Balance Discussuon Thread

[Ircher]

Scumsided. 2-Shot would possibly balanced.

10-3 Mountainous and 7-2 mountainous are both very scumsided.

6 VT, 1 Cop, and 2 Goons is still somewhat scumsided. (But fairly close)

[mhsmith0]

6vt/cop/goons is scumsided in a closed setup, fine in an open or semi-open one like matrix6. Closed is always an edge to scum since they have much more room to strategically fakeclaim as convenient.

Also 9/2, rather than 7/2, is an edge to town. Giving town a 1-shot cop on top of that at least makes it closer to balanced, though yeah I'd probably say that town probably does need a 2-shot cop there to make it fair. 9/2 with just a 1-shot cop is probably fine as an open/semiopen tho.

[Antihero]

Also 9/2, rather than 7/2, is an edge to town.

9v2 is 35% town ev vs 7v2 at 30% town ev. not an enormous difference.

educated guess for town ev in vijarada's setup is a town ev somewhere in the high 30s - low 40s so a little scumsided.

[mhsmith0]

I meant edge to town compared to normal balancing. One extra mislynch is still pretty meaningful and relevant for balancing imo.

[Fro99er]

you live you learn

Exactly. Wasn't criticizing you...sorry if it came off that way. (Didn't even realize you made that comment in BEES until I went back to look at it yesterday)

Was just curious about what people think of it now.

[pieguyn]

15-4
Mafia Goon x 3
Mafia Doctor
------------
Town Gunsmith
Town 2-shot Vigilante
Town Mason x 3
Town Fruit Vendor
VT x 9

[Firebringer]

15-4
Mafia Goon x 3
Mafia Doctor
------------
Town Gunsmith
Town 2-shot Vigilante
Town Mason x 3
Town Fruit Vendor
VT x 9

I'd say scum sided.

Doctor gives false positive.
Vigilante can shoot masons and has lower chance of killing scum with scum.

[mhsmith0]

Hmm, not sure tbh. Mafia doc is a big counter to both gunsmith and vig, but three never lynchables and a gunsmith that SHOULD get a result or two is pretty solid for town. Off the cuff I'd say about balanced, MAYBE you give town just a little bit extra but honestly not much.

[pieguyn]

that setup was actually ran before and it ended in a close game; the game entered 4p MYLO with no clears and me not even being able to predict who'd win. opinions on balance at the time ranged from "balanced" to "very town sided, scum got lucky for it to be this close". at the time, I felt that it was maybe slightly town sided and that I should have removed the third mason.

interesting to see that a year and a half later, I get opinions both ways about it.

[Faraday]

it might be okay, i don't like the doctor though - mostly bc im not a fan of the false positive potentially deciding the game. vig has bad synergy with the masons/fruit vendor too.

i think you could lose the doctor and give scum an x-shot rolecop or something and it'd be fine tho.

it's def not "very town sided", though. did town win? bc that comment makes more sense if they did

[Antihero]

did town win? bc that comment makes more sense if they did

you would be amazed...

[mhsmith0]

Mafia doctor is EXTREMELY strong against the gunsmith and vig combo, effectively a godfather (vs GS) AND a "better than roleblocker" against the vig (since it stops a scum kill but allows a town kill). The rest of town's power isn't garbage but is still limited. I KINDA think that it'd be balanced with the vig being full or the doc being like 1-shot but idk.

[callforjudgement]

I know nothing about setup balance. How do I learn how to be a good setup designer? What heuristics are used when balancing setups?

There are plenty of different ways you can get a sense of what a setup is like, from things which are fast to calculate but somewhat inaccurate, to things that are much more accurate but a pain to work out.

My go-to "quick but not as accurate as you'd like" method of balancing a setup, which I use for most 10:3s, is to try to work out how many alignments town are likely to figure out purely based on nightplay over the course of the game. For example, for a Cop in such a setup, we'd typically put this somewhere in the 2-3 range. You have to take the context of the setup into account when making this sort of judgement; if it's known that there's a Doctor in the setup and the scum have no way to punish the cop for claiming, the value's more like 5 (because the Cop can claim D1 and usefully investigate every night from then on, in addition to most likely confirming themself with the claim; there's a reason that a Follow the Cop combination that the town can reasonably become aware of and take advantage of is almost universally considered a major problem with a setup). Protective roles are converted to confirmations in three ways: they can amplify the power of other roles by keeping them around longer (making those other roles more valuable), they can save confirmed players from the scum nightkill, and they can sometimes buy mislynches (a mislynch is worth around 2 confirmations), although you'll need to take into account even/odd effects here (e.g. saving a confirmed townie from the nightkill if it's 3:1 going into night isn't all that helpful). You can also take into account whether the confirmations are likely to be on town or scum, and on who's controlling who gets confirmed (scum's choice, town's choice, or random players) and when, but I tend not to use that for a rough check. My baseline for designing a 10:3 is to arrange things such that a little over three players will get confirmed purely by night actions, plus an additional 1 player getting confirmed as a result of claims. (Incidentally, this goes a long way towards indicating why setups that the playerbase as a whole believe to be balanced are typically scumsided, and likewise that setups that actually are balanced often feel townsided; if you're playing scum, getting that many players confirmed out from underneath you feels a bit like the mod is screwing you over. But towns are so bad at scumhunting that they typically need the help.)

Another rough method of assessing a setup's balance is an EV check, which is unfortunately often very difficult to calculate unless the setup is very symmetrical. Basically, this consists of determining the best strategy for each player based on only the information they (and players they can communicate with in alignment-confirmed PTs) have, and then assuming that each player acts according to the strategy and makes choices at random when there's no nightplay reason to prefer one over the other (in particular, a random player gets lynched unless someone breaks the symmetry by claiming). Then you work out the probability that town wins. I don't bother with this unless the setup has no power roles (or specifically, that all scum have the same role and all townies have the same role), but it can be quite useful for mechanically unusual setups which are vanilla apart from their weird mechanics (i.e. the sort of thing I run all the time in the micro queue). What values for EV are balanced seems (based on experience in the Open queue) to depend mostly on the nature of the scum nightkill; if scum get a kill per lynch and they have unrestricted choice on who to kill (i.e. the standard Mafia nightkill), this seems to be around 50% (I think 45% is commonly used as a balancing point but think even that might be too low), whereas if scum's only way to eliminate people is the lynch, even EVs as low as 20% can be balanced (and an EV of 50% will almost certainly lead to a town win; a fun example is that a vanilla 12:4 nightless has a 50% EV, but I think almost everyone experienced with balancing setups would expect town to win almost every time).

For accurate comparisons, you'll want to compare to setups which are already well understood. For example, a closed-but-known-to-be-simple game with 6 VTs and 1 Cop against two Mafia Goons is known to be very close to balanced, due to years and years of experience in the Newbie queue (it's probably a little scumsided because town have less information about the setup than they would in an actual Newbie game). If someone asked me to review a Micro setup where all town power was in investigatives, I'd compare it to that setup and see how I thought it compared; for example, we can deduce that 6 VTs and 1 Tracker against two Mafia Goons is probably scumsided. In general, Micro setups are the setups you can be most confident in reviewing because a) the Newbie queue gives so much useful data (and is much easier to analyse than the Mini queues are), and b) you can often enumerate all possible scenarios, which is not very useful in a Mini.

Another important tool when balancing a setup is a swing check; if the game is too swingy, the concept of balance becomes pretty much meaningless (because it'll be overwhelmed by random events). There are two main parts of this. One is an early-game swing check: what's the best possible scenario for town and for scum on day 2 (day 3 too, in larger games), and how likely is it to occur? There are plenty of setups that are otherwise good, but which become really unfun if a Mafia power role gets lynched on day 1 (because the rest of the scum get annihilated by power roles overnight). The other is to check for a concentration of late-game-oriented power roles; if I see a Mini setup with a Town Roleblocker, a Town Jailkeeper, and a Town Tracker, I'm almost certainly going to reject it as too swingy because all three of those roles become a lot more powerful the fewer scum there are alive, and thus the town power is going to be much more useful for town when they're doing well than when they're doing badly, which is kind-of the opposite of what you want. (Meanwhile, in a sufficiently large Large, this would be less of a problem because

three

late-game oriented roles isn't enough to overwhelm random events that happen during the game. Once a game gets large enough, pretty much any careful balancing by the moderator goes out of the window because there's too much going on that could interfere with it. There's a reason I prefer to stick to Minis and Micros for my reviews.)

One thing to note is that I've been focusing on town power roles, and ignoring scum power roles. The reason for that is that almost all scum power roles don't actually do anything in a vacuum; their purpose is to screw over the town power roles, and thus they're best measured via treating the town power roles as less useful, rather than trying to assess their power in their own right. For example, a Cop might typically get 2-3 investigations off in a Mini, but if scum have a Roleblocker, 1-2 is a more reasonable value. (Because a Roleblocker is capable of screwing up almost

every

town role, it often by itself weakens town by enough that you can add another fairly powerful town power role and yet town will still be behind. Being careful with Mafia Roleblockers is a lesson that the set of reviewers as a whole have had to learn.) Along similar lines, if you're reviewing the sort of game which wants to give Mafia a counter to every town power role, you'll find that the resulting weakened power roles struggle to make up enough power to balance the game. A better way to weaken overstrong roles is to place modifiers on them, or to replace them with similar roles that have limitations.

[FakeGod]

Hey, Anti, what about this setup:

3 Scum Gentlemen/Ladies
14 Town Gentlemen/Ladies

One Town player's alignment is confirmed at game start. Players proceed to pair up via asking and answering publicly. The loser who has no dance partner then flips.
Dance pairs become lovers. Up to 2 lynches in Day 1, 1 Nightkill in Night 1, and lynch until game end in Day 2.

[Firebringer]

Hey, Anti, what about this setup:

3 Scum Gentlemen/Ladies
14 Town Gentlemen/Ladies

One Town player's alignment is confirmed at game start. Players proceed to pair up via asking and answering publicly. The loser who has no dance partner then flips.
Dance pairs become lovers. Up to 2 lynches in Day 1, 1 Nightkill in Night 1, and lynch until game end in Day 2.

I know you didn't ask my opinion but since this is effectively nightless with only one power role, plus 1 night kill, think it's pretty scum sided.

Like almost as bad as mountainous.

[mhsmith0]

I'm not anti, but I'd say it's structural balanced as described with two important caveats

1) scum neighbors are notoriously powerful (if there were lovers without PT then I bet it'd be explicitly townsided)
2) towns option to suicide is a poison pill that undisciplined towns love to take, and towns are notoriously undisciplined (indeed, I'd say the poison pill option is the biggest reason scum have been winning, and it's pretty balanced without said poison pill option)

[FakeGod]

Quiet in the peanut gallery please.

Antihero

is the one I asked for.

[mhsmith0]

[Antihero]

Hey, Anti, what about this setup:

3 Scum Gentlemen/Ladies
14 Town Gentlemen/Ladies

One Town player's alignment is confirmed at game start. Players proceed to pair up via asking and answering publicly. The loser who has no dance partner then flips.
Dance pairs become lovers. Up to 2 lynches in Day 1, 1 Nightkill in Night 1, and lynch until game end in Day 2.

no suicide?

[Antihero]

yeah... ditching the suicide mechanic might be the thing that makes this setup balanced

i say run it.

[FakeGod]

What if I told you that even with suicide mechanic, it's HEAVILY townsided?

ehuehuehuehuehuehuehuehuehue

[RadiantCowbells]

I don't like it because lover's dance is better with no inghtkills.

[mhsmith0]

I mean theoretically it SHOULD be townsided (if you presume that a townie gets left out early and the IC pair is town and eats a bullet, neither of which is certain [the first especially is a chance for town], then it's six pairs, three good three tainted, and presuming RNG lynches, you have 50% win odds for town, and town SHOULD exceed its RNG changes), but it hasn't actually turned out that way.

And I do maintain that the reason for this is that neighbors and neighborhoods are extremely powerful weapons for competent scum. Summer Waltz had pie get pocketed by her lover, New Years had scout and kagami get pocketed, and I'd guess the first game had at least one such outcome as well. Towns aren't good at solving in hoods, and scum are good at manipulating in hoods.

In 10v3 mini normal games, scum are UNDEFEATED as neighbors or neighborizers, a perfect 7-0 record:

Spoiler:

Game 1460: Normalville Mafia
Moderator: ac1983fan
Current Update: Mafia Win

Players
PeregrineV, Town Bodyguard, lynched Day 2
Darthe, Vanilla Townie, lynched Day 1
Slandaar, Town Neighbor, died Night 4
Grimgroove, Town Neighbor, died Night 5
Ztife, Mafia Goon, lynched Day 3
Shamrock replaces Nachopappa who replaced thegooner, Vanilla Townie, lynched Day 6
Baezu, Town Odd-Night Cop, died Night 3
Nachomamma8 replaces Kattaze, Vanilla Townie, died Night 1
ChannelDelibird replaces Red Dragon, Vanilla Townie, lynched Day 4
StupendousMan replaces Miss Stranger who replaced BP, Mafia Neighbor, lynched Day 5
Lucky2u replaces chernobylcitybus, Vanilla Townie, endgamed
Elyse, Town Miller Jack of All Trades, died Night 2
SunnyDays replaces Future, Mafia Two-Shot Roleblocker, survives


Game 1463
Moderator: N
Current Update: Mafia Win

Players
yessiree, Town Gunsmith, died Night 4
Generic, Mafia Neighborizer, survives
Mutleyddmc replaces levio, Vanilla Townie, endgamed
Kid A, Vanilla Townie, died Night 5
DCLXVI, Town Bodyguard, died Night 1
Bub Bidderskins, Vanilla Townie, lynched Day 2
Alabaska J, Town Even-Night Vigilante, died Night 3
NotThePope, Mafia Goon, lynched Day 1
TMTOLBTWNTOF, Vanilla Townie, died Night 2
Smudger, Vanilla Townie, lynched Day 3
RadiantCowbells, Vanilla Townie, died Night 2
Alduskkel replaces absta101, Vanilla Townie, lynched Day 6
Om of the Nom replaces Kingsrush, Mafia Goon, lynched Day 5


Mini 1511: Insurgency Emergency
Moderator: JacobSavage replaced Fegelein
Current Update: Mafia Win

Players
Elyse, Mafia Doctor, Survived
Albert B. Rampage, Vanilla Townie, Lynched Day 1
aphix, Mafia Goon, Survived
Titus replaced Ant_to_the_Max, Town Neighbour, Died Night 3
Skelda, Mafia Neighbour, Lynched Day 3
Kublai Khan, Vanilla Townie, Died Night 3
implosion, Two shot Vigilante, Died Night 4
Humble Poirot, Vanilla Townie, Lynched Day 4
Fusion, Town Jailkeeper, Died Night 1
EspeciallytheLies replaced AzBlueM, Vanilla Townie, Endgamed Night 4
evilpacman18, Town Gunsmith, Died Night 2
havingfitz, Vanilla Townie, Endgamed Night 4
RachMarie replaced gorkat, Vanilla Townie, Lynched Day 2


Mini 1666: The Reds
Moderator: pisskop
Current Update: Mafia Win

Players
dragonspawn, Town Tracker, lynched day 2
MonkeyMan576, Vanilla Townie, died night 1
Prolasped Brain replaced Zerediah, Mafia Roleblocker, lynched day 3
Riabi, One-Shot Bulletproof Townie, died night 4
Kitty Galore, Mafia Neighbor, survived
Titus replaced VictorDeAngelo, Town Cop, died night 3
RandomActs, Vanilla Townie, lynched day 4
TheSoldier, Town One-Shot Vigilante Neighbor, endgamed day 5
Lapsa, Vanilla Townie, endgamed day 5
Terrador, Vanilla Townie, lynched day 5
Drixx, Vanilla Townie, lynched day 1
Slandaar, Town Neighbor, died night 2
Shinobi replaced Keybladewielder, Mafia Godfather, survived


Mini 1685: BooneyToonz II - Looney Tunes
Moderator: Boonskiies
Current Update: Mafia Win

Players
Pegasus30 replaced DrDolittle, Town Innocent Child, endgamed day 5
Aneninen, Town Vengeful Neighbor, lynched day 2
BlueBloodedToffee, Vanilla Townie, lynched day 4
pisskop, Town Neighbor, killed day 2
massive, Mafia Neighbor, survived
Varsoon, Town Neighor, survived
Zebulin, Town Neighbor, lynched day 1
Salamence20, Mafia Goon, lynched day 3
RadiantCowbells, Vanilla Townie, endgamed day 5
Mainez, Town Jailkeeper, died night 3
T S O replaced Big Bad Bull, Vanilla Townie, lynched day 5
BBmolla, Town Neighbor, died night 4
evilpacman18, Mafia Encryptor, survived


Mini 1733: Jurassic Monkey
Moderator: Monkeyman576/Farside22
Current Update: Mafia Win

Playerlist
Scorpious, Town Commuter, lynched day 3
Wanderer-nl, Mafia Neighborizer, lynched day 5
The Archmage Ludicrous, Vanilla Townie, endgamed day 6
pistachi0n, Vanilla Townie, lynched day 6
All Alone, Vanilla Townie, died night 5
MarioManiac4, Vanilla Town, lynched day 2
ICEninja, Town Cop, died night 3
Garmr, Vanilla Town, died night 2
Syndesis, Town Innocent Child, died night 4
3dicerolling, Mafia Godfather, survived
Davsto, Mafia Goon, lynched day 1
Firebringer, Vanilla Town, died night 1
Alchemist21 replaced Diego1487, Town Universal Backup, lynched day 4

Mini Normal 1897
Alive (3)
Boonskiies, Mafia Goon, survives and wins.
Something_Smart, Mafia Goon, survives and wins.
horrordude0215 Jinzou, Mafia Neighborizer, survives and wins.

Eliminated (10)
Transcend, Town 1-shot Weak Friendly Neighbor, lynched Day 1
TwoFace, Vanilla Townie, killed Night 1
I Am Innocent, Vanilla Townie, killed Night 1
aronagrundy, Vanilla Townie, lynched Day 2
Eddie Cane, Town 2-shot Vigilante, killed Night 2
hapahauli, Vanilla Townie, killed Night 2
eagerSnake, lmays1999 Town Watcher, lynched Day 3
ThinkBig, Vanilla Townie, endgamed Day 3
Darkshadow64540, Vanilla Townie, endgamed Day 3
rb, Vanilla Townie, endgamed Day 3

It's an aspect of balance that simply isn't fully accounted for in terms of overall site meta.

(peanut gallery can't shut up it seems )

PS as a side note, the ability for players to pick their partners is i think pro-scum, since scum can choose people that they think they can manipulate, and that's probably more reliable than town knowing who they can legitimately sort in a hood.

[Antihero]

What if I told you that even with suicide mechanic, it's HEAVILY townsided?

ehuehuehuehuehuehuehuehuehue

i'd say this is probably one of those times when practical ev diverges from theoretical ev. (like scumhunter speed 8p)

[Powder Punk Girl]

Made a topic about it but I am hoping to get a quicker answer here. What is good ratio for 2 mafias and town.