This question is posted by ACE on behalf of an anonymous user. 

Question

Early next year the Somaliland National Electoral Commission will conduct a biometric voter registration exercise. The voter registration process requires scanning Iris images (two eyes of registrants),  a picture (facial recognition), and demographic data Like Name, Age, Mother’s Name, date of birth etc. This time there is no fingerprint at all. 

Somaliland performed a biometric voter registration exercise in 2008-2009, with fingerprints and facial recognition, and much effort was devoted to cleaning the voting register, but the NEC failed to provide accurate and credible lists. 

This is the first time we are using Iris Biometric voter registration in Somaliland. In June 2014, the Somaliland National Electoral Commission and University of Notre Dame in the USA piloted the system in two districts with a sample of 500 registrants, to test how the matching system will detect multiple registrants. 

Related to the above background, I have the following questions: 

1. Are any country/countries using this or similar Iris biometric voter registration system? 
2. Do you think Iris biometric voter registration is more accurate than fingerprint biometric registration systems? 
3. Are there any visible challenges of using iris biometric? 
4. What are the risks of implementing iris based biometric voter registration in less developed country like Somaliland? 

Please use this link from University of Notre Dame as further background information: 

• https://research.nd.edu/news/49903-notre-dame-biometrics-expert-is-helping-to-ensure-an-honest-election-in-somaliland/
• http://www.cis.fordham.edu/events/Bowyer-iris-talk-2014.pdf
• http://www.biometricupdate.com/201408/notre-dame-researchers-using-iris-recognition-to-improve-accuracy-of-somaliland-election-process

De mucha ayuda seria documentar este proceso y sus resultados, no conozco de otro de iguales características.

El registro de biometría con huellas dactiloscópicas, es un sistema valido, es el menos invasivo y el más económico, en esta tecnologia de identificacion, con el plus de que se encuentra en funcionamiento en varios países, con buenos resultados.

También es mi opinión, que  cualquiera que sea la herramienta que utilicen, sin no se hace bien, tendrán los resultados inesperados y la pérdida de confianza en la conformación de un buen registro electoral.

I was part of the team conducting an options analysis for Somaliland's NEC with regards to the voter registry. As far as I know, Iris has not been used for a voter registry in any country before. 

I don't know if Iris is "more accurate" than fingerprints, but it can be a better solution in a pastoral society where many people may not even have fingerprints. Both are vulnerable to error during the acquisition process and require proper training on equipment. 

International IDEA is now conducting global research on the use of ICT in elections, which includes registration stage. Our data from 106 countries so far show that not any country used Iris technology during registration. Please contact me for more details, I can give you access to the data in the current stage.   

It is interesting to learn that another biometric character is being applied, I had never heard of the same any where, but the flaws may come on with those with visual impairment because there will be only one iris or none at all, I wonder what mitigations are in place.

I suppose iris biometric implementation is not an end in its self it would be used as an alternative to those without fingerprints as David Jandura highlighted.

As  much as I know,the human Iris is a biometric component which is used in airports and other transport agencies.To the best of  my knowledge,I have not known the Iris as a part which can replace the fingerprint.However,since it has been proven beyond all reasonable doubt that the fingerprint biometric which the NEC of Somaliland used between 2008 and 2009 harboured about 30% duplications,then there is a likelihood that  the human Iris is going to be more accurate than the fingerprint-thanks to the research carried out by the University of notre Dame.

I have no knowledge of any country which has used the Iris biometric technology in  voter registration exercise.The possible visible challenges with the Iris biometric voters registration will be how to cope with citizens with visual impairment.The situation will be further compounded in Somaliland because it is a less developed Country. 

I was involved in the evaluation of a pilot Civil/Voter registration project in Afghanistan in 2008 that used both fingerprint and iris biometrics captured. It was not rolled out nationwide (for reasons that had nothing to do with iris recognition).

Can I assume that the hit rate of 450 from 2,124 pilot iris captures was high because you were testing the system? If it reflects real world duplicate registrations then you have problems that biometrics won't solve!! Or you have problem biometrics...

I believe that it is the degradation over time of the iris that presents the greatest challenge to its use in certain VR/Civil Registration paradigms. But with careful study, the benefits of iris over fingerprint may possibly outweigh this and other downsides.

Traditionally, the cost of the specialist hardware to enroll iris biometrics was also a significant negative. There was also the need for two devices - one for the facial image (typically required in VR scenarios), the other for the Iris. I understand that a single device can now be deployed that will capture both iris and facial image.

Accuracy is highly complex in biometric voter registrations - the choice of biometric is only one factor - training, configuration of systems, quality control - these can all have significant impact.

I'm not familiar with Somaliland - what did the pilot tell you about the feasibility? Was the pilot "realistic" in terms of where you chose to run it (remote, rural as well as proximate urban sites  - expert operators versus trained field staff - etc)??

Looking forward to reading more on this question!

I was involved in the evaluation of a pilot Civil/Voter registration project in Afghanistan in 2008 that used both fingerprint and iris biometrics captured. It was not rolled out nationwide (for reasons that had nothing to do with iris recognition).

Can I assume that the hit rate of 450 from 2,124 pilot iris captures was high because you were testing the system? If it reflects real world duplicate registrations then you have problems that biometrics won't solve!! Or you have problem biometrics...

 
I believe that it is the degradation over time of the iris that presents the greatest challenge to its use in certain VR/Civil Registration paradigms. But with careful study, the benefits of iris over fingerprint may possibly outweigh this and other downsides. Indeed, research from NIST in 2013 downplays the degradation issue. http://biometrics.nist.gov/cs_links/iris/irexVI/irex_report.pdf

Traditionally, the cost of the specialist hardware to enroll iris biometrics was also a significant negative. There was also the need for two devices - one for the facial image (typically required in VR scenarios), the other for the Iris. I understand that a single device can now be deployed that will capture both iris and facial image.

Accuracy is highly complex in biometric voter registrations - the choice of biometric is only one factor - training, configuration of systems, quality control - these can all have significant impact.

I'm not familiar with Somaliland - what did the pilot tell you about the feasibility? Was the pilot "realistic" in terms of where you chose to run it (remote, rural as well as proximate urban sites  - expert operators versus trained field staff - etc)??

Looking forward to reading more on this question!

I think that it is better to combine the iris technology and the fingerprint to make the voter registration. In DRC, during the electoral process in 2006 and 2011 the INEC used the biometric registration like picture and fingerprint. However the search of multiple inscriptions uses the fingerprint to match the data. I advise to use the two technologies (iris and fingerprint). The biometric system is very challenged : Cost, logistic deployment, technology mastered etc. the risks of implementing iris based biometric voter registration in less developed country like Somaliland is the lack of finance, operations and security (No state in Somalia).

Do you think Iris biometric voter registration is more accurate than fingerprint biometric registration systems?

The basic principles of biometric registration

A biometric system is an automated process aimed at acquiring data, extracting discriminating information from these data, comparing this information with the one contained in a database -clustering reference templates- to decide whether they match or not, and indicating to the user if authentication or identification is successful or not. This means both to capture the information to be analyzed and processed, and to create through digital files the characteristic elements of a series of images stored on a specific medium.

The information stored is unlikely to be made of the original images, but of a model -mathematical elements- that distinguishes a biometric sample from another. This model is called “template” or “signature”. In this way, small size and easily exploitable files are recorded.

Creating a signature/template occurs during the registration phase. Registration is therefore the critical step generating and recording the identity of the person that usually combines   alphanumeric data (name, surname, address ... etc.) with a biometric image. The image quality of the biometric sample is paramount to ensure a rigorous authentication/verification process.

The performance of biometric systems

It is virtually impossible to obtain an absolute coincidence (100% similarity) between the original file -signature/template file created during registration- and the restitution file -signature/template file created during the check-.

The performance of biometric systems are therefore expressed in

• FRR (False Rejection Rate): percentage of people rejected by error;

• FAR (False Acceptance Rate): percentage of acceptance by error;

• EER (Equal Error Rate) giving the rate on which FAR is equal to FRR;

Each biometric technology has its pros and cons. Their respective performance must take into account not only the effort required to capture biometric data, the intrusiveness or not of the technology, its cost but also the degree of precision and reliability in the verification of voters’ identity. The environment in which the technology is going to be used is also paramount to appreciate the true value of the various biometric technologies.

Among the various biometric identification technologies developed in recent years, only morphological data seems to have found a direct application in the field of electoral identification. Digital images of fingerprint, face and/or iris are indeed commonly associated with biometrics and de facto the most used in electoral matters.

Iris biometric recognition

Iris technology is quite recent. The algorithm identification of the iris wasn’t patented until 1994. The recording mode of biometry using iris is quite simple: any person seeking to be identified must simply look at the lens of a camera that instantly records the drawing of his/her iris. The image capture system must be fast, accurate and should not use a reflective light.

Shooting can be conducted by a classical CCD image sensor camera with a macro lens. Lighting constraints often require the eye to be close to the camera. Moreover, a source of lighting of a wavelength comprised between 700 and 900 nm, invisible to human eyes, is used. This artificial calibrated light –infrared - is used to mitigate as much as possible the natural ambient lighting.

As a matter of fact, the iris has a very dense pattern that is not dictated by genes, and therefore not related to DNA, each eye being unique. The iris thus provides a very high uniqueness, with a remarkable stability until death, hence an extraordinary reliability. Although the color of the iris can change, its pattern remains very stable.

In addition, an image (photo) of the iris with likely more than 200 independent variables makes the probability of confusing two individuals very low. The contraction folds, crypts, and rings of the iris, whose formation is quite chaotic and generates a high variability, are used to distinguish different iris. In this regard, the iris biometric recognition benefits from much more parameters than any other biometric method.  Consequently, by comparison to AFIS (automated fingerprint identification system), the probability of finding identical data is likely to be lower with an iris biometric data processing system.

The advantages of IRIS biometric recognition are therefore:

• Great quality information
• Excellent reliability (FRR and FAR very low)
• Real twins identifiable;

CHALLENGES OF USING BIOMETRIC REGISTRATION and What are the risks of implementing iris-based biometric voter registration in less developed country like Somaliland?

No matter which biometric system is used, the introduction of new technologies in electoral processes requires addressing a series of issues. 

in particular, when setting up a new system of voter registration, criteria of transparency, verification and availability of the voter roll are as important as the technical criteria related to the constitution of a register of voters.

In addition, two questions should be addressed since biometric voter registration represents a major financial and technical undertaking:

• The question of a permanent –regularly updated - voter roll;
• The question of synergies between civil and voter registration.  

In any circumstances, the introduction of a new technology must entail 1) setting up a clear objective, 2) having realistic expectations, 3) adjusting (and/or reforming) the legal framework and 4) receiving the support from all stakeholders.  

The technology must be adapted to the environment in which it is used.

The new technology must be suitable, that is to say compatible with the environment in which it is implemented. There are several criteria to assess the appropriateness of introducing technical innovations into the electoral process in a given environment. These are related to investment, maintenance, operability, added value and sustainability of the new technology put in place. 

Has the EMB the financial resources of its ambitions? Has the EMB the capacity to maintain such system? Is the new technology compatible with systems already in place? Does the technology actually promote democratic progress? And what are the long-term prospects of a technical innovation? are among the many questions to be answered. 

To summarize, the suitability of the introduction of biometric technology is defined by:

•  Its cost-effectiveness,
• Its easiness with regard to its daily management
•  Its modularity, flexibility and interoperability,
• Its standardization,
•  Its permanence.

Any innovation that doesn’t meet these criteria may not be "appropriate." However assessing the adequacy between technology and a given environment is intrinsically marked by the way in which such a technical innovation is implemented. 

A rigorous, precise and realistic implementation timeframe is paramount. 

It might be appropriate to conduct a feasibility study before launching a full-fledged voter registration involving a new system. The participation of different stakeholders may be opportune to assess all issues to be considered and/or addressed, not only those directly related to the immediate voter roll update.

Obviously, the more detailed the specs are the better.  One of the most common challenges is to clearly identify needs. The risk of ending up with a vendor-driven exercise should not be underestimated since it may result in a "dependence" vis-à-vis of solutions proposed by suppliers that may not adequately fit into the local environment. A proper anticipation of all requirements, precisely spelled out in the call for proposal, taking into account all aspects of the set-up of new equipment and/or services, is likely to be a guarantee of success.

The organization of a pilot test is often necessary to verify the relevance of a technological choice. The pilot test not only helps validate technical assumptions but also ensures that the proposed solution is actually implementable. It also provides a unique opportunity to evaluate the ability of the players –principally the election management body (EMB)- to incorporate new technological processes into the electoral framework. It may also provide for public awareness and facilitate the acceptance by all stakeholders of the technical choice made.

The introduction of new election management processes should also be based on a thorough assessment of the capital and human resources required for each sequence of the exercise, including the development, certification, use and evaluation of new equipment and/or services. In this regard, taking account of the "life expectancy" of innovative investments is paramount for the successful introduction of new technologies.

A step-by-step introduction of any innovation can be a wise solution in order to take into account the likely adaptations and adjustments to be made to the initial plan. Developing progressively a new election management system may facilitate adaptation to local constraints and allow for a gradual reorganization of the logistics that are likely to accompany the process of modernization.

Planning the introduction of a new technology requires a proper understanding of the context in which the technical innovation is going to interact. Taking into account the political, economic, social as well as physical environment is fundamental. Advocates of bridging the “technological gap” should not misjudge the attitude of the users, including political actors, supposedly benefiting from the technological innovation.

It is worth stressing that reforming and modernizing electoral systems may be worthless if, firstly the immediate users –meaning EMB staff - are not closely associated to the exercise and, secondly the ultimate beneficiaries, namely political parties, candidates and voters, do not adhere to the choice made.

Internally, taking into account the “human dimension” is a key factor in the successful reorganization of the working environment resulting from the use of new systems. The introduction of a biometric voter registration may require not only that the operational department in charge of voter registration works more closely with the IT department but that consultations are also conducted with other departments, which may have to revise their own organizational arrangements and activities.

Externally, it's about communicating effectively in view of maximizing the adherence of all stakeholders to the exercise. In this regard, biometric voter registration should be seen as a tool to improve the reliability, accuracy and inclusiveness of the voter roll. It can also be appropriate that, in addition to the pilot test, simulation exercises be undertaken to allow the population to become familiar with the new system.

Introducing high innovations into the electoral process requires a comprehensive approach.

An often-overlooked issue is the time necessary to adjust the legal framework, as a result of the implementation of high tech electoral processes. A biometric voter registration may require toileting electoral regulations and procedures. It may also be relevant to conduct a thorough analysis of the different legislations and/or regulations to identify precisely the legal provisions to be amended prior to launching an exercise involving new technological processes.

The modernization of the electoral process, subsequent to the introduction of new technologies, cannot be conceived as an autonomous and a one-dimensional action, limited to the purchase of new equipment. High-tech solutions have an impact on the cost of elections, not only because of the initial investment (capital investment), but also because of long-term operational costs, including service, maintenance and renewal of equipment  (depreciation, obsolescence of materials).

High-tech solutions impact on local infrastructure, including in terms of logistics and transportation (workspace, warehouses, supply energy, equipment maintenance, delivery of consumables…etc.) and require especially qualified electoral staff, for whom tailor-made training must be provided. 

The design and implementation of security measures should also be contemplated. Paradoxically, technical innovation may increase the vulnerability of the systems put in place. Risks assessments, set-up of recurrent technical controls and constant update of software should be adequately and timely anticipated.

Finally, innovative solutions must be aimed at strengthening the credibility of the electoral process. The “public” (voters, candidates and political parties) should be confident in the reliability, accuracy and inclusiveness of the voter registration exercise. They should be regularly informed of any developments made to improve the system. As a mater of fact, associating stakeholders to the modernization of the electoral process is guarantee of success. 

In conclusion, a well-planned, timely implemented, and far-sighting technological innovation policy applied to voter registration is likely to offer tangible results in improving the electoral process, provided that the exercise meets clearly identified needs:

• What critical improvements are expected from the innovation?
• What are the best-suited equipment in terms of cost, required level of infrastructure and assimilating capacity?
• What is the level of political support in order to secure both funding and institutional support required for the successful completion of the exercise?
•  What security measures should be taken to ensure the reliability and security of the new systems put in place?
•  What is the viability and sustainability of new biometric registration in the long term?
• Will the new technology serve the interests of the voters and will it be compatible with the fundamental principles of democracy?

End