[SETUP] Least Sus

[Irrelephant11]

11 v 4

Before each day, scum must split the playerlist into [# of living mafia] groups. These groups will be published at the start of the day, and the mod will highlight which group has the fewest mafia in it ("least sus"). In the case of a tie for least sus group, the mod will choose from among the tied groups randomly.

At no point can mafia create groups which completely match a previous day's groups, minus the dead players, unless it is impossible to not do so.


... is this breakable?

[Something_Smart]

Presumably the best strategy for scum is to make a group with a single town person and split the rest of the groups semi-randomly with at least one scum in each, then just kill the person they confirmed as town the next night.

[Irrelephant11]

Oh right! I forgot to type that each group must have at least 2 players

[Something_Smart]

Except going into lylo where that's impossible, I assume?

Doesn't seem eminently breakable, I don't think. Probably the optimal strategy is to put one scum in each group some percentage of the time and do whatever the hell you want the rest of the time. That way it's pretty hard to get any solid info from the groups and there's a ton of WIFOM. It's probably fairly scumsided given that.

[Irrelephant11]

It's never impossible going into lylo. If you have 3 scum vs 4 town, for example, you're splitting the PL into 3 groups, which can always have at least two players each. In 1v2, all players are in 1 group, etc.

[Irrelephant11]

It does feel scumsided intuitively, but I haven't yet had the brainpower to figure out what town vs scum #'s would lead to the optimal balance between "none of these lists mean anything yet" to "the lists suddenly confirm half the living townies as town" before lylo

[Something_Smart]

It's never impossible going into lylo. If you have 3 scum vs 4 town, for example, you're splitting the PL into 3 groups, which can always have at least two players each. In 1v2, all players are in 1 group, etc.

Oh true I was thinking 4 groups for some reason, nvm.

Is it ever possible for the lists to confirm anyone as town? I guess if you got two scum flips off of the same list, it would confirm that day's least sus list as town.

[Irrelephant11]

If anyone who was ever in the "least sus" group flips red, the others from that group become conftown.

[Irrelephant11]

Also if you ever flip someone red from a least sus group, you confirm that every group from that day has 1 scum. Which totally flips the game on its head, which then gives scum a reason to not spread themselves out too much, which then lightly conftowns anyone in a least sus group

[Something_Smart]

Even with all that I don't think 11:4 is justified. 11:4 mountainous in practice would probably have a town wincon of like 4%, it takes quite a lot to balance it, and if town is flipping people in the least sus group then they are playing super risky and the reward isn't even

that

good.

Scum also have the option of confirming 2 people as town and just leaving them alive for a while, to ensure that the groups give no further info, if they want.

[Irrelephant11]

Totally fair! Any idea what numbers would be fair? I thought too many townies could lead to too many game days could lead to "oh we flipped a scum the day before lylo and now 2/3 of the playerlist is conftown" but whatever number works, I'm happy, if the idea works

[Irrelephant11]

confirming 2 people as town and just leaving them alive for a while, to ensure that the groups give no further info, if they want.

This is interesting. Scum would still be forced to shuffle the other groups while spreading out enough to ensure the same 2 townies get conftowned each day. That shuffling would provide enough information to cross out some scumteams. But you're right that it wouldn't do much more than that

[Something_Smart]

I thought too many townies could lead to too many game days could lead to "oh we flipped a scum the day before lylo and now 2/3 of the playerlist is conftown"

If scum lets this happen it's their own fault, I'd say, because they can preemptively kill the people who they expect might soon become confirmed.

It's really hard to tell what might be balanced just off of intuition. I might try putting this into my EV program to see what it says. Otherwise idk, you could try just running it at 10:3 and see what happens.

[Gamma Emerald]

Once the game goes down to 1 mafia the info becomes stagnant, no new confirms can be gleaned from future days’ groups
That should factor into the balance somehow

[Jingle]

I think as a large it’s an undesirable setup due to inevitable player attrition. 10:3 is probably close to runnable as 11:2 is a scumsided by ev and observation mountainous format and with two scum alive you’re functionally guaranteed a one in each group split. (A pool being least suspicious means it’s all town if there are fewer remaining scum than groups and thus you can’t allow a large least suspicious group as scum. Similarly, one scum remaining basically means scum loses, because scum can’t prevent cop checks on a minimum of four players and four conftown is a town win if they can continue generating conftown at any rate.)

[Irrelephant11]

Once again, that’s not how many groups there would be with one scum left

[Irrelephant11]

Even with all that I don't think 11:4 is justified. 11:4 mountainous in practice would probably have a town wincon of like 4%, it takes quite a lot to balance it, and if town is flipping people in the least sus group then they are playing super risky and the reward isn't even

that

good.

Scum also have the option of confirming 2 people as town and just leaving them alive for a while, to ensure that the groups give no further info, if they want.

I think this actually optimal play for scum, with a slight adjustment. Conftown 2 players, then kill and replace one each night. Makes for a game with 2 conftown every day except Lylo, where there’s WIFOM if scum are still using the same strategy for lylo or not

[Jingle]

Even with all that I don't think 11:4 is justified. 11:4 mountainous in practice would probably have a town wincon of like 4%, it takes quite a lot to balance it, and if town is flipping people in the least sus group then they are playing super risky and the reward isn't even

that

good.

Scum also have the option of confirming 2 people as town and just leaving them alive for a while, to ensure that the groups give no further info, if they want.

I think this actually optimal play for scum, with a slight adjustment. Conftown 2 players, then kill and replace one each night. Makes for a game with 2 conftown every day except Lylo, where there’s WIFOM if scum are still using the same strategy for lylo or not

I somehow missed the # of groups decreasing with scum elims, but I don't know that this is optimal play. There's definitely something to be said for picking strange group comps of roughly even sizes, because 10:3 mountainous is VERY scumsided. (80% scum EV) and that's basically what this is if you don't see two scum flips over the course of the game or flip the entirety of a group. It's probably ~viable given a townsided meta and the potential for useful information, but I think the mechanic works better as a supplementary mechanic to a non open than as a setup by itself.

[Irrelephant11]

I think if you play it out though, there's no way to not give town lots of information, it just takes a couple day phases. If 0 scum ever flip, maybe it does feel super scumsided, but a single red flip reveals tons about previous day's groups. I think. I've only written out a couple ways it could play out, there's a lot of options

[Jingle]

Yeah, that's kinda along the same lines as what I'm saying. You're starting with a very scumleaning setup and there's probably going to be late game townsiding information, so I think 10-3 is right, but probably not quite enough to be truly balanced. If I thought there was a likely chance of no town info from the groups it'd be a clear 11-2 setup (which, incidentally, I don't think the mechanic is super interesting without at least three scum.)

I just happen to also think the setup lends itself as a mechanic in a low power closed game more than a true open, because near mountainous setups where the information gain is backloaded tend to be frustrating for a large portion of the early game. (See FOTM a few years ago where a small arsonist team dominated town because town just didn't have anything happening to keep them invested.)

I think increasing game size is just going to make the mountainous drag more pronounced and thus not a good solution, but 10-3 seems playable at the very least, if still a little bit scumsided. I don't think that going the 2 conftown route is an objective optimal strategy (tbc, this is a good thing) because going three groups of the same size and putting 2-1-0 or 1-1-1 have potential benefits based on the strengths of the teams, making it a more strategically interesting game.

If you wanted to do something closed using this, I'd lean towards prs that are thematic to this setup: "You can't be killed by someone not in your group." or "Target a player: You will instead be in that player's group during the next day phase." as opposed to traditional roles like cop or doc.