TRUST Security Seminar: Electronic Voting Security in India
Seminar | November 4 | 1-2 p.m. | Soda Hall, Wozniak Lounge
Alex Halderman, University of Michigan
Team for Research in Ubiquitous Security Technologies
India uses paperless electronic voting machines (EVMs) nationwide, and the Election Commission of India, the country’s highest election authority, has long maintained that the machines are “perfect,” “infallible,” and “fully tamper-proof.” Despite numerous reports of irregularities and suspicions of electronic fraud, the authorities have never permitted a serious, independent security review.
Earlier this year, an anonymous source approached Hari Prasad (a Hyderabad-based engineer who has been among the most active e-voting researchers in India), and offered him an EVM to study. Rop Gonggrijp, my students, and I joined with Hari to conduct a detailed analysis. We discovered that, far from being tamper-proof, the machines suffer from a variety of weaknesses that could potentially be exploited to change the outcome of national elections. India’s EVMs are considerably different from the electronic voting machines used in the U.S. and Europe, which have been the focus of almost all prior research, and the vulnerabilities we found have some surprising implications for the science of voting system design. We announced these findings in May, and although the Election Commission continues to publicly deny any security problems, the question inspired a hot debate in India’s media and political circles. Then, in late August, the story took a disturbing turn, when my coauthor Hari Prasad was dramatically arrested by Indian authorities.
In this talk, I’ll describe the design and motivations behind India’s electronic voting system, the technical problems our study demonstrated, the political circumstances behind Hari’s arrest, and the implications of these weaknesses for voting technology in India and beyond. I’ll also discuss some of the formidable practical challenges that India and many other democracies face in conducting elections. Designing voting systems that provide transparency and security under these constraints presents many open problems.
jesnjosh@eecs.berkeley.edu, 510-643-5105