So, game was mostly balanced off of a program of mine. It runs a series of randomly generated games and spits out mafia wins vs town wins. It assumes mostly optimal play from both sides, and it does not include backups, mayor, mike/receiver pairs, the fake ID, or the difference between a conscription and an extra mafia at game start. Lucy and I decided that taking into account those differences, and the fact that kills this game would be really obnoxious to make, we wanted the game to be balanced so it was somewhat mafia biased.
Upon realizing that despite being revealed, Dgrazian and I would have still been able to power through and win game, I started to wonder what was up. I ran the game start conditions, and it spat out 53.5% mafia win rate (which accounting for the differences mentioned above, is actually close to being even). However, when I put in the day 2 start conditions (medkit with a mafia, two known innocent children, one known gay knight pair, two fewer investigators), it gives a 89% mafia win rate. If I add the conditions of Jugalde and Victor being inactive (1 fewer desperado, 1 fewer conspiracy theorist), it goes up to 96%. Basically, game is very swingy if, even in a game this small, a single day makes that much of a difference.
So, in light of this, do people have suggestions for rules changes? My suggestions:
1. Compensate for non-optimal play. In real mafia games, there are both new and inactive/mostly inactive players. These disproportionally hit town, because a new mafia will be advised by more experienced mafia, and mafia will poke and prod all but the most resistant inactive mafia into activity. I don't know if there is a better way to do this than to just lower what is considered to be a balanced mafia win rate.
2. Compensate for aspects of gameplay that can't be simulated. The biggest difference here is between theoretically making kills and actually making kills. I tried to initially compensate for this by raising the balanced mafia win rate and by making sure that there were plenty of powers like poison and hire a hitman. However, mafia will additionally compensate for this themselves, for instance, by staking out locations for hours. And I really don't think that should be as much of an optimal strategy as it is, because it makes planning and executing skills take a lot out of the mafia players' real lives. So, I advocate for more strict enforcement of making informative introductions, especially in summer games.
3. Lower the overall power level (and thus lower the swinginess) of game. Especially in small games. Part a. I advocate for fewer total mafia powers, but the same number of options for mafia powers. In a game like this, I'd say something more like 1 use of either frame an innocent or plant evidence, 2 recharging kabooms, 1 recharging set a trap, one use of either scheme or another set a trap, 1 slaughter the weak, 1 use of either poison or hire a hitman. Part b. I advocate for less town power. Particularly with prophet. Simulation-wise, prophet is BY FAR the most powerful town role. I'd be interested in perhaps changing prophet to be more like the old role, seer, with a once-per-game ability to find out someone's role. It would make prophet claims much less clear cut, and it would prevent the incredibly swingy (and broken, in my opinion) "get all the backups to backup prophets" game. I wouldn't want to go back to having vanilla townies, but I think it would also be good to have games with a disproportionate number of conspiracy theorists. I like conspiracy theorists because unlike vanillas, they have something to strategize about, but unlike many town roles, they are low power and low swinginess. They also encourage the mafia to be sneaky and subtle, which is much more in the spirit of the game, in my opinion, than what game would have been had it played out.
4. Misc ideas. Not a solution to any particular problem, but something I think would be cool: some sort of power for manipulating the time of death. As long as people were good about not letting on that they died unless they were directly asked about game (like Kemus in 2014 Spring 1!), I think it would be a neat addition.
|