Towards improving the efficacy of code-based verification in internet voting

Abstract

End-to-end verifiable Internet voting enables a high level of election integrity. Cast-as-intended verification, in particular, allows voters to verify that their vote has been correctly cast, even in the presence of malicious voting devices. One cast-as-intended verification approach is code-based verification, used since 2015 in legally-binding Swiss elections. We evaluated the Swiss paper-based polling sheet and voting interface, focusing on how well it supported voters in verifying their votes. We uncovered several potential issues related to manipulation detection. We improved the paper-based polling sheet and voting interface accordingly. Then, we carried out a between-subjects lab study with 128 participants to compare the original and improved sheet and interface wrt. usability and its effectiveness in supporting manipulation detection. Our improvements significantly enhanced detection. Our study delivered insights into participants’ somewhat ineffectual reactions to detected anomalies, i.e. starting over again and trying to cast the same vote again, or calling the telephone number provided by the interface. This problem is likely to manifest in any verifiable voting system and thus needs to be addressed as future work.