08-09-2009, 08:20 AM
Electronic Voting Machines are not fair, not transparent, have no voter verifiable audit trails
* * *
SPECIAL REPORT | Anupam Saraph and Madhav Nalapat (6 Aug. 2009)
DONâT BANK ON EVMS
Were the results of some Lok Sabha constituencies decided even before polling?
While tracking the lok Sabha elections of 2009 we stumbled upon startling versions of a file on polling information on the Election Commission of Indiaâs [ECI] website.
We were tracking candidate information from an Excel spreadsheet titled âCandidateACâ, which was downloadable from the ECIâs website. The âCandidateACâ spreadsheet has a column for âvotes polledâ. As one would expect, this file when downloaded on 16 April [the first day of polling] and 24 April did not contain any information on votes polled. The file, when downloaded on 6 May, 7 May and 11 May, had coded the candidate name, party name and the number of votes polled [check footnotes 1-5]. The last phase of polling took place on 13 May and, officially, the votes were counted and results declared on 16 May.
This information could have changed in three possible ways:
1. Someone had hacked the file.
2. The data from various EVMs had been uploaded to the private database, and this was used to create the public file.
3. Someone who had access to the private database had actually added the data.
The first possibility is serious, as it implies that the databases hosted by the ECI are not secure and can be compromised. The second is a violation of the ECIâs rule that no counting was to begin till 16 May. It also constitutes a breach of security, as the EVMs were supposed to be in safe custody in a strong room of the DEO [district election officer]. The third would imply that the election results were being decided by whoever had access to the private database. If the results were to be read only from the EVMs, why was a coded spreadsheet prepared? How could this spreadsheet have the data on âvotes polledâ? Check the file downloaded on 15 May. Surprisingly, it does not have any information on âvotes polledâ. More surprisingly, after the media raised the issue on 15 July, the link to the file disappeared from the ECI website.
Sceptics might say that the coded data containing information on the votes polled was test or dummy data. If so, why was this test or dummy data prepared? What was being tested? How was this test data generated? What were the results of the tests? Why was the data generated after the first three phases of polling, but before the fourth and fifth phases? Why was the data removed when there was no intention of loading real data into the file? Why was the dummy data coded? How do 108 winning candidates in the dummy data match the eventual winners?
When the implications are so serious, why has the ECI not clarified the issue despite repeated requests, or constituted a panel to initiate reforms to make the process less vulnerable to such serious implications?
EVMs, like all technology, can be hacked. Hacked EVMs can have Trojan Horses or programs with instructions to receive a coded spreadsheet and translate them into results. It is possible for such spreadsheets to actually prepare a predetermined result. If your bank publishes your account details and claims it is dummy data, will you let them manage your money? To ensure that such manipulations do not ruin our democracy what must the EVMs have? They must be built to leave a transaction trail auditable by anyone in case there is a dispute. To create such transaction trails, every vote cast on every EVM must be distinguishable from votes that might be cast by a Trojan. In the Indian system, the electronic votes are not recorded to indicate that they belong to a particular voter. Imagine a situation where you cannot track the money in a bank account as money deposited through a source: Someone, by gaining access to your account on the bankâs computer deposits an unlimited amount of money or withdraws whatever you have deposited, but you do not have the resources to trace that money or have a third party audit the bankâs books. That is why leaving a transaction trail is important in any audit process.
Now imagine a candidate having an âaccountâ in an EVM. Unless there is a mechanism to track every transaction as an inflow or outflow of votes from a legitimate source, there is no transaction trail and no ability to audit the actual vote balance in the candidateâs account. The votes cast by the voters are indistinguishable from those cast by a Trojan software or hardware. This is a very serious design lapse, and renders the EVMs completely useless for carrying out a costly democratic process. Will you trust your money with a bank that cannot track your accountâs transactions?
The current EVM system banks on faith, not on the accounting of votes â the faith of the voter in the honesty, integrity and fidelity of the EVMs, the people involved and in the process. Will you switch to a faith-based banking system when it comes to your money?
If the system is not faith-based, why not simply give the EVMs â like telephones â to electronic voting booths [EVBs] â like the STD-ISD booths â for anyone to run? Give everyone a week to vote and then collect the votes as you would collect the coins?
The manuals prepared for the chief election officer, district election officer, observers, polling agents, presiding officers, returning officers, candidates and counting agents show that the ECI gives little focus on building trustworthy databases, or on building good accounting practices and clean auditable systems that anyone other than they themselves can audit. Will the RBI permit financial institutions that cannot maintain basic records, or cannot be audited by third parties, to function?
Imagine a situation where you deposit your money in your account without getting any acknowledgement of your transaction specifying your account. In case there is a dispute, will you be able to claim that you had deposited the money into that account? When you deposit a vote for your candidate, do you get any acknowledgement about your deposit that you can use in a dispute? Is there any way that you can check later that your vote has been counted? Will you trust your money with a bank that does not give you a passbook or a statement of your deposits and withdrawals? What makes you trust the EVM with your liberty, and with your money that funds these elections? Finally, imagine a situation where your bank issues statements in which the number of depositors varies from statement to statement, as well as the deposits made. How comfortable does that make you about your bank?
Managing the democratic process of voting requires a database of voters, of candidates, constituencies and of votes deposited in a candidateâs account by legitimate voters. The votersâ banker, the ECI, has failed not just to track its depositors â the voters â but also how many votes they have deposited. Public databases maintained by the ECI indicate that the elections of 2009 had 716,676,063 or 714,103,070 or 713,776,525 voters [check footnotes 6-8]. As for the number of votes polled in these elections, the figure was either 417,158,644 or 417,156,922 [check footnotes 9-10]. Evidently, there is much the ECI can do to reform its technology [¼] Dr Anupam Saraph, an IT and governance expert, has created the worldâs first governance wiki at giki.wikia.com Professor Madhav Nalapat holds the UNESCO Peace Chair and is Director of the Department of Geopolitics at Manipal University
FOOTNOTES
1. http://www.scribd.com/doc/15676183/Candi...downloaded on 16 April 2009
2. http://www.scribd.com/doc/15676724/Candida...1.xlsdownloaded on 24 April 2009
3. http://www.scribd.com/doc/15676840/Candida...2.xlsdownloaded on 6 May 2009
4. http://www.scribd.com/doc/15676045/Candida...4.xlsdownloaded on 7 May 2009
5. http://www.scribd.com/doc/15676489/Candida...5.xlsdownloaded on 11 May 2009
6. 716,676,063 according to http://eci.nic.in/Analysis/ Voter Turnout
7. 714,103,070 according to http://eci.nic.in/eroll&epic/ERoll2009.pdf
8. 713,776,525 according to http://eci.nic.in/Analysis/ Electors Information
9. 417,158,644 according to http://eci.nic.in/Analysis/ Voter Turnout
10. 417,156,922 according to http://eci.nic.in/Analysis/ All Candidates Votes Polled
http://www.covertmagazine.com/anupham.htm
http://sites.google.com/site/hindunew/elec...voting-machines
* * *
SPECIAL REPORT | Anupam Saraph and Madhav Nalapat (6 Aug. 2009)
DONâT BANK ON EVMS
Were the results of some Lok Sabha constituencies decided even before polling?
While tracking the lok Sabha elections of 2009 we stumbled upon startling versions of a file on polling information on the Election Commission of Indiaâs [ECI] website.
We were tracking candidate information from an Excel spreadsheet titled âCandidateACâ, which was downloadable from the ECIâs website. The âCandidateACâ spreadsheet has a column for âvotes polledâ. As one would expect, this file when downloaded on 16 April [the first day of polling] and 24 April did not contain any information on votes polled. The file, when downloaded on 6 May, 7 May and 11 May, had coded the candidate name, party name and the number of votes polled [check footnotes 1-5]. The last phase of polling took place on 13 May and, officially, the votes were counted and results declared on 16 May.
This information could have changed in three possible ways:
1. Someone had hacked the file.
2. The data from various EVMs had been uploaded to the private database, and this was used to create the public file.
3. Someone who had access to the private database had actually added the data.
The first possibility is serious, as it implies that the databases hosted by the ECI are not secure and can be compromised. The second is a violation of the ECIâs rule that no counting was to begin till 16 May. It also constitutes a breach of security, as the EVMs were supposed to be in safe custody in a strong room of the DEO [district election officer]. The third would imply that the election results were being decided by whoever had access to the private database. If the results were to be read only from the EVMs, why was a coded spreadsheet prepared? How could this spreadsheet have the data on âvotes polledâ? Check the file downloaded on 15 May. Surprisingly, it does not have any information on âvotes polledâ. More surprisingly, after the media raised the issue on 15 July, the link to the file disappeared from the ECI website.
Sceptics might say that the coded data containing information on the votes polled was test or dummy data. If so, why was this test or dummy data prepared? What was being tested? How was this test data generated? What were the results of the tests? Why was the data generated after the first three phases of polling, but before the fourth and fifth phases? Why was the data removed when there was no intention of loading real data into the file? Why was the dummy data coded? How do 108 winning candidates in the dummy data match the eventual winners?
When the implications are so serious, why has the ECI not clarified the issue despite repeated requests, or constituted a panel to initiate reforms to make the process less vulnerable to such serious implications?
EVMs, like all technology, can be hacked. Hacked EVMs can have Trojan Horses or programs with instructions to receive a coded spreadsheet and translate them into results. It is possible for such spreadsheets to actually prepare a predetermined result. If your bank publishes your account details and claims it is dummy data, will you let them manage your money? To ensure that such manipulations do not ruin our democracy what must the EVMs have? They must be built to leave a transaction trail auditable by anyone in case there is a dispute. To create such transaction trails, every vote cast on every EVM must be distinguishable from votes that might be cast by a Trojan. In the Indian system, the electronic votes are not recorded to indicate that they belong to a particular voter. Imagine a situation where you cannot track the money in a bank account as money deposited through a source: Someone, by gaining access to your account on the bankâs computer deposits an unlimited amount of money or withdraws whatever you have deposited, but you do not have the resources to trace that money or have a third party audit the bankâs books. That is why leaving a transaction trail is important in any audit process.
Now imagine a candidate having an âaccountâ in an EVM. Unless there is a mechanism to track every transaction as an inflow or outflow of votes from a legitimate source, there is no transaction trail and no ability to audit the actual vote balance in the candidateâs account. The votes cast by the voters are indistinguishable from those cast by a Trojan software or hardware. This is a very serious design lapse, and renders the EVMs completely useless for carrying out a costly democratic process. Will you trust your money with a bank that cannot track your accountâs transactions?
The current EVM system banks on faith, not on the accounting of votes â the faith of the voter in the honesty, integrity and fidelity of the EVMs, the people involved and in the process. Will you switch to a faith-based banking system when it comes to your money?
If the system is not faith-based, why not simply give the EVMs â like telephones â to electronic voting booths [EVBs] â like the STD-ISD booths â for anyone to run? Give everyone a week to vote and then collect the votes as you would collect the coins?
The manuals prepared for the chief election officer, district election officer, observers, polling agents, presiding officers, returning officers, candidates and counting agents show that the ECI gives little focus on building trustworthy databases, or on building good accounting practices and clean auditable systems that anyone other than they themselves can audit. Will the RBI permit financial institutions that cannot maintain basic records, or cannot be audited by third parties, to function?
Imagine a situation where you deposit your money in your account without getting any acknowledgement of your transaction specifying your account. In case there is a dispute, will you be able to claim that you had deposited the money into that account? When you deposit a vote for your candidate, do you get any acknowledgement about your deposit that you can use in a dispute? Is there any way that you can check later that your vote has been counted? Will you trust your money with a bank that does not give you a passbook or a statement of your deposits and withdrawals? What makes you trust the EVM with your liberty, and with your money that funds these elections? Finally, imagine a situation where your bank issues statements in which the number of depositors varies from statement to statement, as well as the deposits made. How comfortable does that make you about your bank?
Managing the democratic process of voting requires a database of voters, of candidates, constituencies and of votes deposited in a candidateâs account by legitimate voters. The votersâ banker, the ECI, has failed not just to track its depositors â the voters â but also how many votes they have deposited. Public databases maintained by the ECI indicate that the elections of 2009 had 716,676,063 or 714,103,070 or 713,776,525 voters [check footnotes 6-8]. As for the number of votes polled in these elections, the figure was either 417,158,644 or 417,156,922 [check footnotes 9-10]. Evidently, there is much the ECI can do to reform its technology [¼] Dr Anupam Saraph, an IT and governance expert, has created the worldâs first governance wiki at giki.wikia.com Professor Madhav Nalapat holds the UNESCO Peace Chair and is Director of the Department of Geopolitics at Manipal University
FOOTNOTES
1. http://www.scribd.com/doc/15676183/Candi...downloaded on 16 April 2009
2. http://www.scribd.com/doc/15676724/Candida...1.xlsdownloaded on 24 April 2009
3. http://www.scribd.com/doc/15676840/Candida...2.xlsdownloaded on 6 May 2009
4. http://www.scribd.com/doc/15676045/Candida...4.xlsdownloaded on 7 May 2009
5. http://www.scribd.com/doc/15676489/Candida...5.xlsdownloaded on 11 May 2009
6. 716,676,063 according to http://eci.nic.in/Analysis/ Voter Turnout
7. 714,103,070 according to http://eci.nic.in/eroll&epic/ERoll2009.pdf
8. 713,776,525 according to http://eci.nic.in/Analysis/ Electors Information
9. 417,158,644 according to http://eci.nic.in/Analysis/ Voter Turnout
10. 417,156,922 according to http://eci.nic.in/Analysis/ All Candidates Votes Polled
http://www.covertmagazine.com/anupham.htm
http://sites.google.com/site/hindunew/elec...voting-machines