Roll For Loot: Procedural Story Generation in Tabletop Games

 Introduction: What Is Procedural Story Generation?

Hi! Today I want to talk about something I've rarely seen discussed in the TTRPG space. I want to cover what it is, what it does, and why its important.

Procedural Story Generation (which I'm going to refer to as PSG going forward) is a system or series of systems which allows for a Game Master to create or inspire a game's story with little to no input. The simplest example is a random table with multiple prompts (to create questions) or answers (to resolve questions). The table, without much context, can create the beginnings of a story for the GM to use.

A set of random tables from Chris McDowall's upcoming game Intergalactic Bastionland

What's So Cool About Randomness?

The thing that makes PSG so compelling to me is actually not the mechanics themselves, although novel ways to do PSG are always entertaining. What really interests and inspires me about this approach to writing games and stories is the way it allows the writer to guide players towards interesting stories, without exerting direct control. Let's break those ideas down.

Writer Guidance

Of course, a PSG system has to be designed and filled out by somebody (hence "procedural"). When you roll on a random table, you aren't just getting something out of the ether; you're consulting the table's designer on what could happen in this scenario. Knowing this, a writer can be selective and deliberate about how they fill out their PSG system, subtly guiding the player while staying hands-off. It should also be noted that a PSG system doesn't have to be used for PSG. If you've ever looked at a random table and thought a little too long about one of the cells, congratulations! You got to experience writer guidance firsthand. Remember, PSG doesn't have to write a whole story for you. In fact, it usually can't. But what it can do is give you a story's beginnings and fill in its gaps.

Interesting Stories

A major takeaway I've had while thinking about PSG is an appreciation for a "do now, explain later" philosophy. I even put a similar phrase in the design bible for the game I'm working on right now!

An excerpt from the design bible, containing the phrase "Do Now, Justify Later"

The example I always think to has to do with rolling for dungeon loot. When I play dungeon-crawling games, I tend to elect to manually choose the loot a monster or other enemy is guarding, because it gives me a level of narrative control that allows me to be purposeful with the loot's backstory. But what if we take that level of thoughtfulness, and remove the intention? That's where we start to get questions. Why does this troll have a fancy sword? How did you loot 3d6 gold coins from a wolf? Where did this kobold get a magic wand, and why did it choose to hold onto it? All of these questions are like miniature puzzles for the players and GM alike, and they allow for a surprising and interesting gameplay experience. And as we've already established, PSG can not only raise questions, but answer them as well. So a carefully-designed PSG system might even have a way to justify its results!

A Really Fantastic Example (That's Not A Table)

Seven-Part Pact, jay Dragon's upcoming game where players embody wizards trapped between their own fantastical powers and the responsibilities and societal expectations that come with them, has a fantastic PSG system in its "orrery". For those who don't know archaic terms, an orrery is a device that tracks and/or approximates the movements of celestial bodies. In Seven-Part Pact, the orrery is essentially the main source of story beats and complications, as I understand it. Based on the relative positions of celestial bodies, the individual minigames played by each player undergo changes. Here's an example from playtest material I was given access to:

An excerpt from the Necromancer's codex of Seven-Part Pact

You may note that this is largely a mechanical outcome, rather than a fictional one. That's ok! Although PSG should influence the fiction, it doesn't have to be entirely fictional. Remember, mechanical results offer opportunities to "do now, explain later". 

Now, you might be wondering why jay chose to make the orrery system, as opposed to doing something like a roll table, dice mechanic, or cartomancy system. I was wondering that as well, so I asked her. Here's what she had to say (edited for syntax): 

"So, I wanted it to be deterministic, something that could reward careful playing and intellect, but I also wanted it to be hard to anticipate, and extremely vulnerable to the butterfly effect. I also wanted it to contain a ton of information from multiple different angles, so that different people can look for different properties (the Necromancer cares about what's in alignment with Saturn, the Warlock cares about the King's zodiac sign, the Mariner cares about the season, etc.) And because everyone's looking for something different, its super easy to make one thing better for you, and accidentally make it worse for someone else." - jay Dragon

Basically, what this boils down to is that the orrery is chaotic but predictable. You could say it has a high entropy. There are other ways to do this as well, like Vincent Baker's Otherkind Dice or Mindstorm Press' Ladder Tables. But the orrery is what jay has determined works best for Seven-Part Pact, for a variety of reasons (I have ideas on what those reasons are, but I'm not jay, so mostly it's just speculation and interpretation). I think there are some good lessons here, which I'll expand on in momentarily.

How Can We Build On This? (Theory Time)

Procedural Story Generation is a framework. Like all frameworks, you can build well within its limits or stray a little further from the beaten path. 

 

Let's start simple, assuming we have a set of six outcomes, and that any one of them could be true when we consult our system (in other words, we have a d6 roll table). One of the clever ways I've seen this classic system iterated on is by using two dice - but not in the way you think. This idea comes from the D&D 5e module Icewind Dale: Rime of the Frostmaiden, and is criminally overlooked. When exploring the frozen wilderness of Icewind Dale, for every day you traveled you would roll for a random encounter. The twist was that you would actually roll the die twice - once for the table, and once to see if the encounter happened in a blizzard. Blizzards occurred based on whether the blizzard die was higher or lower than the encounter die, which meant that different encounters had a different likelihood of being a blizzard encounter, based on their position in the table. If we abstract this idea a little, we could replace "blizzards" with "elite enemies", "rare loot", "magical mishaps", or any number of other things.

 

Ok, augmenting results is fun, but let's try something different. What if you want players to have some control over the outcome they get? The first idea that came to me had to do with cards. I play a lot of Magic: The Gathering, and in MTG, there's a mechanic called "scrying". When you scry a number of cards, you get to look at that many cards from the top of your deck and choose which ones you want on top and which ones you want to put on the bottom, getting to choose their order either way. This is part of a family of mechanics and cards collectively known as "topdeck manipulation". Other highlights from the family include the beloved card "Brainstorm" and the scrying variant called "surveilling", where you discard cards instead of putting them on the bottom, allowing you to later recover them. 

 

But enough about Magic, how can we use this idea? Well, let's say you have a deck of playing cards. Before you start the game, each player gets to look through the deck and find one specific card, based on their character's abilities. Then, they shuffle the deck and go around the table, letting each player put their selected card on the top or bottom of the deck. In this way, players can determine whether the outcome is likely to appear or not, and smart players can try to stack the deck such that they get good cards their enemies get bad cards (for the sake of this example, let's assume the players might come into conflict. Doing so deepens the implications of this idea). Or, maybe the deck starts with no cards in it at all, and pre-game player choices (character building, life paths, etc.) determine which cards (and therefore which outcomes) are shuffled into the deck.

What Am I Doing With PSG?

In the "Psychic weirdos solving mysteries" game I'm currently working on, Psychnoir, I'm incorporating a lot of ideas from Procedural Story Generation. First of all, all the Moves you have access to are formatted as a list of options you get to pick from, with what you pick becoming true and what you don't becoming false. Here's an example: 

A Move from Psychnoir, called "Feel Shit Out"

Also, I've decided to leave cases open ended and let determining the answers to the mysteries be up to the players (a la Brindlewood Bay). That said, its a little difficult to bring actual randomness into a diceless game!

Can You Use The Ideas Here?

To start with, you can use any of my ideas that I've presented in this article. I give you express permission! Just let me know, so I can check out your work. As for other folks' ideas, like jay's orrery, I can't necessarily give you permission to use them. That said, jay Dragon and Chris McDowall, who both generously gave me permission to share their unpublished work here, are both wonderful, open people who I'm sure would be delighted to hear that they inspired you. Just make sure to talk to them first, and get their permission.

 

 

If you like my writing, feel free to toss me a few bucks on my Patreon. In the future, I'm going to be giving subscribers early access to these blog posts, as well as updates on my projects and maybe even some exclusive stuff.