A Center for Correct, Usable, Reliable, Auditable and Transparent Elections (ACCURATE) submitted public comment today to the U.S. Election Assistance Commission on their draft Voluntary Voting System Guidelines, version 1.1 (VVSG v1.1). The VVSG provides national certification requirements and testing protocols for voting systems against which many states require their voting systems to be certified.
ACCURATE’s comments criticize the new draft v1.1 for claiming to include only incremental, modest changes when, in fact, many of the requirements would necessitate substantial revisions and re-engineering of voting systems. We argue that the VVSG II, the previous more-ambitious overhaul of the VVSG, is a better backdrop upon which to require substantial revisions of voting systems than the draft VVSG v1.1.
Most notably omitted is any requirement for software independence, which would require systems to be designed so that undetected flaws in the voting system software could not cause undetectable changes in the vote count. In fact, many of the more onerous, detailed requirements in the VVSG v1.1 bear the full burden, now, of security assurance, when requiring software independent system architectures would bypass the need to rely on, for example, detailed coding requirements. ACCURATE fully supports requiring software independence as the backbone of a robust and comprehensive next-generation voting system certification regime, and we were disappointed to see no evidence of it in the VVSG v1.1. The commentary goes on to emphasize that most of the changes relevant for cryptography, structured vote data and security are not nearly as powerful as they would be in a regime that required software independence.
ACCURATE is more optimistic about the benefits from the testing-related changes, such as those proposed changes to the accuracy and reliability testing framework, the usability and accessibility requirements (and the requirement that manufacturer’s must conduct usability testing) and the required voting system documentation. However, there is a conspicuous absence of volume testing and adversarial vulnerability testing, two types of testing that have shown great promise in state-based testing efforts. Finally, the novel usability benchmarks and benchmark testing are also missing, despite evidence that the general usability and pollworker documentation benchmarks are close to finished if not actually finished.
ACCURATE welcomes the opportunity to participate further as the draft VVSG is modified and readied for adoption.