That's an example of why relying on flagging for decisions like moderating/banning users does not work. I'm giving you a scenario where it's possible for multiple users to be using 1 ip address or it could be one user faking multiple users from a single IP. So we cannot really depend on a self sustaining system as you have suggested.