It's been great reading thoughts on the game and I'm glad in general people seemed to enjoy it! There are some comments I'd like to reply to but first: I promised a restrospective and I wrote too much - this post is entirely just about the game setup and balance, not much specifically about how the game played out! I will write another post about that later
Design ObjectivesOverall I had two main objectives for the setup:
1) Make the debate over rule variants part of the game
2) Create a game that is balanced
I think there was an inherent contradiction in trying to achieve both, but I gave it my best shot!
After most other AYA? games I've played on here the discussion of game balance and rule variations gets brought up (often I admit by myself). There are always proponents on either side, often holding strong opinions (again, myself included!) which I thought would make a good basis for a meta element to a game that could incorporate that debate.
The idea of voting on rule changes as part of the game is one I have had in my head for a long time and has roots in the game Nomic, a game entirely about changing its own rules that I have run twice now on GRcade. More recently in the forum games thread discussion came around to a similar idea (I remember Dan specifically suggesting a combination of Nomic and Mafia) and that encouraged me to flesh my idea out into a fully formed setup.
The players voting on rules each day alongside the game playing out was the core of the design and the parliament themeing followed naturally from that. Defining those laws was then one of the most important aspects, and I especially wanted to include what I saw as the "Big 4" variations that get debated - an open vs closed setup, allowing outside communication, a vote to lynch vs a vote for mayor, and an inactive cull. My secondary concern was then in trying to balance the laws to not be entirely one-sided in their benefits and to make debate over which to prioritise and implement interesting.
It made sense that the initial setup of the game should be at its most basic so that the laws felt like additions rather than reductions to the game. It would have been odd for example to include an inactive cull in the initial setup and give the option of removing it as a law, or to allow who is online with the option to block it. When it came to knowing the roles in play it was necessary that the game start closed with the law there to make the setup open. Even though the laws followed that trend then in "opening up" the game, I was aware of making sure there could be both benefits and downsides to the town and the mafia for each of them.
The two most impactful laws by far in the way the game played out were the
State Secrets Act to make the setup open and the
Private Communications Act to allow all player to communicate in private.
With open vs closed setups (that is, whether the number and nature of roles in play are publicly known), that has a huge impact on game balance and I think creating a game where that changes was the biggest contradiction in trying to achieve balance. Closed setups give space for more roles because there is much more uncertainty in role claims, and having extra roles in a close setup can be very useful to give the town better odds given it is much harder to deduce what is going on and the state of the game. Closed setups allow for more speculation but I also find them much more likely to lead to frustration, especially in the late game, if the town can never feel like they have a grip on what is happening.
What I think went well in the game with the transition from a closed setup to an open setup on day 3 was that the initial closed setup allowed for more speculation in the first couple of days which really helped the game get going and be interesting from the off, but then there was no (or at least very little) frustration in the mid and end game down to the rules of the game being too unclear.
In retrospect it probably would have been more balanced to have only made the game semi-open - that is, a set of roles could be made known but exactly which are in play and how many would still be unknown for people to speculate on. This would have given the mafia more space to hide also.
That squeezing of places to hide was in large part due to the high percentage of roles. As a general rule the hard limit on balance is that there should be more standard town players than those with named roles. That way if a mass role claim is made, night kills can work through the special roles quicker than the lynch can work through the regulars which must contain the mafia. The near fifty-fifty split in MPs to roles was probably fine for a closed setup as the game started but too much for an open setup which the game became.
In choosing the roles I did I considered that role claiming is not a popular strategy on GRcade (in fact I have never seen a mass role claim) and so I was happier to run the risk of including more roles. The way I saw it it was a tradeoff between potential but unlikely imbalance and having more engaged players due to more of them having roles.
That gamble on an "unlikely imbalance" partly held true in that there never was a mass role claim! I think the game as it played out could have been as good as over several days earlier than it did end if there had been one. Even in private it seems not all roles came forwards to the essentially known-town seer circle. Where the game tripped up on the unlikely imbalance was in the first seven kills all being MPs! While the game started with MPs outnumbering special roles that early run of only MPs dying had a significant impact and was one of the major factors in the Conformists losing.
One of the points of balance I paid particular attention to was in maing sure an invincible protection circle couldn't form, no matter the combination of laws in play and without changing the role of a mayor to be out of the ordinary. That's why there was only one doctor who couldn't protect himself and why the Fanatics were included with their one-shot strongmen kills. I think this was succesful - while a seer circle did dominate the game, the Conformists had plenty of opportunity to disrupt it, but chose another strategy in hunting for the seer.
I also thought about "swing" a lot. To make a game feel balanced I think the impact of random chance should be lessened on the game and positively reinforcing actions avoided. Having deputy roles is one of the easiest ways of achieving this - it doesn't overpower the town like giving them an extra concurrent seer or doctor would, but it means a seer and doctor are much more likely to remain in play. The two strongmen kills were similarly divided between two Fanatics rather than given to one so that the power couldn't be lost with one unlucky death.
It's usual on GRcade for the mafia to start the game with more than one night kill, reduced only after losing a certain number of players, but this is inherently swingy - if the town by luck can reduce that kill rate earlier in the game, it has a postively reinforcing effect on the town doing well, whereas if the mafia start out well and keep those kills it accelerates them towards a win.
On the subject of the mafia, I was surprised when everyone thought seven to be a lot. I think this is roughly in line with the recommended ratio, and perhaps the perception of it being a lot was because of the mafia having multiple kills being more normal on GRcade. The general rule I used to decide the ratio of mafia to town was to start the game with a balance between codemns to win and miscondemns to win. To explain, a condemn is a lynching of a mafia and a miscondemn a lynching of a townie. With 7 mafia out of 32 players, the game was roughly balanced so that the first possible town win was in as many days as the first possible mafia win - in fact it was slightly favoured to the town, but with the serial killer having 3 kills that were more likely on odds alone to balance that out.
A higher ratio of mafia I think is just more fun anyway and good for the game as it makes finding them by pure chance more likely - which is often what is needed to get a game going, as it's only once one mafia is found that post histories etc. can be more meaningfully examined. I think seven was a good number!
That covers open vs closed setups and the choice of roles, but it was the combination of an open setup and private communication for the town that really swung the game in their favour. I've made it clear in the past that I myself am very much against the town having private communication for several reasons, and while I didn't set the game up to try and prove a point I think this game bore that out. In thread activity dropped noticeably on the introduction of PMs, they enabled the seer circle that carried the game, and all things considered I think it made the game more boring to spectate and probably for those not directly involved with the seer circle or mafia more boring to play. If this game and the contrast between those early days without PMs and those after helps win others over to the idea of games without PMs, I'm going to consider that a success
I've written an absolute essay when "design principles" was just meant to be the first of several headings in my game retrospective, so I'm going to stop there for the moment. Absolutely though I will write another follow up to this with more of a breakdown of how the game actually played out and moments I want to comment on!