ECI, EVM and other tripwires

A guest post on the dangers lurking ahead in this election cycle

Image courtesy NDTV

ON May 17, a Supreme Court bench of Chief Justice of India (DY Chandrachud and Justices JB Pardiwala and Manoj Misra asked the Election Commission of India what the problem was with putting in the public domain the number of votes polled per constituency.

This was a special hearing in response to a joint application by the Association of Democratic Reforms (ADR) and Common Cause, which argued that in order to maintain voter confidence in the democratic process, the ECI should disclose on its website the number of votes polled per constituency within 48 hours of the completion of polling.

The petitioners were not asking for anything beyond the norm — this, in fact, is the first election in our history where the ECI has not released the final numbers of votes polled within two days of completion of polling in a constituency.

The ECI argued in court that it will take time to compile and publish the figures. Mahua Moitra, Trinamul Congress candidate for Krishnanagar in West Bengal, underlined the ECI’s intransigence when she posted the full details of polling in her constituency, compiled from the Form 17s her polling agents had procured from the various booths:

Krishnanagar had gone to polls on 13 May; Moitra’s post was at 9:21 PM on 14 May, just under 24 hours after polling had closed in her constituency.

The Supreme Court, to put it mildly, whiffed. It gave the ECI a week to file its response, and has scheduled the next hearing for May 24, one day before the sixth phase of polling. The SC is now closed for its summer break, which means that vacation judge Justice Bela Trivedi will hear the case.

Why is the lack of absolute numbers a problem? A security consultant, who has worked as consultant in elections in other countries, provides this explainer on condition of anonymity (mildly edited for size and syntax):

***

Is it possible to rig an election?

Objective: I am writing this article to point to a potential path that exists for the outcome of the ongoing Lok Sabha elections 2024 to be rigged, with minimal resources and without the need for sophisticated technology. I write this in the hope that readers who are influential in their parties can take note, and initiate corrective action.

Background: The most important unit of Indian elections is the polling booth. Irrespective of all else — voter suppression, voter deletion, violence, etc — the fundamental point is that for a free and fair election, the number of votes recorded in each individual polling booth must be known.

For instance, consider a parliamentary constituency which has 1000 polling booths, with 200 voters registered per booth. Assume that it had a 50% voter turn out. The number of votes polled in the constituency is therefore one lakh — it cannot be more, or less, in the final count.

Back in the days of paper ballots, if the aggregate count does not add up, it means that either an extra ballot box(es) was smuggled into the counting room, in which case the counted votes will be more than the registered votes, or that one or more were smuggled out before counting began, in which case the counted votes will be less than the registered votes.

Theoretically, it is possible to argue that in the EVM method, there are checks and balances to keep this from happening. Every EVM is given a serial number and, when polling closes, the booth officer issues Form 17C, which contains the following data:

• List of EVMs and their serial numbers.

• Number of votes registered in the booth, split by sex.

(NB: Mahua Moitra’s post, above, was an aggregate of the Form 17Cs her polling agents collected from the various booths in her constituency).

The expectation is that on counting day, the agents of the various parties will use the respective Form 17Cs to verify the EVM serial numbers against the existing list before allowing the counting to proceed further.

The First Breach:

As described in detail by journalist Poonam Agrawal (Follow her, she is one of the most important voices of these times), there has been consistent discrepancies between votes polled and votes counted since the Madhya Pradesh elections of 2018, and the Lok Sabha elections of 2019.

Given that an EVM can at most hold 2,000 votes, consider the data from Poonam's article with regard to where the highest surpluses were recorded:

1. Kancheepuram - 18,331.

2. Dharmapuri   - 17,871.

3. Sriperumbudur - 14,512.

4. Mathura - 9,906

If you take 2000 votes per EVM as the measure, then the only way to account for the extra votes is to conclude that additional EVMs were added, as under:

1. Kancheepuram - 10

2. Dharmapuri   - 9

3. Sriperumbudur - 8

4. Mathura - 10

Agarwal in her article (which needs to be read in full) points out that when The Quint asked questions at the time, the ECI pulled the relevant data off its website.

What the above data points indicate is that only a small number of EVMs need to be smuggled in (or taken out) from the strong room to rig the result of a particular constituency.

However, the strongrooms where EVMs are kept are under constant CCTV surveillance, as is pointed out here.

Note that Agarwal’s article was published on 31 May 2019, and counting in that general election was done on 23 May 2019. Once the discrepancy was pointed out, the ECI could have gone back to the CCTV footage of the concerned strongrooms, and checked the EVM serial numbers against the official list in the Form 17Cs. Instead, the ECI erased the discrepant data, and then went ahead and matched the counted data.

This permits of only one conclusion: ballot stuffing via inserted EVMs had already happened, and the ECI merely did the cover-up.

An additional catch with the provision of CCTV footage of strongrooms is that these recordings are stored only for 45 days, after which they are deleted.

In this connection, note that the first phase of polling, involving 102 Parliamentary constituencies in 21 states and Union Territories (besides 92 Assembly constituencies of Himachal Pradesh and Sikkim) was on April 19. The votes cast in that phase will be counted, with the rest, on June 4 — the 47th day after polling.

From the above, it is possible to understand the implications of the long election cycle, and the possibility of ballot stuffing via the insertion, deletion, and/or modification of EVMs:

A longer election cycle means strongrooms have to kept under watch for longer periods, providing more time for manipulation. And secondly, the initial phases provide the best chance for ballot stuffing compared to the later stages because there is less likelihood of anyone going back to check CCTV footage from six weeks earlier.

Agarwal’s article also raises a related question: Where does a party intent on ballot stuffing get extra EVMs from? A possible answer lies in an article in Frontline dated 08 May 2019, which cites an RTI response that 20 lakh EVMs which the manufacturers claim to have delivered are missing from the ECIs stores.

While on this, I was told that the polling agents are checking the serial number of the EVMs against the official list merely by looking at the tag printed on each EVM, and that the agents have not been trained to check the hardware serial number. This makes swaps trivial to accomplish.

In this connection Sabyasachi Das, assistant professor of economics at Ashoka University, had in July 2023 published a research paper suggesting that the BJP had manipulated results of some seats in the 2019 elections, and that this manipulation had resulted in the party winning a disproportionate share of closely contested constituencies.

In an unusual move, the university distanced itself from its own faculty member and dissociated itself from the research - which, in sum, backed up what Agarwal had reported.

Both — the journalist and the researcher — pointed in the same direction: that a party with access to the “missing” EVMs merely has to load them with the same symbols as on the kosher one, and add them to the strongroom.

The Rig Veda and the Three Rishis

Given the above, the ECI’s intransigence in refusing to publish the data of the actual numbers of votes polled in the first four phases becomes understandable: (1) The longer the lead time between polling and counting, the more the opportunity to create mischief and, (2) Many of the seats in those phases feature direct contests between the BJP and the Congress, which are expected to be tough and hence can be manipulated via a few additional EVMs introduced into the strongroom via the simple expedient of switching off the CCTV cameras, as for instance happened in Baramati.

Note also that Opposition parties have been drained of resources, and several of them don’t have the infrastructure and resources to put a couple of polling agents in every booth. If there is no polling agent, there is no one to collect the Form 17C on behalf of the party — and therefore they have no evidence to contest whatever results the ECI choses to announce.

This leads to the conclusion that the only reason to deny publication of polled data is to prevent the sort of analysis Agarwal did in the 2019 general elections.

Potential Counter-measures

1. Every political party needs to make sure that in constituencies they are contesting, every booth has a trained polling agent tasked to collect the duly filled in and signed Form 17C.

2. Ensure you are represented in the Supreme Court on 24 May when the vacation bench hears the petition mentioned above, and ask the court to issue directions to the ECI to upload scanned copies of every single Form 17C from the five phases thus far, and publish aggregate figures, so parties and even journalists can collate the information and check for discrepancies. The time to do this is before, not after, the results are announced.

3. Launch a public awareness campaign, through all available channels, to inform the voter about the potential dangers.

(The writer is a security consultant with experience with elections, who wishes to remain anonymous.)

EOM