How political parties can use voter data to manipulate your vote

With the legislative elections in Delhi just a day away, the issue of voter data being used for political campaigning has again become a matter of contention. What kind of data do political parties use in campaigning?

Though the electoral roll is itself meant to be a public record, a lot of useful information that can be easily appropriated for political targeting can be extracted from it. In a country such as India, knowing just the surname of an elector will allow someone, in most cases, to infer his or her religion.

Additional depth could be added where the surname would allow for inference of caste. With information about the elector’s age and area of residence, one would be able to arrive at an indication of the socioeconomic status, which though not always accurate, could be used in conjunction with information from other sources for building a precise profile of potential voters.

Tools for scraping social media websites can be used alongside natural language processing (NLP) tools for performing sentiment analysis on a given keyword. Sentiment analysis refers to computation processes through which subjective sentiment scores can be extracted out of supplied text.

For example, for any given hashtag on a website such as Twitter, sentiment analysis could be used to infer public opinion and voter intent, though, given the volume of artificially generated political discourse on Twitter, this data is likely to provide inflated results, including a lot of false positives.

It is interesting to note that, while 30 years ago this work would have been carried out by people possessing contextual knowledge and on-ground experience (and by hand), it is now possible to automate it, thanks to big data and analytics.

Campaigns such as the missed call campaign for showing support for the Citizenship (Amendment) Act, or CAA, allow political parties to easily collect and classify data. All that would be required is amplification of the phone number on which supportive electors are required to leave a missed call. Following this, all that would have to be added to the existing electoral roll data, is a column that indicates whether or not the elector has left a missed call to the number.

One may ask, that as the electoral roll only includes the name, age, and location of each voter, how would such data be meaningfully collated with mere phone numbers? Reports indicated that the Election Commission’s website, as recently as two months ago, was leaking the phone number and email address of every elector who had provided this information. Discounting this, databases, such as those containing information scraped from caller identification apps like Truecaller can also be useful here.

Just as how information about potential voters can be used by parties looking to further their campaign, information about those who may be inclined to vote for other parties could potentially be abused by way of fraudulent objections submitted to the National Voters’ Service Portal. Submitting copies of the objection form (Form 7) en masse could result in troves of names being struck off the electoral roll.

All that has been discussed so far does not even take into account databases maintained by third parties, such as Modak Analytics, which can be purchased easily and used in harmony with information available otherwise, to glean a more accurate picture (even if slightly) of electors’ voting preferences.

In Delhi, there have been several reports of WhatsApp groups being created, to which unsuspecting residents were added. The admins of these groups stated that they had been created for the purpose of “sharing news items omitted by the media”.

No one knows how the administrators of these groups acquired the phone numbers. It would not be wrong to speculate that this data could have been purchased for very little money from unregulated data brokers. One doesn’t have to look far to come across marketing companies such as Sarv, which help political parties automate targeted messaging through WhatsApp.

The Personal Data Protection Bill would offer some respite, which may affect the way parties interact with data, though it is not likely to be enough to cause a change of attitude in how parties tackle modern elections.

In the end, it should be said that the use of data for elections is not behaviour that should be objected to in and of itself. Rather, it should be objected to only when the data is acquired dubiously or through questionable means, or without the explicit consent of the subjects, or if such data is used to amplify surreptitious campaigns, including those that seek to further the spread of false or misleading news. Voters should be wary of the communication they receive and not only that which is received in the short run-up to the elections.

Karan Saini is a Delhi-based security researcher and technologist.