[OLD] Open Setup Ideas and Discussion

[callforjudgement]

Pressure Mafia

Nightless

3 Mafia (White Flag)

2 Innocent Childs
2 Vanilla Townies

Seems very townsided. The town needs 2 correct lynches (out of 5 targets, 3 of which are scum) to win; the scum need to force 2 mislynches to win, and only have 2 VTs to land them on. If the lynches are done purely randomly, it's equivalent to considering 3 out of the 5 unconfirmed players to be picked at random and lynched, and out of the 10 ways to do that, 3 lead to scum win (VT, VT, Goon * 3), 7 lead to town win (Goon, Goon, Goon *1; Goon, Goon, VT * 6), so it's 70% town win even without scumhunting. Arguably having a scumteam that strong would mean that the winrate would be less than 70% in practice, but I'm not sure, especially with two confirmed townies.

EDIT: the EV isn't 50% like you say, because nobody ever random-lynches an Innocent Child, so that should be factored out of the calculations.
RE-EDIT: Oh right, I missed that 3 scum vs. 3 townies doesn't lead to random-lynches either, but rather to a scum win, so the above calcs are wrong because of the day 1 lylo. I don't really like the concept of a day 1 lylo...

[callforjudgement]

Looks a tad town-sided, but within tolerance. I wouldn't run it without some form of confirmation voting system though - otherwise every town vote risks a quicklynch.

You could use delayed hammer, which is perfect at stopping quicklynches (basically, anyone on the wagon can unhammer via unvoting within 24 hours, and the day continues). It changes the dynamics of the game somewhat, but I suspect a day-1 lylo would translate to it anyway (people would FoS or fake-vote, and then when they were all agreed on the day's vote, only then actually vote).

[callforjudgement]

I think I found a partial breaking strategy in SS3 (although the expected value is still 50% because it's symmetrical between the scum and VT, so the scum can play identically to a VT and get a 50% win chance no matter what): the supersaint is basically equivalent to a reveal-any-time Innocent Child, as with two supersaint claims, everyone unvotes, then the (now-confirmed) VT votes for one and the other hammers, which leads to a town win no matter which way round it happens (and anyone who refuses to follow that strategy is claiming scum). Thus, as supersaint is not a possible fakeclaim for scum, the supersaint can confirm themselves at any time via claiming.

Because the SS is confirmable, optimal strategy is for nobody to claim, then for the players to scumhunt amongst themselves, FoSing rather than voting. Once someone has two FoSes on them, they claim; if they claim VT they die (leading to a town loss if the VT looked scummiest and win if the scum looked scummiest), if they claim SS and are counterclaimed town wins, if they claim SS and are not counterclaimed, the SS gets to pick which of the other two are scummier. Still 50% EV because of the symmetry, but if the town form a list of scummiest -> least scummy, they win if the scum is above the VT on the list, which isn't a ridiculous thing for the town to manage. At least, the strategy, AFAICT, gets the town's winchance in practical games as high as possible.

(Open) SS3 mostly isn't a played setup here anyway, so this is mostly academic, but it's documented and may be played on scumchat, so I thought I'd spill the breaking strategy here for reference purposes.

(I'm currently trying to find a breaking strategy at the other end of the micro/large scale, in Picking Simplicity. I think a modified follow-the-cop strategy is viable, but haven't proved it yet; it depends a lot on how quickly the SK can be effectively brought under town's control, which I think might even be possible day 1 with the right strategy and optimal play on all sides.)

[callforjudgement]

Town's only method of winning is to determine who the Mafia are, and lynch them on the correct day. This seems very broken, as if the scum don't let on which of them is odd-day and which of them is even-day, town have only a 50% win chance even if they figure out all the other information perfectly.

[callforjudgement]

As far as I know, there's no minimum experience requirement for suggesting open set-ups...

It seems reasonable for a 1:4 setup, but scumhunting is generally more interesting if the scumteam has at least two members (otherwise, there's nothing really to look for). I think Vengeful is designed specifically to fit a 2-man scumteam into a 5p game, and this setup misses that point.

[callforjudgement]

If I'm reading this right: the root of the problem is that no one actually wants to play 1:n games?

Pretty much. The problem is that scum have no incentive to do anything but play like a townie. (Whereas in 2:n games and higher, scum are also concerned with trying to keep their scumbuddies alive, and optimal scum play ends up not identical to optimal town play as a result.)

[callforjudgement]

I think must lynch was implied, but not specifically stated. So yes, mustlynch.

Definitely needs plurality deadlines too, for much the same reason.

[callforjudgement]

There's been a pretty interesting mini theme game with basically that mechanic run already: Spare Me Mafia (Mini 1148). I think it showed that the mechanic can work quite well; the major issue is preventing the scum cross-sparing, and I don't see how the setup you have here avoids the problem. (Also, why no scum QT?) You can't simply ban scum from sparing each other, as there'd be a breaking strategy involving repeatedly moving votes around and waiting for votecounts. There may be some other way to fix it, though (the simplest is to ban cross-sparing between any two players, not just the scum).

I'm also not sure how many VTs are needed to make it balanced.

[callforjudgement]

I think restricting cults to even-night or odd-night recruit (effectively what BBmolla's post above does) is one of the most necessary fixes for balancing cults, although it isn't enough by itself. I'd also suggest making the recruit

factional

; this is a large buff for cult, but also makes the game a lot less swingy (and some weakening of the cult would be necessary to counterbalance that).

There are plenty of bad ways to do cults. I think it's also possible to do them well, although that's pretty rare.

And that reminds me of this:

Mountainous Cult

• 2 Even-Night Cult Members (even-night factional recruit, no nightkill or any other abilities but the recruit, nighttalk but not daytalk)
  
• 10 Vanilla Townies

Not sure how balanced it is, but hopefully it brings the swing down somewhat. Perhaps it needs an investigative role or the like in there, to make it less mountainous, as I'm not sure it'd be town-sided with

any

number of VTs.

[callforjudgement]

Intense Mafia

• 5 1-shot Killproof Townies
  
• 2 1-shot Killproof Mafia

I think there's a hole for a small setup that takes a long time to play. The scum aren't allowed to shoot each other (I think the ability to do so would make the game much more scumsided, as shooting townies would no longer confirm them due to the amount of WIFOM involved, even if the scum always aimed at town).

[callforjudgement]

So is it revealed when someone gets hit?

You know how many hitpoints everyone has left, but they don't flip until they die. Sorry, should have made that clearer. (It does matter, otherwise the scum could no-kill and then fakeclaim being damaged.)

[callforjudgement]

Call me back when a town wins this.

Actually, there's a trivial breaking strategy for town (the dayvig confirms by shooting someone, and the town repeatedly no-lynch until the dayvig is the only person alive). Only way the jesters could win would be if they were the last two players remaining, which would be very unlikely.

[callforjudgement]

Lovers Coral

Mafia A

2 Mafia Goons(One at random is in love and not told)

Mafia B

2 Mafia Goons(One at random is in love and not told)

Town

2 Townies in Love(Told they are in love but not with who)
Love Guru
9 Townies

Third Party

Loved SK

If the loved SK dies everyone in love with him dies.
The Love Guru get a yes or no about someone being in love, the loved SK come back as a yes too, because he loves him self.

Isn't optimal town strategy there for the loved townies to volunteer to be lynched? Perhaps on day 2 or 3, so as to allow relational tells to build up first. Then the Love Guru effectively becomes an FBI Agent, which sucks for the SK.

[callforjudgement]

7:2 mountainous is scumsided. 7:2 mountainous no flip is thus even more scumsided, as no flip is definitely a scumsided mechanic. (Not to mention that no flip, and equivalents like what's being suggested here, are unpopular mechanics.)

[callforjudgement]

Basically, because correct play in a 12:2 as town is to no-lynch at some point (either mylo or day 1, not sure which is better), and scum can simply kill the towniest player. So it's effectively an 11:2 where scum have some say over the playerlist, thus town is weaker.

[callforjudgement]

I don't think this is only the case in Opens; scum has a higher win percentage in every sort of setup.

[callforjudgement]

I think part of the problem is that games are generally designed for a perfectly logical, competent town who are all trying to win, and that often isn't the case in practice. (Scum can play badly just like town can, but that tends to have less of an effect in practice for some reason.)

[callforjudgement]

re Phillamon's setup: doesn't it have the WitP problem, where the optimal play is town massclaim (on the basis that you have more than a 50% chance to be on the larger side, if a townie)? WitP dealt with that using hidden daykills, but I don't see a similar solution here.

[callforjudgement]

I mean, there are only two alignments. Scum, and Town. Alignment flip is unnecessary because they will flip town, as there is only scum who have NKs, and scum self killing in a vanilla+investigative game is MORONIC.

I've seen something very similar done offsite (although the setup wasn't quite so simple, it was close; a cop/doc/vig variation). The mafia member in question maintained it was the right play afterwards; they'd killed the cop and were trying to make out that the dead scum was the real cop, and it worked right up to the day after the day the town thought was lylo :)

[callforjudgement]

10 v 3 with possible confirmed town, I'd have thought, would be fairer, though. As for the anti mass-claim, surely that's by definition a problem with all opens? How do most non-variables stop it being a gamebreaker?

Either by having very few power roles, or with a specific anti-mass-claim mechanic (normally involving some sort of scum power).

[callforjudgement]

10 v 3 with possible confirmed town, I'd have thought, would be fairer, though. As for the anti mass-claim, surely that's by definition a problem with all opens? How do most non-variables stop it being a gamebreaker?

What's to stop scum mostly claiming oracle?

[callforjudgement]

That's why I said "mostly"; say, all but one. They're still going to be more likely to hit town than scum lynching within the oracles, and meanwhile, scum can NK the VTs.

It's the strategy I'd be suggesting to my buddies if I were scum in that setup.

[callforjudgement]

Junpei's setup would suck for groupscum, but I'm not sure if the SK is balanced against the town in it.

[callforjudgement]

11:2 isn't a particularly fun play balance; both scum and town get demoralised after a while, unless there's a scumflip early (then it's just scum who get demoralised).

[callforjudgement]

Let's see, with those numbers, you're going to want 2:4 to 2:6, which means that you're going to need a lot of town power.

The "standard" 2:4 is nightless white flag, which makes for a neat mountainous game, if you want something simple. Likewise, for 2:5 there's 7P vengeful (5 Vengeful Townies who can only make their vengekill in 5p lylo, 1 Goon, 1 Godfather who automatically loses his faction the game if he gets lynched, but not if he gets vengekilled), most of the small doc/cop setups (3 VT, doc, macho cop, 2 goons is reasonable).

For weird mechanics, Nomination Mafia has been cut down all the way to 7p because it was townsided with more players. (Basically, it's a 2:5 alternates between day phases where town can lynch anyone, and night phases where people can still talk but must lynch from a pool of 3 players chosen by the scum, which might or might not contain any actual scum. Scum can daytalk as well as nighttalk.) And Fight or Flight is a 2:5 with no regular nightkill, with every single player having a 1-shot commute (action immunity for the night) and 1-shot vig.

[callforjudgement]

If none of the scum know each other, there are no associative tells, and you don't really have an informed minority.

[callforjudgement]

I guess for 6p, nightless with a town power role would work. Perhaps 3 VT + 1 Governor vs. 2 Goons (no NK); most power roles don't work so well in a nightless game.

[callforjudgement]

Town Silencer is effectively an Innocent Child, there (they can't be counterclaimed as they can confirm their role by silencing the counterclaimer). So it's probably balanced or slightly scumsided, and the role might make for some fun apart from that. Giving the scum one as well would make the setup somewhat more scumsided, although not by much.

[callforjudgement]

I've been thinking about this for a while, but after playing in Worst Roles Mafia, it seemed to become particularly viable.

Keep It Balanced Mafia

• 5 Vanilla Townies
• 3 Mafia Goons
• 3 Werewolf Goons

If all players die simultaneously, or all townies die leaving an equal number of Mafia and Wolves, town wins even though they have no surviving members.

Working out the strategy for this one is really confusing. (For instance, in 1:1:1 lylo, it's optimal play for town to self-vote.)

[callforjudgement]

The bargaining between the scumteams. Deciding whether to crosskill or not to crosskill really was a strong part of that game from the scum play, and to a lesser extent, playing the scumteams off against each other (see: the enslavement of saulres) was noticeable in the townplay, towards the end.

[callforjudgement]

Even with the RB, isn't it broken by circle-targeting?

[callforjudgement]

AitP is famously broken; vent standard looks screwed-up to me (in particular, 7p nightstart 2:5 puts the town at mylo to

start with

, so they're obviously going to no-lynch to get another cop result, and they're probably going to do it as follow-the-cop because you didn't stop them; and even then, it's probably going to end up scumsided and mostly based on N0 luck).

[callforjudgement]

HOw is aitp famously broken?

Town's best strategy is for someone (a different person each day to stop the assassin making multiple attempts) to vote a random non-king player in their first post and everyone to sheep them, no discussion. The assassin's only method of winning is to correctly guess the king

and

beat everyone else to first post. Not sheeping or trying to discuss gets you instantly lynched, of course.

This is not a particularly fun strategy to play with.

[callforjudgement]

I assume that cop immunity doesn't work against an FBI agent? Also, is the ninja/strongman kill both ninja and strongman, or SK's choice when making the kill, or SK's choice when purchasing it?

[callforjudgement]

11:2 has never been won by town, right?

The major issue with n:2 mountainous, though, is that it's really demoralizing for both factions. Scum can't do better than lynch a townie every day, day after day. And they're probably going to manage it several days in a row. It may be that scum win it so much, not because of any balance issues, but just because everyone gets bored and trollish around day 4 or 5.

[callforjudgement]

I don't think your setup has the same problems as 11:2; I just elaborating on Hoopla's response to Junpei. It might have different problems, though. (I wouldn't enjoy playing in it because I prefer a small playerlist and plenty of time for discussion, whereas that mechanic seems designed for the opposite; people with different preferences from me might well enjoy it, though.)

[callforjudgement]

In mountainous, each extra scum makes a big difference, in theory. I remember seeing a calculation that you need something like fifty townies to balance three scum in a mountainous (and that setup would suck and should never be run ever). 11v2 is close to theoretically balanced (I can't remember whether it's 11v2 or 13v2 that's better-balanced), but the theory is reasonably dubious because town has never actually won it.

Extra town strength goes a long way to curtailing scum's power, though.

[callforjudgement]

Hoopla, wrt that Near-inspired setup, would it help to cause Mafia to lose their NK for one night after a member is lynched? As in, if a Goon gets lynched, scum can't kill the following night.

[callforjudgement]

The only reason for the scum not to join is that it takes away their second nightkill, right?

OTOH, the second nightkill is pretty powerful. Interesting choice.

[callforjudgement]

Watcher's arguably

stronger

than cop; guessing who's going to die is often easier than guessing who's scum.

Also, multiple scum NKs get more powerful as time goes on.

[callforjudgement]

Survivor is a scum role, that has a chance of winning with town if things go badly, and that can't sacrifice themselves.

Among other things, this means that town should always kill claimed survivors.

[callforjudgement]

If scum had three neighbours, they could definitely find a way to use that to their advantage, I think.

If masons had three neighbours, it would at least give them information (e.g. that all the neighbours were town, = the scum didn't know who they were).

[callforjudgement]

Fwiw, I think neighbourhoods are townsided no matter what the faction of the people in them (except perhaps the extreme case of scum-dominated neigbourhoods). I was in an opposite-scumteams neighbourhood in Worst Roles Mafia (advertised as being masons, but nobody believes that as alignment confirmation in a bastard mod game), and both sides ended up leaking a lot of information about their scumbuddies unintentionally. Fooling the town is hard enough as scum; you have to be very good to fool a neighbourhood as well.

[callforjudgement]

I agree with CFJ; neighbourhoods will almost always favour town massively. Iirc, it was information from the "masonry" that really made the mafia win in the end - but if CFJ had been town, and there were no mafia, it would have been an incredibly easy game for the town.

We used the info from that mostly to wipe out the wolves (I caught two out of three of them just from the QT just after replacing in, and my buddies figured out the third based on their impressions of the game and what I told them). The wolves also had a pretty good idea of who we were, but didn't pick on us for strategic reasons and because they weren't entirely sure there was another scumteam at all; this was probably something of a mistake. I think that if the wolves had tried to defend themselves rather than trying to hurt the town as much as possible, town would have won; the more two scumteams know about each other, the worse it is for the both of them, IMO.

Note that I'm not upset at all; I came into Worst Roles Mafia

expecting

to get a bad role, and I wasn't disappointed ;)

PEDIT: The game wasn't particularly bastard, except in the setup; and I don't think the setup was made for maximum bastardliness, it's just that it had only bastard roles to choose from, so it was going to end up as a bastard setup no matter what.

[callforjudgement]

9of4

• 9 Townies, each of which are (independently randomized) a 1-shot Cop, 1-shot Doctor, Macho 1-shot Jailkeeper, or a VT.
• 1 Mafia Goon
• 1 Mafia Roleblocker
• 1 Mafia Backup Roleblocker

Had this idea a while ago while trying to figure out how to deal with the problem that 2of4 solves (having a known number of VTs/power roles locks scums into unavoidable 1v1s or 1v2s at lylo), and also the problem that it has (having more VTs in a setup can make it somewhat townsided, and likewise a lot of power roles can make it somewhat scumsided). OK, so the 9-townie version is probably not balanced (although still definitely winnable), but it's also mindblowingly unlikely; using a lot of weak power roles seems the best way to diffuse the setup sensitivity. The roleblockers are to make it very difficult for town to coordinate actions and break a specific setup. (The Macho on the JK is to avoid paradoxes; likewise, scum can't both kill and block.)

Probably too complex/large for a newbie game (although you could probably scale it down to 2:5 or 2:6 to make a smaller version, then it'd be the right size for a newbie but still too complex). But I think it might work well as an open.

[callforjudgement]

The version of my setup with a factional roleblock might well work better; it's simpler, which helps. (The presence of a JK means that kill choice matters, so it's not an equivalent setup, but it's probably close enough that the simplification makes it better.)

So we have this:

9of4

• 9 Townies, each of which are (independently randomized) a 1-shot Cop, 1-shot Doctor, Macho 1-shot Jailkeeper, or a VT.
• 3 Mafia Goons.
• The Mafia have a factional roleblock, which cannot be used by the same person as the person who makes the kill.

Seems reasonable, and pretty simple, with only one "unique rule", the factional roleblock. (Anyone noticed how Opens nearly always have a unique rule? I think it's because it's the only really reliable way of preventing massclaim as a breaking strategy that doesn't put huge constraints on the rest of the setup.)

[callforjudgement]

Hi guys. I am new to this forum, but I have played mafia games in several forums quite a bit. I have created a 2-UC setup a while ago to prevent the over-reliance of the investigative roles while increasing the likelihood of mafia coming out with false roles. I have tweaked a few things to balance the game... Hope I can get some of you guys' input as well. thanks!

Not sure how many people will read a post that's that long and "busy", especially with nonstandard names. Let me try to translate that into standard names for roles over here…

2-UC version 3

• 1 Sane Cop ("UC")
  
• 1 Insane Cop ("dUC")
  
• 1 Alien ("Medic")
  
• 1 0-Shot Vigilante Miller ("Vigilante"), gains an extra shot whenever the Mafia's NK flips as a power role
  
• X Vanilla Townies
  
• All players who flip as townies can post one informative post within 12 hours of death
  
• Y Mafia Goons (factional abilities in addition to the NK: is told who is targeted by cops each night both cops are alive, can cause the flip to be anything if one NKs another) OR Y+1 Mafia Goons (factional abilities in addition to the NK: is told who is targeted by the Insane Cop each night both cops are alive, can cause the flip to be anything if one NKs another)

It seems reasonable enough as a basic idea for a setup, although there are several things in it that people over here tend to dislike (the possibility of an inaccurate flip, for instance). I think the extra member for the Mafia is typically much more important than knowing who the Sane Cop investigates; it'd be more likely to be accepted in basic concept if you just removed all the factional abilities from both sides. There are various other issues, to do with timing (12 hours is too short) and playerlist size (25 players is unlikely to fill as an Open over here, and in fact unlikely to fill except as a Large Theme with a particularly popular theme).

It's also likely to be very townsided due to the two investigative roles. An Insane Cop, especially when it's known that they exist in the setup, is almost as powerful as a Sane Cop. (Especially because cops in that setup would likely spend night 1 scanning the obvtowniest person they could see in order to establish their own sanity. I assume you'd ban them from scanning themselves.) It definitely feels like a "large" open setup; with a shorter player list, it wouldn't work because town roles could claim to confirm themselves as town and give enough information from there to lead to a likely town win. But if you don't have multiple scum-controlled kills per night (likely by adding an SK or a second scumteam), the game will last long enough for the investigative roles to solve the setup by themselves, unless they die very early. (Knowing the investigative roles' targets doesn't help the scum much; they could NK them, but that means that the cops effectively get to control the NKs with their scan choices. Optimal cop play in this setup is probably to scan people who you personally think are town but who are under suspicion by the town at large.) You could fix this by adding more scum than usual, but then the game would end up somewhat power-dominated, which is generally no fun for the VTs (and which becomes very swingy, too).

[callforjudgement]

One issue with a setup that's heavily townsided with random voting, and not heavily townsided in practice, is that the town's optimal strategy is probably random voting. You can expect people to try to come up with absurd methods to randomize the vote in a way that scum can't interfere (offsite, I've seen people doing things like agreeing in advance to hash the timestamp of the next post by the moderator and use that to determine who to vote for; this is particularly awkward as it means that the moderator ends up determining the game result by their actions).

[callforjudgement]

The "no provable randomness" rule would help; it leads to it being very risky for the town to try to arrange a randomizing method, because of the risk of scum manipulating the results.

[callforjudgement]

Trying to identify a townie is going to be awkward when the town is almost half scum, though. Both because there are more ways to go wrong, and because there are more players trying to guide you wrong.

I guess it'd play a bit like Conspiracy on day 1, which might be a good thing.

[callforjudgement]

The Godfather means it's not breakable the same way as Dethy, at least.

However, I think it may be breakable the same way most Opens with lots of different roles are; after a couple of days of elimination of who does what, you're beginning to get into 1v1 territory. (It's probably in the town's interest for everyone to scan the same person; and it may be in the Mafia's interest to not use the Framer at all, or at least to randomize whether to use it. The main disadvantage for town is that Naive and Sane cops look much the same unless they actually scan scum, and with only one investigates-guilty player in the setup, that's going to be awkward. The Results Cop helps town quite a bit, but only in determining one cop's sanity at a time; and they're probably the #1 target for the scum nightkill. (However, both town and scum will have issues figuring out who they are.)

This one might need a computer search to find the best strategy…

[callforjudgement]

Wrt the Dethy discussion, there's no known way to get a 100% town win rate, but the town win rate in practice is very very high (you can get a pretty high win rate just with setup-breaking strategies, and then there's scumhunting on

top

of that).

[callforjudgement]

Intense Mafia

• 5 1-shot Killproof Townies
  
• 2 1-shot Killproof Mafia

I think there's a hole for a small setup that takes a long time to play. The scum aren't allowed to shoot each other (I think the ability to do so would make the game much more scumsided, as shooting townies would no longer confirm them due to the amount of WIFOM involved, even if the scum always aimed at town).

(And clarification: whether the killproof shot has been used is public knowledge.)

With an appropriate queue for running this in, I'm beginning to seriously consider running it. Do people have balance/fun suggestions for me before I take it to Hoopla?

[callforjudgement]

Killproof was meant to include lynches.

Thinking about it, it's a little broken, though; I can't see any reason for scum not to use the strategy of never shooting the same player as the lynch, and just targeting the same townie twice. Town don't have much of a reason not to just lynch everyone twice (in fact, spreading around is worse), which basically makes it 2:5 mountainous where town know scum's next NK in advance but only once they've already decided who to lynch.

So the whole thing can be collapsed into 2:5 mountainous with final words, which is not a balanced setup (it's scumsided, probably not gamebreakingly so, but still not that fun to play), and also misses the point of the setup.

Perhaps it could be fixed by barring scum from NKing the same person twice in a row?

[callforjudgement]

Scum can force at least a draw by giving votes to each other. You should probably allow scum to remove only.

[callforjudgement]

Give town two removes, scum one add? And start everyone at two votes?

That way, even if scum exclusively give to buddies, town can outpace them.

[callforjudgement]

@Timeater: I suspect GreyICE was not merely being nonserious, but the entire sentence was dripping with sarcasm.

I think known-survivor is potentially OK in setups, more so than many other people do (as in, known that there's a survivor there). Being lynched the day before lylo (for the same reason you no-lynch in mylo) is what would happen if the survivor claimed, which probably gives them enough incentive not to. (It's 6:1 mountainous with an observer trying to avoid day-before-lylo if they

do

claim; this is townsided, but not townsided enough for the observer to shift it into that situation intentionally and try to win with town.) So the survivor is going to try to stay hidden, which means sitting tight and hoping, or even becoming an honorary member of the scumteam. It's safe for the survivor to claim identity to the Mafia (although not vice versa, obviously), and he has every incentive to do so; it avoids the Mafia whiffing on their kill, which means town get fewer lynches before endgame, which reduces the chance of being lynched. And 2:6 with a weakened scumteam is balanced. Huh, I guess it's a reasonably balanced setup in theory, although I'm not convinced it'd be much fun to play.

[callforjudgement]

Having too many unique roles is a common problem with Opens. In this case, it could probably be fixed by randomizing the exact number of each role somewhat. (If you want the possibility for perfect scum fakeclaims, also let them know how many there are of each role, but that generally powers up scum too much.)

[callforjudgement]

Open Theme games are typically run in the Theme queues not the Open queues, right?

I agree that this is a good thread to discuss the setups in, anyway.

[callforjudgement]

Town's optimal strategy is to direct shots.

Town also have no possible way to punish a shot not used as directed, because the shot user is dead already and just venging (if I've understood the setup correctly). So they can give advice, but they can't direct.

[callforjudgement]

Floating an idea based on one of the scumteams from my last theme game.

Apocalypse

10 VTs
3 mafia - 2 goons, 1 core

One of the mafia is designated as the "core". If the core dies, the mafia instantly lose. The idea is that it's unlikely that the core will be lynched before the other goons just because they will work to protect him, but those interactions will eventually make it easier to find the core. The exact number of VTs can be adjusted to find the optimal balance.

I like the idea. I think Mafia should definitely be able to choose the core pregame; if they do, I'd expect the game to be approximately balanced with the numbers given (using White Flag, which automatically makes the second lynched Mafia a core, as the guideline).

Oh, and I like "Mafia Secretary" as a name for the core (it's more unique than "Vengeful-style Godfather", the most common name for it); the role would specifically be "Mafia loses if this role is

lynched

", which comes to the same thing in your setup (and fits the aim behind your role better, in games without vigs). The idea is that they look after the Mafia's membership list, so if the town lynch them, they immediately learn the rest of the scumteam and endgame them.

[callforjudgement]

Daytalk + no NKs on scum + choose secretary at start would me my preference for that setup, and I'd personally think it's reasonably balanced 10:3 and scumsided 8:3. I might be wrong, though; it's often hard to know about balance until a setup is run, especially when it introduces a new mechanic.

[callforjudgement]

The problem with core swaps, is that the first is effectively free, because it puts the town on evens, and the second gives an extra mislynch which is probably a crippling drawback for scum. So it's reasonably equivalent to being able to swap once, with no cost but the town being informed. And I think I prefer setups in which connections can be drawn and stay all game; also I prefer not to make setups more complex than necessary.

[callforjudgement]

Black Flag Nightless

• 7 Vanilla Townies
  
• 3 Mafia Goons (daytalk, no nightkill)
  
• Town win if the Mafia is reduced to 1 member
  
• Mafia win if the Town is reduced to 3 members

White Flag, but on both factions. And Nightless.

The most fundamental problem with mountainous games is that the scum NK becomes a very powerful weapon, usable to eliminate all the best town players. (In fact, increasing the number of VTs in a mountainous setup may well reduce town's win chance even if you allow for the odd/even thing.) The most obvious fix to this is to go nightless, but then the problem is positive feedback: killing scum gives town an extra mislynch. The above setup is my attempt to fix this problem as well; town have exactly 3 mislynches no matter what the order of deaths is. The White Flag mechanic on the scum is to avoid the normal demoralisation issues that mountainous games have, and to help prevent the game running on too long. Finally, this lets me put 3 scum (the ideal number) in a reasonably small playercount, thus catering for people like me who like games that are smaller but not shorter than average.

The EV/theoretical balance of the setup is identical to White Flag. In practice, the differences are that scum can't eliminate good town players (= more townsided), and that scum have more control over the lynch vote due to the smaller player lists (= more scumsided).

[callforjudgement]

Still, optimal play is for N1, mafia vanillaizes one of themselves, N2, the other one. Town wins with all VTs, a Cop, and maybe a doctor or two.

Vanillizing doesn't affect alignment or factional abilities.

Also, they wouldn't do that even if it worked, because the optimal play for a vanillized-and-town former-scum is to tell town who their buddies were.

[callforjudgement]

What would the rule be if they reach deadline without a lynch?

Same as in any other nightless game. Deadlines in nightless don't really work unless you do something arbitrary to punish town if they're hit. (Or plurality lynch, but where's the fun in that?)

[callforjudgement]

It isn't. The Lyncher can't possibly vote for the Jester, because if he does the Jester will self-hammer and the Lyncher will lose. Thus, the Lyncher can only vote for the Townie. Knowing this, the Jester can't vote for anyone else; if he votes for the Townie, the Lyncher will hammer and the Jester will lose; if he votes for the Lyncher, the Townie will deduce that the Lyncher didn't make the vote, so learn the Jester's identity (and in general everyone will learn everyone else's identity), making it impossible for the Jester to win. Finally, the Townie cannot get the Lyncher lynched because the Jester won't hammer anyone other than himself, and the Jester won't vote first (we just established that), and as such the best the Townie can do is no-lynch. Conclusion: nobody can be lynched. Further conclusion: that game isn't going anywhere.

[callforjudgement]

I was looking at drmyshotgun's post, which implied 3 players and the Lyncher aiming for the Townie.

[callforjudgement]

It might be slightly scumsided, but I'd enjoy something like that. Perhaps it'd be worth making it 2:8 must-lynch in order to make the neighbourhoods simpler, although that'd make it a bit more scumsided.

[callforjudgement]

Town need an extra pick for this to have a reasonable chance of being balanced. 2 town PRs is balanced against 3 goons in a 10:3 only if they're both pretty powerful. With scum power, the town is going to need more power than that. (3 average-strength town PRs against 1 weakish scum PR is the usual way to balance a 10:3, although by far not the only one.)

[callforjudgement]

My attempt to break this: day 1, townies decide who the 2 scummiest people are; they lynch one, and for the other, the remaining players cycle-nominate them (i.e. with players A-H, A gets lynched, B is nominated as the next scummiest, then C claims BD, D claims BE, E claims BF, F claims BG, G claims BH, H claims BC.) If B is scum (and NK resolves after guesses, like it would in NAR), town wins. Otherwise, B gets nightkilled, because they'll be confirmed town on the next day. This technique leaves us as, effectively, 2:6 nightless which is white flag on every second day. And that's mathematically rather townsided.

If scum kill happens first, then if B/C are scum together, C can NK H to avoid automatically losing, and scumteams not involving B can also nightkill non-B for the WIFOM. This version's a bit harder to analyze.

[callforjudgement]

Scum can't claim, though. If either scum is known town autowins the next night. In other words, you effectively have Jesters in the game.

[callforjudgement]

KILL EVERYONE MAFIA

• 4 Town Vigilantes (
  not
  one-shot!)
• 3 Mafia Goons

Day start.

Not

nightless (doing this nightless is mindboggingly townsided with correct strategy, incidentally).
No-lynching is allowed.
Scum can NK each other if they want to.
Scum win if all players die simultaneously.
The game only ends when one faction has no chance of victory; merely equalling the town's numbers is not enough for scum to win.
I recommend using Open Night, so that town can coordinate nightkills if the entire scumteam decides to pile onto a quicklynch.

People say that Texas Justice is broken, and they're right. So I decided to tweak the numbers to take the brokenness into account, and, um, ended up with 3 scum in a 7 player game.

If scum outnumber town at any point, mathematically they win, so you can call the game there. 1:1 is a scum win, as usual; however, 2:2 and 3:3 are both undecided situations. The strategy: 2:2 is a town win 66% of the time; divide the players into two pairs and have them cross-shoot, town wins unless the scum are paired together. 3:3 is rather scumsided, though (any lynches on town that day, or town kills on town that night, lead to a town loss, and also the town need to land at least one shot; their best option is to no-lynch, nominate the towniest player to shoot their top scumread, and hope to get into day 2 as a 2:2). 2:1 with a confirmed townie is a 100% town win; 2:1 with no confirmed townie is 66% a town win (in both cases, the towniest player doesn't shoot, the other two cross-shoot). Town may well want to no-lynch day 1, and probably should no-lynch on future days; I think the best day 1 strategy is to no-lynch, then overnight have the second-scummiest player shoot at the scummiest, which leads to:

• If both nominated players are scum: scumteam's choice of 4:2 with no confirmed players, or 3:3 with one confirmed scum;
  
• If the town was right only about the scummiest: scumteam's choice of 3:2 with one confirmed town, or 4:2 with no confirmed players;
  
• If the town was right only about the second-scummiest: 3:3 with no confirmed players;
  
• If the town was wrong about both players: instant scum win

Town have truly excellent chances in 4:2 (well over 66%); 3:3 with one confirmed scum simplifies to 2:2 the next day (66% town win); 3:2 with a confirmed townie is at least 66% by using the same strategy as for 2:2; 3:3 with no confirmed players is quite bad for town, but not unwinnable. As such, the day 1 accuracy, especially on the scummiest player, makes quite a big difference, which is exactly what you'd expect for a 7 player game.

[callforjudgement]

Yeah, I actually like that 3:3 strategy. If you hit scum, it goes to 2:2; if you hit town, you lose. Probably about the best odds you can get.

(However, you have to be quite careful in trying to come to an agreement on who to target when the town is half scum. It's not like you can hold votes or anything like that; the townies will have to try to identify each other and form an agreement just among themselves.)

[callforjudgement]

If both nominated players are scum: scumteam's choice of 4:2 with no confirmed players, or 3:3 with one confirmed scum;
If the town was right only about the scummiest: scumteam's choice of 3:2 with one confirmed town, or 4:2 with no confirmed players;
If the town was right only about the second-scummiest: 3:3 with no confirmed players;
If the town was wrong about both players: instant scum win

I think if town is right only about the scummiest scum just shoot the player nominated to shoot?

Oh, right. So you get 3:2 with no confirmed players, because you killed the player you would have confirmed. You're right, that's better for the scum.

[callforjudgement]

3of7

• 5 Vanilla Townies
  
• 2 town power roles
  
• 1 Mafia Goon
  
• 1 mafia power role

Power roles are randomly chosen from the following list, without duplicates; any power role can go to town or scum:

• 1-shot Cop
  
• 1-shot Watcher (cannot see blocked actions)
  
• 2-shot Jailkeeper
  
• 2-shot Roleblocker (cannot be chosen at the same time as JK, the entire setup is rerolled if it is)
  
• (full) Doctor
  
• (full) Tracker (cannot see blocked actions)
  
• (full) Rolecop (reveals role but not alignment)

Apart from that, usual mechanics (day start, scum have an NK, etc.); scum power roles can both kill and action regardless of how many scum are alive

I've been thinking about the problem of creating a new newbie setup. Here's my attempt, but before proposing it for newbies, I'd like to know if people think it'll be balanced, simple to understand, and fun. And thus, the Open discussion thread. Not sure what to do about the possibility of scum getting useless roles for them (Cop / Doctor); I think it's reasonable to leave it possible, because at least they'll know the role isn't in the setup and thus is safe to claim. The ban on double roleblocking roles is to avoid awkward interactions that have no obvious resolution.

[callforjudgement]

Using EV to balance setups is not particularly reliable. (But if you

are

trying to balance setups with EV, the right value to aim for in an Open is somewhere between 40% and 50% to town. So 43% seems reasonable.)

I actually think Mafia should start by giving the gun to the strongest town player, thus forcing them to shoot with little information and probably killing themself.

[callforjudgement]

People are going to be very reluctant to hammer in that setup. Especially scum. (Although this may make the setup more townsided. It's probably breakable by repeatedly forcing the second-scummiest player to hammer the scummiest.)

[callforjudgement]

Would Named Townie be enough town power? (I'm guessing yes.)

[callforjudgement]

@Faraday: Named Townie can save themselves from a lynch by claiming, it's an open. Scum who claim it will get counterclaimed because its only purpose is to counterclaim scum who claim it. Thus, it improves town win rates via giving them an extra chance if they try to lynch a particular townie.

(It's basically a slightly less strong Innocent Child, because scum can counterclaim a townie claiming it as a gambit if they think it benefits them.)

[callforjudgement]

You can just work it out for 0% and 100% counterclaim rate (with the town unaware of this in both cases), in order to get an idea of what the optimal tactics for scum would be and the resulting EV. (I'm a bit too tired to do it myself right now.)

[callforjudgement]

Adding any number of obvious treestumps to a setup doesn't change the EV at all, even if it

does

change the win rate. (This is, of course, just another demonstration that EV is not the only thing you have to consider when balancing setups.)

It's impossible to make a power role weaker than a named townie in a fully open setup, unless it's unaware of its own role. (I guess you could make a few of the townies secretly bulletproof, but that's just adding extra swing to the setup which nobody can do anything about. Unless the scum know who's bulletproof but the actual bulletproof players don't.)

[callforjudgement]

You would, because of the lack of vengekill. (Unless scum get a vengekill upon a death miller lynch. Giving the vengekill to the death miller wouldn't work, because he'd just vengekill the counterclaim, whereas a scum-controlled vengekill would vengekill a VT.)

[callforjudgement]

Kitoari: If they flipped as town, they wouldn't be a death miller any more. I think you've invented a "townie that gives scum a vengekill if they die D1" role, which you should probably go add to the worst role ideas thread.

Actually, it might be optimal town play to, if there was a CC'd vengekill-gifter, to wait until day 2 before lynching them. That would turn off the scum vengekill.

[callforjudgement]

Sweet spot is normally somewhere between 40% and 50% to town.

[callforjudgement]

@Kitoari: Not sure if that setup is balanced or not, but all the evidence from mountainous games suggest that it would really suck to play.

[callforjudgement]

This is actually mathematically equivalent to an existing Open setup, "Follow the Cop… Or Not". (Because the JK is effectively a Doctor who makes the Cop paranoid.) IIRC, it didn't work, but I can't remember why.

[callforjudgement]

Only if the message is different from the message for targeting an Ascetic. It'd be "you have been blocked" either way, wouldn't it?

[callforjudgement]

I don't think there is a good open Smalltown setup at the moment, and think that if one could be designed, it'd be a good thing to add to the Open rotation. Perhaps I'll try myself sometime.

[callforjudgement]

KILL EVERYONE MAFIA

• 4 Town Vigilantes (
  not
  one-shot!)
• 3 Mafia Goons

Day start.

Not

nightless (doing this nightless is mindboggingly townsided with correct strategy, incidentally).
No-lynching is allowed.
Scum can NK each other if they want to.
Scum win if all players die simultaneously.
The game only ends when one faction has no chance of victory; merely equalling the town's numbers is not enough for scum to win.
I recommend using Open Night, so that town can coordinate nightkills if the entire scumteam decides to pile onto a quicklynch.

People say that Texas Justice is broken, and they're right. So I decided to tweak the numbers to take the brokenness into account, and, um, ended up with 3 scum in a 7 player game.

By the way, I actually ran this. It doesn't work: it's too swingy day 1, and probably townsided too. The basic problem is that scum are easy to find just via scumhunting when almost half the playerlist is scum, and once one scum is identified, the breaking strategies swing very much in town's favour.

It might even be worth trying at 3:3 (or 4:4?), but the problem there is that you'd need a lot of organization from the town. Something that wasn't there in the game that just ended, resulting in scum winning. (They might have won anyway, though.)

[callforjudgement]

If not by some weird coincidence all of the players sharing a hydra need replacing then there will be no discontinuity because there will still be someone to represent that slot even while the mod is looking for replacements

I predict that around half the hydras would have all three players flake simultaneously at some point (because most of the time, there'd only be one head active in each).

For whatever reason, hydras seem more prone to flaking than single players.

[callforjudgement]

@Leafsnail: I've considered a mechanic along the lines of "At any time in the game, you may send the mod the name of a player slot and a player you think is in the slot. If you're correct

and

that player isn't in your faction, they die. Otherwise (if you guessed wrong

or

the player is in your faction), you lose this ability for the rest of the game (and won't learn whether you guessed wrong about the player or the faction)."

[callforjudgement]

This is pretty much why I thought "staying anonymous" would make an interesting mechanic. It's very difficult/close to impossible, so you'd expect a lot of anonymity slips, and thus a lot of interaction with the mechanics.

[callforjudgement]

The enforcement mechanism would be punisher-style; "if the opposite faction guess who you are, they get to kill you". As usual in setups built around trying to prevent the reveal of information. (It's the only method that works, because it's objective and because it doesn't have the problems that mass modkills have.)

[callforjudgement]

That looks biased against town to me. If they correctly lynch the SK day 1, it's as if they'd mislynched day 1 of a 3:9 with a named townie (which is definitely a very scumsided setup). Their best chance is probably to repeatedly no-lynch and hope the SK takes out scum.

Also, if the SK is caught, their best option is to claim, let the town direct them, and hope that it ends up in a 1:1:1 despite the town directing them, which is entirely possible; this leads to an SK win in that situation with best strategy from all sides, if my 1:1:1 analysis isn't badly off. The result is that the groupscum are mostly helpless; their only real weapon is the vote, and there's an NK actively seeking them out. It may be balanced with respect to the groupscum (although I doubt it), but it probably wouldn't be much fun for them.

[callforjudgement]

I'd like to have a good Mini-size open Smalltown setup, but they're kind-of hard to design.

[callforjudgement]

What are peoples thoughts on a modification of this set-up Duck, Duck, Goose! That instead of limiting it to one shot make it unlimited but in doing so means that it takes (one/two) less vote to lynch.

Maybe have some mechanic where scum can reduce their lynch threshold in order to appear town.

Broken. If it's known that some townies have reduced lynch thresholds but no scum do, people will use pseudovotes, then all pile on at once when there's a majority. (And voting for anyone before that would be treated as a quickhammer attempt and have everyone pile onto you in an attempt to lynch you first.) That makes the lynch threshold thing irrelevant most days, and dependent on who's online when at lylo.

[callforjudgement]

It's broken by massclaim; no matter what scum fakeclaim, at least 4 townies are confirmed, also at that point the Gunsmiths become, effectively, full cops.

Also there's possibly too much town power for an 11:3, even if that problem's fixed (say via making the players not know their role until after they're lynched, or by adding a large amount of randomness to the number of players with each power); in this setup, town get three investigative power role shots + a free effective vig shot (either the vengeful vig or the vengeful tree stump; it's not in town's interest to use both of those powers). That sort of investigative power is within the range of what's normally expected for a game with three mislynches, but with the free effective vig shot, town effectively have four mislynches, which is a pretty big advantage.

So yeah, highly townsided. The idea is interesting, but it needs a lot of tweaking to be fair.

[callforjudgement]

The massclaim is broken based on role counts. Say 3 people claim Cop. You immediately know they're all town; nothing to do with claiming actions or anything like that.

I actually miscounted wrt the 4 confirmations; correct strategy for scum is for one to claim Cop, one to claim Watcher, one to claim Gunsmith. That confirms the vengeful townie and the treestumper, which is only 2 confirmations. Then the town have three groups of 4 and know that there's one scum in each.

Basically, the problem is that in an Open, claiming your role means that either you get counterclaimed, or you're confirmed town, so any role that's guaranteed to belong to exactly one town player immediately becomes confirmable. This is true regardless of any of the rest of the details of the setup. This is the main balance difference between Open and Closed games; you can't give someone a power role unless either enough other people have the same power role that scum can safely hide there, or you're OK with them effectively being an Innocent Child in addition to having their power.

[callforjudgement]

Then you don't have any actual confirmations, but knowing that 5 out of a list of 6 players are town is still a pretty large advantage for town.

[callforjudgement]

What about giving the Mafia some sort of minor crosskill protection, to make up for town's enhanced ability to pick on them? (Something like a Mafia Doctor who can self-target, but can't both protect and kill.)

[callforjudgement]

3:9 mountainous is really really scumsided. To put it in perspective: 2:11 mountainous has never been won by town, and I've forgotten what the number of townies needed to theoretically balance mountainous against 3 scum is, but IIRC it was something ridiculous like 43. Or to put it in perspective yet another way: 3:9

nightless

mountainous is theoretically balanced.

[callforjudgement]

Good thing he asked about nightless mountainous, then.

Sorry, I'm tired at the moment. For anyone who's noticed that I've basically siteflaked, it's because I'm so busy in RL at the moment that I'm not getting time on the site often enough to be able to mod or play in a timely manner, and it'd be unfair to inflict my inactivity on anyone else.

How do you go about determining if a set up is balanced just out of curiosity?

First you have a baseline of well-tested setups where you have some idea where the balance is. (For 9p setups, the newbie setups over the years give the best data because they've been run so often. For 13p, mini normals in general are a good comparison.) Then you see how the setup differs, especially in terms of roles that mostly cancel each other out, and the like.

An alternative method (I use both) is to work out how much information the town can get entirely with their power roles (counting "confirmations"). Then you compare that to the number of confirmations you'd expect for the game size and type (e.g. for a 13p closed normal, typically around 3 would be balanced if there's no scum power at all). You can't blindly apply a formula for that, though, because the power of a role depends a lot on its context. (For instance, a mislynch is worth two confirmations, one town-controlled and one scum-controlled, under normal circumstances. So a doctor becomes a lot better if they're likely to gain the town a mislynch; otherwise they're only worth as much as the players they can save, plus a bit if they can reasonably save themselves by claiming, which is typically true in an open but false in a closed game.)

[callforjudgement]

Several games (rounds)

are played by a group of 8 players, with the same setup and

the same role distribution

. Before each 'round'

anonymous alts are randomly distributed

among players for use in this game, such that players are newly anonymous at the start of each round, but know that everyone has kept the same role. The actual playerlist is hidden until the end (and nameclaims can be disallowed). From round 2 onwards, it is equally useful to identify people by their posts, as it is to scumhunt.

Breaking strategy for town: Game 1, each townie chooses and publicly announces a different number. (The scum have to as well, so as not to give their scuminess away by acting differently from a townie.) In future games, each townie announces their number at the start of day 1; the best scum can do is 1v1 them. This doesn't involve any sort of nameclaim, nor cryptography, so it's not banned under the rules you're listing (and attempting to ban it specifically would be really difficult, along the lines of post restrictions to prevent setup breaking, which never work).

[callforjudgement]

Breaking strategies don't have to get a 100% win. Getting a 1v1 with every scum in a setup that's otherwise designed to be balanced is breaking it; it doesn't outright win you the game, but it makes things a lot easier.

[callforjudgement]

VTs aren't compulsory, but it's quite difficult to design a balanced setup without them in practice. Opens need a large number of duplicates of some role to not be broken by massclaim, and most roles unbalance a game if you use a large number of duplicates of them. VT doesn't, so it fulfils the "role scum claim to blend in" by default.

[callforjudgement]

The problem is anything with a large number of killing roles (Texas Justice, Fight or Flight are two off the top of my head) have been broken mid-game.

Kill Everyone Mafia (4 vigs, 3 scum) also ended up being heavily townsided; scum won the test game, but via the method of killing townies who attempted to break the setup. The dead thread broke it.

[callforjudgement]

Wait Your Turn, Tex may not have been tested sufficiently yet, but the games run so far have not been townsided in any way.

It's broken via No Lynch, No Kill, isn't it? You No Lynch repeatedly, taking no night actions, until all the Mafia have taken their shots (you know how many there are, so you can keep track of this). Then everyone claims odd or even. Then you circle-target; whichever category has more Mafia claiming it has to circle-target first.

This is pretty reliable with the typical version of the Rocks Fall, Everyone Dies rule (which requires scum to break NLNK cycles via killing).

[callforjudgement]

Mafia are not Compulsive in that game. So at worst you'd get Happily Ever After.

Right; I was balancing against the normal rule that Mafia have to move first if nobody is doing anything.

Even so, having a Happily Ever After as an option means that town can't possibly do any worse than 50%, because if the setup is scumsided, town should just take the draw.

[callforjudgement]

Well, shooting is optimal (in fact, mandatory) in the EBHOB… setup. Good point about MAD, though.

And as Kill Everyone Mafia shows, a setup being breakable doesn't necessarily mean that people will have the discipline to break it.

[callforjudgement]

Isn't it in town's interest to lynch a suspected townie day 1 in We Need A Fifth? (Preferably the second-towniest of the 4.) That gives them a confirmed townie going into day 2 if the vengekill hits (which it probably will), and if they mess up and lynch scum instead, that isn't exactly a problem.

I'm a bit wary of setups where aiming to lynch town is better than aiming to lynch scum…

[callforjudgement]

Isn't it in town's interest to lynch a suspected townie day 1 in We Need A Fifth? (Preferably the second-towniest of the 4.) That gives them a confirmed townie going into day 2 if the vengekill hits (which it probably will), and if they mess up and lynch scum instead, that isn't exactly a problem.

I'm a bit wary of setups where aiming to lynch town is better than aiming to lynch scum…

?

You lynch scum you get a confirmed town

You lynch town you get confirmed town, BUT could possibly lose the game

You're welcome to figure a way to provide more incentive to not lynch town

Put it this way: the town make an order from towniest to scummiest, 1/2/3/4.

Normal Mafia play would be to lynch 4 (and have them shoot 3 if they're town); I'm suggesting they should lynch 2 (and have them shoot 4 if they're town).

If the actual town/scum for 1/2/3/4 is:

most likely
TTSS

lynching player 2: go to lylo with player 2 confirmed

; lynching player 4: player 4 dies, go to lylo with no confirmations
TSTS

either strategy: go to lylo with no confirmations

TSST lynching player 2: go to lylo with no confirmations;

lynching player 4: player 3 dies, go to lylo with player 4 confirmed

STTS

lynching player 2: go to lylo with player 2 confirmed

; lynching player 4: go to lylo with no confirmations
STST lynching player 2: scum win;

lynching player 4: player 3 dies, go to lylo with no confirmations

SSTT

lynching player 2: go to lylo with no confirmations

; lynching player 4: scum win
least likely

I've bolded the better outcome in each case. Not only are there more possibilities where lynching player 2 is optimal, the possibilities where lynching player 2 is optimal tend to be more likely.

[callforjudgement]

If a mafia member is lynched, he has to confirm one of the town players as town.

So there's not reason not to lynch mafia.

Oh, I missed that entirely. Right, trying to lynch Mafia is probably better there, because you get two tries.

[callforjudgement]

@Shamrock: doesn't work, you could just show both PMs.

[callforjudgement]

Yeah, the rule is that the queue you use for signups is based on the number of players who need to sign up. (The list mods then get to choose which forum the game is actually played in, but that mostly doesn't matter.)

[callforjudgement]

Slightly more randomness is what Dethy needs. The setup's pretty popular, but also broken in town's favour, so it just needs some way to inject extra confusion to become a good setup.

One potential way to break it would be if town collectively voted on cop targets day 1, aiming both cops at the scummiest person they don't lynch. Day 2, the cops claim. If they get the same result (50% chance), now you have a confirmed townie or a confirmed scum, but don't know which. If they get opposite results, you lynch their target (they were looking scummy anyway, right?), discovering which cop is the sane cop, then the Doctor starts protecting them for the rest of the game. I'm not convinced that that strategy would be any more effective than just playing normally, though; there's a lot that can go wrong. (Not to mention the chance of the scum fakeclaiming cop.)

The other thing to note is that it's an Open with a bunch of power roles, meaning that even if town don't do anything with their power roles at all, it's effectively a setup with 3 Innocent Children if town simply massclaim, unless scum decide to fakeclaim Cop (which gives a bunch of confirmations in a different way). That puts a 2:9 setup at a bit more townsided than I'd want, on top of the chance that town actually manage to accomplish something with the power roles. I'm not sure how easy that is to fix, though.

(EDIT: I meant 2:7, not 2:9.)

[callforjudgement]

Scum could also out the SK in twilight if they were lynched.

I'm not sure if that's a good thing, though.

[callforjudgement]

It's pretty similar (identical?) to No Lynching Town. If it's different, the only difference is in the numbers.

[callforjudgement]

Or just spam 100 contentless posts between every real one (and annoy the QuickTopic admins while they're at it).

[callforjudgement]

Putting a role name that abbreviates to "IC" in a newbie is asking for trouble, given that that has a meaning already.

And it isn't broken, but it's rather affected by who draws which roles. In particular, the skill of the Innocent Child is going to be the main determining factor.

It does have the good feature that it's basically impossible to pull off stupid gambits in, which is something that will help out newbie game quite a bit if implemented in a setup.

[callforjudgement]

I'd assume all the actions in a setup like that would be simultaneous unless a) otherwise stated, or b) it was designed by zoraster (or another mod who consistently and intentionally uses an action resolution method other than NAR).

[callforjudgement]

Vote For Town Mafia

• 5 Vanilla Townie
  
• 2 Mafia Goons (no nightkill)
  
• Nightless
  
• The win conditions are reversed from normal; each faction is trying to lynch its own members.
  
• Town wins when 4 townies are lynched
  
• Mafia wins when 2 Mafia are lynched

The main thing that's bothering me about this is the 2 town : 2 Mafia situation. Much of the time, Mafia will end up getting 3 votes on them "naturally", but when they don't (i.e. both townies refuse to lynch either Mafia), some means is needed for the game to continue. Plurality lynch and random both suck in that situation. Mafia-chooses would work, but might be a bit scumsided; another option is to let the most recently lynched townie break the tie.

Apart from that, the setup seems pretty balanced. (For people who care about the EV, I calculated it as 10/21 or around 47.6%.)

[callforjudgement]

I was planning to flip as normal; I can't see a good reason not to (although would be interested to hear of one if you have one).

[callforjudgement]

I'm not convinced that the game is broken by VCA at all. Flips do make things harder for the scum by making scumhunting easier; however, I don't think that making the game more scumhunting-based is a bad thing, and I don't think town's advantage from flips is enough to unbalance the setup.

[callforjudgement]

Not sure if claiming day 1 in that setup is the

best

strategy for town, but it seems workable. That effectively translates to an Innocent Child and a 1-shot Cop, which in turn is basically a 7:2 mountainous in which the scum's first two kills are determined by the mod pregame and their third is chosen by town. That's not that far off balanced, but it's rather swingy depending on who gets which roles (because the main benefit of the scum kill direction is to prevent the scum killing the towniest townies).

[callforjudgement]

Can't you shoot the cop innocent night 2?

Then the cop stays alive to get another result, and town effectively gets to direct the scum kill for multiple days. Unless scum counterclaim the cop day 2 (which is probably a bad idea), the cop's confirmed town in that situation, and with the miller dead, there's no doubt about the results.

[callforjudgement]

Ah right. In that case you end up with the town-directed kill and mod-directed kill in a different order, and the setup basically stays the same as it would be otherwise.

[callforjudgement]

hayatoBL's setup has the same issues as SS3 does: none of the scum have any more information than the town does (they know they're scum, but that doesn't aid their gameplay in any way). This means that scum playing to their win condition looks indistinguishable from town playing to their win condition; the goal is just "get anyone else lynched", pretty much.

[callforjudgement]

"Neighbour" is a really awful description of that. They're basically just a scumgroup with no kill, and bulletproof until one dies (they have no incentive to do anything with their night action but protect each other forever).

[callforjudgement]

in neighborhoods, my intial thought is to suspect everyone and trust nobody, its even easyier for scums to just dismantle a neighborhood and destroy it.

if you ever take a look into the alicae in wonderland game you will see that the scums just manipulate neighborhoods to an extent, and the one that had none exploded into a fight.

add in daytalk and scums can easily communicate all things going on at once if they get randomized into all the hoods. a neighborhood (imo) is ment to talk about players in secretive. if its known that scums can exist in all of them i would never post in there.

I have a rather different view of neighbourhoods. Last time I was in one with a scum player, I managed to figure out, through neighbourhood conversation, a) that they were scum, b) who one of their scumbuddies was. That's rather more powerful than a cop result.

I think there's some debate about whether neighbourizer-type roles are better used for creating a pseudo-masonry, or as a pseudo-cop scan. IMO the pseudo-masonry method is overrated, but I might be wrong on this.

[callforjudgement]

OK, so if that setup wasn't multiball, optimal strategy would be for people to vote for the two scummiest players, and force one of them to desperado the other (on pain of lynching). This continues throughout day 1; town never lets day end except to lynch someone who refuses to follow the strategy. If desperado'ing someone backfires, the target is confirmed as town.

In order to work out what happens in this setup, we need more information. Do the scum factions have nightkills? Must all Desperado shots be taken in public?

[callforjudgement]

Yeah generally we can juggle 2 quite easily.

Though honestly i think newbies should be heading to micros after the newbie rather then anything else, but that's my opinion.

I think Mini Normals are the best next step up, unless the newbie in question is particularly fond of Opens.

[callforjudgement]

In Herodotus' setup, people in one pod need to be able to discuss the pod opposite. Otherwise, it's not Mafia until day 2, because on day 1, there's no informed minority.

[callforjudgement]

4:8 nightless townsided, seriously?

(That said, I'll buy 4:12 as townsided, given that it has 50% EV and townsided mechanics that would cause town to do better than EV on average in practice, so 4:14 is too much. The true balance is probably somewhere in between.)

[callforjudgement]

I guess this says a lot about what the value of a scum NK is. They never really seem so great at the time, but they tend to cause the playerlist quality to deteriorate over the course of a game, which is what scum want.

The positive feedback behaviour of a normal Nightless is indeed problematic.

[callforjudgement]

Here is a simple 7 player Micro set up. Is this game balanced?

VT

- 5

1 - shot bulletproof

- 1

Mafia Goon

- 1

This setup has the typical problem (for forum Mafia) of setups with a scumteam of 1, which is that scum have no real incentive to act differently from townies. They want to try a little harder than townies do to stay alive, but that's hardly going to have much of an effect, and mostly won't be observable. This specific problem doesn't apply so much in face-to-face Mafia (where you can use psychological tells to see if someone's lying based on their body language), although this setup would be even worse there for reasons explained below.

Also, just lynching randomly gives town a pretty good win rate in this setup! The bulletproof player doesn't affect anything about random lynching (town will no-lynch if it gets to evens, thus making the bulletproof pointless in the random-lynch scenario), so town have three shots to hit the scum at a 1/7, a 1/5 (× 6/7), and a 1/3 (× 6/7 × 4/5) chance; that's a 57/105 chance, better than 50:50, which is not normally where you want to be for completely random play.

Worse (from the balance point of view), town has a better strategy than lynching randomly; the bulletproof player can confirm themself by claiming. (It's not like scum can kill them unless they're either the D1 lynch candidate (and they save themselves by claiming), or the N1 kill attempt, a 2/7 chance. Note that it's actually to scum's benefit to shoot a BP player twice to kill them in this setup, which is another sign that something's gone wrong somewhere.) In an Open, every town power role is effectively an Innocent Child, and making that player bulletproof too is really awkward. The odds of town victory lynching randomly, with a BP player saving themselves by claiming, is a little under 1/6 + 1/4 (× 5/6) + 1/2 (× 5/6 × 3/4), or 33/48: this is better than 2/3 (i.e. 32/48). Note that the scum can't save themselves by fakeclaiming or counterclaiming BP; on day 1 or 2 this leads to a 1v1 which will be resolved by simply lynching both players in turn for a guaranteed town win, and on day 3 it changes the situation from a 1v1 to a different 1v1 (i.e. doesn't make a difference really).

So, this setup is broken. There's no outright guaranteed win strategy for town, but they can get over 2/3 chance, and will pretty much have to given the absence of other information. In face-to-face Mafia, they also have the chance of finding scum via scumhunting, which will drive it even higher.

Normally the recommended way to design a 7p setup is to place two scum in there, then try to give the town sufficiently large buffs that they stand a fighting chance (5:2 is /not/ easy numbers for town to survive, but you can mitigate that with power roles). It's hard to do this in a way that eliminates all the problems with setups of that size, though. (It's not too hard to come up with relatively

balanced

setups like Original Newbie, but that's considered broken because the entire gameplay of the setup revolves around both scum and town knowing the breaking strategies and how to counter them and thus needs very aggressive fakeclaims/counterclaims from scum in order to avoid the game being a town blowout. Likewise, 2:5 nightless is probably close to balanced, but has the typical problems of being a nightless, such as the positive feedback when town get a kill.)

As for the 11p game, a 5:5:1 setup where both the Mafia and Town are vanilla and the Town have no compensating advantages can't possibly be balanced. The setup is symmetrical between scum and town, except that scum have more information (I'm assuming no nightkill (because they'd have no way to arrange a target) and no daytalk, but they still know each other's identities, and importantly, can still immediately chain-vote out all other players if they gain a majority). I think a setup a little like this can be balanced, but the randomly distributed (day?)vig is not an element that's at all easy to balance in a setup (it should probably unconditionally go to town), and the numbers are off. (It's also going to be very hard to make it work at 11p. The correct number of scum is probably 3 with an NK+private topic or 4 without the NK, but I find it hard to see how to make up enough town power to compensate in either case; 2 and 3 would be much too weak for the scumteam.) That said, "survivor who causes everyone else to lose if they win" is a role that I've wanted to see worked into a balanced setup for a while (assuming that it's Open, or at least that the presence of the role is known by everyone all game).

[callforjudgement]

Jester is a great cure for communities that policy lynch too much (as opposed to here on mafiascum.net, which doesn't policy lynch enough), and the players should be aware it exists. Apart from that, it's not a very good role.

[callforjudgement]

No Kill Mafia

7 VT
2 Mafia Goon

nightless
no night kill (or any kill at all) for mafia
no lynch is not an option
no flip when lynched

game will end automatically when win con is fulfilled

This is townsided via random lynching. Given the no-reveal mechanic, it may well be that town decide to do that rather than try to scumhunt. My guess is also that it wouldn't be much fun to play. (Also, do scum have daytalk? Normally they would in a nightless, but the nature of the setup makes it unclear to me.)

[callforjudgement]

Once the Mafia are dead and multiple Druids survive, is it even possible for Town to win? As soon as they lynch all but one of the Druids, the remaining one automatically wins. (Or does lynching the armed druid mean that town share a win with the other druid?)

I fear that this game has the "scumteam of one" problem that makes scumhunting very hard because there are no associative tells. It might be fun for the druids (although they have an uphill battle), but it doesn't seem fun for town.

[callforjudgement]

Town and Mafia are misnamed in this setup.

What you actually have is a five-player scumteam who don't know each other's identities, and an eight-player town, which has a Mason pair (one of which is also a Neopolitan).

[callforjudgement]

This setup still has the Town and Mafia labeled the wrong way round (although as it's an Open, that mostly doesn't matter). Any setup that contains more Traitors than it contains any sort of Townie will necessarily end up like that.

[callforjudgement]

If a Weak Faith Healer targets scum, do they have a 100% or 50% chance of dying?

Something along those lines could work, I think. I'm too tired to have a good idea of whether those specific numbers do, though.

[callforjudgement]

Longer deadlines are normally considered townsided, and shorter deadlines scumsided. Regardless of whether this is actually true or not, picking anything other than the maximum would thus be seen as a scumtell.

Meanwhile, choosing before alignments (and thus before confirmation) means that some players will be stuck in a game with deadlines they don't want.

(The best tactic is thus to advertise day periods during signups, and have a range of different day periods in the queues. Which is pretty much what happens at the moment!)

[callforjudgement]

Clearly the correct approach here is to convince the scum they can only break the setup via self-killing, and then not protect anyone.

(I seem to remember something like this happening to Hoopla once.)

[callforjudgement]

What happens if you hide behind a Mafia Ascetic? (Hider has a ton of variations anyway, and I'm 100% not sure how any of them would work out.)

[callforjudgement]

If it has short deadlines and plurality lynch at deadline, the basic concept should work. (Also, based on experience with this sort of game, I should mention that it's very important that the game does not have a purchasable kill.)

First come, first served is a potential problem, though (especially where Cop and Watcher are concerned, which are easily the two most powerful on the list); it means that timezones have a large influence on the game. Especially if the same player goes for it again after again. Perhaps it'd work best if the players are given role priority numbers in their role PMs, and if multiple players request the same role, the player with the highest number wins. (That way, players with lower numbers would know that they would be unlikely to get the more desirable roles.)

Also 1 night deadlines won't work in the Micro Queue (too much risk of missed night actions), so I think this game would be more suited to the blitz queue (where you can quite legitimately tell people "get online every day or don't show up"). Probably even 3-1 would work (although would definitely count as blitz).

[callforjudgement]

I had posted this setup idea a while back but I think this is a much stronger version:

Docs and Blocks Mafia

• Mafia begins with THREE Mafia Goons
  
• Town begins with SEVEN Vanilla Townies
  
• Town is guaranteed to have THREE VT’s upgraded to a either a Doctor or a Roleblocker (2 of one, 1 one of the other)
  
• During pre-game, Mafia have 72 hours to upgrade up to TWO goons to 1-Shot Strongman. They may choose not to upgrade any.
  
• For each goon upgraded, TWO additional VTs will be upgraded to either a Doctor or a Roleblocker. (must be one of each)
  
• During pre-game, ALL Town players will receive Vanilla Townie role PMs. Once the Mafia has decided how many goons to upgrade, power role PM’s will be sent out.
  
• Town will not be notified how many goons the Mafia chose to upgrade.

Mod would "flip a coin" to decide if its 2 Docs and 1 RB or 1 Doc and 2 RBs to start. Town is guaranteed 1-4 Docs, 1-4 RBs and could have 0-4 VT's

I suspect this is highly scumsided at the moment. If I'm making a 7:3 nightless setup, I typically give an advantage to town to compensate for their numerical disadvantage, i.e. I think 7:3 mountainous nightless is scumsided. Town's abilities here are entirely kill-stopping, thus they're in a worse position than they would be in a 7:3 mountainous nightless; Mafia could no-kill every night and still have the advantage in the setup, and they get a further advantage from the possibility that they might get a kill.

A separate problem is that town gain basically no information from a successful kill; if Mafia upgrade two goons, then that gives them two kills that town can't stop, and town have no way to know who made them via night actions. This means that by day 3, after two mislynches, Mafia will have won (lynch + kill + lynch + kill = 4 dead townies, now it's 3:3). So in the two-upgrades situation, town's power roles are actually entirely pointless unless they manage to lynch scum blind on day 1 or 2. This can happen but it's rare (less than 50% in my experience).

If I wanted to give this a shot which (if any) queue could I try it in?

Mini Theme would be easiest, given the rather abNormal mechanics; you could probably spin a bunch of flavour around it.

Open would also be possible but you'd need to pass a review in the Open Setup Reviews thread first, and I'm not sure it'd pass like that.

To be honest, I think it's worth seeing if you can rebalance the game to work as a 7:2 or 6:2 and make it a micro. Come to think of it, the setup as listed might well be scumsided even at 7:2 (if scum upgrade both goons, the town need to lynch correctly with no help from power roles on day 1, 2, or 3 to avoid losing, and the only information they'll have from claims is that the scum probably won't both claim the same thing). I don't think it's hugely off-balance as a 7:2, although the details might well need tweaking.

[callforjudgement]

I haven't thought about it in detail, but have a feeling that the new setup is townsided (I am far from certain, though). Perhaps it isn't, on the basis that if a kill doesn't go through, it's likely to be highly unclear why. I can see town trying to attempt a breaking strategy involving cross-protection, but there's enough uncertainty in the setup that I'd expect the attempt to fail.

[callforjudgement]

If the scum just simply dodge every day, is there any way to get rid of them?

[callforjudgement]

Training Hospital

• 7 Town Semi-Doctors
  
• 2 Mafia Goons
  
• A player is protected from a kill if at least two semi-Doctors are protecting them. A single semi-Doctor protection does nothing by itself.
  
• As usual, the mod doesn't comment on who Doctors protected or if a kill was prevented, just on who died (if anyone) and their alignment.

I don't think this is breakable; scum can't protect people but they can choose not to shoot a person who would give them away as having not protected. Town can certainly coordinate to keep specific townies alive but that's an intended part of the setup. I believe optimal strategy for town is to direct the protects of the scummiest-looking players but to leave everyone else to make a free choice with their night action. (Optimal play for Mafia is to attempt to kill every night, although taking "long shots" that have a high chance of hitting a protection is reasonable; if Mafia don't kill, that's a 7:2 nightless, which is known to be very townsided.)

This could also work at 10:3, I think. Not 100% sure on balance in either case but it has to be close.

[callforjudgement]

I think that's mostly balanced if the town adopt the strategy of picking the scummiest person and having them make an investigation, with nobody else investigating. If the scum shoot them, it's an effective 7:2 nightless, meaning that scum can't kill the nominated cop. If scum shoot someone else, it's similar to cop+6:2 (which is balanced), but with two exceptions: the town can't know that the "cop" who investigated is town (thus the investigations lead to 1v1s or A-scum-or-B-town results), and if the cop dies, another player is promoted to cop. That's probably slightly scumsided, but not by much.

Town can probably do better by nominating two people to investigate and dealing with potentially 75% sane cop results (which retroactively become 100% sane if one of the investigators eventually turn out to be scum). I'm not sure if this is enough of an advantage to make the setup townsided.

An every-townie-for-themself strategy, on the other hand, would be a disaster; the results would most likely be more misleading than useful.

[callforjudgement]

Top row is pretty townsided; if town can predict the kill correctly on any night (and it isn't the watcher that's killed), that's scum caught. Predicting the kill is normally easier than finding scum.

[callforjudgement]

Top row is pretty townsided; if town can predict the kill correctly on any night (and it isn't the watcher that's killed), that's scum caught. Predicting the kill is normally easier than finding scum.

Top row is equivalent to 2 Mafia:6 Town + 1 Watcher. Watcher has essentially the same chance of finding scum as a Tracker (and can't find innos with 1 scum left), which is worse than Cop. Is it actually townsided?

A Watcher has a better chance of finding scum than a Tracker.

With a Tracker, you have to predict who will make the kill (i.e. the towniest mafioso).
With a Watcher, you have to predict who will be killed (i.e. the towniest townie).
It's normally easier for town to figure out who the towniest townie is than who the towniest mafioso is; the former requires a lot less information.

[callforjudgement]

The extra warning role seems kind-of broken if you can kill someone with it, especially in scum hands.

I suspect the rule investigator is near-worthless except as a named townie (always useful in an open!) because there's not much reason not to just have everyone claim their rules, if the information is relevant.

[callforjudgement]

Land of the Rising Sun:

Mafia Watcher
Mafia Ascetic
Mafia Goon

Town Ninja Tracker
Town Roleblocker
Town Vanilla-Cop
7 Vanilla Townies

• This setup uses a Daystart.

The Ascetic has no reason not to perform every kill, as long as they're alive. In particular, the Town Roleblocker should no-action every night until the Ascetic is dead, making it easier for the Town Tracker to gain information (they will get a "Vanilla-Cop/Watcher", "Ascetic", or "someone else" result via distinguishing between action, blocked, no-action), which is a really good way to narrow it down (another advantage of this is that it means the Watcher can't figure out who the Roleblocker is by watching them block someone). Existence of the Mafia Watcher means that town can't break the setup via directing the Vanilla-Cop target, which is a good thing here.

I'd need more thought to figure out if the town has the right amount of power here. Their power is basically entirely investigative; they have a

lot

of it but it's also quite awkward to use. I think that the town probably has too much power, come to think of it; two of the town roles can detect the Mafia Ascetic (so town is getting two investigations for that player every night), and once the Ascetic is dead (along with, probably, the townie that caught them), the town is

still

getting two partial investigations per night (from the Roleblocker, who starts actioning, and whichever other power role the scum didn't kill). That's probably balanced if the town doesn't find scum any other way, but the probability that town find scum via dayplay seems to outweigh the probability that scum kill town power roles overnight before they have useful results. (It's also very swingy, depending on how early the Ascetic dies.) Because this is an Open, the town also have the possibility of forcing "not vanilla" results to claim, taking advantage of the "named townie" aspect of all the town roles being different; I'm not sure when they should use it (and/or whether the Vanilla-Cop should breadcrumb results), but the possibility can only help.

[callforjudgement]

We Need Another 4P

• 2 Mafia Lovers
  
• 2 Vanilla Townies
  
• Must-lynch (no no-lynching allowed)
  
• If a townie is lynched, that player gets one try to call the scumteam; if they get it right, town win, otherwise scum win

This is about the simplest 4p I could come up with. It also has the interesting property that everything happens within one game day, which is probably a downside. This game mostly defies EV calculations, but if you used the (banned due to provable randomness rules) strategy of "pick a random player, everyone votes for them" you get a 50% town win rate (if the chosen player is scum, the reluctance of the other scum to vote for them despite the strategy will confirm them both).

Note that it is possible for scum to vote for each other in this setup! (Although not to hammer, obviously.) Obviously, they have to hope that the lynch doesn't go through, but it may be necessary to try to seem town.

[callforjudgement]

That's got to be scumsided; you put a scumsided restriction on a scumsided setup.

The idea has merit, though. It might be possible to add power roles or tweak the numbers. (Scum should probably also choose a Gladiator pregame.)

[callforjudgement]

If Mafia just pick someone entirely at random, the game is highly scumsided.

They can probably do better than picking at random.

[callforjudgement]

Give Town some power roles, add a mechanic like White Flag, and/or change the faction sizes. (Much the same things you'd do to balance a Mountainous.)

It's probably still scumsided at 10:3 White Flag but would be much closer to balanced than anything you've listed above.

[callforjudgement]

Is there an equivalent thread for closed set-ups?

"The New Set-Up Review Thread" allows you to find a reviewer for a Closed game. (I used to be pretty active there, although more recently I've mostly been doing closed Normal reviews rather than closed Theme reviews, and thus have been going via the NRG.)

[callforjudgement]

I just had to get this idea out there... :P

Friend or Foe

==== Town ====
Town Neighborizer
VT x9
==== Mafia ====
Mafia Neighborizer
Goon x2

- Mafia members don't know the identities of other mafia members.
- Mafia has one factional night kill per night.
- Each mafia member can choose to submit a night kill target to the mod or to take no action.
- The factional night kill will resolve to the player targeted the most, or targeted by the most recent pm in the case of a tie.

Town's nowhere near powerful enough for a 10:3, and the fact that scum don't know each others' identities makes scumhunting basically impossible; there's no way to draw associatives. I'm inclined to call this "not Mafia". (Besides, a decent proportion of the value in a town Neighbourizer is in neighbourizing scum and figuring out who their buddies are via seeing how they react to various suggestions of yours.)

[callforjudgement]

Scum have zero incentive to act any differently from town.

Think of it this way: normally scum have to play a little nontownishly to help out their partners, but if they don't know who their partners are, this incentive isn't there. So scum can just play exactly as though they were town; there is no reason for them to do anything else. (In fact, it would be optimal to not read your role PM in that sort of setup, except for the fact that you have to know whether you need to send in a nightkill.)

[callforjudgement]

Dubious Idea Mafia

• 8 Town 1-Shot Day Vigilantes
  
• 3 Mafia 1-Shot Day Vigilantes
  
• If a player is lynched and still has their shot left, they can vengekill a player with it if they like (although they must make the decision very soon after the lynch, it can't be saved for later)
  
• Daykills are made privately via PM to the moderator, and it's not public who made them
  
• Attempts by people who have already been daykilled to kill people fail, even if the moderator hasn't been online to flip them yet
  
• Nightless

Numbers are flexible and I have no idea whether they're in the right place or not at the moment (it should be close though). This is a Bad Idea variant, or viewed another way, a Texas Justice variant. I can't see a way to break it due to the fact that actions happen in sequence rather than being simultaneous.

[callforjudgement]

It's more like a 6:3 than a 5:3 if scum all shoot first. If the setup ends at 2:2 or 1:1, a townie can daykill scum in order to keep town's chances alive, thus town have one extra mislynch. (Not to mention that if town

knows

that scum have used their shots, they can pick two people for the lynch and ask them to vig each other; if both are town only one would die, if one were scum then they'd be killed by the townie, and if both were scum it'd become obvious when neither could shoot. As such, scum would have to claim their daykills on townies in order to prevent it being obvious that it was a scum shot, which makes them easier to read and rather restricts their targets.)

[callforjudgement]

Lynch or No Lynch

• 9 Vanilla Townies
  
• 3 Mafia Goons
  
• The first time the town no-lynch, the Mafia have to kill one of their own members overnight, rather than killing a townie
  
• Town=Scum is not a Mafia victory if town still have their no-lynch available
  
• To save time in this situation, the no-lynch is used automatically and the day phase is skipped

Basically, the idea is that town have to estimate whether having the scummiest scum revealed would be more useful in the future or now. If there's obvscum, you lynch them as if they're actually scum, no lynching now would give scum a relatively cheap sacrifice. If there isn't, it may well be that a flip would make the situation clearer, in which case it's probably in your interests to get a scumflip now rather than risking lynching someone who makes the scum's sacrifice decision easier. This reminds me of Deal or No Deal a lot (you try to estimate the best time to "take the offer" as the probabilities change, you can only take the offer once, but even after you do you play the game out), so the name is very fitting :-)

The idea originally came about when trying to fix an issue in White Flag (that it's hard to get associatives with dead scum in a mountainous and the White Flag mechanic just makes it harder, thus forcing you to scum

team

hunt rather than scumhunt in the normal way). Giving town the first scum death for free rather than the last seemed like the way to go, but you can't just flip one scum at the start of the game because town wouldn't gain anything from it. Instead, by adding the possibility that scum can be lynched

first

, scum are encouraged to defend each other and thus create associative tells.

The EV (town win percentage with random lynching) is highest if town delay their no-lynch to the last possible moment (to increase the chance of a random lynch hitting scum), at 48.646%.

[callforjudgement]

I thought about that but decided that there wasn't much of a reason not to leave it up to the moderator. It doesn't have a balance effect, just an effect on town activity, so it falls into the realm of moderator discretion.

[callforjudgement]

I don't think scum can force a nolynch D1 in this setup any more easily than they can force a nolynch D1 in any other setup. (Let's face it, they would if they could even in something like White Flag or Matrix6.)

[callforjudgement]

Hmm, perhaps we should mandate a plurality lynch at deadline (in addition to hammers ending the day early, as usual) in order to get around these problems, then. I don't think it affects the setup's balance, but if people

think

it does, then that'll be enough to throw the balance off by itself.

Plurality lynch at deadline also encourages stalling to some extent (which is why it's unpopular) but it would at least work around this problem.

[callforjudgement]

What's the limitation on using the cards? Once per night at night? Without any limitation it's clearly very townsided as passing one card between every player will reveal both scum and scum will give themselves away if they try to break the chain.

[callforjudgement]

OK. So in the most common scenario (scum aren't lynched day 1 or day 2), the players have the choice of claiming results day 2 or waiting until day 3. If you claim day 2, one of the people who was investigated is likely to get NKed (both because they'll be at least somewhat confirmed, and to prevent anyone learning what their N2 investigation result was). This effectively means that in an average situation, town has a 3-way Parity Cop result D3. That's townsided if all the claims are truthful, but take into account the possibility that scum got the cards and fakeclaimed their results, and I think this is in the ballpark of balanced. If I had to guess at a possible imbalance I'd say it's townsided (because if it goes to D4 scum are highly limited in who they can kill without giving their identity away through night actions) but I might well be wrong on that.

I think it may be a good idea to have the players who originally got the cards D1 announced by the moderator. This means that if scum do get a card D1, they can't use the strategy of claiming they didn't get it (which is undisprovable/unobservable as townies will claim that too), thus effectively removing the card from play. (Even if you make using the cards compulsive, they could just pass it to a buddy.)

Is it worth using some different sort of flavour than Sane/Insane? Because you can just swap the results from one card and it should always be clear which card is which from night to night (except in the unfortunate case where the same player gets investigated by both cards, which should be avoidable via town coordination) it's mathematically equivalent to there being two cards, each of which give results that have unknown meaning but are consistent for a given faction. For example, we could do the setup like this, which is a little more flavourful and a little simpler:

• 2 Town, 7 Mafia
  
• 2 Ladies of the Lake, 7 Vanilla/Goon (roles are randomized independently of alignment)
  
• Lady of the Lake PM is:
  You are a «
  Town/Mafia
  » Lady of the Lake. At the start of Day 1, your role (but not alignment) will be announced by the moderator.
  
  At night, you may investigate another player. That player will become a Lady of the Lake (this will not be announced by the moderator), and you will become vanilla (goon if Mafia).
  If you die during the day, or your target dies at night, a random vanilla/goon player will become a Lady of the Lake.
  Your result will be of the form "water" or "wind", and indicates the faction of the investigated player. Which word refers to which faction is chosen randomly by the moderator at the start of the game, but will be the same for all Ladies of the Lake all game.
  
  «
  win condition, factional abilities if Mafia, etc.
  »

[callforjudgement]

Mafia will figure out which of them is the vig very quickly, at which point they basically have two kills per night, + will be able to protect their vig from the town vig. I don't think the town PRs compensate for this.

EDIT: That said, the Town have a ton of power too; the Cop will be self-aware by D2, and has a backup so that they can safely claim. The Vig will probably also be self-aware by D2. I still don't think it's enough to compensate for the scum controlling two kills, but it'd be enough to unbalance any normal setup (especially because Follow the Cop is possible by D3; you get one of the remaining town PRs to claim, the other protects the Cop, there's a 50% chance it's the Doctor, or a 50% chance it's the JK in which case the Cop can claim being blocked, the Doctor probably gets NKed, then the (hidden) Power Grabber takes over the Doc role).

[callforjudgement]

The mafia doesn't have a group NK. Also I don't think mafia will always shoot day 1. The vig may think he is a doc and kill his own teammate.

Oh, I missed that. In that case it's highly townsided for reasons I mentioned earlier.

Also, scum will definitely shoot night 1. They'll just shoot two different townies, one with the vig, one with the doc. Doc'ing a townie you aren't killing has very little downside for Mafia. This isn't enough to make the setup not-townsided, though.

[callforjudgement]

The setup you have there is close to balanced if mountainous (given that Mafia can only kill once). As such, I suspect giving the town power roles, even relatively weak ones, makes it townsided.
If I misinterpreted you and you mean that the Mafia collectively have one kill per night, it's scumsided; I don't think town gets colour investigations quickly enough to be able to conclude much from them. (The best case for town is if all the investigated players are the same colour, but that's vanishingly unlikely.)

I suspect the best way to balance the setup would be to give the Mafia one factional kill every night and to make some but not all of the townies colour-aware. I'm not sure how many would have to know their own colour, but I suspect that there's some value that is balanced. (You could potentially even make the "colour investigation" a town factional ability, controlled by voting in the thread, rather than relying on power roles; this would avoid the "named townie effect" which lets power roles clear themselves by claiming. That'd make the setup more scumsided, but it could be balanced around, and would make the setup purer in a way.)

[callforjudgement]

Claiming your colour hardly makes you confirmed. Think about it this way: if every player knew their colour, you'd basically have three games of Mafia running in parallel, a "red game", a "green game", and a "blue game" (with a final victory calculated on the winning margins town/scum had in each of the three games). Vi has run setups similar to this in the past. If you consider the example of a 2 red town/2 green town/2 blue town/1 red scum/1 green scum/1 blue scum setup, you basically have three independent LYLOs where town needs to win two out of three to win (because they only have one mislynch). The main reason that this setup doesn't work (and why Vi never ran it exactly like that) is that because the three pods are independent you don't really get associative tells.

Having some players know their colour, and other players only discovering later, is a really good way to get around the associative tell issue, and also allows for a lot of fine balance tuning simply by changing the number of self-aware townies. As such, I'm convinced that there is an interesting balanced setup along these lines, and would like to see it played/become successful. That said, I don't know the specific numbers at which it would be balanced; if all the townies are self-aware it's clearly townsided, if none are I believe it's scumsided, and so the balance lies somewhere in between but I'm not sure where.

[callforjudgement]

Oh, I was seeing it as a cleaner version of Triplicate (that neatly handles balance, and associative tell issues); perhaps not in its original form, but once it's been cleaned up a bit.

[callforjudgement]

Well, it's a 9:2. You wouldn't expect town to have a lot of power in one of those.

[callforjudgement]

After the Micro version of Vote for Town was tested, with the game working out pretty well, I've been considering a larger version:

Vote for Town 16p

• 11 Vanilla Townies
• 5 Mafia (no nightkill)
• Nightless, daytalk
• Instead of voting to lynch, you're voting for a player who gets to "escape" the game, helping out their faction (thus townies want to vote for town)
• Self-hammering is allowed; non-hammer self-votes are banned
• Plurality lynch at deadline (or earlier if a player is hammered)
• Town win if half the townies (i.e. 6 townies) escape
• Mafia win if half the Mafia (i.e. 3 Mafia) escape

This has an EV of 50% exactly. (This may be slightly too high, but given that the mechanic tends to "naturally" get rid of the stronger townies I'm not convinced, and adding two VTs to counteract this would have issues of its own.) The "you win if half your faction escapes" wording is something I thought up today, and produces a Flag mechanical effect in the most intuitive way I've seen. The flags are larger than normal; this is intended to counter the effect where an inactive or obvious Mafia member can make the a Vote for Town game very hard to win for scum (who can afford to "effectively sacrifice" two members). It also means that the game will last at most 8 days (compared to Vote for Town 8p's 6), so despite the size, it's unlikely to drag on.

There are a couple of other nice theoretical properties too: although I haven't found a way to mathematically quantify swing yet (which is something I'm thinking about), I'm pretty sure it's at or very close to the mathematically optimum value; and unlike most setups, it can tolerate a small amount of flaking even without the slots being replaced (which means that the game can survive just fine while waiting for a replacement).

[callforjudgement]

It's better than going for 14, which is what the default would be for a 16p nightless. (Or in other words: it's not intended as a selling point but as a sanity check, showing that a Nightless that large is potentially viable.)

[callforjudgement]

I have a suspicion that the 9:4 versions of the setup are much more scumsided than the 9:2:2 versions even though the scumteam doesn't have perfect information. (Also, the 9:2:2 versions clearly favour scumteam 1 over scumteam 2 except possibly when endgame screwiness is involved.)

[callforjudgement]

So are scum 1. There aren't any roles in the setup that have the ability to kill them at night.

[callforjudgement]

I mean, in the 9:2:2 setup,

both

scumteams have equal NK immunity.

The kill-less scumteam has bulletproof, so it can't be killed by the other scumteam.
The scumteam with a kill doesn't have bulletproof, but they can't be killed by the other scumteam because the other scumteam don't have a kill, and there are no other killing roles in the setup that could kill them.

[callforjudgement]

Something very, very similar has already been done, I think.

We Need A Fifth. Actually, it's the setup Jungle of Bullshit was based on.

[callforjudgement]

Simple, Give a gun to the person below you on the player list, 3 people won't get guns, the person above them is a mafia member.

Scum can cause confusion via claiming not to have received the guns either. I suspect the setup is still broken but it's less obvious.

For example, a pattern of TSTST, one possible outcome is for players 3, 4, and 5 to claim no gun, and player 2 to claim a gun. With TTSST, exactly the same sequence of claims could happen. With TTSTS, again the same sequence of claims is possible (and it has no scum in common with the first pattern!), although this would require scum nearby afterwards in order to prevent the sixth player giving it away. You also can't make people shoot to confirm that they have a gun; in each case, everyone claiming that they have a gun actually does (it's just that some people with guns falsely claimed not to receive them).

I actually think that the only playerlist pattern which actually 100% gives away the scum is TTSTTSTTS, but this is pretty unlikely to occur. You'd need a computer search to work out what the EV in terms of setup breaking is.

Other possibilities involve passing guns in other cyclic patterns, e.g. groups of three players swapping in a triangle. I suspect there's probably a broken combination somewhere, but currently am not sure where it is.

[callforjudgement]

Unnamed Combo Variant reminds me of the White Flag Gambit. (That is, the worse bussing becomes in a setup, the more value scum get if they pull it off successfully; in this case, you'd be hoping that the bus lynch doesn't go through despite how hard you're pushing it.)

[callforjudgement]

With respect to shos' setup: in the setup without VTs, town can just repeatedly no-lynch and wait for the scum to kill each other, then try to hunt down the remaining scumbag. This gives them a 50% chance of winning (which is the EV of a 3:1 nightless).

If the VTs are present, then this isn't obviously the optimal strategy for town, as the VTs would end up dying. But given that it has a 50% EV, town presumably have better win odds if there's a better strategy out there.

[callforjudgement]

The Townbloc

• 6 Vanilla Townies
  
• 4 Mafia Goons (special nightkill rules, see below)
  
• Instead of lynching, players vote on a "townbloc" forming (exactly) half the playerlist. (Most likely, they'd vote individual players onto and off the bloc, then vote to hammer the bloc.)
  
• If a townbloc that contains no scum is ever hammered, town win.
  
• If a townbloc that contains every member of the scumteam is ever hammered, scum win.
  
• After three attempts to form a townbloc (i.e. three townblocs are hammered, none of which are no-scum or all-scum), the game goes to a night phase. This repeats until game over (i.e. it's basically a Triple Day mechanic).
  
• At night, scum kills a town player, but also sacrifices a member of their own scumteam.
  
• Additionally, there's a restriction on kill targets; either both the killed townie and sacrificed scum have to be on the most recent townbloc, or neither of them are.

One interesting fact about this setup is that it has 50% EV no matter what the number of scum and town are (so long as there's an even number of total players; otherwise, "half the playerlist" is ill-defined). However, the actual numbers of scum and town likely make a big difference to the actual

winrate

(not to mention swing, enjoyability, etc.), which is mostly just an indicator that EV is not the only thing that matters! 6:4 seems like it's in a fairly interesting place (and after nine indecisive townblocs, leads to a 3:1 ending that's guaranteed to end one way or another on the next attempt; I imagine deadlines would be set for entire days (i.e. three townbloc attempts), rather than individual attempts).

[callforjudgement]

I was reasoning on the basis that if a bloc's indecisive, there's not enough information to tell why. I'd missed that that wasn't the case when only 2 scum were left. (The scum controlling both kills was intended to allow scum to eliminate players who had been confirmed as town or scum.)

It might need a small mechanical tweak to allow for the balance issue, or might not. An earlier version of the setup just had scum win if town failed to hammer a townbloc at :2, and that would probably fix things here. Not sure though.

[callforjudgement]

It should also be noted that unlike games with normal day/night cycles, where towns frequently underperform their EVs (especially in low-power-role setups), towns normally overperform EV in a Nightless. (Most likely this is because weak-at-being-town players are more likely to be lynched, whereas scum have no way to get rid of strong-at-being-town players.)

[callforjudgement]

24 is probably too fast, considering weekend V/LAs and the like. Apart from that, I like the setup, and agree that it probably shouldn't have a scum instant win condition.

One interesting property I note here is that a scum townbloc member has no incentive to pick anyone in particular other than trying to look town, in much the same way that scum in a (two-faction) normal game have no incentive to scumhunt other than keeping up appearances. I consider this to be a feature, rather than a bug.

[callforjudgement]

I'd say having an unaware modifier isn't Normal, but it's entirely acceptable to have a "your role is either X or Y but you don't know which" style of role in an Open. (Dethy is a good example of that; it's imbalanced and emphasizes puzzle solving over scumhunting, but I don't have any rules-based issues with it being in the Open queue.)

Does the rule that Open setups have to be Normal technically still exist? Because that isn't really being enforced atm (e.g. I'm pretty sure White Flag would fail a Normal review under the current Normal guidelines).

[callforjudgement]

I agree that Wolf-Sorc-VT-VT is a scum win. The Sorc isn't going to vote for the wolf, thus the wolf can't be lynched, and in order to force a no-lynch, both VTs have to give themself away as town, allowing the Wolf to nightkill corectly.

[callforjudgement]

How many Cliq members are there total? A fixed number, or does it vary?

[callforjudgement]

Gambler Nightmare

Rolelist

Mafia Goon
Mafia Goon

Town Vigilante
Vanilla Town
Vanilla Town

Special mechanisms

Town wins if:
- Both mafia are dead
- Mafia shoots a vanilla town
Mafia wins if:
- The vigilante is killed
- The vigilante shoots town
- There's only the vigilante left
(One of the win conditions is enough)

If the vigilante decides to kill someone, mafia won't use their factional kill.
Mafia may choose to not kill.

Here's the best I can do in terms of maximizing the EV for town:

Two players volunteer to be part of a lynch pool (the vig doesn't volunteer; VTs volunteer, but not necessarily immediately to help hide the vig; scum can also volunteer). The other three players are a vig pool. Lynch the scummier lynch pool member. If you lynched a VT, the Vig shoots overnight, and you lynch outside the lynch pool with 65% probability the next day. If you lynched a Goon, the vig doesn't shoot, and you lynch outside the lynch pool every time if the Vig dies overnight, or with 57% probability if scum don't shoot. If the game goes to N2, the vig shoots, and aims in the vig pool.

With 0 scum in the lynch pool: you lynch a VT, the vig has a 66% chance of hitting scum overnight. If they miss, scum win. If they hit, after the claim, town win with probability 65% (if you try to lynch the vig, they claim to save themselves, and if two players claim vig you no-lynch and let them kill each other, thus the victory depends on whether you lynch inside the lynch pool or not). 43% EV for town.

With 1 scum in the lynch pool: There's a 50% chance of lynching a VT, in which case town can only win via vigging a Goon (33% chance), then lynching inside the lynch pool (35% chance); the probability of a town win from this branch is only about 6%. On the other hand, there's a 50% chance of lynching a Goon. If scum shoot in this situation, they have a 50% chance of autolosing, and a 50% chance (1/4 total) of hitting the vig, at which point town lynch in the vig pool which now has a townie and Goon in and so win half the time. Town have a 75% chance of winning in this situation, and so this contributes 37% towards town's victory odds if scum adopt this strategy; thus town EV is 43% in this case too. If scum don't shoot, then they have a 28% chance of being lynched right away and losing (because the vig can claim to save themselves, thus the lynch is aiming for their pool 57% of the time and they get lynched over the VT half the time); the rest of the time, a VT gets lynched, and town win anyway half the time if the vig hits overnight. With this scum strategy, town EV is 6% from mislynch+vig+lynch, 28% from lynch+lynch, and 18% from lynch+mislynch+vig, a total of 52%. As such, scum are going to shoot overnight in this scenario, holding the town to their 43% EV.

With 2 scum in the lynch pool: A goon gets lynched N1. If they shoot overnight, they have a 2/3 chance of hitting a VT (as they have no more information than when they started), so they don't shoot (this is worse than their other options, such as simply not volunteering). In the daytime, town wins with 43% chance when they lynch the other vig pool member, and otherwise lynches a VT and autoloses when the other VT gets vigged. EV for town is 43%.

So unless I'm missing something (e.g. one of the strategies listed is suboptimal; perhaps three members is a better lynchpool size, and this size is achievable because town have an autowin if both scum and the vig refuse to join the lynchpool), this setup is fairly balanced when broken correctly, as 43% is pretty much right where we'd expect it.

[callforjudgement]

Oh, I was analysing it as though the Mafia won if the Vig was

lynched

.

If kills count too, it's almost certainly scumsided, given that I couldn't get it to come out townsided even with the more lenient win conditions.

[callforjudgement]

OK, so what about this updated version of Creature's Gambler Nightmare setup?

Gambler Nightmare 2

• Mafia Goon
  
  
• Mafia Goon
  
  
• Town Vigilante Global Roleblocker
  
  
• Town Bulletproof Paranoid Gun Owner
  
  
• Town Bulletproof Paranoid Gun Owner
  
  
• If the Vigilante makes a kill, they globally block the scum nightkill for that night. Blocked actions don't trigger PGO responses.
  
• The Mafia endgame a lone Vigilante day or night, and (as usual) they endgame if scum >= town and the Vigilante is dead.

I made the town somewhat more powerful in the hope of balancing the setup. Looking at the situation where there's no attempt to break the setup: if town lynch scum D1, most likely nobody will shoot overnight (unless the Vigilante was outed in the process), and they thus have two chances (a lynch + a vig) to find scum. If a townie is lynched D1, scum get a "free shot" at the vig overnight (automatically winning if it hits, and going to a 50% EV endgame otherwise). If I've done the maths correctly, the EV is therefore 35% without a setup-breaking attempt (this figure takes into the account that possibility that the Vig is put up for lynch first and has to claim to save themself). If you do try to break the setup using similar principles to the ones I posted above, I think town can get their EV somewhat higher but not unbalancedly so; I haven't worked out the details yet.

@LicketyQuickety: The easiest way to coordinate in this case is for the Cop to investigate only scummy-looking players, and claim their investigation in advance. The scum can't kill the Cop through Doctor protection, and killing the Cop target to deny the investigation would effectively be making a town-directed kill, which is bad for them.

[callforjudgement]

Do the Mafia win together? If not, are they even really Mafia at that point?

[callforjudgement]

If a townie is lynched D1, scum get a "free shot" at the vig overnight (automatically winning if it hits, and going to a 50% EV endgame otherwise).

Maybe I'm missing something in the setup, but if townie is lynched D1, the Vig has a 1/3 shot of hitting both scum if he shoots, while letting scum shoot has an EV of 1/4.

Also, Vig shooting in the scum lynch case has the same EV - 1/3 hit, plus 2/3*1/2 lynch. Doesn't matter if Vig is outed or not, or if the lynch is scum or town, Vig can shoot. EV is exactly 50%. I doubt they can improve on this.

Ah right, I missed that the vig would shoot after a townie lynch (which is weird, because I considered that case correctly in my previous analysis).
After a scum lynch, though, the vig dies if they shoot and miss, leading to a 1/3 rather than 1/2 chance of lynching correctly, so the EV in the "lynch scum then vig someone" case is 1/3 + (2/3*1/3) or 4/9. (Unless I've missed something again, which is quite possible.)

[callforjudgement]

I am enjoying the discussion.

I am just assuming in a setup like this that playing things by the numbers is the right play without considering overall Optimal play? What I mean by Optimal is taking into consideration the psychology of the game as well as the numbers. Obviously depending on the setup the Optimal play is going to change. Take for example a Vanilla game where most of the game is itt. Is there a point where weighing in on the psychological components outweighs the EV?

The EV is basically the "baseline" play that town use if they don't have anything better. You can normally assume that town do better than the EV in practice. (For example, if you're in 2:1 lylo with one confirmed town, EV says that town have a 50% chance of winning, but in practice the confirmed townie gets to choose who to lynch and thus (you'd hope) town actually have a better than 50% chance.)

When you have a breaking strategy present, the simplest thing you can do is to follow the strategy, but use scumhunting to decide which choice of player you make to lynch/vig/investigate/etc. when you the strategy gives you a free choice of player. This is a pretty common play (and leads to town scoring well above EV in Nightless games; oddly, this doesn't seem to happen to so much of an extent in games with a normal day/night cycle). Much riskier is deviating from the strategy because your reads are strong enough that a normally low-percentage play becomes a good one. It's rare for players to be this confident in their scumhunting ability, and even players who are that confident are often incorrect to be, so it doesn't happen very much and doesn't succeed all that much when it does.

In your example of a vanilla/mountainous game, EV theory just says "lynch any player", which in practice becomes "lynch the scummiest player", and thus theoretically allows town to win above EV. (In practice, town do well below EV in vanilla-heavy setups with a normal day/night cycle such as 11:2 Mountainous and (as of the most recent statistics, although I suspect they're out of date) White Flag, but that may be the result of a small sample size.)

[callforjudgement]

First, that's not Mafia.

Second, it's broken (100% town EV). BP townie claims. (If they're counterclaimed, you lynch both claims and dayvig the survivor. Or dayvig them both, either works.) Assuming no counterclaims, dayvig claims (and counterclaims, if necessary, with a bullet), then you lynch every unclaimed player twice. Anyone who survives must be the Survivor and so gets dayvigged, and there's no chance of town losing as there are at least two townies (the BP townie and the dayvig) alive, so you have to end up lynching the Survivor eventually.

[callforjudgement]

Finally, even though I called it survivor here, I could very well just call it a Mafia Goon in a nightless game and then it would be considered mafia.

No it wouldn't.

When people say "not Mafia" in this thread, that normally means "there isn't a scumteam consisting of 2 or more members". (It can also mean "town isn't the largest faction", but that's much rarer and mostly happens with AitP variants.) What the roles are

called

is irrelevant. (In particular, the town is typically considered to be the faction with the least information. This is why town/scum tend to reverse roles in 1:1:1 endgames; the scum don't necessarily know which of the other players is the townie, but the townie knows both the other players are scum, and thus knows more.)