[FEBRUARY CHALLENGE] Voting!

[callforjudgement]

That's basically Rarefaction adapted to ten players (which means the Micro Queue is probably the wrong place for it :-P)

The original version of the setup cancelled the D1 lynch with no flip and then split its nine players into three sets of three, but that obviously doesn't work as an Open, so changing it to two sets of three was done for the Open version. Changing it to ten players lets you go back to the three-sets-of-three version.

[callforjudgement]

The Representative

20 players; 6 scum, 14 town. These are arranged by the moderator into 4 groups of 5 players each; one group contains 3 scum, one contains 2 scum, one contains 1 scum, one contains 0 scum. Which players share a group with which other players is public information, chosen either during signups or during role randomization, but the number of scum in each group is not told to the town (i.e. the moderator randomly chooses which group is the 3-scum group, which is the 2-scum group, and so on).

Scum can privately communicate with each other at any time; there are no nightkills. The groups have neighbourhood functionality, with players able to talk with other members of their group via a Private Topic for the group, but only during Private phases.

The game alternates between Private phases (where it starts) and Public phases. During a Private phase, the main game thread is locked, each group elects a "representative" for the group via voting. This is compulsive; due to the potential for ties when groups can be scum-dominated, scum may specify in advance of the deadline how they wish any ties they're involved in to be broken, and if a tie happens entirely among town or scum fail to specify a tiebreak before deadline, the moderator breaks the tie secretly at random.

During a Public phase, the neighbourhoods are locked, and the identity of each group's representative is made public. Only these representatives may talk in the main game thread (although they can certainly read through their locked private topics to find points that they want the rest of their group to say). As usual, posts from the private topic cannot be posted directly in the public thread, only paraphrased.

The representatives can talk among themselves in the public game thread, and vote for an entire group to eliminate (this is compulsive, with scum able to break ties they're involved in, just as with the private discussions earlier). Once a group has been chosen, then every player in that group is eliminated; all the representatives are also eliminated, taking no further part in the game. (So at the end of Public Phase 1, eight players will be eliminated; the four representatives, plus the other four players in the eliminated group). The eliminated players all flip, but in different locations: an eliminated representative flips only in their group's private topic (without the flip being posted publicly), and a player from an eliminated group (other than the representative) flips publicly in the main game thread. (This means in turn that the representative of the eliminated group doesn't flip in a useful way at all, with only eliminated players being able to see it.)

This process repeats for six phases total, except that in Private Phase 3, each group elects two representatives rather than one (so Public Phase 3 has four players posting, not two). This means that 8 players are eliminated in Public Phase 1, 6 players in Public Phase 2, and 5 players in Public Phase 3. At this point, there will only be one player left remaining; that player's faction wins. (The game can be called early if one faction is mathematically incapable of winning, e.g. if both non-representative players in Public Phase 3 have the same alignment.)

This is a pretty confusing setup to think about. The scum will have control over many of the votes, but at least for private votes (and possibly also public votes?) they have an incentive to manipulate the votes without giving themselves away in the process (they'll have to send a town representative sometime, and they really won't want them saying "I think the rest of my group is scum"; that said, a scum representative would also be saying that about a town group!). As such, I'm not 100% certain on the balance, but it could easily be varied by adjusting the mix of groups (e.g. you could move to 3, 2, 2, 0 or 3, 2, 0, 0 to make it more scumsided or townsided respectively). The game is, however, somewhat self-balancing (as whenever scum have control over the public vote, that's fewer scum left in the groups to be potentially chosen as the final player).

For what it's worth, although the strategy in the private vote is confusing, in the

public

vote, you're just trying to get rid of as many scum as possible if town (or keep as many scum as possible alive if scum). That means that it's clearly helpful to be able to scumhunt, both as to which group contains most scum, and to figure out how many scum are in your own group.

[callforjudgement]

Town could probably break the Face Eaters setup by dividing the roles into three groups of three, then having every player claim which group they're in. This gives town a huge amount of information, without giving the Face Eaters much additional opportunity to get a large set of guesses correct at once.

[callforjudgement]

Re: Tough on Crime, if a Parole Officer adds someone to the parole list, when does town discover that the player in question has been paroled? Start of the next day?

[callforjudgement]

Favourite at the top. (For what it's worth, there are no setups here that really caught my imagination, including my own; the top of the ranking is more "null-to-good" rather than "this setup is great, it should be run".)

Rivalries

: I like the idea behind this one. I suspect that the numbers are wrong, though (although I'm not certain): even with the 12.5% instant-win chance accounted for, it doesn't feel like town have enough information to balance the setup. (However, the technique I used to check this is fairly experimental at the moment, without much rigour, and involves guessing at how some things would go, so my calculations may be off; at least, it doesn't seem ridiculously scumsided, and it's plausible that it could be balanced.)

Romeo, Romeo

: I'd expect this to be townsided, but again, I like the general idea behind it. A 9:3 nightless is townsided, and I don't think that scum can scumside it far enough with the pairings (especially as Mercutio can act as an Innocent Child if necessary, although the unpair power is probably more useful if it ends up triggering when Mercutio's partner is lynched); but maybe I'm wrong on that! In a way, the pairing mechanic is a method of giving scum compensation for the lack of nightkill, as they can pair themselves (or scummy-looking townies) to powerful town players.

The Representative

: I still like the idea behind my own setup, but it's a bit clunkier than I'd like, and I was unhappy with the details even when I posted it; I just couldn't find a combination that worked better. (Maybe it'd be better off working this sort of idea into a Large Social Game rather than Mafia?)

Recruitment

: I think this is townsided. With a successful recruit, scum make this into 9:3 White Flag – a somewhat scumsided setup – but failing the recruit makes them lose outright, and the setup is basically equivalent to an 8:2 Lovers setup without recruiting (as the applicants cancel each other out). Incidentally, it's to the scum applicant's advantage not to claim, and the town applicant's advantage to pretend to be a scum applicant (!), so it's likely that scum won't have any basis better than chance to select their kill.

House Party

: I dislike the general idea of this; throughout much of the game, scum are highly likely to have a kill that they can make safely (i.e. without giving away too much information). As such, it's comparable to a White Flag where the town has a small number of factional JKs available (probably on the order of 2 to 3). It's probably balanced, but the mechanic feels like it'll feel pointless from town's point of view and frustrating from scum's point of view, until late game where it may end up randomly winning the game for town. Incidentally, this is likely to be very comparable in how it plays out to Tough on Crime, but I've placed it higher because some players may like a game in which the strong townies can be protected for a while.

Tough on Crime

: Surprisingly similar to House Party; again, it has a mechanic that won't matter through much of the game, but can become very relevant at the end. I believe this is probably townsided, due to a combination of town's power roles and town's endgaming power. I ranked this below House Party because paroling can't be used to much effect earlier in the game (it sometimes restricts who

makes

the scum kill but scum normally have a lot of choice in that), whereas partying at least does something (restricting who the scum kills).

Three Pods

: Not new, this is basically a minor tweak on Rarefaction to run with 10 players. It may well be better than the original, but Rarefaction isn't massively popular as a setup, and I've ranked this setup lower because I think the monthly challenge would be better as a breeding pool for new mechanics.

The Boardroom 3

: I think it's still possible for town to work around most the mechanic? Town work out who they want to lynch, first, then select a player who's responsible for implementing the lynch, sending that player to the boardroom (who then invites someone they believe to be town, and also the player that town wants to lynch). The lynched player self-votes (if they refuse to, the extra player votes them), and the implementor hammers. If the wrong person ends up dying, then either the lynch was on scum and the implementor invited scum (which would be a pretty inept thing to do, although not impossible), or else the implementor is scum. So scum do get something of an advantage from the mechanic, but not much of one, and it may not come up in any given game. I expect the setup to be balanced, but the mechanic to not be running to its full potential.

Love 'em and Leave 'em

: Probably townsided. The presence of the JK helps to alleviate the usual skew problems seen in a 10:2 (in that they give a plausible "partial good result" for town that falls short of a lynch, making the average case more distinct from the perfect-for-scum case). However, this is still nonetheless a 10½:2 (almost equivalent to a 10:2 or 11:2 based on which scum dies first) with a couple of confirmable townies and the chance to block kills and push the setup back onto odds, and that seems likely to outweigh the normal town disadvantages in 2-scum Minis.

City Council

: Isn't it correct for town to just systematically shoot inside the Councillors, lynch outside them? That seems to break the setup; if the Sergeant-At-Arms survives long enough, town need only make a single correct decision (kill or lynch), with all the others being incorrect, and still manage to end up in a 3:1 lylo. (Not to mention that if the scum Councillor dies early, the town Councillors become Masons.)

Three Strikes

: 5/7×4/6×3/5 = 2/7 chance of town missing all three times, so town EV is 5/7. That's way too high as it is, and the Innocent Child has no influence on the setup and is clearly just there to make up the numbers, making it rather inelegant. Additionally, this is functionally a no-flip setup (because a scumflip would end the game), which is dubious in its own right.

Guidance

: It took me some time to even work out how to count the number of setup randomization possibilities for this one (what we're looking for is a derangement of the players for which no scum maps to another). Anyway, it's unclear what a "co-jailkeeper" is, but as scum control the jailkeeps of half the town, any plausible meaning allows scum to arrange the jailkeeps to block their own kill (unless town try to organize to all target the same player, in which case the kill will fail for a different reason). That in turn means that this is effectively a vanilla setup in which town know that certain players can't be scum together, but that isn't nearly enough information to balance an 8:4.

The Wilderness

: Broken for town. Every day, no-lynch, and vote a scummy-looking player into the Wilderness. The game is now equivalent to 11:3 Nightless (maybe 10:3 if the scum guess the Headhunter correctly). (When a player is alone in the Wilderness, they have no actions that don't commit suicide. You forgot to add in the "Shoot air" action from RAF. Even fixing that, though, the game is still townsided unless the Mafia Head can guess the Headhunter.)

Face Eaters

: Doesn't comply with the challenge restrictions. Also, I expect it to be broken, but haven't spent much time looking at it (beyond what I've already posted in-thread) because invalid entries surely have to come last.