Design Principles for Mafia Games Aside From Balance, Etc?

[northsidegal]

Great thread!

I personally think that the two "highest level" concepts that setups should strive to get right are balance and fun. Some might argue that fun is the only one that really matters, but I think that in most cases a setup being significantly unbalanced will tend to lead towards it being unfun. (In this sense you could also say that they're actually the same thing, but whatever, it's useful for the sake of discussion I think to keep them separate)

One aspect that falls under the balance side that I think doesn't get thought of or discussed enough is "skew", a property that I think CFJ first talked about. Skew is the tendency for one faction to be able to make massive gains if they play well or get lucky, while the other faction does not have that ability. If a game is skewed equally, then good play or lucky actions from either town or mafia should benefit their team about the same. I think that most games of mafia tend to by default be skewed in town's favor, and especially normal games. I also think that this natural town skew is what's responsible for a lot of anger over "townsidedness", again especially in normal games. Here's some of my justification on that from a thread where a lot of people thought that normal games were too townsided:

The very peak of scum mechanical play mostly looks like killing the best people each night, probably very good PR hunting and

maybe

good roleblocking (if scum have a roleblocker), whereas the very peak of town mechanical play (where we imagine every town PR makes the best decisions possible) would almost certainly end with town victories in most setups. I say this because Normal games tend not to be balanced around PRs making perfect checks every night, because that's not how most games go.

...

1. Normal games aren't balanced around TPRs making perfect checks each night, because this doesn't tend to happen.
   
2. If TPRs took perfect (or at least very good) actions each night, they would probably win a lot of setups.
   
3. The same ability does not exist for scum, because their PRs tend to be based around disrupting town PRs. Perfect nightkills and roleblocks would stem the tide of perfect TPR actions, but would still likely give town the advantage in most games.
   
4. Thus, we might say that these games are skewed, because the town can make more gains through great actions than the scum can through similarly great actions.

(Most people have probably heard a lot more about "swing" as a balance property, and might consider skew the same thing. I think that there are a lot of cases where they match up, but there are still some meaningful distinctions. For example, I would consider the Newbie A1 RB vs Cop + Doc to be a swingy setup, but not a very skewed one. Again, probably too tangential at this point)

Despite the fact that the data didn't actually bear out normal games being townsided (winrates were almost exactly 50%), the perception existed in a lot of people and seemed to be decently strong, indicating that there was at least something there. My diagnosis was and is that it was a skew problem – if you have a scumteam which is playing excellently during the day but which gets decimated by just a few night actions (as I've seen happen), it can feel both like the setup was way too townsided

and

that there was nothing else that you could have done, which is a really bad feeling.

I think it's difficult to combat skew problems because I think scum will almost always be inherently really skewed against due to their smaller numbers and more precarious position, but I think that it's worth keeping in mind ways that a setup can reward scum for good play and trying to minimize the possibility that a scumteam playing well will still get destroyed due to a few lucky actions.