I’m attending the…
Panel on Indian Electronic Voting Machines (EVMs)
Moderator: Joseph Lorenzo Hall, University of California, Berkeley and Princeton University
Panelists: P.V. Indiresan, Former Director, IIT-Madras; G.V.L Narasimha Rao, Citizens for Verifiability, Transparency, and Accountability in Elections, VeTA; Alok Shukla, Election Commission of India; J. Alex Halderman, University of Michigan
The first speaker was G.V.L. Narasimha Rao, who is also a blogger on the topic of elections. He is a staunch opponent of Electonic Voting Machines (EVMs). He gave a summary of voting in India — until 1996, all voting was with paper ballots and hand counting. In 1998 there were some EVMs introduced in urban areas, and then in 2004 it moved entirely to EVMs. Vote confirmation was given by a beep, and there were several complaints of machine failure. His claim is that exit polling was accurate prior to 2004 and then after the introduction of EVMs, the exit polls diverged widely from the actual results. In these elections I believe the BJP got a drubbing from Congress (Rao probably got suspicious since he appears to be a BJP political analyst).
Next up was Alok Shukla, the Deputy Election Commissioner of India. He gave an overview of the EVMs in use in India. He gave a review of how India decided to move to EVMs (the Parliament ended up approving the use of EVMs). He claimed that a paper trail was not the solution (mostly due to infeasibility/cost/remoteness of polling locations, etc), and said solutions lie in better transparency and administrative oversight. His main answer to claims that the EVMs have been hacked is that the attacks are infeasible and detectable by election officials. Finally, he said essentially “different systems for different people” (or different strokes for different folks?).
The third speaker was J. Alex Halderman, who is one of the people who attacked the Indian EVM. He described how he got hold of an EVM and showed details on the insides. The first problem is that the devices can be duplicated (or fake ones could be substituted). Another issue is that verifying the code in the EVM is not possible (so they can be tampered with at the time of manufacture). Finally, the reported counts are stored in two EEPROMS which can be swapped out. There are two attacks (at least) that they performed. The first is to hack the display so that false counts are displayed on the LED. A bluetooth radio lets a mobile user select who should win. The second is to clip on a device to reprogram the EEPROMS. Full details will appear at CCS. Halderman’s last bit of news was that one of their co-authors in India, Hari K. Prasad, has been summoned by the police as a result of a criminal complaint that he stole the EVM, which seems like an attempt by the government of India to silence their critics. He called upon Shukla to drop the suit, who was rather upset by this public accusation.
The last panelist was P.V. Indiresan, who is on the advisory committee to the government. He discussed some new security features in EVMs, such as signatures to prevent tampering with the cable between the ballot unit (where people push buttons) and the control unit (which counts the ballots). He claimed that most of the attacks proposed so far are farfetched. Much of his latter complaints were to the effect that to break the EVM is a criminal act (which is a claim of security through obscurity). He ended with a plea to ask researchers to stop (!) hacking the EVMs because they “are working.”
To sum up : the Indian government says the system works and that there is no actual evidence of tampering (with the exception of Prasad, who apparently received stolen goods). Halderman says the attacks show that the system as a whole are not secure, and Rao says that the results are suspicious.
Shukla responded to critics that the Election Commission of India is willing to listen to critics and said that the only kind of attack that is of interest is one on a sealed machine. He reiterated the statement that Prasad was in receipt of stolen government property and needs to be questioned.
The Q&A was quite contentious. I might have more to say about it later… but wow.