Mafia Theory, balance and MYTOWN

[Apache]

So i've thought a bit about the theory of a mafia game, especially the balancing of games and examined it.

Factors of a game outcome

an important question is: what matters to the outcome of a game. basically there are only 3 things that affect the whole game
1) the setup
2) the player skill
3) the luck

these 3 factors determine all other factors
see this diagram i've made:


(here is an advanced complete diagram, but it looks very complex and is not needed for my further points (wasnt sure if i should post this masterpiece of art): https://s14.directupload.net/images/140310/wrqd2q2d.png)

however as you can see in the diagram, all comes down to lynches, kills and the setup.


Balance of setups

so the goal is to determine how balanced a setup is. you can see that the 2 main factors the setup influences are the outcome itself (by how many mislynches are allowed if no kill fails) and the night actions thus votes (example: sheriff targeting mafia will increase votes on that mafia usually) thus lynches (so in the end the setup affects how likely a mislynch is to happen) and thus ofc the outcome again
additionally, since the setup affects night actions it will also affect the night kills (so it affects how likely is a fail kill to happen, for example if there are more than 1 maf team or lots of protectives or how likely is it a vigi shoots a maf/town)

the MYTOWN scale

let's look at the 'how many mislynches are allowed'.
for this i'll need to introduce the MYTOWN scale (mislynches to win)

the MYTOWN value can be determined easily in every conventional setup (means town vs one or multiple scum teams vs neutrals, nothing special like 5 sided)
to determine the MYTOWN value we have to check, how many mislynches must theoretically happen until the scums win, assuming every kill goes through and nothing special like vigi kills happen. so how many mislynches till scum number and mafia number are equal or scums are more (all scum teams and killing neutrals count as scum, nonkilling neutrals don't count)
of course the MYTOWN value changes as the game progress (see below).

let's take a setup with only town and maf as a simple example. lets say it's d1 (lynch day) and there are 8 town vs 3 mafia. let's see, after n1 6:3, after n2 4:3. then a nolynch would be enough, which counts as 0.5 (see below)
so the MYTOWN value here would be 2.5

definition: the starting MYTOWN value is the MYTOWN value a game has at the beginning and can be determined for every conventional setup (or aproximately if the rolelist is somewhat unknown)

conlusion:
the mafia wins once it's 0 (happens if town number=scum number) or below (happens if scum are more than town). however it doesn't have to be a mafia win because:
a) there may be no mafia member left, a 0:0 has the MYTOWN of 0 but noone wins
b) it may be night start, so the kill just has to fail to recover MYTOWN (see below)
c) scum may have equal or more players than town, but they might be split among more teams (or killing neutrals) so no factions controls the vote and the game goes on
d) there may be a nonkilling neutral, so even if scum number = town number the scum doesn't control the lynch and game goes on
e) some hosts may keep the game going despite one scum faction controlling the lynch if town can still recover because of vet, vigi, bd etc (see below)

so the town can still recover (get the MYTOWN higher than 0) and win though that's unlikely if it's once 0 or below


now how do events change the MYTOWN value

let's start with such simple setups as the example:
mislynch: -1 (obviously, the scale is based on this)
no lynch: -0.5
failed kill: +0.5
kill on their own: +1
succesful lynch: +/- 0
vigi/jailor/vet/kidnapper kill on town: -0.5
vigi/jailor/vet kill on mafia: +0.5

now we look at it with non killing neutrals:
lynching a jester: -(amount of town suicides - amount of mafia suicides +1) * 0.5
lynching a survivor, exe, amne, student, etc: -0.5

sk/second mafia (more KPN):
the additional kpn will change the starting MYTOWN val. just calculate it through with 2 deaths each night
example: 11 town vs 3 maf vs sk. after n1: 8:3:1 after n2 5:3:1 =>1 kill enough for no town majority => nolynch and 1 failed kill enough=> starting MYTOWN val = 2.33 (see below)

most gains and drops depend on the kpn so here are general formulas (all can still be applied for 1 kpn aswell):
mislynch still -1
even a successful lynch will now remove (kpn-1)/(kpn+1)
a nolynch will now remove KPN/(KPN+1)
a failed kill will only give 1/(kpn+1)
kill on their own/other scum faction will give only 2/(Kpn+1)
vigi/jailor/vet kill on mafia will give 1/(kpn+1)
vigi/jailor/vet/kidnapper kill on town will remove 1/(kpn+1)
lynching surv etc removes KPN/(KPN+1)
lynching jester removes (town suicides - mafia suicides + KPN)/(KPN+1)

removing a kpn will force you to recalculate the MYTOWN val

you can see additional kpn in a setup will make it a LOT of harder, because the starting MYTOWN is much lower and additionally many events pull more/grant less. and even lynching successful EVERY time is not enough anymore if you have already lost too many points which wouldnt be possible with just 1 kpn


one thing is still missing, the cultist conversion. depending on how many conversions will most likely occur in the setup, you can either see it as a kpn or not.
if counted as kpn: successful conversion -1/(kpn+1); not succesful: +1/(kpn+1)
not counted: succesful: -2/(kpn+1); not succesful: nothing

arsonist is difficult because it has no fixed kpn, it can either have almost 1 kpn or none at all if lynched soon enough
the witch: either you count it as nonkilling neutral (lynch counts as surv lynch) or you count it as mafia member (reduces starting MYTOWN but lynch counts as successful)

whether d1 lynch is allowed or not or game starts at n1, doesnt affect the starting MYTOWN, it makes it just impossible to avoid a drop on d1 if nolynch or n1 start. but even if d1 lynch is allowed, a drop is almost certain because most likely ppl wont lynch and if they do, it will be a mislynch


for convenience another definition: one mislynch equivalent are actions that reduce MYTOWN by the same amount than a mislynch which is 1. example: 2 no lynches are a mislynch equivalent in a game with 1 kpn. conclusion: the MYTOWN tells us how many mislynch equivalents are needed so the mafia wins


so now what's the point of the scale. first it can be used during a game, to determine what has to happen so you win or to judge plans (for example would it help to lynch the jester)
second and most importantly it can be used to determine how balanced a setup is. since in most of our setups the chance of mislynch is pretty much the same and the chance of fail kill too (as long only one mafia team because more would increase the chance they kill each other) the MYTOWN is the most important factor of a setup. since luck is random and player skill is equal on both sides (unless you deliberately put all good players in one team or you'd say on a site all players are one sided talented, so eg in general much worse as town than as mafia compared to all other sites. in this case you'd have to make setups more town sided specifically on this one sided gifted site) the MYTOWN is the most important factor for the whole outcome of any game

take fm 20 for example:
23 townies
11 scums (without student who became scum)
3 kpn at start (without arso) (5 from n4)
that's 3 starting MYTOWN (not accounting n4 kpn increase so in reality it's even lower
and that's also how the lynches went pretty much, 3 mislynch equivalents happened because of nolynch, student and jester lynch, vigi-/jailor-/modkill, marshall mislynch. normally there should have been a bit more allowed than the starting MYTOWN because of failed night kills and the reduce of kpn on d2 with sashas lynch, both would increase the MYTOWN but the increase of kpn in n4 decreased it again

optimal starting MYTOWN

so now the actual question is: what starting MYTOWN values should setups have? obviously, the more players the more mislynches must be allowed because there will be more lynches in total. maybe 2 for sfm, 3 for mfm and 4 for big fm (however less if 2 big scum teams). for exact values we'd need detailed statistics




thoughts? mistakes? discuss

[louiswill]

There should also be a link between setup and luck. Setup decide the maximized chance of productive night action.

There also missing a square called "Talk"---meta and popularity does have influence upon people.
---unless you throw those into skills too.

[Apache]

There should also be a link between setup and luck. Setup decide the maximized chance of productive night action.

There also missing a square called "Talk"---meta and popularity does have influence upon people.
---unless you throw those into skills too.

it depends on how you define setup. is it what you see in the setup thread (with hidden town, hidden mafia etc) because that's not affected by luck at all and solely by what the host wants. in this case luck would not affect what we define as setup, but it could still determine what roles are in game (if host randoms), thus what's actually affected are the night actions (example: will there be a doctor heal or a sheriff check for the hidden town)

on the other hand if it is what finally will be played, with all hidden roles determined, it can be affected by luck if the host really randoms the hidden roles instead of choosing them (though it would be optimal if all likely outcomes of the randoming have the same starting MYTOWN and nearly the same mislynch and failkill chance (or mislynch and failkill chance change mutually, example sheriff instead of doctor will decrease both chances so it may be ok if proportion is right), so luck is a minor factor of the setup).

so what definition makes more sense:

for the host himself:
if the hidden roles are picked, the latter definition makes sense, because you know the hidden roles and can use this info to balance the setup
if they are randomed, the first definition makes more sense, as the host doesnt know what will be randomed so the setup has to be balanced without the knowledge for every likely outcome

for others:
no matter if picked or random, the first one makes sense, because you don't know the picked/random roles and have to judge the balance without that info

so you can see if the roles are randomed, the second definition makes no sense that's why i didn't make a connection
we could also add both definitions to the diagram, the visible setup and the final setup i call them now, with the final setup affected by the visible setup and luck (only if host randoms) and both, final and visible setup affecting pretty much all other factors (the final setup directly and the visible setup because that's what players analyze)



and you're right, past games (by what is ppl's opinion on others) are a factor and it would be on the top. this is a factor that shouldn't be there in an optimal game and it rather favors the town, because the more ppl know each other, the better they can tell who acts differently. so this factor can be used along with skill to form opinions on player's alignments or skill and thus determines almost everything (example: someone acts differently => votes on him or someone is known to be a good player => kill on him)
if anonymous accounts are used, past games are a less important factor but can still be used with the skill to hunt COMs or prevented from being used by the skill to hide your COM

however you can also say past games are only a factor of skill and nothing else (and are thus included in skill) because it improves the skill of the players to find the scums or to find smart townies for the kill

[Yayap]

Well I was going to write up a big post about FM balancing... but I see someone has beaten me to most of it. I guess you leave me with balancing specific roles/mechanics in a setup.

[Apache]

well if you have anything to add it would be great though

[Yayap]

So Apache has taken care of the basics of balancing the mislynches to win aspect of making a setup. Unfortunately most hosts make the mistake of ending their balance testing there. The other aspect of balancing involves probability of causing a mislynch throughout the game which takes a much more in dept analysis of a setup as well as forethought.

It's nice to say that all scum need to do is cause.. say 3-4 mislynches to win. Sounds simple enough, doesn't it? The mistake most hosts make is not factoring in how hard that becomes with their selection of PRs.

Lets start off with the biggest factor: Investigation ability.

I clearly call it investigation ability and not investigative roles for a reason, that is because roles from other non investigative groups do have some investigative abilities. Here is a general list of common roles that have some sort of investigative ability:

Sheriff/Insane sheriff: finds alignment.

Investigator: finds role/ability possibilities

Detective: finds the target and can deduce the exact role if the person targeted gets feedback and shares that info. Example: person was witched.. detectives target is most likely the witch.

Lookout: similar to a detective except in reverse, will get all the names of the people who targeted the person the lookout watched. This is even more powerful than the detective because (going back to the example in the detectives) a lookout knows if there a second person that could be the witch if multiple people visited.

Journalist: if the interviewee claims a role, the journalist knows who it refers to.

Jailor: Jailed targets talk allot, usually forced into claiming a role if no scum jailor exists. Also - because of the night chat, jailors may learn codded messages sent by other investigative PRs they detain.

Architect: Similar to the jailor night chat - can learn codes and organize a town in a hurry.

Escort: If all scum actions are accounted for by people claiming feedbacks during the day, escorts can deduce their target was town. Or if certain actions are missing, repeated blocking can confirm if the rb was responsible for the missing action.

Witch: In similar fashion to the escort, the witch can deduce what his target is based on feedback reported the next day.

Mason Leader/Recruiter: Confirms citizens if recruited.

Cult/Cult Leader: Gains and shares all info with every new recruit they get.

Executioner: if the setting that target is always town is on - he knows that person is obviously town.

Ninja and soul thief: exact role investigator (but not alignment)

Mayors/Marshalls/Students: can easily prove their roles and thus eliminating themselves from the equation.



So to cause a true mislynch, the idea is that none of the above roles detected scum.. now if you factor in that most setups have a near 3:1 ratio between town and non town.. you can see that a sheriff alone has a 33% chance of checking non town at the start, factor in all the rest of the investigative abilities and its nearly impossible for all scum to remain undetected and all it takes is 1 scum to be found to eliminate the chances of a mislynch.

Now if scum need to cause 4 of these mission impossible feats... you can see that scum have little chance.

Of course, this does not factor in that not all PRs reveal as soon as they have info.. but eliminating a mislynch on the day they get the info or holding onto it for later still eliminates 1 mislynch, the difference being later might have more than 1 target up for hanging at the same time.

Another miscalculation most hosts make is the time needed for full coverage of investigation roles. Meaning they calculate the amount of time it would take all the invest roles to check all the players for information. If we take the example of our sheriff, he does not need to check 100% of the players to complete this goal, because KPN eliminates some players as well as process of elimination aka if the sheriff finds 4 townies and there is only 4 town left alive, even if he hasn't checked anyone else, he can deduce that they are all scum (ignoring any immunity).

The more investigative roles in a setup, the less time it will take for full town coverage.

There are 2 phases in most FMs:
1- acquiring information and the guessing games
2- clean up

In the best balanced setups, phase 1 should never end. Meta on this sites setups however show that it usually ends around day 5-6. What do I mean by that? I mean by day 5-6, every role is known and exposed and its just a matter of night actions/open negotiating with other players - aka full town information coverage has been reached.

The problem is too much information is piped through the PRs to the players too fast.

So how should hosts correct his problem? Usually by either reducing the amount of information roles in their setups or nerfing them with either counter roles or charge limitations. Unfortunately, this is not what most hosts are doing, instead they are creating reverse mafia setups with high KPN and a very low MYTOWN... where a single mistake on towns part spells doom.



Next part will deal with counter roles/mechanics used to nerf invest abilities.

[Voss]

please continue with this! i'd like to read more about peoples way to balance mafia games!

[Yayap]

been busy irl... will get to this when I have time to think and write an hours worth... dont worry, I haven't forgotten

[Apache]

also not only mayor etc can confirm themselves, many prs can if their can be no counter role in the mafia (or if ppl know there is none because the mafia roles are known because of their feedback)

so that basicly means we should make setups with few prs, so few info will be found and many citizens, to keep the MYTOWN high enough and to prevent mass roleclaims (which also often lead to phase 2)
ppl don't like being citizen so much, but it will also increase the skill to scumhunt if there are more of them and ppl cant rely on prs

[Apache]

so yayap started to deal with the "how likely is a mislynch to happen" aspect of a setup. i will pick up on this

but first about a little different topic: how many prs are needed to find all scums. apparently the "complete coverage" theory is bullshit, so here is a different theory.

a sheriff in average finds one scum in 13 ppl fms, estimatingly. some reveal once they found one and die (not accounting protectives). some wait until they have 2 scums but probably half of them die without revealing the info and thus get no scum at all. so the mean is 1 [1*x + (0*0.5 + 2 * 0.5)*(1-x) with x being the probability a sheriff reveals instantly or waits]. in bigger fms they find more, because they live longer. i'd say in 30 ppl fms it's maybe 2.
this is of course without last will. with last will, they tend to wait longer until they reveal because they can put all info in lw. and no info can be lost through a kill so average finding increases significantly to maybe 1.5 with 13 ppl and 2.5-3 in 30 ppl

so what does that tell us. the amound of roles with investigative abilities in a setup should be so, that in average not all players would be detected by them (we want some to be found by post analysis too of course). that means for example in a sfm with 8 town, 3 mafia and 2 neutral (example jester and survivor) (starting MYTOWN = 2.5 which is good) town should consist of 2 sheriffs, if no lw allowed. 2 mafias will be found in average and 1 must be found by analysis. rest of town would be citizens only and gf (sheriff immune) + mafiosos in mafia.

of course we dont want games of only sheriffs and citizens. different roles have different scum finding abilities. the sheriff is probably the most successful. but invest, det, lookout, escort etc all have such abilities too. and even a doc would increase the investigation ability of town because he can make sheriffs etc more productive, maybe increasing the investigation ability of town by 0.5
maybe add a doc to our example setup so 2.5 of the 3 will be found by prs in average, depends on how good the player skill is to find scum without prs

so besides the mytown i will now introduce the next setup factor: town's investigation ability
almost every town role that is in the game will increase this score. total investigation ability score should be always lower than the number of scums.
the score of each single role is preset but is determined by factors like game size, role options, last will allowed or not etc
citizens and killing roles are pretty much the only roles that do not increase the score.
mafia and neutral roles can decrease the score. example: janitor who cleans lw of investigative role, consort that blocks them or even a jester who puts town in general in a worse slot by decreasing their mytown and thus also lowering the time investigatives have to find scum before game ends/they die (death more likely because less targets)

something else we found out now is that bigger games require more roles in general but NOT more prs. so the new roles should be filled up with citizens to the most part. that is because the average finding chance of all prs will increase proportional with the game size (due to more meatshields of cits) which completely compensates the increased scum number


however we can also introduce another scale to setups. let's go back to the original "how likely is a mislynch to happen". i call it the mislynch chance. it is affected by the night actions and thus alive roles and the town:scum ratio (and by the skill of ppl). for example if there are more sheriffs, a mislynch is less likely to happen as yayap pointed out. it may increase or decrease while the game goes on, but usually it increases, because the town:scum ratio remains pretty much equal until the game is decided but the amount of investigative roles will lower in late game

each setup then has a starting mislynch chance. it is determined by the roles that are in a setup (assuming skill wouldnt matter) and the scum:town ratio. every role with investigation ability affects it. the starting mislynch chance describes the mislynch chance of d2 assuming no one was lynched d1. basicly to factors of roles are important: the confirm ability and the find ability.

let's create an example with 13 ppl, 8 town, 3 scum, jester and survivor. town has 2 sheriffs and a doc, like in the example above and one citizen with confirm ability and rest citizens. mafia has a gf who is immune to sheriff and mafiosos. we count jester lynch and survivor lynch as mislynches too (though surv would actually be only 0.5 mislynch equivalents, similar to nolynch but we won't account for this now)

the starting mislynch chance can simply be calculated as follows
first of all we take out all confirmable roles of of the equation, leaves 3 successful lynch targets and 9 mislynch targets. that returns a mislynch chance of 9/12=3/4. so much for the confirm ability.
but we still have the find ability. in night 1, one sheriff has 2/12=1/6 (excluding gf in numerator and himself in denominator) to find a scum. we have 2 sheriffs, the chance at least one of them finds a scum would be 1-(5/6)^2=0.31
but i feel like it's better to use the mean here, because the above doesn't differ between 1 find or 2 finds, which wouldn't play a role on d2 but it would carry over to d3.
the mean would be 0.33, but actually 0.32 because they could both find the same scum (but you can ignore this aswell since it doesnt really matter). unfortunately using the mean will make the actual chance to mislynch on d2 slightly wrong but i think using it allows us to better judge and compare the balance of setups (difference gets bigger 0.01 with the number of investigative roles). ofc mean could also be more than 1 but in this case your setup is flawed anyway.
so we have a chance of 0.32 to lynch a scum right away which would of course prevent the mislynch. in the other 68% of cases we still have our 75% mislynch chance.
so the total mislynch chance would be: 0.68*0.75 = 0.51

since we figured out that the used setup is balanced above (one confirmable citizen doesn't really influence the result), 0.5 must be a good value. since we also figured out that bigger fms shouldnt have more investigative roles, we can conclude that 0.5 is a good value for all game sizes (let's say +/- 0.05 can be tolerated). with a third sheriff in our example we would be below 0.4 which is too low

that 0.5 is a good value can be backed up by a simple calculation: the setup has a mytown of 2.5. each day the mytown will be decreased by an average of 0.5, so the game would end after 5 (2.5/0.5) days with a mytown of 0 and left mafia of 3-5*(1-0.5)=0.5
0 would be the perfect value (perfect draw) so both sides have an equal chance of winning. i said the mislynch chance would increase over time, so the remaining mafia would even be more in reality. but we didn't account for one important thing: lynches aren't random. people have skill too. this compensates the too high result.
or should at least, if i am right about the site's skill to lynch the scums by analysis. if i'm wrong about the skill, the mislynch chance must be lower or higher (changing this value by for example adding prs will also change the investigation ability but it must be closer or further from the actual scum number anyway if we want to adapt to the skill so that's good)

it's obvious that if for some reason you increase the mislynch chance (for example cit only game) you also have to increase the starting mytown. this calculation gives you a good estimation of how much, as it shows how mytown and the new value should relate to each other.
it also becomes clear that if you allow lws, you will have less prs according to the first factor and thus obviously the mislynch chance will and must be higher in such setups but that's ok, because the last will can prevent mislynches once a sheriff dies

mafia roles also affect the mislynch chance, take for example a framer who adds a chance to cause an instant mislynch or just a consort who can block a sheriff.

it is disputable what the chance of investigative roles like det or invest is. the important question here is how many of the scums can it detect. for detective definitely the gf, but also scums like the janitor or maybe even consort. for invest it depends on the pairings. but for all roles you can say, certainly it doesn't get as high as a sheriff, who is definitely the most powerful investigative role (unless you make all roles immune to sheriff lol)
in our example a det would only be able to detect the gf and is thus half as effective as a sheriff who can detect both mafiosos

the value can be calculated for all setups, though it might be very complex and some things may be estimated.

protective roles do not affect the mislynch chance (at least by the protection itself, they may still be confirmable) but they help to stop the increase of it





both new factors have their advantages and disadvantages. the first for example accounts protective roles and lws better. the second however accounts the confirm ability which the first doesnt.

as always, exact data requires statistics. i'd start some sfms to see if my estimated values are right, but since ppl wouldn't play it it's pointless
things i'm unsure of are for example how much scums a sheriff finds in average or the skill of the players here
in particular i'd like to know how my example game would play out

still i feel like this is a huge step forward for balancing my own games and i hope it helps you aswell. there are now 3 values you can check for the balance of your setup
as always, feedback appreciated. and sry for the broken english

note:
roles with investigative ability refers to what yayap introduced

[Apache]

so in general you could say that the following values are ok:
mytown a bit lower than scum number (scum = all scum teams + nk + witch)
investigation ability a bit lower than scum number
starting mislynch chance about 50% if no lw, if lws are active maybe 55%

[Yayap]

I said I'd post this so apologies if I rambling on about something Apache already covered.

So now that investigative abilities has been covered, we will look at the different ways hosts counter those abilities for scum not to be defenseless.


The first and probably the most obvious is investigative immunity mechanics, this includes GFs, Beguilers hiding behind town roles, Framers that can hide their own mafia, as well as any custom mechanics that grant immunity to the mafia that the host might put in.

This type is most effective against Sheriffs, however when looking at investigators, lookouts and detectives, this can pin a scum into the citizen claim and if there are few citizens in the setup and if masons present, this is also a death trap. Town is less hesitant to lynch/night kill a citizen claimer than an important PR.


Mirrored roles: This would be all the scum roles that are an almost duplicate of a town version of the role. Consigs, Consorts, Kidnappers, Chauffeurs, Morticians, Trackers, Arms Dealers, Evil Journalists, even Mafiosos.

These roles have little or no investigation protection against sheriffs but have more options if detected by investigators, lookouts and detectives. Their alignments can't be proven and night actions/post analysis come into play as evidence to determine that, good liars can usually delay a lynch or even cause a mislynch on their counter-parts with CCs. The other thing mirror roles do is add uncertainty to real town roles when they claim, adding doubt to the mix instead of 100% follow PR claims. Keep in mind that any mafia can fake being a sheriff because they already know all the results the sheriff would get... and even if they say that an neutral killer is clean... that neutral isn't going to argue that fact.


Multiple Sheriff versions: (Naive, Corrupt, Insane.. etc) This is more of a delay mechanic than a counter, but in some cases it can lead to a mislynch in early game. But once the sheriffs figure out how to translate their results, its retro-active, in other words, they can translate the results they got from early game as well. I'm not particularly fond of this mechanic myself and it usually is used in smaller setups rather that big games.


Framers: To be honest, framers are more of a fear tactic role than anything else. They mostly prevent investigative roles from targeting the obvious choice, or if they do, the accuracy of their results are in doubt. Framers are lucky if their night actions do coincide with an investigative role but then have to deal with the obvious defense of "I was framed". And if that person was the obvious investigative choice, the town might believe him and let that person live a bit longer. The other problem framers have is that if the person being framed has a solid claim he can rely on or has information only that kind of role would have, the framing is almost useless. In my opinion, a default settings framer can be one of the weakest counters in mafia in a game with a wide variety of PRs.


Disguiser: Ignoring post analysis and codes, disguising in a late game has the effect that all previous investigation results are null until the disguiser is found. Sounds good doesn't it? Now lets factor in the post analysis and codes.. I've yet to see a disguiser accomplish this task and not be discovered almost immediately. Thus Disguisers are more effective in sc2 where very little is said during the day. In FMs, they are mostly used to get a boost in kpn in the early stages of the game before player's personalities emerge. The only benefit the disguiser usually gets is if his target is cleaned and there is no coroner to verify anything.


Graveyard Alterating Roles: Although less of a counter to investigative roles, this counters more post-analysis than anything. These roles include: Disguisers that clean, Janitors, Actresses, Fabricators, Lawyers and Tailors. These are the bane of the mathematicians that keep track of what roles are in the game and what possibilities are left. Keeping these players off their game is essential as well for a scum team to succeed. However keep in mind that a single Coroner can usually negate everything these roles accomplish.


Direct Interference Roles: All Killing Roles, Kidnappers, Consorts, Drug Dealers, Witches. These are the roles that can directly prevent an investigative role from gaining any credible information at all... which in turn causes more protective/investigative roles to watch each others back. The Lookout is the nemesis of most of these roles unless they have some sort of immunity which brings us back to the first counter.


Enforcers: I've always considered the masons to be one of the biggest threats to any scum team. Various methods have been tried to limit their power but in recent games, they are gaining more power and less risk which is a concern for me. What I mean by that: in the early versions of FMs, if a mason tried to recruit a mafia - the mason was killed... this hasn't been the case in a long time. Now we have been depending on Enforcers to prevent conversions. Enforcers have to be really lucky to target the same citizen a mason is going to recruit. This limits the number of citizens mafia can hide behind - especially those that are pinned to a cit claim because of immunities and it creates an informed town that can hunt down the scum rather quickly. Hosts have been adjusting how many players an enforcer can target each night to make it easier for the enforcer to be successful.


The Infamous Cult: This is probably the most hated counter in big FMs. They undermine every single role in the game, and until they are purged, no role is safe and no role is confirmed. Even the mafia are not safe with this group running around. A single converted mafia = entire team is exposed.


Special mechanics limiting use of PRs: Limited number of shots for a Vig, limited alerts for a Vet, sound familiar? This not only limits those roles but can also confuse detectives who would get information similar to a citizen. This mechanic is becoming more and more popular as time goes on, expanding it to other roles as well. This mechanic will be used extensively in my next FM.



The next thing to discuss in this interesting thread would be Towns Killing Power.