e-voting - india and the philippines · one of the latest incidents of hacking to take place was...
TRANSCRIPT
e‐ Voting: India and the Philippines – A Comparative Analysis for Possible Adaptation in Africa
Page 1 of 32
e‐Voting
IndiaandthePhilippines–AComparativeAnalysisforPossibleAdaptationinAfrica
Writtenby
SurendraThakur
SouthAfrica
e‐ Voting: India and the Philippines – A Comparative Analysis for Possible Adaptation in Africa
Page 2 of 32
Abstract
This paper contributes to the debate around e‐voting by describing and discussing the experiences of two adopting countries, India and the Philippines with careful, qualified, comparison and to gain from their experience for possible adoption in Africa. The region often finds itself as the centre of attention with regards to elections. This regrettably relates to extreme violence, civil disobedience and fraudulent activities. Will e‐voting be the panacea for her problems?
A challenge with electronic voting is that the voting machines are susceptible to sometimes undetectable changes by insiders, outsiders and hackers who may or may not have ulterior motives.
One of the latest incidents of hacking to take place was during the Zambian election in 2011 when the Electoral Commission of Zambia’s (ECZ) website was hacked and struck an embarrassing blow to democracy questioning the validity of the elections.
1.0Introduction
There are some countries have acknowledged and abandoned e‐voting after adopting or piloting. On the other hand many other countries continue to adopt, pilot or trial e‐voting.
India is the world’s second most populated country with 1.1 billion people. India has complex election‐related challenges such as being the world’s largest democracy with 671 million voters; multi‐lingual illiteracy; a practiced though illegal caste system; rural off‐grid geographic areas and booth capturing (where booths are ‘taken over’ by politically hired thugs).
The previous paper‐based elections took weeks to administer and months to resolve, rendering this massive country almost without governance after elections. India has a coalition government which is an interesting characteristic of many current adopting e‐voting countries.
The Philippines is the world’s second largest archipelago with over 7000 islands and a population of 90 million people. Philippines also experiences election‐related challenges such as a monumental complex island logistic; extreme violence exacerbated by delay in results and counting; ballot stuffing; illiteracy and areas with no telecommunication and a high number of invalid votes.
Elections here took weeks to administer because the logistics of counting and collating results was very complex. This led to extreme election‐related violence and tension occurring in the Philippines because of the complex logistics of counting and collating results extending the voting administration period.
Notwithstanding the Indian and Filipino e‐voting experiences, both have largely been positive elections and have certainly helped speed up the results with a drastic reduction in violence. They thus serve as enticing models for study and evaluation.
The key word here is “context”. This defines the different implementation strategies that India and
the Philippines used that helped towards their elections being successful.
e‐ Voting: India and the Philippines – A Comparative Analysis for Possible Adaptation in Africa
Page 3 of 32
The research method is desktop literature review which comprised EMB websites, journals and media articles. In addition experts were interviewed or canvassed for validation and experiences.
Background
The use of Information and Communication Technologies (ICT) is pervasive and its increasing use in elections is unsurprising. ICT is now used, inter alia, in back office Electoral Management Boards (EMB) operational activities; communication for information dissemination such as election results and voter education and Geographic Information Systems(GIS) to assist in demarcating voter districts.
It has also helped in capturing data management system(s) to create, store, and update voter registration information; to capture and tally votes (e‐voting); auditing to statistically analyse results for voter irregularities, predictive algorithms and most recently as a tool in observing elections through NetBooks, tablets, cameras, video cameras and digital voice recorders1.
E‐voting is a term encompassing several different types of voting, embracing both electronic means of capturing a vote and electronic means of counting votes. An important and difficult requirement of e‐voting systems is that the vote cast is secret, that there is no link whatsoever between the vote cast and the vote captured2, and the integrity of the vote is secure in that the vote cast is the vote counted. E‐voting machines are referred to by many names. In the West they are called Direct Recording Electronics (DRE), or DRE Voter Verifiable Audit Trail (DRE/VVAT) ; in South America urnas; in India Electronic Voting machines (EVM) and in the Philippines, Precinct Count Optical Scan(PCOS).
Intriguingly as each successive wave of automation introduces new advantages such as speed and accurate counting (lever machines), ease of use and multi‐lingual facilities (DRE),auditability and speed (DRE with VVAT), it also introduced serious disadvantages such as mechanical failure and non‐validation (lever machines), software and hardware errors (DRE), software hardware and printer failure (DRE VVAT).
India along with Brazil and the Philippines are the only three countries to have full precinct based electronic voting but with subtle differences. It is for this reason that two of these countries, India and Philippines, have been chosen for an in‐debt analysis.
It is accepted that direct comparisons is a dangerous and almost impossible task, so this paper focuses on each individual country’s efforts and point to experiences where relevant to Africa rather than each‐other. Context in every country is extremely important. This is the second such comparative bilateral study with the first by Pieters and van Haren (2007) which undertook a comparison of the electronic voting experience of Netherlands and the UK3.4 Indeed they recommended more comparisons in different contexts.
In Mexico there were losses of approximately $10 billion in 2006 caused by strikes, protests and riots from various sectors after Lopez Obrador and Calderon clashed with each other in elections. Thus a few vote swings can have serious economic consequences. A reliable and secure voting technology
1 (Thakur, 2011a) 2 Most Software Programs and Operating Systems record the date and time a record is created or give it a sequence
number to allow for system audits to assist when a system crashes. Both of these events can be replayed and with appropriate video monitoring outside a booth, one can predict who cast a vote for whom for most of the day.
3 (Pieters and van Haren, 2007) 4 Pieters is author of a seminal thesis: La Volont´ e Machinale Understanding the Electronic Voting Controversy.
e‐ Voting: India and the Philippines – A Comparative Analysis for Possible Adaptation in Africa
Page 4 of 32
will reduce this violent possibility and that results, in spite of being too close, are accepted by all contenders5.
The next section (2.0) presents international adoption patterns of e‐voting. Section (3.0) describes India with respect to e‐voting. It commences with a description of the Political System (3.1), The Electoral Management Board (3.2), a description of the actual Electoral Voting Machine(EVM) (3.3) It then describe the Indian e‐voting experience with to challenges and how EVMs mitigates these (4.0). The Indian experience is critically evaluated against fifteen critical success factors namely: Voter Registration, (2) Rural, elderly and illiterate voters (3) Access (4) Multiple Elections (5) Voter Turnout (6) Overvotes, undervotes and invalid votes (7) Voter acceptance and trust (8) Regional Cooperation (9) Vendor Selection (10) Black and white box testing (11) Security (12) Cost of elections (13) Post‐election tabulation and transmission of results (14) Accountability, recount and transparency and finally (15) Observer reaction. The Indian section concludes with a description of current Indian electoral research efforts (5.0) and the perceived advantages of EVMs (6.0). This data is summarised in Table 3.0. This entire process is repeated for Philippines. Both countries are comparatively analysed for lesson learned in Africa and a Conclusion is provided. The Appendix provides some costing to provide an economic context.
2.0 The international adoption patterns of e-voting
There is a statistical trend towards e‐voting adoption as Table 1.0 shows. Indeed Smartmatic the contracted vendor for e‐voting in the Philippines emphatically asserts on its website that “Electronic voting is an irreversible tendency.” Yet five countries who have adopted e‐voting have now revoked this choice with enormous political risks and economic cost.
Adopting e‐voting is a major political decision. Abandoning e‐voting is an even bigger decision with enormous political consequences. Therefore some notes are provided on the decision of these countries. Australia discontinued its universally hailed most open e‐voting software effort due to high costs. Germany's highest court had in 2009 ruled the use of electronic voting in the last general election was unconstitutional as it was not transparent and the count is not publically observable. The Irish finally dumped €50m of e‐voting equipment in 2010 after the 2004 decision over lack of public trust and security fears. The Dutch abandoned e‐voting in 2008 after a Justice Court decertified the machines following a group called Wij vertrouwen stemcomputers niet (“We do not trust voting computers”) showed the machines could be compromised. England, in 2008, said e‐voting pilots were "extremely expensive” and that: “Serious concerns persist about the security and transparency of e‐voting systems and their vulnerability to organised fraud6."
5 (Saba, 2011) 6(The Register, 2008)
e‐ Voting: India and the Philippines – A Comparative Analysis for Possible Adaptation in Africa
Page 5 of 32
Table 1.0 ‐ Data on elections and e‐voting (Thakur, 2011:48)
Status quo Country
Twelve countries with legally binding poll‐
site electronic voting
Belgium, Brazil, France, India, Japan, Kazakhstan,
Russia, United States of America, United Arab
Emirates, Philippines, Paraguay, Venezuela
Six countries with legally binding Internet
Voting
Austria, Canada, Estonia, France, Japan,
Switzerland
Twenty countries planning, trials, non‐legally
binding e‐voting
Argentina, Azerbaijan, Belarus, Chile, Czech
Republic, Costa Rica, Finland, Greece, Italy,
Indonesia, Lithuania, Mexico, Nepal, Norway,
Poland, Portugal, Romania, Slovenia, South
Korea, Sweden
Five countries with Pilots Peru, Kenya, Ecuador, Slovenia, and Spain
One country with a Trial suspended and one
Pilot suspended Bulgaria (Pilot) and Latvia (Trial)
Five countries with e‐voting projects
stopped
Australia (2009) Germany (2009), Ireland (2004),
Netherlands (2008), and England (2008)
The next section describes India, its political and electoral board as well as some critical success factors in the context of e‐voting.
3.0India
India is the world’s largest democracy. The number of eligible voters for the 2009 elections was 741 million voters (671 m in 2004). This number is greater than the total populations of Europe and the USA put together. India used 8800 metric tonnes of paper to handle the 1996 elections (7700 tonnes in 1999) when it gradually started to use e‐voting7. India began experimenting with EVMs in 19828 and went fully electronic in 2004. The first fully electronic election was in 2004 with 1075 million EVMS deployed.
7 (ECI, 2011) 8 NarhimaRao (2010) says 1982 as he means any effort but the ECI website talks of the EVM in 1989 as largely used today.
e‐ Voting: India and the Philippines – A Comparative Analysis for Possible Adaptation in Africa
Page 6 of 32
3.1 Indian Political System
India functions as a federal constitutional republic. The President of India is head of state and the Prime Minister as the head of government. Executive power is exercised by the President. Legislative power is vested in the government and the two chambers of the Parliament: the Lok Sabha and the Rajya Sabha. Federal and state elections take place within a multi‐party system. India has been led by coalition governments for the last 30 years9.
Since 1991 no political party has had a clear majority. In 1998 two alliances of parties were formed: the National Democratic Alliance (NDA) with several non‐congress10 related parties and the Indian National Congress (INC) led coalition called the United Progressive Alliance (UPA). The NDA became the first non‐Congress government to complete a full five‐year term in 1998. In the 2004 elections, the UPA/INC won the largest number of Lok Sabha seats and formed a government11 .
3.2 India Electoral Management Board
The Election Commission of India (ECI) is an autonomous, quasi‐judiciary constitutional body. Its mission is to conduct free and fair elections in India. It was established in 1950 under Article 324 of the Constitution of India. The Commission enjoys complete autonomy and is insulated from any kind of executive interference. The ECI also functions as a quasi‐judiciary body in matters of an electoral nature. Its recommendations and opinions are binding12.
Indian Elections are events that involve political mobilisation and organisational complexity of an
amazing scale. To schedule its elections India needs to consider, inter alia, weather, agriculture
cycles, festivals, exams schedules and public holidays. In the 1996 elections for the Lok Sabha there
was over 12 000 candidates while over 592 million people voted. The Election Commission employed
almost 4 million people to run this election13.
3.3 The Indian Electronic Voting Machine (EVM)
Figure1.0 – The Indian EVM developed by BEL and ECIL.
The voting machine has two parts a control unit (controlled by the electoral officer) and the ballot unit which the voter has access to.
9 (India Profile, 2007) 10 The Indian National Congress (INC) has led India since 1948. 11 Ibid 12 (ECI, 2011) 13 (ECI, 2011)
e‐ Voting: India and the Philippines – A Comparative Analysis for Possible Adaptation in Africa
Page 7 of 32
In order to vote, a person must present sufficient identification14. The voter then signs an official record and is assigned a serial number. The electoral officer presses the Ballot button on the control unit to release one vote. This battery‐powered unit is connected to the voting unit by a five‐meter cable. A green (Ready?) lamp lights up on the voting unit which is in a screened‐off area. The voter has to press the button next to the name and party symbol of the desired candidate. A red lamp glows briefly once the button is pressed and the machine emits a beep indicating that a vote has been cast.
The machines were reused in 5‐phases over a period of four weeks across the country. As per direction of ECI, randomisation of EVMs had been done in order to have transparency in the allotment of EVMs to polling stations15.
Each ballot unit allows for 16 candidates. The system allows for a maximum of 64 candidates16.
The next section looks at the perceived benefits of the EVMs.
4.0 The challenge India faced ‐ How EVMs mitigates these
Prior to e‐voting India faced many challenges with respect to elections namely i) massive democracy resulting in lengthy elections and charged atmospheres, ii)multi‐lingual illiteracy, iii)deep off‐grid rurality and iv)booth capturing‐voting mitigated some of these challenges. However in so doing it simultaneously introduced additional challenges such as technical subversion which is difficult to detect17. These challenges and some are discussed below.
i) Lengthy Elections
The counting of ballot papers used to take many hours creating a charged and sometimes violent atmosphere for the counting officials, candidates and political parties. This was aggravated sometimes by the demand for recount especially where low margin of difference of votes between the top two candidates occurred. This was also complicated by large numbers of invalid ballots. Country wide voting used to take weeks and dispute resolution months.
With EVMs the elections takes four weeks as the machines are reused in a rolling election across the country. The speed of counting is very quick and the result can be declared within two2 to three3 hours as compared to 30‐40 hours, on an average, under the conventional system.
ii) Multi lingual illiteracy
Mumbai, the capital of India, alone has 31 languages. So the complexity is conceivably difficult to comprehend. This challenge was greatly simplified by the use of symbols and official colours.
14 This is now being replaced by the Photo‐Id project 15(Kerala, 2010) 16 There is a recorded case of 1033 candidates for a single seat in Tamil Nadu in 1996. The ballot paper was produced in
the form of a booklet (ECI, 2011) 17 It also introduced errors and threats that any computer hardware and software may experience such as hacking,
monitoring, denial‐of‐service etc.
e‐ Voting: India and the Philippines – A Comparative Analysis for Possible Adaptation in Africa
Page 8 of 32
iii) Deep off‐grid rurality India is geographically the seventh largest country. It is a country in development and has a deep urban rural divide. Most of the country has no electricity and infrastructure is limited to urban areas.
The Indians countered this problem by using minimal energy consumption for the voting equipment. The machines operate with AA alkaline batteries. These 6V batteries cannot electrocute a human and are green. As the machines are small, logistics in elections are not that difficult. Smartmatic praised the Indian ingenuity. This is certainly a useful option in Africa.
iv) Booth capturing and Ballot Stuffing This is a type of electoral fraud practised mostly in India where party loyalists "capture" a voting station and actually vote in place of legitimate voters to ensure their candidate wins. This is a far less subtle form of electoral fraud than chain‐voting.
The upsurge in democracy has also cruelly manifested itself in the criminalisation of politics. The EVM has consequently been configured to make it harder and more time consuming for booth capturers to stuff ballot boxes. This is achieved by only allowing five votes per minute or by enforcing a 12 second delay between each vote entered. With ballot papers it is possible to orchestrate a stampede and stuff hundreds of votes simultaneously, a practice commonly referred to as Ballot Stuffing. The EVMs further possess a "close" button which can, at the discretion of the polling officer, be used to deactivate the machine. Despite this, booth capturing continues to happen, albeit at a much reduced rate. Even though it is more difficult to do so many candidates who lose elections in India regularly complain that their opponents indulged in booth capturing to win.
The Election Commission of India suggests EVMs have helped to curb booth capturing18.
The next section examines the Indian Elections from a series of Critical Success Factors.
5.0 Critical Success factors and Indian Elections
1. Voter Registration
The photo‐electoral rolls were first introduced in India in 2005. Photo‐electoral roll is one in which all relevant details and a photograph of the voter is recorded. This electoral roll is more reliable as it assists polling staff to correctly identify the voter, reducing impersonation. Intriguingly, Photo‐Ids were implemented after the first full EVM election of 2004. Repeat voting is prevented by the use of indelible ink on voter’s finger tips.
It was only in the 2009 elections that the entire nation witnessed the use of photo‐electoral rolls for the first time. It was the most significant change in the Indian elections procedure after the introduction of the EVMs19.
2. Rural, elderly and illiterate voters
The EVMs are considered simpler compared to the paper ballot where one has to put the voting mark on or near the symbol of the candidate of one’s choice, fold it first vertically, then
18(ECI Resolution, 2011) 19(ECI, 2011)
e‐ Voting: India and the Philippines – A Comparative Analysis for Possible Adaptation in Africa
Page 9 of 32
horizontally, and thereafter put it into the ballot box20. In EVMs, the voter simply has to press the blue button against the candidate and symbol of their choice and the vote is duly recorded. A beep indicates the vote is recorded.21
3. Access
1.378 million EVMs were deployed for the 2009 elections to ensure equitable and fair access to the electorate22. However, even this enormous number is insufficient and the machines had to be deployed in a five‐phase approach in a process that took four weeks to complete.
To prevent sustained access the EVMs are programmed in such a way the machines can record only five votes in a minute which prevents sustained ballot stuffing. It is, however, still much faster than paper balloting23. Additionally, only a maximum of 3840 ballots per machine is possible although in practise only 1500 voters are permitted per poll site.
4. Multiple Elections
These EVMs was used for simultaneous elections for both Parliament and the State Legislative Assembly. The fact that the machines may be used for simultaneous elections suits the trend towards such elections in Africa.
5. Voter Turnout
Although the voter percentage has remained in a narrow band between 58 and 60%, the numeric number has grown significantly. More importantly the election period was drastically reduced from months to weeks.
6. Overvotes, undervotes and invalid votes
An overvote is when votes cast exceed the number of candidates allowed for a particular position.
India states that invalid votes are not possible with their design. The minimal recorded and reported invalid votes are due to paper ballots in exceptional cases. The issue of overvotes and undervotes have either not been reported or occurred.
7. Voter acceptance and trust
Despite sustained though limited opposition to the EVMs, election results have been generally accepted. The Indian adoption of e‐voting is seen as a pragmatic solution to a major mega‐democracy dilemma: How to deliver elections with so many people in a reasonable time frame with accurate results.
The next three points also contribute significantly to Indian voter acceptance and trust.
8. Regional Cooperation
A fascinating trend is that countries who have embraced e‐voting have served as regional champions of this opportunity. India has offered its EVMs to its neighbours and Nepal has already run a pilot with EVMs. These machines were also used in the Afghanistan elections and India offered these machines to Egypt for their Arab spring elections.
20 (ECI Faq, 2011) 21 Ibid
22Ibid 23Ibid
e‐ Voting: India and the Philippines – A Comparative Analysis for Possible Adaptation in Africa
Page 10 of 32
Figure2.0 – Map of India
9. Vendor Selection
The Indians chose an entirely Indian solution. Two companies were selected, namely Bharat Electronics Limited (BEL) and Electronics Corporation of India Ltd. (ECIL). Both are wholly government owned central public sector undertakings under the administrative control of the government of India24.
10. Black and white box testing
The Expert Committee of the Election Commission which approved the EVMs currently in use in elections has only done "Black Box testing"25. This means the Committee did not examine or rather could not certify the software program in the EVMs. White box testing refers to examining the source software to verify functionality.
By not examining the software and merely relying on functionality tests, the Expert Committee has left a gaping hole in the security of the EVMs. The ECI asserts the software used to run the machine is burned into the microprocessor so there is no room for manipulation once it is made and installed26. This is a horrifying lack of security (NarasimhaRao27).
As it is the software in the EVMs that drives all its functions, can and has already began to have an impact on the trust by Indians.
24 (NarasimhaRao, 2010:24) 25 Gained from anRight to Information (RTI) question and answer by the Election Commission. 26 (IEEE, 2004) 27 (NarasimhaRao, 2010:155)
e‐ Voting: India and the Philippines – A Comparative Analysis for Possible Adaptation in Africa
Page 11 of 32
11. Security
There has been a reasonable amount of press stating how low‐tech the Indian EVMs are. Low‐tech solutions obviously have fewer features. On the other hand, they are cheaper, easier to manufacture and may be easier to “defend” against attacks. The design requirement of operability by a 6V battery is strength as it allows for off‐grid functionality. It is also a limitation for the design of the EVM making it smaller with fewer features.
The EVMs were compromised in 2010. An EVM machine was spirited away in a James Bond type heist by Hari K Prasad and a team of scientists. The researchers found they could hack the security of the EVM. This made sensational news28. This is unlikely in a modern monitored democracy as sustained access to a disconnected machine is required. The knee jerk reaction of the Indian government to arrest the researcher is exactly the kind of reaction displayed by Diebold and other earlier e‐voting companies and is perceived to be a primary reason that activists distrust e‐voting. This author argues that preserving the security of an e‐voting machine through secrecy and obfuscation is almost a tacit admission by e‐voting manufacturers that they lack faith in the robustness of their product. So they keep both the machine and its code secret by limiting access.
However, as the machines are off‐line during voting, hacking is difficult and limited to polling stations accessed.
12. Cost of elections
The 2004 elections cost Rs 13 billion or USD 280 million for 671 million voters.
India has been using the EVMs for over ten years and has had little or no violence or discontent over elections. The EVMs reduced the election period to just four weeks which saves time and money. The political parties have also demonstrated a high level of maturity and generally accepted the results. Indeed the ECI asserts the elections thus far have been “flawless, fully electronic voting.” The price at about USD 170 is appealing from an African perspective29. India does lend expertise and equipment to neighbours such as Nepal and Afghanistan. And finally the fact that the machines haves been (re)used over a period of 10 years further reduces the cost‐of‐ownership.
13. Post‐election tabulation and transmission of results
No transmission occurs. This is regardless of the fact that a simple and unconditionally secure protocol exists to transmit. It was a deliberate EVM design feature to intentionally serve as stand‐alone units. This prevents any intrusion during electronic transmission of results. Instead, the votes are collected in booths and tallied on the assigned counting day(s) in the presence of polling agents of the candidates. The results are stored electronically so the electoral officer merely had to press a button on the control unit to display the totals.
Thirty copies of the results are printed, posted and distributed to various stakeholders such as the ECI.
28 (Prasad, 2010) 29 (ECI, 2011)
e‐ Voting: India and the Philippines – A Comparative Analysis for Possible Adaptation in Africa
Page 12 of 32
14. Accountability, recount and transparency
The strength of electronic voting is the fast accurate speed at which vote counting may occur. A major disadvantage of a EVM is its inability to perform a recount. To accommodate some manipulation claims, Indian officials have required the machines to be able to produce a hard copy of the votes once voting has ended.
However, if empowered by a court, election officials can use the serial numbers to identify each voter and to find out who he or she has chosen. ECI points out that although technically possible actual breaches of voter privacy are extremely rare. During normal counting, officials look at only the total number of votes polled and the number cast in favour of each candidate30 (IEEE). This offers transparency and provability. However this violates secrecy, as it provides a real post‐election link between vote and voter whether used or not!
15. Observer reaction
India does not believe in observers. It consequently neither invites nor deploys observers to elections. It is therefore has not easy to find independent observer reactions.
India has not stopped her e‐voting efforts. The next section presents some of these which mitigate some challenges in the CSFs of the previous section.
6.0 Current Indian efforts
6.1. Voter Verifiable Audit Trails (VVAT)31 A VVAT is a system that electronically captures a vote and then produces a printout that the voter can look at to ‘verify’ their voting choice. If they accept, the vote is captured and the corresponding manual ticket drops into a ballot box for audits and contestations. VVAT is meant to counteract any subversive, vote altering code that may have been (unobtrusively) placed in the system.
A mock trial on 27 July 2011 was conducted with over 20,000 voters in two polls in Calcutta. Here the VVAT option was trialled with both an open and closed printer system. Anecdotal evidence is that voters preferred the closed printer; however, they found the process very slow32.
The above demonstrates a proactive EMB that is constantly exercising an environment scan as it responds to ‘shifts in voter trust prompted by greater awareness and advocacy groups efforts’ and changes in voter behaviour particularly over local elections. This demonstrates that even in India the e‐voting journey is a work in progress.
6.2. The e‐voting roadmap in India The Gujarat State, to encourage greater voter participation for the local or district elections permitted online voting. They used traditional EVMs for poll‐site e‐voting. In addition they developed a remote online i‐voting system with Tata systems costing Rs 34 crore33. In the end only 124 actually voted online although those that did touted the
30 (IEEE, 2004) 31(Mercuri, 2000) The VVAT concept was pioneered by Mercuri in this thesis. 32 Ibid 331 Crore is 10 million. This is about USD 7.11 million in November 2011.
e‐ Voting: India and the Philippines – A Comparative Analysis for Possible Adaptation in Africa
Page 13 of 32
geographic mobility and expediency i‐voting provided with one student voting from Mumbai. It is clear the Indian EMBs see no contradiction in using a mixed mode operation. A further important observation is that Australia, the USA and Croatia have used remote i‐voting and have similarly experienced very little, even negligible, voter participation.
6.3 Example of Indian pragmatism The Indian state of Gujarat for its 2010 panchayat (village level or local elections) used a combination of EVMs and paper ballots. This is simply due to the insufficient number of EVMs. Gujarat needs about 200,000 EVMs for its State elections as compared to the 40,000 it has currently. Gujarat considered using its new online voting but felt this was not practical at this stage due to digital divide issues34.
Figure 3.0 – (MediaNama, 2011)35
Do the Indians find e‐voting? The next section presents some perceived benefits of e‐voting to the Indians.
7.0 What are the advantages in using EVMs?
The most important advantage is:
i) The printing of millions of ballot papers can be dispensed with because only one ballot, is required in each Balloting Unit per individual elector. This results in huge savings by way of cost of paper, printing, transportation, storage and distribution.
ii) The low cost of the machines and the reuse reduces the cost‐of‐ownership.
34 (Jha, 2011) 35 (MediaNama, 2011)
e‐ Voting: India and the Philippines – A Comparative Analysis for Possible Adaptation in Africa
Page 14 of 32
iii) The speedy, almost instantaneous counting of ballots through EVMs reduces election related tension and violence.
iv) There are no invalid votes in India under the system of voting under EVMs36. The importance of this will be better appreciated if it is remembered that in every General Election the number of invalid votes is more than the winning margin between the winning candidate and the second candidate in a number of constituencies. To this extent, the choice of the electorate will be more correctly reflected when EVMs are used37.
v) EVMs remove ballot stuffing, make booth capturing difficult, assist the rural‐illiterate elderly and make logistics in areas with poor infrastructure easier.
vi) EVMS have improved voter trust in India, is assisting in building regional stability with the deployment of EVMs to Nepal, Afghanistan and recently Egypt.
vii) The randomisation of the allotment of the machines removed the perception of machine rigging.
The above advantages point emphatically to a general acceptance of the credibility of free and fair elections using EVMs.
The next section (8.0) now describes Philippines with respect to e‐voting. It commences with a description of the Political System (8.1), The Electoral Management Board (8.2), a description of the actual Precinct‐Count Optical Scanner(PCOS) (8.3) It then describe the Indian e‐voting experience with to challenges and how e‐voting mitigates these (8.4). Like India, the Filipino experience is then critically evaluated against fifteen critical success factors (9.0). They are: Voter Registration, (2) Rural, elderly and illiterate voters (3) Access (4) Multiple Elections (5) Voter Turnout (6) Overvotes, undervotes and invalid votes (7) Voter acceptance and trust (8) Regional Cooperation (9) Vendor Selection (10) Black and white box testing (11) Security (12) Cost of elections (13) Post‐election tabulation and transmission of results (14) Accountability, recount and transparency and finally (15) Observer reaction. The Philippines section concludes with a description of the perceived advantages of PCOS (10.0). Comparative data for both countries is provided as a summary in Section 11.0. Both countries are comparatively analysed for lesson learned in Africa (12.0) and a Conclusion is provided in Section 13.0.
8.0Philippines
The Philippines is the ninth most populated democracy in the world with a population of 94 million people. It is the world’s second largest archipelago comprising a complex and logistically intimidating geographical system of over 7,000 tropical islands.
The Philippines started its efforts at automating the electoral process in 1992.It piloted the first automated election system using the Optical Mark Reader (OMR) technology in the 1996 elections and subsequently phased in implementations gradually. We refer to this as Optical Scan (OS) more
36There is a school of thought that regards the spoilt ballot as a None Of The Above Vote (NOTA) and a strategy for disenchantment (Thakur, 2010) 37 (ECI, 2011)
e‐ Voting: India and the Philippines – A Comparative Analysis for Possible Adaptation in Africa
Page 15 of 32
recently. In the 2010 elections, the Philippines conducted the country’s first nationwide fully automated elections.
The Indian EVM is a simple punch‐button vote tabulator. Voters push a button and a vote is recorded. The Filipino PCOS is an Optical Scan machine that takes the voters paper ballot, scans and stores the image, and records voter choice through intelligent pattern recognition.
Elections for all positions in the Philippines above the barangay (or village) were held on Monday, May 10, 2010. The elected president became the 15th President of the Philippines, succeeding the incumbent who was barred from seeking re‐election due to term restrictions. The legislators elected in the 2010 elections joined the senators of the 2007 elections and comprise the 15th Congress of the Philippines.
8.1 The Filipino Political system
The Philippines is a democratic republic with a President and Vice‐President. It has 80 provinces each with a Governor. The Philippines functions as a multi‐party democracy with legislative power invested in government and a two‐chamber congress: the Senate (the upper chamber) and the House of Representatives (the lower chamber). The political parties are very close in ideology which results in candidates frequently switching between parties. Local elections were held in all provinces, cities and municipalities for provincial governors, vice governors and board members, and city/municipal mayors, vice mayors and councillors.
8.2 The Filipino Electoral Management Board
The EMB is called The Commission on Elections (COMELEC). The COMELEC is an independent constitutional body created in 1940.With respect to Electoral matters; the Commission exercises not only administrative and quasi‐judicial powers, but judicial power as well.38
The 2010 election was administered by the Comelec. It was the first national computerized election in the history of the Philippines. This included the counting of votes and the transmission of election results39. Nonetheless voters still used a ballot paper which was scanned and counted automatically.
8.3 The Precinct‐Count Optical Scanner
Figure 4.0 Precinct‐count Optical Scanners (PCOS) Model 180040. This is a Linux based operating system.
38 (Comelec, 2011) 39 (Comelec, 2011a)
e‐ Voting: India and the Philippines – A Comparative Analysis for Possible Adaptation in Africa
Page 16 of 32
Every citizen above the age of 18 on Election Day may vote. In order to vote, a citizen has to register. The COMELEC has a registration period a few months prior to the election. Once a registered voter’s name and precinct are found in the voters' list, he queues for a ballot. Prior to the 2010 elections voters had to write the names of the candidates next to the positions in which they were running. COMELEC‐approved nicknames could be used by the voters in writing the names. After the polling period ends, the ballots were counted by hand. Once all the ballots are counted the election returns are sent to the COMELEC. For the 2010 election, the voters shaded the oval before their chosen candidate's name, and a voting machine automatically counted each ballot as it was fed into it. The results were then printed as the election return and sent electronically to the COMELEC.
A total of over 76,340 PCOS machines (or Precinct‐count Optical Scanners), some 5000 back‐up units and some 1700 servers were deployed in the country's first nation‐wide fully automated elections from counting of votes to transmission and canvassing of election results. There were more than 85,000 candidates for 17,000 national and local positions.
8.4 Philippines Challenges – and how PCOS mitigates them
The Philippines faced the following challenges, prior to e‐voting i) Complex island logistics; ii) Extreme violence exacerbated by delay in results and counting; iii)Ballot Stuffing; iv) Illiteracy and non‐connected areas; and v) High number of invalid votes.
i) Complex island logistics
The 7,000 island system creates a complex outbound and inbound logistic. This was the main cause of delays of inbound results resulting in tension and violence. The combination of Optical Mark Readers or Optical Scans with transmission capabilities solved this inbound logistical problem to the extent that many results were released on Election Day.
ii) Extreme Violence and delay in results and counting
Elections in these tropical islands have been characterised by extreme violence exacerbated by complex logistic which induces lengthy delays in collating results. This delay heightens tensions and rumours are inevitably fuelled feeding and spiralling violence.
Various reports place the number of election‐related fatalities at about 140 in 2004. Thakur postulated in a 2010 study that there may be a correlation between fragile democracies and the introduction of e‐voting41.
Prior to the automation of elections in the Philippines, it took several weeks to know the results of an election, which in itself was a source of social unrest, fraud allegations, and violence when final results were finally announced.
iii) Ballot Stuffing Due to the manual count system, this country routinely encountered ballot stuffing, ballot box snatching, substitution or stealing, vote padding and shaving and stuffing, and election return
40 (Smartmatic, 2011) 41 (Thakur, 2011)
e‐ Voting: India and the Philippines – A Comparative Analysis for Possible Adaptation in Africa
Page 17 of 32
falsification. This was made possible by the lengthy tedious nature of the elections42. PCOS has reduced this error due to the sophistication of the machines.
iv) Illiteracy and non‐connected areas There are many areas that have no network coverage. In addition, many areas are difficult to access and with low literacy levels. The addition of satellite phones and intensive education assisted here.
v) High number of invalid votes
In previous elections, the ballots contained blank lines above which voters wrote the names of the candidates. The 2004 elections had an invalid count of 12% or 4.3 million votes. This is a high number statistically and numerically and is perceived as a direct contributor to the distrust and tensions. In the machine‐readable ballot of the new automated system there was an oval beside each name of a candidate which the voter shades or blackens in order to register a vote
The invalid votes were also high for the 2010 elections. The values were President (5.7% or 2.01 million) and Vice–President (7.8% or 2.98 million) and similar numbers for each of the other four simultaneous elections. This was due to the complexity of the elections and confusing ballot papers.
It was the delay in results and election‐related fatalities that persuaded this country to adopt e‐voting.
9.0 Critical success factors and Filipino elections
1. Voter Registration
The voter registration requires a valid ID, address and at least 18 years of age. A form must be filled and biometrics (thumb print, photograph and signature is taken). The number of new registered voters increased year‐on‐year in 2010 by 10.5%. This was the highest increase ever, which indicates positive voter attitude and hope in the new system.
The National Statistical Coordination Board (NSCB) postulated that more than the current 38 million Filipinos would have registered and voted had the voter registration and the actual voting process been less tedious. This means security impinging on access, which demonstrates the continual conflict between CSFs as attempts are made to maximise security. Interestingly, when the COMELEC tried to simplify processes closer to elections (to be discussed later), after they gained experience, the changes were vehemently opposed. The COMELEC started an Automated Fingerprint Information System (AFIS) project in 2004 which is expected to improve the voters roll.
2. Rural, elderly and illiterate voters
The 50 million voters had never seen such a ballot in prior elections. Voters had previously filled in a blank ballot form to indicate their choice. The paper ballot to be used in the automated election system (AES) was about two foot long, with each side accommodating 300 names. The
42 (Smartmatic, 2010)
e‐ Voting: India and the Philippines – A Comparative Analysis for Possible Adaptation in Africa
Page 18 of 32
new system was confusing to the elderly and possibly even more confusing to the illiterate. This slowed voting. This shows that even with automation, pragmatic ballot design is essential.
Figure 5.0 – A Map of Philippines43
3. Access
82,200 specially configured machines were deployed to maximise geographic coverage for the elections. A full range of equipment is listed in Appendix 1. This is given so a potential adopting country will have an idea of the Bill of Materials in an electronic election.
4. Multiple Elections
The Philippines had a multiple election in 2010. There were several simultaneous elections to be voted, inter alia, the President, Vice‐President, Governors, Mayors, and councillors etc. In all 17,000 candidates were to be elected. This made for a very busy and possibly confusing ballot paper. The ballot design was not intuitive44.
5. Voter Turnout
The NSCB said the 38 million voter turnout in the 2010 elections was the highest actual voter participation since 1978, a period of more than 30 years. The 38 million Filipinos who voted represented 75% of the 50.7‐million registered voters in 2010.
43 (Graphic Maps, 2011) 44 (Telibert, 2009)
e‐ Voting: India and the Philippines – A Comparative Analysis for Possible Adaptation in Africa
Page 19 of 32
There are documented reports of long queues both during voter registration and the elections. There were also technical glitches related to malfunctioning Precinct Count Optical Scan machines, which may have discouraged some voters from voting.
The percentage of voters was statistically higher in 2004 at 77% (or 33.5 million) of registered voters. This lends credibility that PCOS was accepted by the electorate.
6. Overvotes, undervotes and invalid votes
With a PCOS an overvote occurs when two ovals are shaded or marked for a lone position of president for example. This invalidates the vote for that position. However, all other valid votes for all other positions would be counted.
Where an undervote condition exists, like voting only for seven instead of 12 senators, all the seven votes would be counted. The absence of shading or a mark in any of the elective positions is a non‐vote and there are no penalties for it. Any marks placed outside any oval is not going to be recognized by the machine and would constitute a non‐vote as far as the machine is concerned.
Both of these have occurred substantially in the Philippines and were discussed earlier.
7. Voter acceptance and trust
Social Weather Stations, an independent pollster found that 75% of the voters were “very satisfied” with the condition, speed and credibility of the election. These elections allowed the Filipino people to know who was elected president on Election Day for the first time in their history. The speed of announcing results certainly helped address post‐election violence which was very low.
8. Regional Cooperation
Filipinos were involved in skills transfer and training for voter registration in the Middle East. They have also had in‐depth discussions with the Interim Independent Electoral Commission (IIEC) of Kenya over possible electoral process overhaul. The exchange facilitated by IFES “included demonstrations of the Philippines’ Precinct Count Optical Scan (PCOS) voting system and Canvassing and Consolidation System to alleviate registration problems, voter fraud and violence that had marred previous elections.45”
9. Vendor Selection
Here the multinational Smartmatic established a joint venture with Total Information Management Corporation (TIM) a medium sized Filipino company to address the perceptions of, off‐shoring or even outsourcing democracy46. This is becoming an international practice.
10. Black and white box testing
The Filipino contracted an external neutral group who was allowed limited source code access. This was a small positive, though arguably insufficient move by the COMELEC but stakeholders demanded full access even if limited to experts.
45 (IFES, 2011) 46 (Comelec, 2011a)
e‐ Voting: India and the Philippines – A Comparative Analysis for Possible Adaptation in Africa
Page 20 of 32
Hall, in a PhD study, agrees that full source disclosure to an expertise a reasonable compromise. He argues that as white box testing is complex expert testing is a sensible reasonable compromise47. Access limited to an expert is a reasonable middle ground between code secrecy and code openness. This will help to counter both government concerns of hackers and civic organisations concerns about subversion through software.
11. Security
The PCOS from Smartmatic/TIM operates under a similar non‐discloser methodology as the Indian EVM. This is a source of much discontent and is discussed in this paper. An Ubuntu forum already claims that full access to PCOS is possible through a serial cable and a laptop48.
12. Cost of Elections
The computed cost of a vote in May 2010 election was at PhP486 or about USD10. This is computed by a publicised cost of PhP17 billion divided by 38 million voters49. Telibert (2011) reveals this cost is misleading as it does not include the costs borne by civil society organizations doing support work to the Comelec in voter information, outreach to vulnerable sectors, reporting on election‐related violence and others that are not covered by state funds. The machines are a one‐off cost and may be reused.
13. Post‐election tabulation and transmission of results
The solution is designed to transmit electoral data securely and unidirectional with no need for human intervention which is a requirement of the COMELEC. Wherever transmission is not possible, compact flash memory units are carried to the respective municipality and the results are imported into the canvassing system.
After the elections closed and transmissions from PCOS machines began arriving en masse, the COMELEC was able to publish the first partial results. Many former doubts and concerns vanished, replaced by astonishment due to the unprecedented speed of the tally. 57% of the results were in by midnight of the first day50.
14. Accountability, recount and transparency
The use of paper ballots makes the Filipino choice more accountable than India because it is possible to undertake an automatic recount on a different machine or undertake a manual recount. This is a useful feature in really close elections.
The optical scanner records a digital image of each ballot. It adds this record along with a log entry to clearly indicate how the system processed that specific ballot. This reduces the error margin and allows for better auditing and accounting. The optical scanner unit is designed to register and tally with maximum security the votes marked on electoral ballots51.
Although there were a few cases of PCOS machine failures, there was no postponement of elections since most technical issues were resolved by Election Day. Despite the fact that some provinces have reported failure of elections, these have not surpassed the 0.50% of the total
47 (Hall, 2008) 48(Ubuntu Forum, 2010) 49This figure is disputed by the vendors who indicated a figure of USD 150 million. 50 (Smartmatic, 2010) 51 (Coloma, 2010)
e‐ Voting: India and the Philippines – A Comparative Analysis for Possible Adaptation in Africa
Page 21 of 32
number of PCOS machines and most were replaced on time. As a result of the delays, the COMELEC extended voting hours from 6:00 pm. to 7:00 pm.
15. Observer Reaction
The saying that "guns, goons and gold" lord it over Philippine elections may no longer be totally true after a new weapon, the microchip, entered the scene. Election Day had live full coverage from Radio and TV52.
2.1. Positive Observer Reaction The United States and the European Union praised the republic for the smooth elections. The US embassy was one of the first to hail the general elections. The EU did not send an observer mission. However, EU Ambassador Alistair MacDonald described the patience and the number of people who turned up to vote in the elections53.
He “was impressed by the manner in which this first nationwide automated election was conducted. Despite the intense heat, the long lines and the inevitable unfamiliarity of a new process, our observations suggested that this process was carried out smoothly, and the results transmitted rapidly, in the great majority of cases. He also appreciated the teacher's hard work in facilitating the elections54.
2.2. Negative Observer Reaction The following is an important deviation and is presented in some detail due to its importance and to show the level of scrutiny an e‐election undergoes. The observers here are Filipinos ex‐pats who advocate responsible technology adoption. A report entitled GFN 2010 Election Observers Team55(2010) presented arguments questioning the May 10, 2010 elections as summarized below:
1. The election results transmitted from the precincts did not have digital signatures of the Board of Election Inspectors (BEI).
2. The number of disenfranchised voters was sufficient to affect greatly the results of the elections.
3. The Automated Election System (AES) was implemented without the appropriate field testing, and law‐specified testing in actual elections.
4. The source code review was not completed and initial findings were not addressed.
5. No audit was done on the AES prior to the elections. There was only a mandated random manual audit which was not yet done at the time the report was written (May 27, 2010).
6. Several voter and security features were disabled prior to elections.
Due to the exact nature of the arguments they are considered in detail below.
There has been a noticeable improvement in the peace and order aspects of the elections compared to past national elections. There was initial satisfaction with the early voting results. But later events put to question the authenticity, integrity, confidentiality, veracity and accuracy of the vote. Thus, no one (both perceived winners and losers) can be sure whether the vote results are true and correct and reflect the real will of the Filipino people. Accordingly, the Election Observers Team of Global Filipino Nation challenges the legitimacy of the election
52 Ibid 53(Lee‐Brago, 2010) 54 Ibid 55 (CenPeg, 2010)
e‐ Voting: India and the Philippines – A Comparative Analysis for Possible Adaptation in Africa
Page 22 of 32
results.
To resolve this very critical issue, GFN recommended the following:
1. Impound the PCOS machines, the memory and CF cards, and perform forensics on these using the actual ballots.
2. An independent non‐partisan qualified party conduct a full‐blown audit of the Automated System. The audit must cover compliance with law, reasonableness of pricing and expenses, technology evaluation, system evaluation, and staff evaluation (TIM, Smartmatic and Comelec).
3. With the lessons learned in the automation of 2010 the following projects should be pursued in time for the 2013 elections:
a. As a priority complete the computerization of the Automated Fingerprint Information System (AFIS), started by Comelec several years ago, to complete and purge the Registered Voters List.
b. Complete the computerization of the Voters Registration Information System (VRIS) and that of the Project of Precincts (POP) in order to prevent disenfranchisement, "flying and ghost" voters and "ghost" precincts.
The next section provides a short list of perceived advantages to the Filipinos.
10.0 What are the advantages in using PCOS?
The most important advantage is:
i) The full penetration of the machine across the country.
ii) The speedy almost instantaneous counting of ballots through PCOS which reduced election violence. Results were released within 48 hours.
iii) PCOS remove ballot stuffing, ballot theft, intimidation and made in‐bound logistics in areas with poor infrastructure easier.
iv) PCOS have improved voter trust in Philippines with improved Voter Registration rates. It is assisting Kenya and the Middle East.
The above advantages point emphatically to a general acceptance of the credibility of free and fair elections using PCOS.
The next section tabulates some comparative data for convenience.
11.0Summary
Table 3.0 Summary
India Philippines
Political climate Relatively stable with isolated incidents during elections
Volatile and tense during elections
e‐ Voting: India and the Philippines – A Comparative Analysis for Possible Adaptation in Africa
Page 23 of 32
Time from Trial to full‐adoption
12 years
18years
Procurement Practice
Two Government owned public companies
International/local Joint Venture or partnership
Economy (GPD)56 9th 43rd
Size and demography57
1.3 billion (17.4 %) 2nd
World’s largest democracy
94 million(1.4 %)12th
World’s second largest archipelago. It has over 7000 tropical islands
Ballot Paper used 8800 metric tonnes before EVM, now almost paperless
1500 tonnes with OMR. This is significant compared to India’s 8800 tonnes used in 1996 for 671 million voters
Machine Type EVM/DRE without paper Paper ballots, with Optical Mark Reader
Code type Embedded microcode Open source but secret
White Box Testing No Limited
Black Box Testing Yes Yes
Cost Approx 1303 Crore (this excludes equipment purchase)
P 7.1 billion (this includes equip purchase –once off)
Voter Registration Introduced EVM (full) in 2004, then introduced Photo registration in 2005
Started AFIS in 2004, but got OMR s implemented in 2010
Rural and elderly experience
Positive – better than paper Mixed – ballot paper was large at two feet and confusing
Access 1378 EVMs deployed – still insufficient and election held in 5 phases over 4 weeks
82200 PCOSs see Table 2.0 for full list of resources
Multiple Elections Possible, but not implemented 2010 election was multiple
Turnout 60% ± 2% typically 75% highest numeric number in history
Vendor
Two Indian companies asked to jointly develop the solutionvizECIL/BEL
Tender won by a JV between local TIM and the International Company Smartmatic. Smartmatic/TIM
Full Use 2004 2010
56 (World Bank, 2010) 57 (US Census Bureau World Population Clock, 2011)
e‐ Voting: India and the Philippines – A Comparative Analysis for Possible Adaptation in Africa
Page 24 of 32
Government Coalition Majority
CIVIC organisation opposed to e‐voting adoption
VeTa (Qualified opposition Demands VVAT)
CenPeg58Demands more diligence to process
Overvotes Not possible on EVM Possible due to ballot
Undervote Possible Possible
Invalid votes Not possible Very high in 2010 (±10%)
Voter acceptance Yes 3rd election with EVM Poll indicated a75% acceptance
Human Resource Deployed
4 million officials and 2.5 million security officials
48,000 trained Technicians, and 904 Smartmatic experts
Transmission method
The control units do not electronically transmit their results back the Election Commission, EVMs are purposely designed as stand‐alone units to prevent any intrusion during electronic transmission of results. Instead, the EVMs are collected in counting booths and tallied on the assigned counting day(s) in the presence of polling agents of the candidates. Tallies are sent and 30 copies printed
Unidirectional transmission after elections
Regional Corporation
Nepal and Afghanistan (with an offer to Egypt)
Middle East and Kenya have been assisted
Having considered India and the Philippines the next section undertakes a detailed evaluation of each country and its context and attempts to transpose this to the African perspective.
12.0LessonsforAfrica
It is difficult to make a blanket recommendation to all and indeed any of the 53 African states simply because this great continent is in development mode. There are a huge number of social projects with higher priorities that compete for the same limited funding.
That noted, each country is different and will have sufficient nuances to make their election unique and special. So it should be. The paper makes salient points, about the Indian and Filipino experience that one should consider when embarking on this journey.
58Centre for People Empowerment in Governance
e‐ Voting: India and the Philippines – A Comparative Analysis for Possible Adaptation in Africa
Page 25 of 32
1. Voter Registration
Sophisticated Voter Registration drives using such technologies as biometrics is increasing in many African countries. Many of countries receive additional support from the UNDP. As in the Philippines, equipment used in voter registration drives may be reused in other voting activities such as the e‐voting cycle or the results transmission cycle.
One does not need an automated voter registration before adopting e‐voting. This is an interesting observation as the Indians simply adopted e‐voting and then started modernising the voters roll as a parallel activity which is now being completed. A country can introduce the e‐counting or tallying aspect of e‐voting as the Philippines did and perhaps work simultaneously on its AFIS system. In Africa many countries have already resorted to automatic transmission of results.
Thus a process towards adopting electronic voting could start immediately with parallel activities such as testing, trials, and piloting.
2. Rural, elderly and illiterate voters
2.1 Ballot Design
The ballot design was criticised by many stakeholders of the Filipino elections. The invalid votes contributed between 6‐12% of the total votes. Ballot Design, regardless of the medium is an often overlooked area of improvement for a comparatively low investment. Africa certainly has the experience here.
Ballot Design is possibly easier with EVM than ballot paper. Some EVMs have optional sound and multi‐lingual facilities. India, had an almost non‐existent invalid vote due to EVMs.
2.2 Elderly and disabled
The Indian experience shows that the elderly and the illiterate actually benefit from simple EVM design. Machines may, for example, be configured with sound and Braille for the blind.
3. Access
The all‐or‐none‐law has to apply if one decides to use e‐voting. The Indians used elephants, camels, and all possible forms of transport to get the EVMs to deep‐rural areas. Off‐grid areas were mitigated by the use of 6V batteries. The Filipino used, at great cost, satellite phones and modems where 2G/3G connectivity was not possible. This is a non‐negotiable design consideration. The entire country must be considered, not footprints or ‘areas of opportunity. ’The EMBs of India and the Philippines ensure that the entire country can have full e‐voting access. This inclusivity builds trust with the electorate.
4. Multiple Elections
It is possible to conduct multiple elections using EVMs and PCOS or any voting technology for that matter. However extreme care must be taken to design the input medium (paper or screen) so that it does not add to voter confusion.
The Filipino experience was a source of frustration to the voters as they battled with large intimidating ballot papers.
e‐ Voting: India and the Philippines – A Comparative Analysis for Possible Adaptation in Africa
Page 26 of 32
5. Voter Turnout
It is important to point that e‐voting internationally has not led to an increase in voter participation59. The Indian and Filipino experience is a little different. The Philippines had numerically the highest number of voters at over 38 million, though not by percentage. It is also probable that India by using e‐voting sustained the voting numbers which in itself is a considerable achievement in the face of international declines. If one wants to increase participation then perhaps education or other forms of motivation may be considered as well.
On the other hand, voter interest and participation may be reinforced by the knowledge that their vote will count and be difficult to tamper with.
6. Overvotes, undervotes and invalid votes
These three figures have been found to consistently reduce with e‐voting. For example invalid votes reduce to zero60 with India’s EVMs. However this was not the case with PCOS as the invalid votes were still high as discussed in the Filipino case.
Push button EVMs remove errors by enforcing a legal choice. Undervotes will (and perhaps) should be allowed to occur. On the other hand Overvotes are easily prevented on EVMs but not PCOS.
Awareness of the design requirement will influence the technology choice. A middle of the road approach which will certainly be a less paper intensive and greener may be an EVM with a VVAT, which will be less paper intensive. India is piloting such a system now and African countries should evaluate other countries experiences in this process.
7. Voter acceptance 61 and trust
Both countries had a reasonably long period of between 10‐18 years before full scale implementation of e‐voting. This helped to build the awareness both within the EMB (as the champions) and stakeholders (the political parties, the electorate, civic and media). It is suggested that the road to automation must be a phased, deliberate and transparent process.
The Irish found that imposing a voting system will lead to rejection by the electorate. This is an important lesson learned. In India, the group Verifiability, Transparency and Accountability in Indian Elections (Veta) is against the EVM as it is implemented, not e‐voting per se. In the Philippines, the CenPeg‐led opposition is over compliance and deviations from procedures not e‐voting per se. Most researchers are against the adoption of DREs or EVMs without VVAT. This is the issue.
This boils down to auditability and the repeatability of the count of the ballot which is why the paper trail becomes so important. Voter education is paramount to build this trust and make election day efficient. This was carefully followed in India and less so in the Philippines which is why the queues got long.
8. Regional adoption
There is emerging evidence of innovative countries adopting e‐voting and then becoming regional champions.
59 (Oostveen, 2011:144) This experiment showed that e‐voting will not increase participation. 60Some literature argue that an EVM removes a (None Of The Above) NOTA option. 61 (Oostveen and van den Besselaar 2005:11)
e‐ Voting: India and the Philippines – A Comparative Analysis for Possible Adaptation in Africa
Page 27 of 32
It may be possible for Africa to use regional cooperation, as the start to study and trial e‐voting in respective member countries context(s). This has resonance in Africa as she has the regional blocks: namely Southern African Development Community (SADC), Common Market for Eastern and Southern Africa (COMESA) and Economic Community of West African States (ECOWAS). This may assist to co‐fund elections.
9. Vendor Selection
There is a fundamental difference in the technology adopted: The Philippines used a ballot paper along with an optical scanner while India uses a poll‐site paperless EVM. Needless, the amount of paper used in Philippines was significant. The 50 million ballots required 1 500 tonnes of paper. However, the requirement of auditability and accountability was fundamental in the Philippines. The negative fearful consequence of loss of lives for non‐provability of results was (and remains) sufficient reason to use ballot papers. This shows that context does matter62.
Both India and the Philippines have local companies with India in‐sourcing to two state‐owned public enterprises and the Philippines awarding a Joint Venture between a international company and a local Filipino company TIM.
This is also an important deliberate step to show to the electorate that the democratic process is not outsourced or even off‐shored.
10. White Box and Black Box Testing
The election‐ware and code must be evaluated by independent non‐partisan experts. The evaluation and corrective recommendations must be audited and monitored. This act would have satisfied the 2010 Filipino Election Observers Team.
This suggestion is made even if the option is limited to expert companies and individuals only. The Filipinos have limited disclosure, whilst the Indians have no disclosure. So limited or no white box testing occurs in these countries which is a source of debate in both contexts.
Full source code disclosure is an important step in winning voter trust, even if they do not understand programming. This will help to counter both government concerns of hackers and civic organisations concerns about subversion through software. A key feature of a digital democratic electoral system must be transparency through certification and validation: that what went in (the vote cast), stays in (the vote is stored not altered) and is counted (the correct vote is counted). The Australian system, though discontinued due to cost reasons is cited as the most open transparent election system. The Dutch discontinued e‐voting due to the non‐certification of machines.
11. Security
This reduces to risk management. EVMs elevates security threats to a more subtle difficult to trace
form.
If the logistics involved in the Indian elections are considered, then it seems to tip the scale in favour
of the EVM.63The cost is very low. This logic reduces the number of potential criminals but
introduces a more sophisticated criminal.
62 This is expounded articulately in seminal thesis called Context Matters (Oostveen, 2007) 63(Schneir, 2011)
e‐ Voting: India and the Philippines – A Comparative Analysis for Possible Adaptation in Africa
Page 28 of 32
The PCOS offers multiple verification, through hard copy paper ballot, soft copy stored image, and
PCOS audit trails. The soft copy stored image offer the feature of being able to analyse for sustained
ballot stuffing or other fraudulent activities. Where there is intense even fatal distrust, the PCOS is
method that is recommended.
12. Cost of Elections
The Indian EVM at $170 is an easy economic choice, but even India country is evaluating other methods such as EVM with VVAT and Remote Internet Voting. It also must be remembered that the 6V battery, while a green solution, introduces limitations on the design of the voting unit, with respect to functionality (screen size, compact drive, etc.) and duration.
The Philippine solution added an enormous once‐off cost to the election for the purchase of the PCOS and related hardware. It does seems that the vendor hardware and software choice is context driven, that is, one must understand the social‐economic, legislative and political and technological situation of a country before embarking on a choice.
13. Post‐election tabulation and transmission of results
Even with manual or automatic counts, a desirable feature is the electronic transmission system for the results, which will greatly speed tallying and reduce tensions64. The Indians automated the voting and counting, and then downloaded the results to a flash disk for transmission.
The post‐election reduction in tension and anger in both countries is palpable and proudly celebrated by both EMBs.
14. Accountability, recount and transparency
Every solution must emerge and evolve from with in a country. This drives the transparency and accountability component. The technology must not be elevated to a black box, but rather simplified to explain and even consult to the electorate as the choice is made.
Australia did this by setting on‐line portals and even published the software code. The public made the software code stronger by finding errors!
A statistical audit of a random number of machines must be undertaken to determine for evidence of ‘organised or co‐ordinated’ electronic ballot stuffing. This will have other benefits to reveal other sources of errors which may be systematic or systemic error and failures65.
Continual environmental scan The fact that India a confirmed committed adopter of e‐voting and continues to experiment with EVMs with VVAT and i‐voting shows that e‐voting technology adoption requires constant evaluation and testing so that the trust of the nation is retained and transparency is maintained.
The next country provides a short summary of lessons learned from these two countries.
Summary
Having said that, a country seeking to adopt e‐voting, should:
64 This has been successfully introduced in Zambia and South Africa. 65(Thakur , 2010)
e‐ Voting: India and the Philippines – A Comparative Analysis for Possible Adaptation in Africa
Page 29 of 32
i. Go through a public transparent decision and consultative process ii. Undergo trial(s) whose outcomes must be clearly communicated iii. Undergo pilot(s) before embarking on full elections iv. Endeavour to include a local technology partner v. Get the code verified by a non‐partisan company – White Box Testing vi. Get the machines verified by (preferably) another company in the presence of stakeholders
Black Box Testing vii. Have built in protection to prevent ballot stuffing viii. Preferably have no transmission during elections, and have unidirectional transmission to
post result. ix. Have parallel audits on Election Day x. Be inclusive and transparent.
13.0Conclusion
It is clear that electronic voting is an emotive topic. Hall and Alvarez (2004) argue that “passions can
overcome rational debate” and that we should move to a level of scientific study so this debate
becomes more rational66.
E‐voting really comes to the fold when there are close elections, where the degree of mistrust is
high, electoral fraud is rampant; and accuracy becomes paramount with deadly consequences. E‐
voting should be implemented in a consultative, open, inclusive, and transparent manner.
The vendor, and voting technology, must be informed by the socio‐political, technological and legal
framework of the country making the decision.
India demonstrated that a country with mass illiteracy and poor grid‐connectivity can implement a
massive e‐voting system. The Philippines showed that with incredible logistical challenges and
constant threat of violence, an acceptable implementation is possible. The differences between the
implementation strategies are based on one important word: “context.”
Reference
Alvarez, R.M. and Hall, T.E. (2004) Point, Click, and Vote – The Future of Internet Voting. Brookings Press
CenPeg. (2011). Interim report by GFN Foreign Observers Challenge Election legitimacy, Centre for People Empowerment in Governance, Available at: http://cenpeg.org/2010%20layout%20pages/POL%20PARTIES%20AND%20ELECTIONS_2010%20layout/may%202010/Foreign%20Observers%20Challenge%20Election%20Legitimacy.html [Accessed on 8 October 2011]
Coloma, R.(2010). Automated voting jolts Philippine politics into digital age, Agence France‐Presse (AFP)
Comelec. (2011a).The Commission on Elections, Available at: http://www.comelec.gov.ph/aboutus_/bkground.html [Accessed 29 October 2011]
66 (Hall and Alvarez, 2004)
e‐ Voting: India and the Philippines – A Comparative Analysis for Possible Adaptation in Africa
Page 30 of 32
Comelec. (2011b). Automation of the 2011 Elections, The Comelec, Available at: http://www.comelec.gov.ph/modernization/automation_of_the_election_process.html [Accessed on 29 October 2011]
ECI Resolution. (2011).Available at: http://www.indianevm.com/resolution.php[Accessed on 29 October 2011]
ECI Correspondence.(2009). ELECTION COMMISSION OF INDIA, Information under RTI on EVMs, No. RTI/2009‐EMS/ 39, 21st July, 2009
Election Trivia. (2011).Election Trivia, Available at: http://www.indian‐elections.com/election‐trivia/[Accessed on 29 October 2011]
Graphic Maps (2011). Map of Philippines, GraphicsMap.com, Available at: http://www.omeriani.com/maps/philippines‐map.html [Accessed 31 October 2011]
Hall, J,L. (2005). Policy Mechanisms for increasing Transparency in Electronic Voting. PhD Thesis. University Of California at Berkeley Information management and Systems
IEEE Spectrum. (2004). Electronic Voting Eases India Elections, IEEE Spectrum, Available at:http://spectrum.ieee.org/computing/software/electronic‐voting‐eases‐india‐elections [Accessed 20 October 2011]
IFES. (2011). Kenya, Philippines Share Lessons in Election Technology and Management , IFES Voting Technology and Election Management, Available at: http://www.ifes.org/Content/Publications/News‐in‐Brief/2011/August/Kenyan‐IIEC‐Explores‐Elections‐Technology‐Management‐in‐the‐Philippines.aspx [Accessed 31 October 2011]
India Profile. (2007). "Country Profile: India", Library of Congress ‐ Federal Research Division, December 2004
Jha, P. (2011). Gujarat panchayat poll: Ballot boxes back in fashion, Daily News and Analysis (DNA), 1 August 2011
Kapur, A (2010). In India, Sometimes News Is Just a Product Placement, New York Times, Asia Specific, 7 May 2010, Available at: http://www.nytimes.com/2010/05/08/world/asia/08iht‐letter.html?pagewanted=print [Accessed 20 October 2011]
Kerala (2010).Electronic Voting Machines. Chief Electoral Officer, Kerala , Available at: http://www.ceo.kerala.gov.in/evm.html[Accessed18/11/2011]
Lee‐Brago, P. (2010). US, EU praise RP for milestone elections, The Philippines Star. 12 May 2011
Iztúriz, M. (2010). Smartmatic. Available at: www.smartmmatic.com [Accessed 1 November 2011]
MediaNama. (2011).Gujarat Implements e‐Voting For Municipal Elections; News and Analysis of Digital media in India, Available at: http://www.medianama.com/2011/03/223‐gujarat‐implements‐e‐voting‐for‐municipal‐elections‐1582‐e‐voters‐register/ [Accessed 20 October 2011]
Mercuri, R. (2000). Electronic Vote Tabulation Checks and Balances. PhD Thesis. University Of Pennsylvania, School of Engineering and Applied Science, Department of Computer and Information Science
NarasimhaRao, GVL. (2010). Democracy at Risk! Can we trust our Electronic voting machines? Veta Publications, New Delhi
Oostveen A.M. (2007). Context matters, A Social Informatics Perspective on the Design and
Implications of Large‐Scale e‐Government Systems. University Of Amsterdam. Phd Thesis. University Of Amsterdam
e‐ Voting: India and the Philippines – A Comparative Analysis for Possible Adaptation in Africa
Page 31 of 32
Oostveen, A. and P. van den Besselaar (2005) The Effects of Voting Technologies on Voting Behaviour: Issues of Trust and Social Identity. In: Social Science Computer Review, Vol.23 No. 3, Fall 2005, p304‐311, Sage Publications
Pai Panandiker, V.A. (unknown). Problems of the Electoral Process in India, Democracy and Empowerment in South Asia pp 82‐85
Pieters, W and van Haren, R (2007) "Temptations of turnout and modernisation: E‐voting discourses in the UK and The Netherlands", Journal of Information, Communication and Ethics in Society, Vol. 5 Iss: 4, pp. 276 – 292
Pieters, W. (2007).La Volont´ e Machinale Understanding the Electronic Voting Controversy, PhD Thesis, Radboud University Nijmegen
Prasad, H.K. ,Halderman, J.A. , Gonggrijp, R. , Wolchok, S. , Wustrow, E. , Kankipati, A. , Sakhamuri, S.K. andYagati V. (2010). NetIndia, P. HyderabadThe University of Michigan Security Analysis of India’s Electronic Voting Machines
Schneier, B. (2010).Security Analysis of India's Electronic Voting Machines, Schneier on Security.3 May 2010, Available at:http://www.schneier.com/blog/archives/2010/05/security_analys_2.html[Accessed on 16 October 2011]
Saba, S. (2011). Smartmatic Blog, 2011.Available at: www.smartmatic.com[Accessed on 29 October 2011]
Smartmatic (2010).Dossier: General Election 2010, The Philippines, Available at: www.smartmatic.com[Accessed on 29 October 2011]
Smartmatic (2011a).Case Study: The Philippines ‐ General Elections 2010, Available at: http://www.smartmatic.com/case‐studies/view/article/the‐philippines‐general‐elections‐2010/, [Accessed on 29 October 2011]
Smartmatic (2011b). The SAES Voting Machine, Picture, Available at: http://www.smartmatic.com/case‐studies/view/article/technology‐the‐philippines‐general‐elections/ [Accessed 20 October 2011]
Thakur (2010).Unpublished PhD Thesis. Durban University of Technology
Thakur (2011).E‐voting for good governance and a green world. Africa Digital Week 2011 Proceedings
The Hindu Business (2011). India to offer EVMs to Egypt during elections, The Hindu Business Line Available at: http://www.thehindubusinessline.com/industry‐and‐economy/government‐and‐policy/article1718000.ece [Accessed 28 April 2011]
The Register (2008). UK confirms e‐voting death. The Register. 27October 2008
Ubuntu Forum. (2010). The PCOS machines' Linux OS can be accessed through a serial port. Ubuntu Forums. Available at: http://ubuntuforums.org/archive/index.php/t‐1518340.html [Accessed 18 November 2011]
Additional reading Section
Key terms and definitions
e‐ Voting: India and the Philippines – A Comparative Analysis for Possible Adaptation in Africa
Page 32 of 32
Appendix1
Table 2.0 – Bill Of Materials: Equipment deployed in Philippines
EQUIPMENT ITEM QUANTITY NOTE
SAES‐1800 counting machines 82 200 Each has a battery guaranteed for 16 hour for continuous backup operations
Canvassing and consolidation servers and printers
1 722 Each has own printer and power generator
Paper Rolls 338 750 Paper Rolls
30 copies of election returns per precinct
High‐Speed digital printers 4 delivered to COMELECfor ballot printing
Ballot Paper 1 500 metric tons For the voters
Ballot Paper 50 million With security marks with invisible ultraviolet mark and unique barcode
Ink 9 380 litres For voting process
Voting centres 65% With constant and reliable coverage through the public mobile networks.
Mobile Satellite Antennas 65% of 5 500 Deployed nationwide for transmission of results and the polling and canvassing centres after closure VSAT (Very Small Aperture Terminal) 680
Modems 48 000 For direct transmission of election returns
SIM Modules 46 000
Data Centres 2 Redundant nationwide backup sites
SAES‐1800 counting machines 5 853 Surplus for Redundancy
Results predicted 48 hours After close of election