Here I have a pseudo-algorimth for Murderer/Sheriff selection: Let players vote if they want to be a Murderer, Sheriff or Innocent(default) If 0 Players want to be Murderer/Sheriff, it will be picked randomly as it currently is. If 1 Player wants to be Murderer/Sheriff, (s)he will be picked. If 2 or more Players want to be Murderer/Sheriff, it will be picked from them accordingly to how often they've been Murderer/Sheriff. The more often the lower chance. Example: 5 Players want to be Murderer. And they've been murderer this amount of times (Player | Count): Player 1 | 5 times Player 2 | 3 times Player 3 | 3 times Player 4 | 4 times Player 5 | 2 times The sum of times they've all been murderer is 5+3+3+4+2 = 17 times. Player 1 has been Murderer most of times, hence, he has the lowest chance to be picked. Player 2 and 3 have the same chance to be picked. Player 4 has low chance to be picked Player 5 has the highest chance to be picked. totalChance = sum * (nPlayers - 1) = 17 * (5 - 1) = 17 * 4 = 68 chancePerPlayer = (sum - count) / totalChance so each player wil have a chance of: Player1 | (17 - 5) / 68 = 12 / 68 = 17.65% Player2 | (17 - 3) / 68 = 14 / 68 = 20.59% Player3 | (17 - 3) / 68 = 14 / 68 = 20.59% Player4 | (17 - 4) / 68 = 13 / 68 = 19.12% Player5 | (17 - 2) / 68 = 15 / 68 = 22.06% Also, for any player that has never been murderer, the formula gets reduced to: 1 / (nPlayers - 1) which in above's example would be 1 / 4 = 25% Condition: If totalChance = 0 (basically, all players who applied have never been Murderer/Sheriff) then they will be picked randomly from the list of users who applied. Curious things of this algorimth: 1) If all players except 1 have never been Murderer/Sheriff, the player who has been murderer is basically excluded. 2) This formula won't work if all players have never been Murderer/Sheriff because of Division by Zero. Hence, this condition has to be handled separately 3) Generally, the probabily is very fair with all players, so if there are 2 players and one has been murderer 10 times and the other only 2 times, then the chance is: Player1 | (12 - 10) / 12 = 2 / 12 = 16.67% Player2 | (12 - 2) / 10 = 10 / 12 = 83.33% Another example assuming a player with 12 times, another with 8 times and another with 0: Player1 | (20 - 12) / 40 = 8 / 40 = 20% Player2 | (20 - 8) / 40 = 12 / 40 = 30% Player3 | (20 - 0) / 40 = 20 / 40 = 50% As you can see, the player that was never a Murderer/Sheriff has 50% of chance, while the others still have a 20 and 30% chance respectively which is good and fair.

Great idea- I think it coupd be inspiring for any new games to come. However, MurderMystery is fun as it always is randomized. If it weren't, everyone would want to be Murderer; that would be too predictable and by then all the Innos will know who's Murd. Nice suggestion!