EVM hacking

Just a small note on this article. It claims to prove “mathematically” that it is almost impossible that EVMs are hacked in India.

Its argument is the following: there are approximately 3002 EVMs in each constituency in India. After the electronic votes are polled, 5 EVMs are selected at random, and the total number of votes polled is compared with the paper ballots. If there is 100% agreement of the EVMs with the paper ballots, only then are the votes polled in those EVMs considered legal.

Let us suppose that a party only hacks 1% of the EVMs, and does so only in 50 constituencies. These numbers are low, and I find them to be acceptable. 1% of 3002 would be around 30. Hence there would be 2972 unhacked EVMs in each of those 50 constituencies. If 5% of the EVMs are selected randomly for checking, then 5% of 3002 turns out to be around 38. Hence, the probability of selecting only unhacked EVMs is ${38\choose 2972}/{38\choose 3002}\approx 0.68$. Hence, the probability of selecting only unhacked EVMs from each of those 50 constituencies is $(0.68)^{50}\approx 0$.

This would suggest that it is almost impossible for EVM hacking to go undetected.

However, there are a couple of assumptions made in this argument that are simply untrue:

• If a political party has enough influence to hack EVMs, can it not also influence the EVMs selected for testing?!
• Discrepancies between EVM counts and paper ballots are actually common. See this news article for instance. Hence, the statement that a discrepancy in even a single machine in any constituency renders the whole election void is simply untrue.

I don’t know if EVM hacking is a reality in India. However, it is most definitely a possibility (at least mathematically).