On the notion of ‘software independence’ in voting systems

Abstract

This paper defines and explores the notion of ‘software independence’ in voting systems: ‘A voting system is software independent if an (undetected) change or error in its software cannot cause an undetectable change or error in an election outcome’. For example, optical scan and some cryptographically based voting systems are software independent. Variations and implications of this definition are explored. It is proposed that software-independent voting systems should be preferred, and software-dependent voting systems should be avoided. An initial version of this paper was prepared for use by the Technical Guidelines Development Committee in their development of the Voluntary Voting System Guidelines, which will specify the requirements that the USA voting systems must meet to receive certification.