Creating Setups

Designing setups that make players want to join

In most essays, people would cite the numerous particulars about how to make your setup work in a fun and interesting way. “The game must be entirely balanced, and all roles must be individually interesting.” The problem is that this perspective is fundamentally flawed. You need to play to you audience more than anything else.

When I say I want to make a bastard game, you bet your ass that I’m going to go all in.
But I’m not going to do it in a way that is going to frustrate people. They dedicate a lot of time by joining this game, and many play FM to cool down from their daily lives.
I do this by increasing the uncertainty of interactions and what abilities actually do to such an extent that anything happening is hilarious.

I know you’re not here to read about how I create a bastard game though.
You’re here to hear about me screeching about math.
And I can do that.

Every role that has mechanical impact on the game has an expected value (or EV).
Every role has a category in which they impact as well.

These categories are:

Killing (removing players from the game)
Investigative (determining alignments)
Protective (hard counter to killing powers, either by stopping kills or by reducing their impact)
Disruption (interfering with other roles)
JOAT (crossover powers that don’t fit into other categories)

Power (absurdly strong roles that warp the setup around them. Prince is the best example for something that crazy)

In these, you need to calculate the value per night, or VPN if you think you’re funny.
A cop can check 1 player per night.
This will determine their alignment.

This has an EV of 1 since they are able to mechanically determine as many alignments as there are nights in the game.

For most roles, this is how the averages work out. There are two important points to divert from though:

Limited roles
Weaker roles

Limited roles simply follow the EV calculator. These include limited shot roles, odd/even restrictions, etc. Conditionals are very hard to calculate with this and generally are just eyeballed.
There are x nights, and this role can act for y of those nights
y/x = EV for that role

Weaker roles are roles like bodyguard (guards player -> if they would die then you will die instead), tracker, and suicide bombers.
These roles are dependent on the setup itself to determine their usefulness.
In a setup with two extremely powerful town PRs, a bodyguard would be useful to enable those roles to last an additional night if they outed.

Balance using values and math rather than trying to fill out the common roles in the game haphazardly.


This is where semi-closed falls the hardest in my eyes, and why closed ends up being the best form of expression for hosts…alongside just being more fun

1 Like

i get that your trying to mostly use terms that are already fm terms, but i feel it would be better to have more vague categories.

Removal includes killing, but allows for more of a spectrum by including other things, like silencer, which for the purposes of reads is essentially just a weak kill.
Information is also more broad, openly including information from all places, even side-eyed ones. yes almost all information goes toward determining alignments anyway, if a bit roundabout, but this is more open about that
Support shows that it goes toward all of promoting your team. this covers both preventing/fixing bad effects, as well as providing good ones
Offense is the direct counterpart to support, that it goes toward all of harming the enemy, covering both countering/destroying their good effects, as well as providing your own bad ones.

joat is just crossover between these, i.e. JOAT (50% removal, 50% information)

power cant really mix into a joat because of how it works, but power can be implicitly based off these categories, for instance id personally place prince into POWER (25% removal, 50% information, 25% support)
then you can just have their normal EV, or their split EVs into categories. then you could be able to find a rough numbers for good balances between the 4 above categories, say in a 10 player game a good balance could be 2:1.5:3:2
please dont quote me on that, im doing this all in theory and ive no idea how that actually works in a 10p game

and if you want to go super broad, removal and information are just slightly specified offense and support, respectively

bookmarking this thread once candlelight has run

as ill prob explain my thought process behind it

For example
A treestumper is a killing action

dae stump

you know what a stump is right

completely removes all abilities including voting

1 Like

Not exactly
A watcher is a protective

id say that falls more under offensive with removal undertones.
they can still talk which can count for a lot, depending on the target

It’s completely removal
A treestumper kills a slot but allows the voice to still exist

well, id put it under information in my system, but as i said, removal and information are just versions of support and offense

Your system encompasses a broad range that gets under more roles, but doesn’t work as well with the EV system
That’s mainly how the system I blatantly stole is designed

yes. it removes them mechanically. but the voice lingers, which can mean a lot, depending on the players.
perhaps the other way around, with offensive undertones, but almost everything happens to be a slight mixture, except for pure abilities, like kill, cop, heal, occupy,

Not really

Just a few examples of things that it also works for:

Bus Driver
1-shot BP
Tracker (worth roughly half of a cop)
Bodyguard (worth roughly half of a doctor)

It’s a variation on vig, when you get down to it

ima just stop discussing this since we both agree on the facts: it effectively kills them, just leaving a voice, but beyond that we disagree on what the facts mean, so i dont think we’ll get anywhere trying to convince each other.

meh, agree to disagree in how things cross over, but yeah the main thing is that it allows joats and powers to be within the 4 categories kinda, to add that to the calculations

1 Like

Think I’ll just stick to my ‘randonly making roles and hope they get approved’ method, thanks