How to balance games?

[Mr. Flay]

I haven't modded yet, so take these with a grain of salt. But I have written a bunch of games I hope to eventually test...

1) Random numbers to simulate choices is a good early step to see all of the possibilities.
1b) If the game isn't too large you can make a pairwise matrix: Put all roles across the top and the side of a spreadsheet, and where they intersect figure out the outcome. Make sure to keep in mind what happens if either party is roleblocked, doc-protected, mason-recruited, etc... I generally keep that around for resolving night actions when the game is actually running, sort of like a cheat sheet.
2) Random effects (partial-chance protections, random investigations, etc) are harder for players to accept even if you feel they'll "add" to the game, or help balance things. Have you read the metabalance thread? As long as all sides have some reasonable chance that they can survive/win, you've probably made a decent game. Hobbit Mafia is a good example of an imbalanced game...
3) Making one role that the town/mafia's survival depends upon is tricky. Go through your entire game scenario assuming that Role X got killed Night One before anyone ever got to post. Do both sides still have some chance, or is it a foregone conclusion?
4) I second the motion to put some of these tips out there, either in the FAQ or on the Wiki...

[Mr. Flay]

Something along these lines is what I use (mileage may vary)

Target -->

[col]

Bodyguard

[col]

Roleblocker

[col]

Cultleader

Bodyguard

[col]X[col]Protected from one night kill attempt[col]Protected from one night kill attempt

Roleblocker

[col]Night action fails[col]X[col]Night action fails

Cultleader

[col]Recruited to cult's side[col]Recruited to cult's side (unless blocked)[col]X Make sure that if you have two of any one role, you put them both in the table, to account for double-kills, double-roleblocks, etc...

[Mr. Flay]

Well, you're talking about a little bit more than X*X situations (X being your number of roles), but not *that* many. That's why I tend to word protection roles in terms of how many attacks they count against, and kill attempts in terms of how many attempts they count as. If kills>protections, the person dies.

In answer to your second question, that's why precedence of actions is important. Do investigations always go through first? Protections? Recruiting? Blocking? Does it depend on when they send in their PM (note: I don't think this is used very often)?

*checks notes* For example, I've got this in one of my game setups:

Night actions (generally speaking, night actions will be resolved in the following order: role-blockers, protection, investigative, killing/recruitments [but see below]):

It makes sense to me to do it that way; role-blockers are useless if they resolve after anyone else, protection should probably be considered 'simultaneous' with kill attempts, investigations should happen even if the person dies 'later' in the night...

[Mr. Flay]

Flay: if there are 5 night choices targeting any of 12 people, that is 60 possible nights, then look at 12 possible lynches, then another 50 or so possible night choice combos etc and you need to multiply them together. The first 2 days yield roughly 60x12x50x11 = 396000 possible games. And then some minis go on to days 5 or 6, so you have many millions of possible games. Look at the way all of the beginner mafias play differently, with just a basic 7 person setup.

Nah, you're misunderestimating me. I'm really only interested in two things:
a) does it break the game to lose Role X, no matter how cool I might think it is?
b) do Roles M-W have a chance in hell of being effective/fun/winning?

My spreadsheet, which I suspect you're actually referring to, is merely a tool for logistical issues. Does investigation happen prior to killing? Does it make a difference if a killer refrains from killing that night? Does the doctor do anything if they protect the roleblocker who blocks the person trying to kill the roleblocker? I'm not foolish enough to try to map out the entire game ...though as the OP mentioned, I do generally run a few 'random number generator' games to see if the outcomes get weird/skewed.