EVM (Electronic Voting Machine) is in news again. That it can be hacked and allegations about hacking in earlier elections. No need to get in to which party is saying what. We have seen every party worth its name has taken either side of the fence depending on whether they are in the ruling or opposition, or whether they won or lost the election. Let’s leave that to news channels for their primetime fodder. Election commission has taken a stand that EVM-VVPAT can’t be hacked and I would say rightfully so, since so far there is no hard evidence that it has been hacked or tampered with.
But there seems to be a nagging perception that our EVMs can be fixed, can be hacked or tampered with. In a democracy, similar to justice system, where it is not just justice is delivered, it needs to be perceived to be delivered, Elections has to not just free and fair but needs to perceived that way too by the public. With the existing publicly available data, I took my stab at possible steps that could be taken to get the perception right that our EVMs are fool-proof and there is no need to go back to ballot papers.
Now to some facts about our EVM-VVPAT setup. It has 3 units, a control unit, a ballot unit, and a VVPAT unit connected by a cable. In a booth this is what we will see, control unit with the presiding officer, and other two in the place where we vote in privacy. For a vote to be cast in the ballot unit, it has to be enabled in the control unit first. Once the vote is cast by pressing the button, registering of the vote is signaled by a beep, and the voter can check the correctness in the VVPAT display. This process continues for every vote. Simple function.
None of the units have any other connectivity – wireless, Bluetooth, IR, USB, or internet. There is no OS (Operating System) that runs in the machine. Source code is burnt into the chip. Our EVM, from what we know, is a very simple machine. Does it meet the democratic criteria of transparency, verifiability, and secrecy? Are the processes followed by EC meets those criteria? Can it be hacked or tampered with?
Coming to transparency, process followed in a polling booth or counting centers is the same with EVM or with ballot paper. With EVM, before the booth opens for voting, a mock poll gets conducted to verify the setup with party representatives present. This makes sure there is no tampering with the equipment, and we start with a count of zero. For every vote, there is a paper trail with VVPAT in a sealed box. If someone thinks EVM counts incorrectly or it is hacked, there is a fall back to VVPAT counting. On the secrecy, similar to ballot paper, there is no track to link the voter and the party he/she voted for.
Coming to hacking, contrary to claims, since there is no connectivity, there is no way one can take over the machine from outside or insert a malicious code from outside. Since there is no OS and connectivity, there are no entry points from outside. So, only way for anyone to hack is to modify the source code that is burnt in to the chip. For this to happen, it has to be an insider job and many ducks have to line up. One needs to line up willing people from Election Commission, ECIL, BEL, and the third-party company that burns the source code. Chip is Read only once it is burnt. Which means the hacked code has to be burnt in the chip during manufacturing. Assuming, just for our discussion, hacked code is burnt, it has to pass the diagnostics. What you are looking at is a Trojan code that passes the diagnostics, which is not an easy thing unless the verification engineers are also part of the insiders, passes the mock poll but wakes up either during polling or counting to skew the results. Not only that, it has to hack the VVPAT machine, meaning what it shows the voter what he/she voted but register and print a different result. And yes, the assignment of machines to polling booths are done randomly. So, for this hacking to work even at a polling booth level, the hacker or the BIG BOSS has to line up so many people. When you take it to the constituency level, considering the number of polling booths, EVMs, VVPATs, and number of officers that need to be part of the hack network, it is a non-starter. A party is better off buying the voters than taking the hacking route!!
Other possible way of getting in to the machine, that is being claimed, is to install a receiver in the machine, a RF receiver or a BT receiver. This needs both inserting a new hardware in to the machine plus some amount of source code to be modified. Again, requires insider support to make this work. IMO, a non-starter considering the sheer number of people who need to be part of the conspiracy. This is much bigger scale than what VW did with emission test. And they got caught.
I would safely rule out hacking possibility considering the track record of our EVMs so far. We have gained a lot moving to EVM from ballot papers. Firstly, it eliminated the invalid votes completely. Refer to invalid vote statistical analysis by Srinivasan Ramani that shows how invalid votes could have impacted outcome in Assembly elections between 1961 to 2003. Secondly, it brought down booth capturing, ballot stuffing and illegal votes. Yes, booth capturing still happens, but with EVM ballot stuffing is not possible, or very limited since it has an inherent limitation of casting only five votes every minute. Then there is the option of close button the presiding officer can use to lock the units if there is a threat.
While the process is fool-proof, for strong democracy, EC does need to address the perception issues with EVM. They may not be true but as mentioned in the beginning, public need to perceive it is free and fair and there is no tampering or interference. To that front, I would list out few points EC has to look in to.
Malfunctioning of EVM-VVPAT on election day has to be brought down drastically. With EVMs it has come down to 1-2% on poll day but VVPAT, the rate is still high, around 11%. Considering the number of EVM-VVPAT that is deployed and conditions they need to work with, this may be a bearable number, but not acceptable considering the fact that this does open up the process to conspiracy theorists. EC needs to look in to measures to get this down close to 0 or a low %.
Currently, EC limits the VVPAT counting to one randomly picked booth per assembly constituency. Statistical experts have pointed out this is not enough. Based on experts’ input, EC needs to increase the sampling points based on population density, and voting %. To make it more transparent and participative, EC may look in to the option of giving each party a choice of X booths and Y constituencies for VVPAT count post the election on top of statistical booth selection.
Lastly, currently counting is based on per booth, meaning one can find out the voting pattern in a booth. Considering the sensitivity and ghettoization, which still exists in our society, this needs to be avoided. Data could be misused by the parties. With ballot papers, ballots from booths get mixed before they get counted. Similar to that, EC can put another layer of abstraction to count the votes from booths without revealing the booth-wise pattern.
References: