Wednesday 25 July 2012

BIOMETRICS ELECTIONS IN ZIMBABWE: A MARCH TOWARDS TRANSPARENT AND CREDIBLE ELECTIONS


Background

 Transparency and accountability in election processes can be enhanced by harnessing and utilising biometrics technology. Biometrics, refers to computerised automatic identification of people based on how they look (physical characteristics, for example, fingerprint and face), and how they behave (behavioural characteristics, for example voice and signature).



Few elections in Zimbabwe and Africa pass without charges of vote rigging and manipulation. This article advances the argument that embracing new, simple and cost-effective technology can provide a solution to some of the problems afflicting the electoral process.

Concerns have been raised in past elections about ‘zombie’ (deceased voters apparently ‘voting’ from the grave), individuals engaging in double or multiple voting and inflated voting figures. It is important that allegations and the incidence of fraud, double or multiple voting, etc raised in the last presidential and parliamentary elections are not repeated in the next crucial elections.

A growing number of countries such as Ghana, Zambia, South Africa, Nigeria, Namibia, and Mozambique among others have either started using biometrics in the election process or are preparing to do so in the near future. The Zimbabwe Electoral Commission, the body charged with the constitutional role of conducting and supervising elections in Zimbabwe, should seriously consider investing in biometrics technology, particularly when it takes charge of voter registration and maintenance of the voters’ rolls.  

The election process does not begin and end on the actual polling day. The polling day might be the climax of the process but for every individual voter, the voting process actually begins at the point of registration to vote. At that point, new eligible voters register for the first time or existing voters verify that they are registered. This is an understated but critical stage of the voting process because unless one is registered he or she will not be able to exercise his or her constitutional right to vote. The main rationale for registering is to confirm the eligibility of the person as a voter and to ensure the correct identity of the voter. It means that only the person who is registered and whose details correspond to the details on the voters’ roll can vote.

In the past there have been problems with the voter registration exercise and the voters’ roll. There has been criticism that the voters’ roll is a shambles – persons who are long deceased are still on the voters’ rolls; the details of voters are sometimes missing or incorrect – the accuracy of the voters’ roll has been widely questioned. The result has been allegations of ‘ghost voters’ and the disenfranchisement of otherwise eligible voters who discover on election day that they are not registered; that their details are missing or incorrectly recorded, etc. This is a clear denial of the constitutional right to vote.

It follows, however, that voter identity and accuracy of voter details are critical aspects of the voting process which can affect the entire election. As this paper argues, establishing the correct identity of the voter can be easily and cheaply solved by investing in biometrics technology.   The right to vote is a very basic right for which the liberation struggle was waged – therefore it is vital that it be facilitated and protected from fraudulent conduct.

A new voter registration and voting system based on biometrics can and should be introduced in Zimbabwe. It is important to give a brief and basic overview of biometrics and how it works.

Biometrics Overview


As human beings we easily recognise each other by observing the way we look and processing this information in our brains. We do so sub-consciously by collecting a variety of information, processing it in our brains and reaching a conclusion about the identity of individuals. We gather information about for example, a person’s height (‘murefu/mupfupi’ – short/tall’), body size (‘mukobvu/mutete’ – fat/thin), even about ears/eyes size (ane mazinzeve mahombe/ ane maziso mahombe’ – he has big ears /large eyes) etc which enable us to recognise individual. Biometrics is simply an attempt by a ‘computer’ to do similar things.

However, in order for the computer to do so, it has to be presented with information regarding an individual’s physical or behavioural properties in a language which it understands such as numerical distance between the eyes, the depth/size of the nose, size of the mouth, etc. Of course, more complex data is gathered in practice, but it is not the aim of this article to go deep into the complexities of biometrics. 



Biometrics - identification based on distinctive personal traits, has the potential to become an irreplaceable part of any identification system in all spheres in the long term. Biometric identifiers cannot be shared, misplaced, and they intrinsically represent the individual's identity. This is technology which many have now embraced and is geared towards making society safer, to minimise fraudulent activities and improve convenience.

In general and which is important for our present purposes, biometrics can be used for positive identification, that is, to prove that an individual is who they claim to be. It can also be used on a large scale to verify whether the person is in the database or not.  Biometrics can also be used for screening people, for example, to check whether someone is on a police “wanted list” or to prevent/allow access to facilities. The simplest and most basic form of biometric screening can be seen at a nursery school where security of children is of utmost importance. The door to the nursery playing area would be high enough to allow toddlers to run through, but would be too short for an adult to pass through. The biometric trait being utilised here is ‘height’.

The key step in the biometrics process is for a user of the system to be enrolled (registered) by having their biometrics captured and stored. This essentially means giving the computer the physical measurements of the individual to enable it to recognise the individual in the future. This can be the individual’s fingerprint, iris of the eye, face or voice. Positive identification (also called authentication or verification) verifies the authenticity of the identity claimed. For example, a person claims that he is Nelson Mandela to the authentication system and offers his fingerprint; the system then either agrees or disagrees with the claim by comparing what is in store and what has been presented. Now, in the voting process, this technology can be applied to eliminate the issue of ‘ghost’ or ‘zombie’ voters where a living person tries to vote using a deceased person’s identity, therefore enabling him/her to fraudulently cast multiple votes.

For checking whether a user is in a database, an input biometric sample is presented to the system which determines if the pattern is associated with any of a large number of enrolled identities. In this voting process, this can answer the question of whether or not one is a registered voter.

Biometrics can also be used for screening users. Screening applications can covertly and unobtrusively determine whether a person belongs to a ‘wanted’ list. In the voting process, this can be used to determine whether one is prohibited from voting – for example because he or she is disqualified from voting for any reason.

Some examples of usage of biometrics in modern gadgets are computers/phones were you can login using your fingerprint, cars which you can open and start using your fingerprint and phones which can recognise your voice. A lot of advanced countries have also now adopted biometric passports which have a chip containing an individual’s face and fingerprint information. It makes it easier to check, verify and establish identify more efficiently than using the traditional procedures.

Biometrics Devices

In general, biometric devices consist of a reader or scanning device (this captures the biometric information, a simple example being a camera), some software that converts the scanned information into digital form (information or in a language that can be ‘understood’ by a computer’) and a database that stores the biometric data for comparison (to later compare and prove that the individual presenting the biometric information is indeed the person whose information is stored).




Simple/cheap and portable fingerprint scanners


To convert the biometric input, a software application (a computer program) is used to identify specific characteristics of the data (such as distance between the eyes, skin texture, distance between eyes and centre of nose etc for example in face recognition) as match points. The match points in the database are processed using a computer program that translates that information into a numeric value. The database value is compared with the biometric input the end user has entered into the scanner and authentication is either approved or denied. The level of security can be changed by adjusting certain values in the computer programs to make it more difficult for impostors to be accepted by the system depending on where the biometric system is used.

Biometrics in Action


The regular reference to computing technology might cause some to imagine that this is implausible and impracticable in the context of a developing country whose level of technological sophistication is limited. I will conclude this part with a few comments on how good and effective the technology is.

Although this might be overlooked fingerprint technology has been widely used for years in electoral process and policing. Error rates vary considerably between vendors and a small percentage of people are unable to use these systems at all because of unsuitable fingerprints. The main advantage of this method, however, is that people have multiple fingers, each with a different fingerprint. By requiring the use of multiple fingerprints, error rates can be reduced for those able to use the system. Experiments have been carried out using fingerprints and the number of genuine users falsely rejected on multiple attempts can be as low as 1 in 100 000 cases. The chances of imposters being accepted can be as low as 0. One has to be really determined to cheat the system to try several times for acceptance, especially given that measures will be in place to arrest and prosecute such individuals, therefore the chances of that happening can be as low as 0%.

Whilst working for the University of Kent, the writer took part in a 3-year multi-national EU Project on 3D Face Recognition for automatic border control. The project involved several key players in the security technology industry including Sagem Sécurité, Bundesdruckerei (German Security Company), Philips Research, Cognitec, L-1, Polygon Technology, Fraunhofer IGD, Hochschule Darmstadt, Computer Graphics Centre, University of Twente, Berlin Airport, National Research Council, Bundeskriminalamt (German Police), Salzburg Airport and Joint Research Centre. More information regarding this project can be obtained from www.3dface.org.

3DFace Project: Liveness testing at Salzburg Airport
In this project 3D facial data was collected at the University of Kent, at Sagem in France and the Centre for Graphics and Computer Vision in Germany. Practical trials were carried out at Berlin Airport and Salzburg Airport which included “liveness tests”, that is testing whether someone had presented a proper picture or a dummy. The results showed that the chance of someone being falsely rejected by the system was less than 2% and for someone being falsely accepted by the system was less than 0.25%. This was by using face information only in an automated way; results which can be improved by human intervention. Improvements can be obtained by combining information from different biometrics or different algorithms to reach a decision on whether to accept or reject a voter

Prof  Raymond Veldhuis (Biometrics Expert; Netherlands),  Alexander Nouak (Director: Computer Graphics Centre, Germany) and Dr Samuel Chindaro at the EU Commission Biometrics Review meeting in Salzburg

Adopting this technology can lead the way for integration of Electronic Voting, which is a different topic beyond the scope of this article suffice to state that it can help to improve speed in the electoral process through facilitating instant counting of votes and therefore elimination of the risk of manipulation. For example, an electronic voting system was launched in Brazil back in the middle 1990’s, rapidly becoming an international benchmark because of its accuracy, accountability and security. The results of the elections of 2008, when more than 130 million Brazilian citizens in 5.560 municipalities voted, were officially informed just 3 hours after its closure.

In the next part, I will delve specifically into the use of biometrics technology in the voting process and also discuss the important issue of feasibility of using this technology in the context of Zimbabwe.