2015-03-25 Railroading

I recently had the most amazing conversation on Google+ about railroading. I’m going to try and collect the things I said but if you want to know more, I definitely suggest checking out the thread on G+... Just kidding. Google+ is gone. I’m still angry, Google!

It all started with my recommendation to read Justin Alexander’s blog posts on the same topic: If you roll behind the screen, or if you like to fudge a rare dice roll, or let players always succeed, or like to think that illusionism is ok, i.e. if you think players not knowing makes your tiny railroading nudges ok, then maybe you should read it. I’ve read the first four parts and liked part two and three in particular. Much better than the first part! I suggest skimming part two and reading the bold text. If that doesn’t make you want to read more, then you can skip the articles.

In a reshare of that post somebody started talking about having an interesting location no matter where people went, but making sure that investigating the location was optional. I said that the term many people use for this is illusionism. It’s the illusion of choice. Once players start to suspect, how will you ever regain their trust?

Alternatively, would you agree to tell them openly that there’s an interesting location nearby and they can choose to ignore it? If so, no problem. Skip the part where your players make decisions that have no effect. Skip going through the motions of choosing a road when every road leads to Rome.

Then again, what about improvising a world? Neither I nor the players know what lies to the east. If you go east, I’ll roll some dice and we’ll find out. If the world grows as players explore, that can work – it’s not misleading players – but when I realize that the referee is improvising as we explore, rolling on random tables as we explore, when there is no actual world to explore, then why are we playing a game about exploration? In these cases, I’d prefer the referee to be open about it and then we can agree that the structure is not “explore the map” but “choose our next adventure location” and we can skip the map. The ref can just show up and say, “hey murder hobos, I bought Castle Dragonstein and the 59 Shades of Doom and I think I’m going to run it!” And we’ll be all like “Hell Yeah!”

In D&D, the question of trust is linked to the question of procedures. If I implicitly “promise” to run a game about exploring a map and having random encounters between interesting locations, I’m “betraying” that trust if the map doesn’t exist or if interesting locations move into our path at my whim. Yes, it’s possible to run a game without making this promise but then I need to be open about it and not tell my players that this will be a game of map exploration. If it is agreed at the table (perhaps not even said out loud but simply by a wink and a nod) that we will have a road trip from A to B, then players are not making decisions about where to go next since they are on the road from A to B, and therefore players aren’t bothered by locations coming up in a predetermined sequence.

The problem of trust, the problem of railroading, of invalidating player choice, of illusionism, is when I tell my players that there is a map to be explored but behind the screen, I treat it as a road trip and no matter what they choose, the next location is the one I want there to be.

As Justin Alexander says: “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.” (part two)

Perhaps there exists an algorithm so good that it would be impossible for players to ever discover its existence? Sure. What about procedurally generated content? My idea is this: I’m going to imagine a world for your characters to explore and we’ll pretend that it is real and that actions will have consequences and it’ll be cool. In reality, of course, the world has not always been there and its defined boundaries don’t extend very far. It is understood that I will be adding to the world as we go, that I will do a certain amount of just in time preparation. It is therefore impossible to have a perfect imaginary realm that has always existed and extends further than we’ll ever be able to explore. At the other end of the spectrum is the referee freely improvising the entire world and just saying what they think would make an interesting adventure. Procedural generation is somewhere on this line, preparing 10 minutes before the game starts is further along this line, and so on. And the more I realize this, the more I see that there is no actual world, imaginary or not, to be explored, the more I will be disappointed if and only if exploration was part of the point of our game.

Sure, we can have procedurally generated background for the things that don’t matter. But if I know that you’ll roll on that table of things the dead have in their pockets, looting the dead is no longer interesting. It’s grinding. I want the gold but I don’t really care what they have in their pocket because it’s not really there. We can also have procedurally generated content for parts that don’t matter.

In a Wicked Age is a game where the map doesn’t matter – our games have always been about our goals and relationships. We can have a randomly generated city for In a Wicked Age. But we can’t have randomly generated characters and goals in this game! Because that’s what the game is about!

You might say that players killing a dragon in my game and all of us rolling on the random treasure table is the same kind of procedural content generation and you’d be right. I don’t care, however, because the game is not about the kind of treasure the dragon has. We’re rolling because it’s fun to roll high and be rewarded for it, it’s a mini-game utterly divorced from the rest of the game. Exploring the map is not utterly divorced from the game—or if it is, nobody told me. I’m playing under the assumption that exploring the map makes sense because there is a map to explore. If it is not, I might expect us all to roll for the content of the next hex together, at the table, looking up the results, cheering and weeping as we go and it will be cool. But it’s not the same game as before. I might not enjoy it as much.

That’s where the feeling of betrayal and the question of trust comes in. Do you feel an implicit promise about an existing world was made? Was a promise about a map worth exploring made? If I think there was and it turns out there isn’t, I’m going to wonder what I’ve been doing at the table. That moment of discovery will be a sad one. I’ll be as interested to play as when we tried the Mythic Gamemaster Emulator.

If we implicitly agree to explore a map, then don’t generate the map. Since we also implicitly agreed that there would be wandering monsters, rolling for wandering monsters is not a problem. As you can see, I will not prepare each and every thing the characters may interact with in advance. This is not my point.

The point I’m making is also independent of the rules. Even if the rules said to do the very thing I’m arguing against, that doesn’t change my point. You could then claim that I’m not playing by the rules, but the value judgment still stands, the argument still stands.

This is not about interpreting the rules, this is about describing the kind of procedural game content generation that is OK and the kind of procedural or improvised game content generation is not OK because it frustrates players. It’s also about the nature of this frustration. Why are people frustrated? Maybe not all players, obviously. But why me, for example?

If you say that this is not a problem for you and you don’t see the frustration and neither do your players, then that’s fine. I guess there’s no need to continue this conversation with people that don’t see the signs of a problem.

Rolling on random tables is cool and I do it myself, but only when we all implicitly know what is going on. My players know: there are random monster encounters and my rolls depend on them taking too long and making extra noise. They also know that going in particular directions doesn’t change that, so all they need to decide is whether they want to keep bashing those doors, whether they want to use those silence spells, and they don’t need to decide whether to go left or right as far random encounters are concerned. My main beef is that as a player I feel I’m exploring a map and it matters where I go and latter I discover that the referee was rolling on table for random hexes (it’s better if these hexes stay where they are once rolled up, obviously). But the net effect is like playing endless random dungeons in computer games. I’m not interested, or at least: significantly less interested. But if my GM says, “people, these are the cities and travel between these cities takes a week and every week I’m going to roll on this table and every time you are on the road I’m going to roll on this other table” then that’s perfectly fine. I might still have an opinion about the game (and I might like those tables) but at least I’m not wondering whether to go north or south in order to trigger a particular event. I already know that these will be random and I will act accordingly.

If you’re thinking about players leaving the area you prepared and improvise the area – something I do myself, all the time! – then a first step would be to be open about the situation. I think my players know that they too can step so far away from the prepared stuff that I’ll try my best to improvise and maintain the illusion but we all know that this is not how I want to run the game. I will take steps to avoid this. For example, I’ll ask at the end of the session about plans for the next session and I’ll prepare for that. That works often enough. Not always, though. 🙂

As for switching one monster for another because the one determined by a random table don’t seem exciting enough, my goal would be to use the first monster – e.g. kobolds – because that’s what my encounter table told me but I’d try to make it as entertaining as I can. Are they tough bastards with barrels of flaming oil, Tucker-style? Or perhaps they’ll grovel and beg and promise to lead them to a dungeon or help them ambush a greater common foe? Or perhaps they’ll simply run and we’ll laugh for a few seconds, imagining their antics before moving on.

You might think that you’re doing the best by switching to were rats but it made me think of Oblivion where your surroundings level up with you. It was great at first, but later I was a bit disappointed. I didn’t like it after all. And that’s why my level 1 dungeon remains a level 1 dungeon full of kobolds and I expect my players to delve deeper very quickly and not fight 50 kobolds just because they can. It’s boring and we all know it. To replace every kobold lair with gnolls or hill giants or red dragons ... eventually this method will break down, right?

How did we get here? Why do I assume “promises” and “trust” and why do I feel “betrayed” when railroaded? Where did I get all these ideas if they’re not explicitly in the rules?

When I learned to play, a few of us read the AD&D Dungeon Master’s Guide and tried their hand at running a game. Some of us only ran a single sessions, others ended up DM for life. But reading those rules back then, or reading the blogs these days, and talking about it, and talking about the experiences we had, and the things we enjoyed, I think all of this produces a culture of shared assumption and norms in the circle of people I actually play with. So, the principles we grew up with was:

  1. there is a map
  2. the map has a key
  3. there are monsters and lairs
  4. there are wandering monsters
  5. there are “Saturday Specials” (bosses, ends of campaign arcs, …)
  6. there is random treasure
  7. sometimes treasure gets placed

The kind of railroad we experienced when we started playing were maps that had invisible walls that we could not pass no matter what we tried, monsters that could not be beat no matter what we tried, treasure that could not be found no matter what we tried, non-player characters we could not kill no matter what we tried, and those experiences were intensely frustrating. There are of course lesser evils.

So, this is where the notion of trust and promise and betrayal of expectations comes from. If you want to use different words for it, no problem. The experiences still stand, however. My claim is that where as the game may or may not have claimed something in particular, those were the issues I had and I’m actually not that interested in discussing whether my expectations were justified or not, or whether my words are the precise words to use. If you need to use different words to talk about the same issue that’s fine.

Back to players leaving the area I have prepared for. What if they enter a forest I know nothing about? Is my decision about the encounters there equivalent to springing an encounter I though of no matter in which direction my players leave town? I think the answer to the question of where I stand is that both situations are indicative of a game I don’t like as much as the ideal game I strive for. Do I need to choose between those two evils or can I compare each to my ideals?

If the forest is unknown and I have nothing prepared, then perhaps nothing needs to happen. “You slowly make your way across the forest but there are no incidents and you arrive safely at Oathcomb seven days later.”

Less ideal solutions:

If there is a particular encounter you want your players to have, then that’s fine. After all, sometimes the sun shines and sometimes it rains and that’s not about railroading the players. It’s railroading if the players get to choose things that have no effect. If there is an information indicating that a dragon is waiting for them outside the city and no matter what they do, they’ll meet the dragon. That’s frustrating. If you manage to pull it off such that the players believe they had a choice but blew it, then congratulations, you were lucky once. Every good referee can pull this off every now and then, for a session or two. But sooner or later – or so I claim – the players will grow suspicious and the game will feel a bit more hollow because of it.

Do I want to be surprised by an encounter; do I like using random tables? Yes I do. I like the surprise of random encounters, but only if the players understand that particular kinds of events are determined by random tables and the players like it as well, eg. random treasure, wandering monsters.

Do I mind moving an encounter I had in mind from one road to another? It depends. If the encounter was not centrally related to the kinds of choices players made at the moment – if players think they’re exploring an area and that it matters whether they go north or south, it bothers me to get the same encounter no matter what I do; if a small number of encounters are placed by the referee no matter where players go then that’s fine; if the encounter is placed because it kicks a new campaign arc and is therefore peripheral to the activity of exploring (going north or south) then that’s fine; it’s not fine when I discover that I picked seven directions and realize that I would have gotten the exact same sequence of events had I picked seven different directions. Why did you waste my time and tell me that I needed to pick seven directions if if doesn’t matter what I do. Just tell me that we’re on a road trip and bring it on. It might be a good game without cheating me.

At one point Jason went back to the rules. His point that if something isn’t in the rules, then assuming said thing and realizing that the referee was not providing it – such a map prepared ahead of time – wasn’t “betrayal” because nothing had been “promised”. The discussion continued and he wondered why I was blaming the referees instead of the rules themselves.

To me, this rule based approach is weird and confusing. I’m thinking of the railroading and player choice invalidation problem as problems that are unrelated to any rules. In the abstract, role-playing games provide a framework for players to make decisions and influence the events at the table. This breaks down as more and more decisions made have no effect on the events at the table.

Why was the person doing that? Power dynamics? Lack of empathy? Ignorance? Were they doing so because of the rules? Perhaps. Who cares, though? A bad game is a bad game even if the rules as written say that this is within the rules. I only care about the why and how and improving my game (and the game of people I play with).

Jason then tried to frame it as an equivalence. Both I and the imagined railroading referee I’m complaining about got our values from somewhere. Was I prepared to say something positive about these other values that weren’t mine?

I said that I don’t attribute negative motivation to that other person; instead I attribute bad technique and describe unintended consequences. I believe people when they say they are doing what I am criticizing for the enjoyment of the table. It’s just that I think that they are failing to do the best job they can and based on my own experience, I have a hypothesis for how this comes to be.

Now, given that I believe we’re talking about bad technique, there’s no need to identify positive motivations to that person. “Your backswing isn’t great, I think you’re holding the club to close too your body. If you want to hit that ball, you need to let it drop on the way down.” Bad technique doesn’t need balanced perspectives. Your golfing can be improved, that’s all I’m saying.

Later, Harald had some questions. His first question was about randomly generating content and placing it on the map, and keeping it there, and integrating it with the rest. Is this taking away player agency?

The idea of randomly generating placing content and then keeping it there is cool. But does it provide the best experience at the table if this is how we run the game? I don’t think so. A thought experiment: given the choice between doing (A) what Harald described and (B) providing players with a list of rumors about interesting locations and directions of how to get there, and implicitly acknowledging that on the way to these locations random encounters will happen and some will stay on the map (essentially option A), then I’ll say that option B is clearly superior. I’m making the decision to go to a particular location, I’m making the decision to take a certain amount of risk (based on the distance traveled, based on the regions we’ll travel through), I’m making the decision to continue or to turn back, and these all have consequences. If we’re using just option A, then some of these decisions aren’t as great. I can’t choose to go anywhere in particular. I can’t choose a different route to avoid the Griffin Mountains. All I can do, in effect is to continue or to turn back. Yes, it’s still cool. But it could be even better!

Harald also asked about the difference between randomly generating map content vs. randomly generating the treasure some undead might carry since I had mentioned that I didn’t enjoy looting corpses.

I had the same answer as before. Yes, it can be cool. Let’s call it option A. But it sounds an awful lot like grinding in World of Warcraft (which I have never played so this is based on stuff I read). It could be even better! Give me some information that allows me to make a better decision than “let’s kill more undead until we find the bone mirror”. Option B examples: a rumor saying that Vlad the Impaler has one. That Vlad can found in Castle Wittgenstein. On the fifth level. Have a captured kobold offer to guide me to the secret entrance to Vlad’s lair. All these decisions are not possible if we limit ourself to random loot on bodies. Yes, option A is not bad, but all the options B are better. And given option B, looting the undead will always be a minor activity. So minor, in fact, that as a referee I won’t even bother to look at a random table with weird things the undead might be carrying because I already know that it is useless detail. It can be funny for ten seconds. It might be funny in half an hour if an unusual application for the pine cones found on the first level of dungeon can be found. But this entertainment is so tangential, so unimportant, so meaningless in the larger context, that I might as well not bother.

Harald continued this line of inquiry and asked about the difference between randomly generating map content vs. the essentially random outcome of combat.

The way I would analyze this involves thought experiments and better alternatives. So, the situation is a confrontation between party and monsters. Option A is: roll for initiative, fight to win or TPK. This is how we played it when I was a teenager. The only meaningful decisions we could make was “what spells are we going to use?” Option B is: add more tactical stuff. This is what D&D 3, 4 and Pathfinder did. More small decisions, maneuvers, abilities to use, effects to have. Option C is: add more strategic stuff. This what I like best. Do we avoid, fight, run, or talk? Can we learn how to avoid or seek out these encounters in the future? Can we improve our chances of talking i.e. influence the reaction roll? Can we disengage? Can we force morale checks and have the other side disengage?

In short, if encounters turn out to be just randomized outcomes, then this is a poor game indeed. It’s not without it’s charm. I know the excitement of a pitched battle. Will we make it? We need just one more lucky roll! Come on, have the dice cursed us? Nooooo! Fuuuuuh! People shouting and throwing their hands up into the air. These are great moments and they happen even if we pick option A. I’m just saying the game could be even better.

And to return to the matter of railroading: providing the illusion of talking and then have the monsters always attack after a number of rounds (as was common in the Adventure Paths and modules I read by Paizo) reduces the talk option to “ask a few questions to learn about the setting” (much like “loot a few undead to learn about the setting”) when it could be so much more.

So, does it matter whether the map (or whatever the environment of our player agency is) was built from non-random first principles or not? No it does not. I’m sure you’ll smooth out any inconsistencies and provide a coherent experience.

You could of course fake it, rolling quickly, and players might believe it for a while, but when they discover it, the experience will feel hollow. Having done the preparation ahead of time solves the problem. It would have made a difference if we had gone north instead of south. If we later learn this, we will feel satisfaction, we feel like we understand the consequences of our actions and it will be good.

A tangent introduced by Chris referred to the limits of railroading. If railroading is about invalidating player choice, then having meaningful player choice is a prerequisite. So, without information to act upon, all acts are random, all choice bereft of meaning. I like to avoid that. Random actions having consequences is not good enough in the long run for the kind of game I’m looking for. If there was no meaningful decision in the first place, there is no way you can invalidate it. It was already invalid. There was no railroading. There was simply bad design for that particular obstacle.

Random Wizard wanted to know more about my preferences and suggested that perhaps I “prefer a game where details are written out before hand?” He thought that perhaps I wanted to “keep the DM from making up things on the fly” or to “prevent nonsensical things from happening from random tables.” I’m not so sure. The reasons he suggests don’t resonate. I know that referees need to make up a lot of things on the fly and I know they give their best to give coherence to the results of dice rolls. Thus, I wouldn’t describe myself as preferring a game where details are written out before hand, in a general sense. If we’re playing a game were exploration matters, then I’d like to make meaningful choices regarding said exploration, and if these meaningful choices involve movement on a map, then I’d prefer that map to be written out before hand. The list of prerequisites is long. At the same time, I think improvisation makes sense when talking to non-player characters. I think a reaction roll table injects some interesting randomness. I like random wandering monsters. I like road trip adventures or river travel adventures where the obvious route is going to take use through a fixed sequence of locations and events (plus some additional random encounters). I don’t think it’s so simple to nail it down unless you accept a very abstract description such as: I like meaningful choice.

To provide some more context, I’ll try and think of a game that doesn’t involve a map. Let’s say this is an urban setting. The city has 12 factions. The party leaves to meet one of them. I wouldn’t like it if we went to visit the 12 headquarters without knowing what to expect and without a way to learn it. Without information, choice is meaningless. If there is information, then clearly something has been prepared before we arrived on site. It doesn’t even need a written form. Perhaps the referee has eidetic memory. In addition to that, if the referee determines the characteristics of the faction as we enter their head quarters, I will loose interested quickly because I don’t feel like I’m exploring something that already exists. I feel like I’m exploring a randomly generated dungeon in a computer game that offers nothing but random dungeons. I don’t like it. But perhaps we don’t need to choose a headquarter to visit. Perhaps there is a party at the palace and every faction sends an envoy and we get to talk to them. Clearly, no decision needs to be made. We’ll just talk to them all, in whatever order the referee provides them in, and we’ll see how it goes. No fake decisions so far, no problem so far.

It’s weird to read just my side of the story. I did some minimal editing. I’m not sure how easy it is to read without the other G+ comments. There is a lot of repetition.



Also relevant: The Mediocrity of Improvisation.

– Alex Schroeder 2015-03-26 07:10 UTC

On Google+, I talked about this one shot using the Mass Effect Fate RPG and remembered how I decided to switch from Traveller to Fate back in 2010. I picked Diaspora as my Fate system of choice, mostly because it was shorter than Starblazer Adventures (no longer available) and I hadn’t heard of Bulldogs!, and Mindjammer didn’t exist back then. On the day I was supposed to run the Mass Effect one shot, I discovered another file in my Mass Effect folder: Hacking Diaspora to Mass Effect. It was based on Diaspora where as Mass Effect Fate RPG is based on Bulldogs!. Hacking Diaspora to Mass Effect is an add on and therefore shorter (less than 100 pages) where as Mass Effect Fate RPG is around 250 pages but stands alone. Hacking Diaspora to Mass Effect is more conversational. If I ever run a Mass Effect session again, I know what I’ll use!

Or perhaps I’ll just use Stars Without Number. 🙂

Here’s the author explaining how he’d use Stars Without Number to run Mass Effect. And here are some more extensive notes by

– Alex Schroeder 2015-04-08 06:10 UTC

In another Google+ discussion, Jack Gulick talked about the difference between Fate games and traditional games. In another thread, we were talking about how to avoid an unsatisfying “players save all their fate points and nova burst the final boss” issue. More about that thread over here. In this new thread, he explained where player agency resides. He said it is no railroad when you change the stats or the number of enemies in a boss fight. This is supposed to be more engaging and challenging. His point was that the aspects chosen are essentially a sign for the game master – “poke me here”. Essentially, the game is no longer about puzzles and beating challenges, I guess, but about how these things affect your characters and the world around them.

At character creation (and every time you revise you character, which is potentially after every session), you select the 5 (or more) aspects that are always true about your character. These are your strengths and weaknesses, the places the drama is most likely to press about you.

But since they are dynamic things, able to be changed as your character grows and reinterpreted as the story develops, I don’t think everything is decided at chargen. Most of the meaning of aspects is discovered in play as the interaction of various facts with the character and the evolution of the character proper occurs.


[In] Fate, a lot more of the emphasis is on how your characters major attributes (their aspects primarily, but also stunts and skills) impact the world they’re in.

Because, you see, the world also has aspects, and those get changed by what happpens. In effect, the world is a character (for the sake of this topic), as are the NPCs. All of them are changing because of the intersection with the PCs.

PCs can change a lot or only a little, based on if you want a dramatic or an iconic character pattern (dramatic has big change-filled development arcs, iconic change the world more than they themselves are changed), as can the world based on if the GM wants a “bigger than you” world or a “you are making a huge difference” world.

– Jack Gulick

Definitely food for thought!

– Alex Schroeder 2015-04-08 06:23 UTC

Over on his blog, Ben L writes about the joy of discovery and how making things up on the spot is basically denying players that joy. Recommended reading.

So here is one of the pleasures of the OSR style of play: it makes available the pleasure of discovery about the world of the game and the things in it. This fact depends on a differentiation of roles between DMs and players. The DM in designing the world, building the sandbox, creating and stocking the dungeons, making up the NPCs, and so on, acquires one secret after another. On the OSR style of play, prep is essentially stocking a storehouse of secrets. The players, by contrast, go into the situations they choose to enter in a state of ignorance.

– Ben L.

Important note at the end, after discussing low prep play, random encounter tables, locations that are undetermined when play begins and a lot of other interesting aspects:

I’m not so keen on the raving enthusiasm for low-prep gaming that is expressed episodically in certain strands of OSR culture.

– Ben L.

– Alex Schroeder 2019-04-21 19:11 UTC

Please make sure you contribute only your own work, or work licensed under the GNU Free Documentation License. Note: in order to facilitate peer review and fight vandalism, we will store your IP number for a number of days. See Privacy Policy for more information. See Info for text formatting rules. You can edit the comment page if you need to fix typos. You can subscribe to new comments by email without leaving a comment.

To save this page you must answer this question:

Please say HELLO.