Still, so long as it's up to judges to pick between 8 different teams, there is at least somewhat of an advantage to getting more scum games than town games. It's just easier to stick out as a good player that way. For one thing, scum have a higher survival rate, even with a vigilante in the mix. I tried to take into consideration that going out early COULD be the best thing for a team (i.e. VT dying N1), but that's hard to control for. For another thing, if I'm evaluating (as I did to a degree) by choosing the best player from each team, just by random chance a scum player has a better shot at being MVP for his side, and thus getting some kudos from the judges.
So even though points wise, you're absolutely right -- getting scum is equal risk and reward -- at least under the way we did it this time, it rewards teams with more scum roles even if the individual games are perfectly balanced.
Let me be clear in case I haven't been recently: random choice is the only way I see to assign roles unless the game was designed wildly differently. That's perfectly fine with me, and it's in keeping with mafiascum as a whole. But it does describe a reason (one of several) that I'm in favor of trying to get away from overall subjective judgments to decide the winner by instituting a point system that takes the end decision away from me.
That's not to say a future point system couldn't involve a measure of subjectivity, but it should be subjectivity that's narrow in focus. For example, +2 points for being the driving force on a successful lynch on a mafia member as town* involves subjectivity, but it's far easier for a judge to determine.
*Please note: this is not an actual suggestion for how points will be structured; it's only an example of the way it could use subjectivity in a narrower way.
.