Undervote and Pattern Voting: Vulnerability and a mitigation technique

Highly secure and auditable voting systems have been developed recently that open up the possibility of not only voting systems that claim to be secure, but ones that can actually be verified as secure and audited to the satisfaction of third party observers. Yet, these systems are still in development and do not yet handle all of the cases that arise in real-world elections. This extended abstract presents one specific vulnerability or limitation that arose in several of these voting system: Undervote and pattern vote tracking; This issue arose not so much because of flaws in the systems themselves, but rather, they are cases in which the strong protections provided by these systems did not take into account behaviours such as not voting on every contest. We present these issues in the context of PunchScan, a state of the art voting system. We present a technique for solving the issue using contest partitioning and analyze their impact on performance.