This is not a guide to playing the game of Mafia per se, and you may find it useful for a wide variety of forms of competition. There isn’t going to be anything in here about how to detect scum as town, nor how to manipulate town as scum. This is also unlikely to interest you if you do not put a great deal of weight on improving as a player in mafia. There are plenty of guides out there on how to do those things, and there are many people who consider themselves extremely competent at one or the other who would be happy to help you out. Most of these people will give you well intentioned advice with few to none setting out to manipulate you but at the end of the day you are going to have a pile of advice that all contradicts itself. Even ‘good’ players advice will contradict each other. But that statement begs the question from which I want to start this article, what exactly does it mean to be a ‘good’ player?
If this were for example Chess, the answer would be considered obvious. Each match has a winner and a loser and the winner gains the elo that the loser loses with rates differing based on how much they had previously won and loss before. But in the end it all comes down to wins and losses. League of legends works the same way. Winner gains elo and ranks, loser loses elo and ranks. Sports tournaments are handled slightly differently, where the team that wins the most* (oversimplification but not important) time versus the other teams is considered the winner of that tournament. Winners are assumed to be better than losers. This standard is considered universally applicable except on Mafiascum historically arguments based on winrate are more likely to elicit groans than nods. In some of the older scummies judging threads it was made clear that arguments to winrate were not wanted in terms of determining who warranted body of work award for scum & town play.
The human mind is extremely good at self deception and that is going to be a major theme in this guide. We mentally fudge the game in our mind where we had accurate reads on scum despite scum winning, while not assigning ourselves similar penalties in the game where we defended scum to the end. In our mind that salient game where we weren’t listened to by the town after we correctly caught scum and were nightkilled stands testament that we deserved that win, while we see nothing wrong with the fact that townies lost in the game where we removed the obstacles to our win condition from the game. That level of subjectivity lets us mentally put our proficiency at the game wherever we wish in a fuzzy range so that we don’t have to ask ourselves difficult questions. Social deduction unlike Chess gives us lots of outs to make excuses for ourselves, and also makes it a lot harder to see where we are going wrong or right.
I propose instead that, given a large enough sample size, winrate is in fact an accurate heuristic for judging performance against other members of a community, a statement that would be exceedingly noncontroversial were it not made about mafia. On one hand each individual does have only a limited amount of control over the games. But the same is true of league of legends. To be clear, I have not played League in a very long time but I think it is a solid comparison that is relevant to the audience of this guide. Nightkilling does remove a townies’ voice for the duration of the game, but then again so does getting chain ganked in League of Legends. That elo and rank are not a reasonably accurate metric of performance with a sufficient sample size is not a position taken seriously in league outside of by a rather sizable contingent of low elo players who complain of being stuck in “ELO hell” and complain for various reasons that their winrate isn’t tracking their performance. It is generally accepted by league of legends players that these people are delusional and speaking as someone who was near the top of Masters back when I played I was easily capable of troll picking mid lanes like AP Nasus in Gold and having a near 100% winrate. Based on the trajectory of my own town winrate steadily, linearly increasing it should be abundantly clear that 2019 Paragon me placed into my 2015 games would win a much larger percentage of them. Northsidegal placed into the position of a 45% town winrate player would not have a 45% winrate in all the same games that player played, and no one would dispute that.
When we think from this point of view, our original question starts having a clear answer. A good player is one that makes actions that lead towards their team winning more of the time. A bad player is a player that is less likely to make those actions and makes actions that lead to their losing more of the time. Any individual game’s win or loss comes down to a myriad of factors but better players make better decisions over a larger sample size of games and as a result win more of those games. The correct action to take in any individual game, then, becomes the one that is most likely to lead your team to victory. Whether you had scum correctly pinned but couldn’t eliminate them is irrelevant. Another name for the ‘best player in a losing team’ is loser. I don’t say this to discredit people who have given valiant defeats while their team thwarted them. Better players are more likely to be these people who are doing their best and failing to save their team. But in the end the objective, the way to play ‘good’ in a game is to win it and the way to determine which actions are better and worse is by how they interact with your odds of winning.
This does not preclude taking other metrics of performance into consideration. For example, if two players with the exact same winrate as town exist and one is horrible at explaining their reads and tends to get miselimed while the other is charming and charismatic, the first one is likely to be the superior scumhunter by virtue of their lack of other skills. But if two players have equal scum winrate and one of them tends to get elimed midway through the games and the other tends to endgame, that is not in any way an assertion that the survivor is a better player. One game that has always stuck with me in terms of how I mentally model scum play is this one. MariaR spent the entire early game not distancing excessively but establishing interaction patterns that felt normal with her scumteam. Her scumteam played generally well but a few people, most notably Titus, had caught the entire scumteam together as a bloc barring her with her seen as not!paired with the team. In response to this she reined in her play, playing less accurately and more scummy, and in doing so allowed her teammates to have her limmed giving her team, effectively, no real way to lose the game. Looking at individual play it’s not immediately obvious that she gave the strongest performance as opposed to Drealmerz or Vecna who endgamed, but when you consider her scum winrate over a long period of time it is a lot more obvious that across games she is consistently doing a lot to give her teams higher odds of winning If you want to improve at the game, it follows that you should also focus less on flashy things like solo endgames and heroic catches and just do what puts your team closer to winning. A simple, yet incredibly complicated concept.