Society of Dreamers: Randomness through mad-libs


Society of Dreamers is an “indie” roleplaying/storytelling game by Matthijs Holter, author of Archipelago amongst others. In some ways it’s closer to improv, in the same way that Cleopatra’s reign is closer to the moon landings than the building of the Pyramid of Giza1. I’ve been running/playing it with a group of friends for the past few weeks: while Holter suggests that you can run the whole thing in one session, we’re now up to our third.

The premise of the game is this: at some point, in Europe in the 19th century, a group of people form a society whose goal is to hunt down one or more mnemosites: creatures that live in peoples’ dreams. The number and nature of the mnemosites is unknown, and much of the game is dedicated to finding out more about them. The Society also knows of a mnemosite host, and can participate in dream diving: a form of lucid dreaming in which the diver experiences the host’s dreams. The setting has elements of Victorian-era adventure, supernatural horror, and suspense.

The game itself is excellent for flexing some muscles that go underused in traditional gaming. One of the elements that immediately got our group going was the shared character-creation method: something that’s not necessarily at the core of the game, but was still pretty cool. I thought it would be interesting to explore it in more detail.

This and other images in this post taken from the Society of Dreamers rulebook.

How we make ourselves

Characters in Society are not defined by their ability to do things. After character creation, the only concrete things written down about your character are their nationality, gender/sexual preference, age, and profession. You pick these categories by what boils down to an elaborate game of mad-libs:

  1. Every player writes down two possibilities for each category on scraps of paper (i.e. two nationalities, two genders or sexual preferences, two ages, and two professions).
  2. These are put into piles, one for each type, and shuffled. At this point you’ll have a pile of nationalities, genders of sexual preferences, etc., with 2n cards in each pile for a given n players.
  3. Each player draws one card from each pile. This is your character.

In effect, you’re getting the players to define the “possibility space” of characters. This is a fun trick - simple once you know it, halfway between rolling on a table for random character traits and just writing down whatever you want. Unlike completely random generation, the party makeup will inevitably be influenced by the players at the table, their relationships with each other, and their mood during character creation. Our group is a collection of relatively young, sexually liberal people with a scattering of practical (manservant, priest) and “romantic” (inheritor, writer, con artist) professions, which reflects the demographic of the group as a whole and what they want from the game. One could imagine that a different group could result, for example, in a much older party with professions like carpentry, soldiering, or even lacking a profession.

The random element means you’re faced with a character you may never have considered playing otherwise, which in my experience leads to better play experiences. My character, a twenty-two year old lesbian Scottish con artist, is an example of this2. Constraint breeds creativity - and some of the best characters turn up when you’re forced to play something you wouldn’t have considered had you just filled in a couple of boxes.

Using other people as random generators isn’t unique to Society: Microscope, for example, relies on this method of randomness generation to keep the game interesting, and elements of it exist in games like Fiasco. What makes Society‘s use of other players’ inputs interesting (in my mind) is the blind recombination of a number of different elements with the assumption that the final product will be consistent - and the strange narrative twists you sometimes have to pull in order to make it that way.

Adapting the method

Let’s not just leave this system be: let’s see how else we can use it! Potentially anything where you need a little creative input is fodder for mad-libs style creation. If you’re playing a fantasy-styled RPG, maybe you want things like race and class, rather than nationality and sexual preference. The main thing, I think, is that your categories are relevant to the type of game you’re playing3 and somewhat central to each characters’ identity. There’s a couple of situations where this method might not be so handy, however:

Ranking qualities: obviously you can’t really use mad-libs style generation for ability scores or similar. There’s too much dissatisfaction when you pull “terrible” in all four categories. Ideally you should be excited with the result of every pull from a pile, not secretly dreading a dud card.

Tightly-coupled values: if only elves can really be mages in your setting, and you’re mad-libbing both race and profession, you know your party will end up 80% non-elven mages. It should be possible to explain the vast majority of quality pairings you could end up with.

On the plus side, the system is incredibly flexible: you’re not limited just to character creation. Let’s have a go at making a few examples.

Character-level: Super heros

The mad-libs generation method really needs something viceral to hinge on. Super heroes are a great example of this, with their larger-than-life origins, super-powers and weaknesses, and nemeses. While Society really encourages mundane choices to make quiet but detailed characters, super heroes allow you to be as gung-ho as you want.

  • Origin story (bitten by radioactive cockroach, alien from another planet, exposed to gamma-rays, government experiment)
  • Super power (flight, invisibility, regeneration, telekinesis)
  • Weakness (cold iron, Farady cages, alcoholism, mirrors)
  • Nemesis (super-villain, corrupt millionaire, government agents, threat from beyond reality)

Community-level: Colonial America

This is inspired by Dogs in the Vineyard and the idyllic towns the characters tend to pass through. Each village is like its own little island: while trade occurs, the inhabitants of these towns can’t rely on it. Nature is prominent, and people rely on its bounty to survive. Still, living is hard, and every town has something it’s people has to endure. The inclusion of sin makes this a crowd-sourced session generation tool for the GM too: they immediately know what trouble will be plaguing the town. You could deal these cards all out in front of the players if they’re fine with knowing the main theme of the town beforehand, or you could determine one or more of the village’s qualities behind the GM screen.

  • Natural feature (river confluence, hill, small wood, flood plain)
  • Abundant resource (corn, wood, grazing land, furs)
  • Hardship (cold winters, native tribes, little fresh water, high winds)
  • Sin (polygamy, envy, nepotism, greed, wrath)

Game world-level: Planets

Science fiction games like Traveller and its recent cousin Disapora thrive on having a wealth of different worlds for you to explore. While each planet likely varies from pole to equator in its character, you can still pick out a planet because of its sun, or because of the climate, or the people who live there.

  • Star (binary, ominous and red, tiny and dim, conspicuously absent)
  • Culture (strongly patriotic, colonists and natives, balkanised nations, alien)
  • Abundance (minerals, food, gasses, technology)
  • Scarcity (oxygen, water, food, technology)


So that was a fun foray into mad-libs style character creation, or even world-building. I should stress that I haven’t yet had a chance to try this out in anything outside of the original context, so everything here should be taken with a grain of salt.

Should I get a chance to try character creation by this method outside of Society, or use it to help build a world, I’ll make sure to write something up.

  1. By which I guess I mean that conceptually and theoretically, by the way it’s laid out and by the provenance of the designer, you consider this thing to be an RPG; but really it’s closer to improv theatre in terms of the techniques you use and the skills it stretches. 

  2. Originally, she was seventeen, but faced with a bevy of late teenagers already in the group, I ended up discarding my age card to get a bit older. 

  3. And maybe by specifying what gets generated in this manner, you’re implicitly making these qualities relevant.