A Network Game Theory
Play as one of two competing agents: the attacker moving between nodes in a graph to collect points, and the defender trying to guess where the attacker will move next to catch them. Compete against the calculated, game-theoretic, optimal strategy and see if you can achieve your agent's optimal score.
Comments
Log in with itch.io to leave a comment.
I’ve studied some AI and decision trees so I like the concept. Although computing expected value from probability weight and reward is not very fun for a human, as we tend to like being intuitive and getting the feel for it rather than doing the math. An exercise to force ourselves to get a rational mind?
I noticed that the random setup can give odd rule combinations. In particular, when there is no “if the attacker is caught once, the game ends”, there is no way to end the game and check the final score (and I tend to just get worse and worse). I also got a setup where Reward = -(If caught) everywhere, so everything was about reading probability. On this game I could catch the attacker, so it ended pretty quickly (generally speaking I just end ignoring the score and focusing on probability to end things faster).
This game has a lot of potential, but it needs better instructions. They're comprehensive from a programmatic point of view, but confusing to read.
I really enjoyed this game, though. Good job.