I only figured out what had gone so horribly wrong in the postmortem.)

Of course, it gets even more obvious once the players start demonstrating Abused Gamer Syndrome.

Go to Part 2: Methods of the Railroad

Share on TumblrTweet about this on TwitterShare on StumbleUponShare on FacebookShare on RedditShare on Google+Digg this

12 Responses to “The Railroading Manifesto”

  1. AMX says:

    Speaking about excuses, what about “This DM just totally sucks at improvisation, so going off the rails results in a crappy game until he’s had time to re-prepare?”

  2. Justin Alexander says:

    Check out GM Improvisation 101. A good quote from it:

    “A lot of times when I hear gamers talk about improvising, it’s spoken of like it’s this super amazing skill that is rare and difficult – like Bruce Lee’s 1 inch punch or something. But here’s the thing: it’s only difficult because most rpg advice for GMs is the exact opposite of improvisation.”

    You might also find Don’t Prep Plots useful.

  3. Brendan says:

    I’ll be honest, I’d like to hear the story about the group that expected there to be a GM NPC. It sounds interesting, in a “gawking at the car crash” sort of way.

  4. AMX says:

    Some of that does look useful, although this bit does not bode well:

    “If you were playing a Batman game, could you take 5 minutes and simply make up a dastardly plan for the Joker? Sure. You get what kind of problems the Joker creates. You can probably make up Joker action and reactions on the spot. It doesn’t have to be perfect, but it’s easy enough to come up with something.”

    No. I don’t understand the Joker – I have no idea what he would want, or how he would go about achieving it.
    I could probably mine old issues for ideas, but that’d be the exact opposite of improvising.

  5. Justin Alexander says:

    @Brendan: Bumbling in Freeport. Meant to include a link to that, actually.

  6. Justin Alexander says:

    @AMX: I actually had the same reaction to that bit about the Joker! I’d feel pretty comfortable playing him from moment-to-moment, but actually concocting one of his schemes is not something I’d generally want to attempt on the fly. (If I absolutely needed to, you could certainly boil a fair portion of his schemes down to “poison X with Joker gas”. Filling in an arbitrary value for X is relatively easy; the problem I’d have is coming up with a scheme that would feel uniquely Joker-ish to accomplish that.)

    What I think that really means is that the Joker is not a great Source of Problems for you and me.

    This comes back to a concept I’ve referred to in the past as smart prep: You want to create a toolbox of stuff that you can play in the moment (the same way that the players are playing their characters). That step is pretty similar to the Source of Problems concept. Then you want to focus your prep on stuff that can’t be reliably improvised during play while adding a high value to play.

    What constitutes “can’t be reliably improvised during play” will vary from GM to GM and also from system to system. One example that’s pretty universally true, though, would be an awesome prop or handout.

    The other thing to consider is that improvisation is a skill. And, like any skill, it can be practiced. For example, you could take a city supplement like San Angelo: City of Heroes, flip through it select a random location, and then use the “poison X with Joker Gas” construction to give you the seed of the Joker’s scheme. Then see if you can spin off 4-5 cool or twisted or ironic ways that the Joker could do that in under 2 minutes. (It’ll be useful to think about why the Joker wants to poison this target.)

  7. Brotherwilli says:

    Your comments about the Joker made me think about the difficulty of playing the “mad genius” villain, which in turn made me think about some advice about playing genius villains. I don’t recall what supplement it was, but I think it was either a Dragon Magazine towards the end of its run at TSR or the Eye Tyrant Book. The gist of the advice was that while a DM during preparation may not be a genius and may not anticipate the PC’s plans, the genius villain would do so. The DM, therefore, could retroactively hold that the villain had prepared the right counter to some or all of the PC’s plans, because the villain anticipated them.

    It struck my young self at the time that players would not like that one bit. This advice may not fit perfectly with your definition above, as it is not forcing a preordained outcome (the PCs may still win the fight), but it still smacks of railroading.

    I am curious, however, if you have any other thoughts on playing genius villains – those with Intelligences so high and experience so deep that a DM couldn’t possibly match them.

  8. Justin Alexander says:

    I actually thought I’d written a bit about this here at the Alexandrian in the past, but apparently my memory is playing tricks on me.

    What I’ve generally concluded over the years is that players are rarely if ever disappointed when they find an Achilles’ heel to the villain’s plan, whether I intended for that weakness to exist or not. So I generally just let them succeed.

    In situations where that’s less than satisfactory, I find the “retcon planning” solution to be your best bet. But I’d temper it one of two ways:

    First, generally default to the response to the PCs’ plan be an obstacle to their plan and not a complete impediment. For example, if they decide to go into the sewers under the bad guy’s base and use a spell to drill up into the base, then the bad guy’s anticipation of that possibility is to put a bunch of guards in the sewer (which the PCs can defeat) or an alarm spell (which they might be able to detect and disable).

    Second, filter it through a mechanic. How clever is the idea that the players have come up with? Set a DC accordingly and then make an Intelligence check for the bad guy. In Numenera I might have the PCs make an Intellect test opposed by the bad guy’s level (“Did they outwit him?”) and then use a GM intrusion to instantiate his precautions if the check failed.

    Use the two in combination and you’re probably golden.

  9. guest says:

    That article on “abused gamer syndrome” starts off by defining Killer DM as one who “wont do anything to prevent PC death”…

    : /

    Lost me with that.

  10. Anton and Erwin are taking the train (or not) | Spriggan's Den says:

    […] Alexandrian has been writing (a lot) about railroading over the last weeks and I read an older article by the Angry DM from a couple of years back a few […]

  11. The Evil They Do – Strange Flight says:

    […] for the players to intervene in the course of events in meaningful ways, or you risk running a railroad on […]

  12. Gamosopher says:

    About the “railroad creates specific situations”, I read somewhere (can’t remember where; maybe it was a post here that if this is what you want to do, the GM should just quickly describe the stuff happening between those interesting scenes to the players, and then drop them in the situation before asking “ok, what do you do”. This way, it’s clear when the players have agency, and clear what the game is about.

    I think the example was for a series of combat encounters that happen in a linear dungeon; some people were saying that a super linear dungeon was a good way to set up cool combat encounters. Someone else said that at this point, the map is not even needed : at the end of an encounter, if the players decide to go on, just describe the PCs walking in tunnels and whatnot instead of make them go through the motions. It saves time, and if the game you want to play is tactical combat, this way you get more of what you want in a single session.

    It could be taken as some kind of reductio ad absurdum, but to me, that’s a perfectly valid way to play. Of course, if you play like that, the illusion of player agency between combat encounters is broken. I’m under the impression that GMs (and maybe even players) that actually enjoy this kind of game still want to keep this illusion alive.

Leave a Reply



Recent Posts

Recent Comments