During this election campaign, voters have been threatened with consequences if they don’t vote for the Bharatiya Janata Party (BJP). The BJP candidate for Sultanpur (UP), Menaka Gandhi told a predominantly Muslim crowd that she would know how many people voted for her, and she would not be inclined to help them find employment if they did not vote for her. She also said she would categorise villages “A, B, C, D” depending on the number of votes she received and structure development accordingly. Another BJP legislator, Gujarat MLA, Ramesh Katara, also reminded voters there were CCTVs in booths.
The threats were odious. But there is a deeper issue here. Citizens have the right to a secret ballot. However, the technology and processes used in Indian elections actually makes secrecy almost impossible to maintain.
It is now possible to make very accurate guesses about the votes of individuals. While few politicians will be crass enough to say this as bluntly as Gandhi or Katara, every major political party does its own data-analysis to work this out in as fine-grained detail, as possible.
The earlier system of paper ballots stuffed into boxes made it impossible to judge the order in which votes were cast and tally that with individual voters. In addition, ballot boxes were collected at a single location (as are EVMs) and the papers mixed up together before counting. So the vote was secret.
This is not how things work in the modern electoral system. The voters’ rolls are open and accessible online, and include details like name, father’s name, address, gender. The voter is tied to the voter ID, the assigned polling booth, and increasingly, to the Aadhaar number.
The electoral rolls have to be open in fact, since this gives citizens a chance to check if their names have been deleted, or to change address. Of course, electoral rolls have also enabled communal riots targeting one community or another. That started with 1984 long before rolls were online.
Voters go to their assigned booth, where their ID is checked and their names entered in a register before they vote. EVMs are used, where votes are registered in the order cast and there’s a VVPAT record as well. If the register could be tallied with the order of votes cast, it would indeed be possible to figure out the vote cast by every individual voter. We have to rely on the integrity of the Election Commission in this respect.
Since each EVM services a specific booth, which services a specific neighbourhood, it is possible to figure out how entire neighbourhoods voted. This information is available publicly. The candidates have representatives observing the counting process. The ECI releases Form 20 Data for elections online, and this contains breakups for votes received down to EVM-level.
Political parties do a lot of big data analysis to guess religion, caste and other preferences of voters. Local workers usually have a good idea of economic status. There are new tools available, given the prevalence of social media and the linkages made possible through Aadhaar.
Aadhaar for example, gives access to mobile numbers, and it is also tied to IT Returns, welfare schemes, etc. It is easy enough to start mining for more information such as presences on Facebook, WhatsApp, Twitter, the educational background, employment pattern, family dynamics and so on. Very few people scrupulously sweep their online presence clear of their political biases, and family connections, and indeed, why should they?
When all this information is matched to Form20, algorithms can make accurate guesses as to how individuals, families, communities and villages have voted. Local party workers can flesh in details, create WhatsApp and Facebook groups and get cracking as “influencers”.
Once “non-supporters” have been identified, skullduggery such as attempting to delete the names of non-supporters, or simply intimidating them to stay away on election day, is possible. So are carrot-and-stick threats like the ones Gandhi and Katara employed.
There has already been a scandal about a private firm mining Aadhaar data in Andhra Pradesh to profile voters. The data for almost 8 million persons has been copied and misused. This is probably just the tip of the Aadhaar iceberg. The penalties for this seem close to nil in practice, and the potential returns are obviously, large.
There are no easy ways to anonymise voting data beyond a point, without allowing massive scope for post-poll rigging. The Supreme Court vetoed a proposal in 2017 to “totalise” by clubbing together data from different EVMs. This is just another example of how the legal system has been overtaken by technology.