Cool! If you see any way to improve the scoring process for paranoid debating, it would be neat to fold the improvements into code so other people have an easier time getting started.

At a meetup on southern california we speculated that the game might be useful as a test bed for trying out discussion strategies and directly measuring what worked best, but I expect that it will take some work to figure out a process for playing and scoring in a way that can detect significant differences in playing strategies.