Here's a quick summary of the players' roles (more detailed information is below):
1. BTD6_Maker,
Town One-and-a-Half Voter
2. BulletNLynchproof,
Mafia Triplevoter
3. drealmerz7,
Town 1-Shot Informed Half Voter
4. Secret Agent Jin,
Town Doublet Voter
5. The MM,
Mafia Doublet Voter
6. Realeo,
Town N1 Modified Sensor
7. Eggman,
Town Voteless D2 Census-Taker
8. Gamma Emerald,
Mafia Doublet Voter
9. Human Sequencer,
Town One-and-a-Half Voter
10. JasonWazza,
Town Doublevoter
11. Not_Mafia,
Town Half Voter
12. Kain Tepes,
Town Doublevoter
13. ɀefiend,
Town Doublevoter
The vast majority of players were vanilla apart from how their vote was counted.
A Voteless player has no vote, obviously.
A Half Voter has a vote weight of ½; a Voter has a vote weight of 1; a One-and-a-Half Voter has a vote weight of 1½; a Doublevoter has a vote weight of 2; a Triplevoter has a vote weight of 3. All these types of player have only a single vote at first choice, second choice, and third choice (the vote weight determines the extent to which the vote contributes towards a lynch).
A Doublet Voter has two first-choice votes, which count as half a vote each. Their second choice and third choice votes work the same way as with a regular voter.
There were three power roles:
- drealmerz7's ability gave information about how the lynch for a particular day (of drealmerz7's choice) was determined. This would have helped town considerably with understanding the game, if they hadn't quickhammered him first :-P Unfortunately, it seems that there was a mistake in the role PM that meant that drealmerz7 was unaware of when he could actually activate the role, and was lynched before the mod had a chance to correct it. (Additionally, the role PM didn't state what the ability actually did, only that it gave information.)
- Realeo could, during night 1, learn the total vote weight of all scum on the day 1 lynch wagon. In a pretty precise targeting choice, the scum managed to nightkill him before he had a chance to tell anyone this. (This was arguably fortunate; the role could have potentially broken the game when The MM flipped, depending on what probability the town assigned to the possibility of a voteless scum on the lynch wagon. Scum really needed more votes on that wagon in order to avoid being exposed via the Sensor ability, but I'm not sure they had enough information to tell that.)
- Eggman's ability determined the total vote weight of all players in the game, although he didn't know what it did. However, he apparently completely misread the PM with the results (which apparently said "15 and 1/2", although I haven't seen the PM in question), missing the "15 and" section, and interpreted it as meaning that there was half a scum vote on the drealmerz wagon. As it happens, there actually
was
half a scum vote on the drealmerz wagon, so the misread PM, purely by coincidence, gave town the information that they were denied when Realeo got nightkilled, and may well have ended up winning town the game by accident. The only conclusion I can draw from this is that more moderators should learn how to type Unicode; "15½" would be much harder to misread.
Now, onto the game mechanics. The key to the game, which is something that maybe should have been explained to town directly (it was intended to be implied via drealmerz7's role, which could have been activated D1), is that neither the town's lynch nor the scum's kill is as reliable as in a regular game. (Ircher's basic idea for the game was to base it heavily around voting systems other than the regular one, but that mostly just comes down to "maybe a lynch will hit the wrong player due to the leading wagon not having enough vote weight". The reduction in power of the scum kill was added to compensate, as town clearly don't have enough power to have a chance against a regular scumteam.) For the scum kill, this was very simple: it failed every other night, and the scum were aware of this. (Actually, nobody had any useful night actions N2, but because there were abilities that could have been activated then – on dead players, but still – the night still had to happen to avoid confirming the absence of further night-activated power roles.) For the town's lynch, the way things worked was to take a real-world voting system (many of which were designed to avoid problems with straightforward plurality) and apply it to the votes, to see what happened. Because a hammer required half the votes (regardless of vote weight), and all plausible voting systems give the victory to a player who captures half the vote weight, this mostly only matters if the leading wagon had less than half the vote weight. Additionally, there was also a White Flag mechanic in play. (According to the notes I was given to recover the game if I needed to take over, this was apparently a secret, but
also
was listed in everyone's role PMs. I'm assuming there's an inaccuracy there somewhere.) Scum's major power was the triplevoter, which had a high chance of being able to get a counterwagon lynched if the town didn't agree on a single wagon to push.
On day 1, the voting system was Condorcet, which is actually really complex. The basic idea is that player A is considered a better lynch choice than player B if more players voted player A ahead of player B than player B ahead of player A. (So a player's first choice was considered to be ahead of everyone else from that player's point of view; a player's second choice was ahead of everyone but their first choice, and their third choice was ahead of everyone but their first two choices.) If a player's considered a better lynch choice than everyone else, then they get lynched. (There's also some kind of tiebreak but I don't understand how it works.) I confirmed to my satisfaction that the right player
was
lynched, but it was an "existence proof", so to speak, doing only enough analysis to be sure that nobody else was a better lynch choice; I don't know who came second.
On day 2, the voting system was absolute majority, which is rather simpler: if a player gets more half the first choice vote weight on them, they get lynched; otherwise, nobody gets lynched. (Second and third choice votes there didn't count.) The MM was only half a vote away from surviving the lynch, so if the wagon composition were slightly different, there could have been something of a shock.
On day 3, we were using instant runoff voting; basically, if no player has enough first choice votes to be lynched, we work through the players in order from the fewest to the most first choice votes, and cancel votes on them. Cancelled first-choice votes promote the second-choice votes below them to first choice, and so on. This method was highly likely to lead to the lynch of the hammered player unless there was a strong counterwagon. (I'm not sure it was ever defined how this works with Doublet Voters.)
In general, this was a game that I thought was fairly risky to run (I backed it up only because i was requested to, rather than because I was particularly confident in the setup); it needs a particular sort of player to enjoy it in the first place (to the extent that I thought it might not fill), and on top of that it's somewhat swingy. I didn't expect a quickhammer followed by corner-case behaviour of the power roles, though, and that probably made it impossible that anyone could figure out how the game actually worked. All in all, it's probably for the best that it was abandoned.