BrianMcQueso wrote:LML's role is sketchy (brilliant insight, eh?) While a tracker with a one-use self-target is realistic as a role, I'd say it's also somewhat unlikely. Enabling pre-list discussion combined with the ability to target oneself a single time conveniently leads to the exact plan LML has proposed. If genuine, it could be potentially game-breaking. I'm skeptical that a moderator would include a specific role function that would essentially end the game before the first lynch.
Target lists should not be random. Day 0 is a huge boon to creating your targeting list. A cop can start with the scummiest person and work their way down towards the townliest. A doctor should pick a few of the townliest players to protect, and have some repeat protections (if the doctor does not, and the Mafia missed a kill, then they could reasonably assume that their target is less likely to receive protection the next night. Of course, now that I've brought this to everyone's attention, I've opened a big bottle of WIFOM, and the doctor should protect however the heck they damn well feel they should). Without the benefit of Day 0, then everyone would be forced into random lists, but we have information to guide our choices now, no matter how limited it might be.
We were given two entire weeks for discussion, so let's make good use of that time and get a lot of information that will help our power roles. Day 0 should not be considered a kick-back day that doesn't really matter because there is no lynch. Day 0 could be the most significant day of the game.
I can think of a good reason why a tracker role would be included, but not so much for a watcher.
A formulaic approach to target lists (e.g. BMQ's suggestions above) will work better for some roles than others.
For roles that interact with the scum directly or indirectly (e.g. doctor, roleblocker) the individual with the role really needs to make the decision for themselves regarding what method to apply. Simply protecting the towniest players first allows the scum to get free kills on the scummy-looking players early in the game if they choose to take that path with their list. The counter-scum are playing double-bluff as much as anything. I'd probably advocate a weighted random approach, where certain players (e.g the most townie for the doctor) are preferred, but there is still an element of unpredictability in the selection process.
A cop, on the other hand, can and should investigate players in order from most to least scummy. Presumably scum can only stop the cop's investigation by finding the cop themselves, not by finding their targets. Hence, no need to take probable scum actions into account.