Balancing a Game Philosophy

[Something_Smart]

The question was about how often town should win, that's a moral question. I personally don't find games balanced for enjoyment because I don't enjoy empty wins and I don't enjoy spending almost every scumgame fighting uphill.

This seems to be a minority opinion to me, because I certainly would not call winning a fair game "an empty win" not would I call being scum in a fair game "fighting uphill". I guess I can understand if you're so strong as town that winning a fair game is too easy to even be proud of, but you're so weak as scum that playing a fair game is too difficult to be fun, but that just means you're atypical. By definition, if a game is fair according to current site meta, an average person will have a 50% chance of winning, which should make their wins feel meaningful and their losses feel deserved, on the balance.

Just achieving so that people win 50% of the time isn't necessarily balancing for enjoyment either.

No, but see what I said to Isis-- it sure seems like the default, because it makes sense for people to want win chances to be dependent on skill rather than luck. The burden of proof is on the people saying they don't.

[callforjudgement]

For what it's worth, I believe scumsided setups are more fun than townsided setups (there is normally more town agency in scumsided setups than there is scum agency in townsided setups, because even the most scumsided of setups can be won by sufficiently good scumhunting; and one of the main tools to improve town win rates is by adding town power, which has an unfortunate tendency to get lucky and ruin the game for scum even when they play well, especially as quite a lot of town power is needed to reinforce it against lucky N1 kills and forced D1 claims). This implies to me that it might be preferable for the town winrate to be a little below 50%. However, I nonetheless aim for 50% in balance reviews, because it's what players seem to want and expect. (It might be interesting to allow moderators to intentionally run games that are balanced to a different standard, with a signup advertisement so that players who want to play, e.g., a scumsided game can do so.)

In my experience at least, scum wins are often in spite of the cosmic dice rather than because of them. I've seen/had games that felt like everything was an incredibly uphill struggle against setups that were just deeply setup to favour town. Correspondingly, I've had town games where I've had to do basically nothing to win because we had one or two PRs who got lucky and cleared out half the team.

I would agree with this, and it's a natural consequence of trying to balance games for 50:50 win rates. It's easy to create a game where town

can

win if they outplay the scum. The problem is, they normally don't – there are a lot of players on this site who are fairly useless when they draw town – so town need to be given extra help so that they don't necessarily need to play well in order to win, especially given that scum can nightkill the best town players but town don't normally policy-eliminate the best scum players. (I'm not necessarily sure I agree with the rest of what you wrote, though.)

This problem may be getting less bad over time. A few years ago, towns were doing so badly in Normals that the Normal Review Group intentionally gave them extra power to help level the win rates a bit. More recently, we've reversed most of the balance changes we made then, going back to an older balance standard, and the reason is that towns have gotten better and they don't need as much help as they used to.

[Gamma Emerald]

Agree scumsided setups are more fun than townsided
Source: got pinned in a townsided setup recently as scum and felt endless despair

[StarToAYoungCulture]

How do you even count the winning equity

[OkaPoka]

How do you even count the winning equity

https://wiki.mafiascum.net/index.php?ti ... EV_Project

this is the "standard", they use random elims with a few caveats (like with ICs and cops having fixed behaviors for claim strats), then through random eliminations you get %'s

then there is:

viewtopic.php?f=5&t=81311

which assumes elims are not random but you elim in a sorted order, everyone is at some fixed position on the list

and finally for closed setups with prs they guess basically

[Hoopla]

i agree with isis & co. the sweet spot for balance is probably about 40-45% EV for town.

at various times in the past, town winrates hovered around 35% (notably in the 3:9 era and early 3:10 era). it seemed too much hinged on cops/vigs performing well to get town over the line, that when these roles were outed/NK'ed early, it was an uphill battle and town would have to play to a high level to beat scum. it was still possible, but it was often an uphill battle (especially during the 3:9 games where the town was afforded one less mis-elim chance).

since becoming collectively mindful of balance, the groupthink seems to favour aiming for 50%. at least, when i was running the NRG group that's what we were aiming for, and from what i've seen in recent years, it seems to hover around 50-55%. but like others have said, there are just too many games where scum's path to victory is extremely narrow and PoE dominates the mid-game calculations.

if you have two mechanical clears on D3 and other roles still un-outed (not an uncommon scenario), it essentially forces scum's NK's into a specific sequence for the rest of the game to prevent PoE. if the town gets lucky that the mech-clears fall on otherwise probable mis-elim targets, scum have to go through the stronger/more obvtown looking players to win. it can feel very claustrophobic as scum seeing endgame three phases away, and knowing there are only two or three realistic mis-elims available in the PoE.

but in order to get a 50% winrate as town, you have to have some percentage of no-win games for scum. there's just no way around it. at least in slightly scumsided setups, there is a better balance of agency for scum (ie; more options in NK's, more possibilities to bus and be creative) without taking away too much for town.

interestingly, when we first started trying to balance games at 50%, yosarian was very vocal in maintaining a slight scum bias stay built into setups, and i would argue extensively with him about it. it's taken a long time for me to come around to this view, but it seems yos saw the forest for the trees very early on.

[Isis]

I agree with Yosarian2

[Hoopla]

an interesting unrelated thought:

we determine balance by observing the results of a broad collection of similar games. if we see town win 50% of games over 100 games (for example), we determine we've been balancing individual games well.

but it's still possible that every individual game isn't balanced, and that half of these games have say, a 60/40 scumlean, and half have a 60/40 townlean. when played out it equates to 50% town wins, despite two sets of imbalanced setups. this would give us the veneer of observed balance in the macro, while being unbalanced in the micro.

an example of this was when i was analysing mini normals winrates by specific roles included in the setups. games that featured a mafia roleblocker had a much lower town winrate than games with other scum roles (which had a higher than average winrate). but if we just looked at all the mini normals, without filtering for specific roles, we saw a different win percentage.