I suggested to my GM a campaign idea where one of the player characters would secretly be the final enemy boss and the player would play and campaign alongside the group trying to further his goals using the party as means to his ends. Eventually the party would find out and finish him off (or be killed off by the villain after a while).
My GM thought it was a very interesting idea. We plan to keep the premise secret to avoid metagaming, but have discussed this in a general sense and are confident players won't be put off by this kind of campaign concept.
We have never played such a campaign before though, and the primary problem seems to be not having the conflict come to a head too early, for the villain PC to simply kill everyone in their sleep, etc. We want the campaign to go a while before the conflict comes, and also want it to end with - even if not a straight up tactical fight, at least something that will be a interesting conflict.
What kinds of things should we take into account for both forcing the timing of and balancing the inevitable final battle?
Things like forcing a moral compass to prevent underhanded bad guy wins by killing everyone in their sleep. Would a level buff to the villain to the equivalent of matching the CR the party would face be useful? At what level would be a good point to tell the bad guy he can should try dropping hints, if the party figures it out early and moves to take him out then he’d be at a disadvantage.
If you've run a campaign like this before, how did you arrange it to solve this problem?