Presenting, with some help from
@Pawn Lelouch, the Terrabrand Mafia Design Guide.
This is not a guide on how to play the game. I assume the prospective designer is already familiar with basic play. If you don't already have at least a couple games under your belt, you probably shouldn't be designing one. Some things can't be really explained in the abstract that you get a feel for by playing as such. If you want to learn how to play, we have play guides elsewhere in the megathread.
Mafia is a social deduction game. At its heart, it's about an uninformed majority squaring off against an informed minority.
Games of Mafia can run a huge range, but in designing them you need to remember a few key concepts and keep in mind some basic paradigms. This is a guide on How To Design a Mafia Game.
The Basic Form
If you aren't doing anything fancy, you want a scumgroup- the mafia- who possess a secret chat and the ability to have any member make a kill at night, once across the whole team. That's the basic form.
The important thing to keep in mind whether you get fancy or not is Scum need a crowd to hide in. This is easier to naturally achieve in a simpler form, and so you'll need to manually create such if you are doing something fancy.
The Crowd
Scum need a crowd to hide in. Your Town must necessarily not be able to pin down the scum solely by mass-claiming, or at some point Scum will be pinned down and killed off through no fault of their own. In the basic form, the crowd is the Vanillas, the players with no actual powers who thus cannot cleanly prove they are what they say they are.
If you go power madness, that winds up meaning you need to give every scum player a credibly town power, or else they'll be unable to claim and will stand out as 'the vanillas that appear to not exist'. Alternatively, provide the scum with strongly supported fake-claims. By providing a safeclaim that, itself, is believably Town, they don't necessarily need to look town in a vacuum.
Critically and insidiously, avoid your setup being too clean- if the Town can simply go 'well you aren't part of *my* mason', say, and that winds up revealing all the scum right there, because of all Town players being in a Mason, then scum has no chance out the gate.
Relatedly, if your game is derivative of a existing fiction or reality, merely being familiar with the source materials should not be enough to tell what claims are Town and what claims are Scum.
Bigger Is Better
One of the easiest mistakes to make is simple; the larger a single group is, the more powerful it is.
One six man scum group is more likely to cause a town defeat than two three man scum groups or three two man scum groups. In fact, I would go so far as to say it's actually more likely to win than two four man scum groups or three three man scum groups or five two man scum groups, and so on.
This
also applies to masons. There's a number of reasons for it, but the long and short of it is a larger block has better ability to control the vote and coordinate any secondary powers they have.
For example, a Cop and a Doctor is very standard for Town. In principle, the Cop can claim and get the Doctors protection, allowing him to reveal Town or Scum status of players indefinitely. However, if they stand alone, the doctor might not believe the Cop's claim, and the Cop might not realize there is a doctor. If they are in a mason, they can coordinate to start healing the Cop off the bat.
This goes further! If that mason has a third member, who has no powers of their own,
that player can feed the cop results to town- now even if Scum have a Strongman or Roleblocker, the Cop can give his information to Town from obscurity.
Likewise, multiple scum groups may push a lynch against each other or target each other with a nightkill. Also, they may achieve majority over town
without realizing it- a six strong scum group that whittles the total players down to 11 knows they have majority. If you had instead three two man scum groups, then
even if those scum groups
can win with each other, they won't
know they can know control the lynch freely. And they'll almost certainly crossfire by that point or sooner.
Context is Everything
You might say to me "Terrabrand, I have an idea for a role that is overpowered." I would say to you, you are wrong. I don't even need to know what the role is, I can create a setup where it is, in fact, completely worthless and cannot win. There are roles that are inherently bastard. There's no role too strong
in a vacuum. However, even common roles become too strong in the right configurations.
But let me expand on my comment, in case you don't believe me. Say you have a role that is permanently nightkill proof, permanently lynchproof, a godfather, and automatically wins night three. That seems completely unfair, right? In fact, we'll say the role is permanently strongman proof. But wait. You might realize something. A strongman is a kill that can't be blocked by nightkill proof; so being strongman proof is 'impossible' from a certain perspective. But that's the thing. You can always iterate such qualifiers. Your invincible guy who wins night 3? Worthless if there's a oneshot lynchproof guy who can't be jailed or roleblocked with a nightkill that kills even strongman proof targets and an objective to kill Mr Invincible with said kill night 1. Mr Invincible will simply die. That's frankly horrifically bastard, but it makes my point- there is no such thing as a role card that will
always in any context be overpowered. There is always a counter you can make.
Now, like I said, the point here is context is everything. A doctor is a fair role featured in most setups. Two doctors can create a loop where Scum can't kill either because as long as both believe the other is town they just heal each other. Two doctors in a mason just heal each other right off the bat- a doctor in a mason, while strong, is not actually unbeatable for scum. Twelve doctors in six two man masons could constitute a game where scum can't kill more or less at all. Of course, those twelve doctors in six two man masons don't do anything to stop scum if all scum kills are Strongman.
It is the
complete picture that matters. Powers for Town can constitute a hard stop to one scum group, but slot in one with different powers and all that strength vanishes. Everyone is bulletproof? Who cares, the scum are all strongmen. Everyone passively watches themselves for all visitors? Scum are all ninja. The list goes on.
Thus, one of the most critical things in setup design is looking at how things fit together. For example. Let's consider making a fourteen man setup. Can a fourteen man setup survive a four man scum group? Sure, that's pretty harsh on town but with appropriate roles on the town side it'll be fine. A serial killer to go with your scum? Definitely, it'll speed things up but it's not remotely out of the question. A second scum group? Rough, but not outside reason. An executioner aimed at Town's best role? No problem at all. A vigilante? Sure, that'll be fast.
Now put it all together.
1: Scum Group A
2: Scum Group A
3: Scum Group A
4: Scum Group A
5: Serial Killer
6: Scum Group B
7: Scum Group B (it's not a scum group without being a group!)
8: Executioner
9: Vigilante
10: Town's Best Role
11-14 Other town roles.
Right, so first of all we have five kills a cycle in a 14 player game. Town could be completely gone
at the end of day 2. Second of all, town barely has a plurality, not a majority.
Also, scum group B is just plain screwed compared to A unless they have amazing powers.
I could technically fill out the info to favor town anyways- give all the town roles multiple shots of bulletproof, all town roles are double voters, etc, but that only drives home my point further. There's a huge range of things you can reasonably feature in a given player count, but equally a huge range of combinations that cause huge problems. You have to pay attention to how things combine, not just how they stand on their own.
Specific things to watch for include without being limited to...
- Town should generally start as the majority. They are the uninformed majority. It's not really a proper town if they aren't the majority. You can have a townless game, of course, but don't feature a Town faction if they aren't the majority.
- Watch the kills per 'cycle'. After a day and a night, how many players are alive? Most likely case? Best case? Worst case? If player count goes down too fast, Town can't win unless there's multiple scum groups and those scum groups wind up crossfiring. If the players die too slowly, meanwhile, town will slowly close the noose around any scum, especially if they have any investigatives like cops to root them out with.
- If you have multiple scum groups, you need to not only give town a reasonable loss chance, you need all scum to have a credible chance of victory. If one scum group is four player and one is two, the two player scum group better have better powers or easier objectives. It's plain unfair if one scum group is just better than the other by a significant margin, and is always aggravating when you're the one who gets the no-win chance role.
- Watch for how town roles can combine with each other. Ditto for scum roles. And watch how roles interact in general. A lightning rod may or may not be a problem. A lightning rod in a setup with a doctor, a watcher, and a bunch of scum players with night powers, none of whom are invisible or whatever? That's a recipe for scum to all be out in the open after night 1.
Balance the setup, make individual roles fun
Balance is mostly a setup-level issue, as I said. But individual roles should all have merit. Some players think of Vanillas as boring, because they don't do anything special. But that just frees them up to go all in on stuff like scum hunting.
In general, any faction, be it scum or third party, should have its influence spread
reasonably evenly across the faction. A godfather and a roleblocker scumgroup has both halves bringing something to the table. Either could die and they could carry one without the other.
A generic goon paired with a godfather roleblocker, on the other hand, is basically doomed if they lose the latter. Not strictly, but the former is so much weaker than the latter that trying to win
as the former if the latter gets killed off is just not fun. Meanwhile, the latter can lose the former with comparatively little concern.
Not only is there the fun factor, but there's a much greater degree of swinginess. The more power tied up in a single role, the more it matters who exactly gets that role, and the more one unlucky nightkill or investigation or lynch can swing the outcome of the game.
Now, I speak on a faction level for a reason. Scum roles
should generally be more powerful than individual town- a Town Vanilla can vote and talk. A scum vanilla, that is a mafia goon, can vote, talk, talk in private with his allies, and perform a nightkill. He also knows who his allies are.
Obviously, a town of 'four three man masons where each mason can perform a nightkill per night' is gonna tend to trash a single three man scum group of goons.
Likewise, standalone roles like Serial Killers and Lynchers are compensated by either greater power or easier objectives. A Serial Killer
should be stronger than any one member of the main scum group- he doesn't have support of one or more allies. A Lyncher tends to be weaker, but makes up the difference by winning if they can get a single player lynched. No need to survive to the end or eliminate several other players, just get the one guy killed off and bam, you win. Lynchers can execute gambits normal scum would be insane to try- for example, they can claim false cop results or counterclaim what their target claims, and as long as they get their target lynched, the inevitable wrath of town for lying
doesn't matter.
Factions need to be balanced against other factions. So the natural result is that smaller factions
have to pack more power into each role.
I have a great idea...
For a role or power or whatever. Great! Put it in. But
make sure it doesn't break your setup. Interesting ideas are fine, but do consider them critically. Does it sound fun to play? Does it sound fun to deal with when you aren't it? Does it sound highly influential?
Does it break core precepts of the game? This is a big one.
For example, you could propose a medium who can unlimitedly keep talking while dead. That's an interesting idea right there. But if you make that a
town role, you've removed one of the fundamental precepts of Mafia. The scum can remove
whoever is the biggest threat. Town roles shouldn't generally get advantages like permanent nightkill immunity (exception: Scum has a strongman kill), and especially not while also being GM verifiable or lynchproof, because then scum is
unable to take them out if the role goes to a good player. This prevents the Mafia from executing basic strategy, and indeed can wind up
punishing them for trying.
Fluff Says X
Great. Use that as a starting point.
But keep in mind game balance. If some character should be able to kill literally every other character in a straight fight at the same time? They still need to be lynchable/nightkill-able/something, or else everyone not them can't win.
Powers need to work in context of the overall setup. What the fluff says
should be true takes a backseat to game design. If that offends you, don't go putting characters in that by fluff break the game.
Likewise, as I addressed before, you shouldn't be able to discern alignments and objectives by just looking at character claims and canon. Don't just make the bad guys the scum and the good guys Town. Also don't just flip it around. There needs to be the ability to doubt even once someone claims without being counterclaimed.
Role PMs, Night PMs, and Day Messages
So you have a reasonably solid setup. You're going to have a doctor and a watcher, and a two goon scum group, in a ten player setup, with the rest as town vanillas. This is
more or less balanced. So now you can just halfass the role cards and night PMs and so on, right?
Well, no. Actually, the role cards, night PMs, lynch messages, and so on are actually
more important than the raw mechanics. A serial killer whose role card leads him to believe he's town or neutral will play incorrectly, believing he wins differently than he actually does. A Watcher who causes people to receive Night PMs with fluff about eyes on them all night
has evidence for their claims that one who doesn't does not, but conversely reveals his actions to a degree even when he doesn't want to. A doctor who learns what kill flavor hit their victim through provided fluff also acts as an informational role in a way that a doctor who doesn't simply does not.
People judge likelihood of fake claims by things like the name breaking patterns. Or the fluff breaking thematics. In a game of Bob the Farmer, John the Blacksmith, Alice the Butcher, Kirk The Starfleet Captain, and The Village Hunter, everyone is gonna think the last two are
lying if they honestly claim. They don't sound like they're from the same role-list. They just don't fit in.
Fluff can hint at other roles, lending credence to or undermining the claims of others. Being told your character has a brother somewhere around here leads you to believe any claims that match. In a Yu-Gi-Oh game, telling the Seto player that Mokuba is around here will instantly make any non-counter claimed Mokuba claims sound more credible. Or alternatively, being told you are the
only priest in the village will cause you to believe anyone else claiming priest is lying- even if they're telling the truth.
This is especially important if you offer a safeclaim to a scum role, but even in general it matters quite a bit.
Think about what conclusions can be drawn from the info available to players. In their role cards, in their night PMs, in the opening and any day messages. These things allow players to narrow down details of the setup and make educated guesses about claims.
