Scaling Privacy Guarantees in Code-Verification Elections
Aggelos Kiayias and Anthi Orfanou
E-Voting and Identify, 4th International Conference (Vote-ID 2013)
Springer 2013 Lecture Notes in Computer Science, pp. 1-24
July 17-19, 2013, Guildford, UK


Preventing the corruption of the voting platform is a major issue for any e-voting scheme. To address this, a number of recent protocols enable voters to validate the operation of their platform by utilizing a platform independent feedback: the voting system reaches out to the voter to convince her that the vote was cast as intended. This poses two major problems: first, the system should not learn the actual vote; second, the voter should be able to validate the system’s response without performing a mathematically complex protocol (we call this property “human verifiability”). Current solutions with convincing privacy guarantees suffer from trust scalability problems: either a small coalition of servers can entirely break privacy or the platform has a secret key which prevents the privacy from being breached. In this work we demonstrate how it is possible to provide better trust distribution without platform side secrets by increasing the number of feedback messages back to the voter. The main challenge of our approach is to maintain human verifiability: to solve this we provide new techniques that are based on either simple mathematical calculations or a novel visual cryptography technique that we call visual sharing of shape descriptions, which may be of independent interest.