[setup] The Turing Test

[Alianna]

I don't know math, but I did some math anyway. If my math is wrong, please tell me.
The problem I see is that town have pretty good odds of winning if they just put someone into the test randomly each day and then vote the mafia's pick. With each day, the town's chances increase. I have zero clue how to calculate an expected winrate, but here's the chance of town getting the point each day if they follow this strategy.

With your numbers, D1 is 10v5, meaning 66.67% chance of town getting the point.
D2 is 9v4, meaning 69.23% chance.
D3 is 8v3, meaning 72.73% chance.
If it gets to D4, it's 7v2, meaning 77.78% chance.
If it somehow gets to D5 (it won't), it's 6v1, meaning town has an 85.71% chance to get that last point and secure the win.

In my opinion, town should not be able to vote randomly and still have high and ever-increasing chances of winning.
This strategy becomes less viable if you lower the ratio of town to scum, but that ratio would have to be 1:1 if you wanted even equal chances for both sides. Which is not informed minority vs. uninformed majority, therefore not mafia.
Of course, the players could also just play the game as intended and forget about math, but there's a pretty good incentive not to do that.

[Alianna]

well, true randomness isn't allowed, but voting in the towniest player each day would make it so town has to have 3 townreads correct out of 5? which sounds... likely to happen, i guess?

hmm. not sure if there's an easy fix for it.

Somehow I forgot that rule existed. The point still generally holds though.

[Alianna]

Idk how you fix that.