Last week, at USENIX EVT/WOTE’14, in the beautiful city of San Diego, I presented a paper that was jointly co-authored with my former colleague at Thales (Mr Matthew Kreeger) and colleagues at Newcastle University (Prof Brian Randell, Dr Dylan Clarke, Dr Siamak Shahandashti, Peter Hyun-Jeen Lee). The title of our joint paper is “Every Vote Counts: Ensuring Integrity in Large-Scale Electronic Voting” (presentation slides here).
In this paper, we first highlight a significant gap in the e-voting research field that many people seem to have ignored: while the End-to-End (E2E) e-voting systems have been extensively researched for over twenty years and have been commonly heralded as a rescuer to many controversies in e-voting, in practice few of those systems have actually been implemented and almost none of them used in real-world national elections.
We are motivated to find out the root cause and to narrow the gap. Our hypothesis is that the existing E2E systems’ universal dependence on a set of tallying authorities (who are assumed to be from parties of conflicting interests, be expert in cryptographic key management and be expert in computing) presents a significant hurdle towards the practical deployment of those systems.
We then show that the involvement of tallying authorities is not strictly necessary at least in some election scenarios. In particular, we focus on DRE-based (Direct Recording Electronic) elections conducted at supervised polling stations. This is perhaps the most common election scenario in national elections around the world, e.g., USA, India and Brazil. We present a new cryptographic voting protocol called Direct Recording Electronic with Integrity (DRE-i). The DRE-i protocol provides the same E2E verifiability as other E2E voting protocols, but without involving any tallying authorities. Hence, the system is “self-enforcing”. By comparing with related E2E protocols that are dependent on tallying authorities, we demonstrate that a self-enforcing e-voting system is significantly simpler, earlier to implement, more efficient and has better usability – all of this is achieved without degrading security.
We welcome interested readers to scrutinize our paper, point out any error or discrepancy that you can find, and feel free to write your feedback in the “Comments” below.
Do you have an example of a receipt the voter would get after casting a vote? I assume it contains a cryptographic hash that she can match with the public bulletin board when walking out of the polling station, but how does she know that this hash represents the right candidate without a way to prove this to others?
The receipt could be a truncated hash with a digital signature. It doesn’t show anything about whom she had voted for (otherwise, she would be able to prove to a coercer). But she knows (probabilistically) it’s for the right candidate through user-initiated auditing (i.e., pressing the cancel button). The third paragraph in Section 3.3 in the paper gives more explanation on this. Hope this clarifies. If not, let me know.