I must create a system, or be enslaved by another man's. I will not reason and compare: my business is to create.

- William Blake

Monday, January 24, 2022

Improvisation, railroading, illusionism and the man behind the curtain

I tried to make this point a while ago, without much success. I'll try again from a different angle.

Here is the idea: sometimes, improvisation (i.e., coming up with things on the fly) leads to railroading (or quantum ogres, illusionism and similar things), and sticking to pre-written material (settings, mechanics, etc.) is a good defense against this.

Let's define some terms before we begin. This is the best/oldest definitions I could find, and they seem decent enough:
  1. Improvisation: the art or act of improvising, or of composing, uttering, executing, or arranging anything without previous preparation (source).
  2. Railroading: Railroads happen when the GM negates a player’s choice in order to enforce a preconceived outcome (source).
  3. Illusionism: A term for styles where the GM has control over the storyline, by a variety of means, and the players do not recognize this control (adapted from this source).
It is easy to see how closely related railroading and illusionism are. 

It is harder to see how railroading  ("preconceived outcome") can coexist with improvisation ("without previous preparation"). But, as I'll try to show, the GM often has preconceived outcomes ("in the end, the PCs find the villain and beat him") that goes AGAINST previous preparation ("the villain is Lord Sanydun, he has 50 HP").

One of the main problems of railroading and illusionism is removing agency from players. Not only do their choices cease to matter, but also they are tricked into believing that they do. To use someone else's analogy, it is like letting your little brother play Street Fighter with you, but giving him a joystick that is not connected to the console.

I decided to write this after watching a video from a popular creator (whom I like) with these kinds of GM advice:

1. "If the players mention they suspect an innocent person, maybe you can decide that now HE is the culprit!"
2. "Fudge your die rolls or HP if the encounter proves too difficult".
3. "If a player rolls very well when searching for something that isn't there, maybe it is!"
4. "If the players are talking too much, throw an encounter at them".

Also, to sum it up, something to the effect of "never let the players see the man behind the curtain" - which sounds very close to illusionism.

Now, if that is what rocks your boat, fine. I just want to add that this is not the only style of play and is, in fact, anathema to another style which sees illusionism and railroading as things to avoid.

Let's analyze the advice above.

Advice 1 is the classic example of "improv". But it completely devalues any mystery, any clues you throw at the PCs. You may argue that this is not railroading because the GM hadn't conceived an outcome beforehand - the DM thought the culprit was A, but when the PCs accused B he changed. However, in this case, the outcome enforced by the GM is the PCs find the right culprit; he is negating the player's choice of accusing the wrong person!

Come to think of it, as long as the culprit gets caught, the fact that he was NPC A or NPC B is inconsequential - you cannot claim that changing the culprit is not a "preconceived outcome" because, again, the outcome you are forcing is that the PCs get the bad guy.

Advice 2 will make the players believe they can win any encounter, or worse, they can win any encounter if you let them. It removes player agency. Again, the outcome the GM is forcing is "the PCs win the next battle".

Maybe the only caveat here is if you throw a random encounter at the PCs, through no choice of their own... but even then, I'd prefer letting the Pcs run away or negotiate than fudging the mechanics. For one, it will make they have a better grasp of how powerful a dragon or ogre is.

Advices 3 and 4, again, make the setting feel artificial - as if responding to what players, not PCs, do. Notice that number 3 is a thing that could happen in Dungeon World, for example (IIRC), but DW at least assign consequences for failing your search roll - otherwise, everyone would be searching for treasure everywhere.

Number 4 deserves a caveat: IF the PCs are in a dungeon where you roll for random encounters every 30 minutes, and the players talk for 30 minutes in-character, you should obviously ROLL for random encounters. Likewise, if someone would hear them, etc.

It is not illusionism if the players know

It is not illusionism if there is no illusion. If your players know that fudging dice and HP is the DM's prerogative in this campaign, or that you'll decide whodunnit is as you go, or that a good dice roll will let you find treasure where anywhere, this is not illusionism, it is a style of play.

This is a very important distinction because, as we'll see, DMs must make things up as they go in both styles.

Some different advice

Let me try some alternative advice to the "man behind the curtain" method described above.

- The GM must present an internally coherent setting for the PCs to interact and explore.
- The GM must believe in the authenticity of the setting as much as the players.
- The GM should not alter the realities of the setting (at least not DURING PLAY) to accommodate, entertain, defy, reward or punish players, but only because of things that happen WITHIN the setting. In other words, the setting is defined by PCs and NPCs and not about GM and players.

Another tips I mentioned before that might be related:

- Let the dice push you out of your comfort zone. Your PCs all failed their saving throws - now what? Your important NPC was killed before he could start his plan - what happens now?
- Expect the unexpected from your players. Do not expect them to follow a predictable path, or always find the right culprit, or only pick fights they can win, etc.

How to AVOID illusionism?

Let's say you and I prefer the same style of play - how to avoid illusionism, railroading, etc.?

Well, one idea is use a published adventure, or write your own.

If you follow it to the letter, without improvising, you cannot execute any of the four advices mentioned above.

(BTW, having a plot telling you what happens if the PCs fail or do nothing will help you tremendously. It will relieve you of the temptation of enforcing the preconceived notion that the must win).

Of course, if your PCs stray from the course, you must improvise. However, do NOT improvise a reason to force than back into the adventure. That is exactly what railroading is. Just think of the logical consequences of their choices. 

Also worth remembering that pre-written adventures should NOT dictate what the PCs do, but should only describe NPCs, places, events, and possible ramifications. "In game railroading" is also terrible advice IMO - for example, "if the PCs refuse the mission, the king arrests them until they accept". Here is a good post about the subject.

And what if they enter a random town, far from the original adventure site? Well, then you improvise, but it is ALSO okay to say "I hadn't prepared this, let's take a small break". Remember, it is NOT illusionism when they know you're making things up on the fly.

When to "railroad"

It is okay to talk to your players before the game about a specific premise or even adventure. I recommend GMs that dislike improvising too much do exactly that.

This is NOT exactly railroading, because the players are aware of their choices (or lack of choices) - although the PCs will have no choice in the matter. For example, "hey folks, I have this cool dungeon we can play, what about we run that on Thursday?".

Giving SOME choice to the players is even better. "Okay, I imagine the PCs can go to places A, B or C next Thursday... What do you think? Maybe somewhere else?". If you're mid-campaign, a simple "what you think you're doing next?" will often suffice.

Of course, if you give them just one choice, the players can respond with "why would my PC go into this dangerous place", etc. But after the thing has been settled, it would be rude for the players to just go off-grid without a reason, forcing the GM to improvise on the spot (unless, I guess, the GM is cool with that too!).

As general advice, I'd say that this "railroading" is expected at the very beginning of a campaign - even implicitly. Are we playing Tomb of Annihilation? Yeah, I do not expect the PCs to be disinterested in Chult in the very first session.

Is it okay to "railroad" beginner players so they can learn the game? Well, I can see not wanting to kill every PC in the first fight. "Training wheels" could be useful. But how will they learn that a dragon is dangerous if they beat it at first level? How will they learn their choices matter... if they don't? Why would they play attention to clues if there is no "right" answer to the mystery?

I'm undecided, but I'm inclined to say "no railroads for beginners". In my first Demon Lord adventure, a couple of random thugs almost caused a total party kill. I thought the adventure was too hard - but it did set the tone for the whole campaign.

Should we improvise, then?

Of course. You need at least some improv to run RPGs. What I'm saying is: do it carefully.

For example, you often need to improvise to find out how the NPCs reacts to the PCs. How the events unfold. Etc. 

You also need to improvise to find out things about the setting you hadn't established before. But when you do so, answer your own questions using the setting's internal logic, not the necessities of the players, the moment or the "plot". 

For example, if the player asks "can I full plate armor in this town?", ask yourself "how big is this town?", not "how bad does the PC need this for the next adventure".

You do not change inanimate things and past events because the PCs had an idea, desire, or particular die roll (unless, again, the PC could change the world in such way with his or her actions).

You also improvise anything that players expect you to improvise, of course - what is the blacksmith's name? 

But you do not improvise freely when the answer should be found in the setting - "is there a blacksmith in this town"? When the players ask you that, they do not expect you to be creative, but to give your honest assessment of what you be expected in the setting. Here, you are not looking to be creative, but genuine.

You can also improvise (or, at least, create) anything when your players expect you to do so. For example, between sessions. Or when you ask for a 5 minute break. Or when they break into a random house. Etc.

Its often good to improvise "HOW" but not "IF" (the monster dies, the castle breaks, the NPC escapes, etc.). This is probably worth an entire post.

Oracles and random tables

These are not improvisation. They require previous preparation. If you have a table for random encounters, and you get a dragon encounter, throwing a dragon against your players from nowhere is not improvisation.

But what if these tables are more like oracles? "17 - an ally is revealed as a traitor". Still not improvisation while you roll, but you'll have to interpret the result to the best of your ability - which does require some improvisation. Again, just let the players know that this is the kind of game you're playing - an ally can betray them at random. If there is a transparent mechanic for that (e.g., morale rules), the outcome is not improvisation (although you might have to improvise motives, etc.).

Creation x discovery

Actual improvisation usually trains actors, musicians, etc., to come up with unexpected responses. They are still constrained by other factors - the beat, harmony, or previously established facts. Which is a fine way of playing RPGs*.

There's another way of doing things, however. Many writers describe their processes not as creation, but discovery; i.e., there is a deeper truth about their settings and even characters that they must respect. The goal is not to come up with something unexpected, but to be faithful to the "true nature" of setting in character. 

This is an equally valid approach.

(*One good comparison I got from my friend Jens is jazz - because in jazz you can go "off-key" and even play around with rhythm. But, still, you do not do it alone - you take cues from the band, from the harmony, from the existing beat, etc. You do not set the tone and rhythm for the whole band).

But I LIKE fudging dice!

Again, if that's your preference, that's fine. I'd advice you be transparent about this - let your players know that fudging dice and HP is your prerogative, or that you'll decide whodunnit is as you go. If everyone is on the same page, that's okay (just not my preferred style).

I'd argue, however, that it is useful to let people know both styles exist - at least so people can try both ways and see what they prefer.

Recommended reading:


  1. Great post, man. Lots to unpack, lots to think about!

    I think the dictionary definition of improv actually doesn‘t do it justice, because it mainly describes the lowest form of improv like, say, a child playing with an instrument it knows nothing about. There may be (very young?) DMs out there who play rpgs like that, but I‘d say the gross of DMs actually know the rules, the setting and the adventure enough to, say, extrapolate from that. For me, that‘s improvisation in an rpg context (and you point that out later, so thanks for that!).

    As for railroading and agency, I hold the unpopular opinion that the limits of agency, if agreed upon, are totally fine across all possible outcomes. The DM wants to be on a railroad so he runs a strict adventure, that‘s fine, the players want to be on a railroad because their focus is on getting together and roll some dice to see the prepared fireworks, that‘s totally fine, too. It‘s where people don‘t get to agree (within reason) where it gets difficult.

    And whoever that creator was, the advice they gave isn‘t good :D At least not in general. There are role-playing games out there that work like that (PbtA games comes to mind, lightweight OSR games can go there as well), so the advice might have been gleaned from playing those games? This is where I‘d have loved to see some distinctions between rpg systems and how they manifest narratives. The discrepancies are huge there! I took a close look at Dungeon World at some point and you couldn‘t play something like a published OSR dungeon crawl with it. Totally different assumptions of play (which thee PbtA crowd really doesn‘t appreciate, for some reason).

    You do a good job in pointing out why it might be bad advice. The advice you give after that is a good example for another, very distinct style of play (the well prepared campaign with a system that carries it in all its minutae, basically something that peaked with AD&D, maybe)

    I liked your tips on the meta game aspect of rpgs. DMs should plan and adjust with the players between games. Always helps.

    Again, thanks for sharing (and thanks for the shoutouts!).

    1. Hey Jens! Yup, "improvisation" has multiple meanings, and jazz improv is very different from, say, comedy improv.

      I do agree with your assessment about agency - "if agreed upon", anything goes.

      and well, for my style of play, it is indeed very bad advice. I do understand that he might have a different style, but I think the caveats I added are important in order for people to choose if they want a "illusion"-style game.

      Thanks for the comments and posts!

  2. I agree with you, absolutely. The more prep you do and the more open you are about that prep, the more "genuine" your style can be. I do also think that with the proper intention, additions are perhaps the best way to go about improvisation. Good example: I sometimes think of an ability for a monster halfway through an encounter with it, and I often will implement it, because I feel it will make the world more unique or memorable. I had a big-jawed demon recently, and I decided near the beginning of the encounter that his big tongue would be particularly sticky, just so that someone may get stuck. But I didn't make the bite automatically hit, and I only did it insofar as I could be "sporting" about what I had already presented the PC's. Ultimately, it increased their "tactial infinity"--one of the PC's did get stuck, yes, but he used that stickiness to stick part of a bedframe into the demon's mouth so that he couldn't shut it all the way, effectively incapacitating it.
    I also often think of a piece of information in an investigation or a hint to a vague, self-written puzzle that I might implement if I feel the situation is unfair. This isn't comparable to instantly solving the puzzle, or forcing it to be solved at all--merely to make it theoretically solvable. Though when I do this, I typically tell the players that I've made a mistake in order to keep that authenticity. "My bad, I forgot to mention that there is some broken glassware in that cabinet as well." I can't constantly be admitting my mistakes, so instead of not telling the players, I just try not to make them.
    My intention with these somewhat meta (e.g. make this encounter cooler), but I also find myself to be the most creative in the process of running the game, as I pace about and get in "the zone", and want to channel that energy. Had I had the idea earlier, I would absolutely have written it down. And for the latter portion of puzzles and investigation, I think a small "retcon" is justified, because in reality a puzzle or murder WOULD usually be solvable with the right clues, that I didn't notice with my limited brain--even if it would have been better if I had. So I feel that these sorts of improvisations, ones that expand the situation rather than resolving it, are desirable.

    1. Thanks!

      Yeah, I think there is improv that preserves player agency, and there is improv that negates it.

      For example, changing a monster slightly before the fight begins is not the same as changing it mid-combat because combat is going too well (or too poorly!).

      I agree that it is better to own up to your mistakes, and these little mistakes are very common - I seem to commit one or two in every game ("you see a monster lurking in the..., I mean, you see a vague shape in the shadows"). No big deals, we usually laugh about it.

      Good improvisation can certainly improve the game, we just have to be careful to avoid negating player choices and their consequences.

  3. This is very similar to how I play my games. When I defer to an oracle because I didn't plan something and genuinely don't know what would be most likely, I'll tell my players, let them know the odds and roll in the open. It might not be everyone's cup of tea, but verbalising these thought helps me structure then and ensures people at the table know what I am doing and get a chance to correct me if I make any mistakes or contradict myself or the established world. In my experience, players like this, though it probably isn't for everyone.

    1. This is my favorite way too, and I think everyone should try it!