Original question:

This question was posed by Mark Stevens who is a member of the Practitioners' Network. To view his profile, please click the "Members of the Practitioners' Network" link to the left of the Workspace page.

Posted on 23 May, 2007

Biometrics are used to measure physical characteristics and behaviours, for example fingerprints, speech, face, iris, and hand geometry. The basic task of all biometric systems is recognizing patterns to distinguish those that match closely enough to be considered identical, and those that have great enough variation to be considered non-identical. To do this, the system must first register and store the characteristics (for example the fingerprint), and then be able to match this with new information. The system can either be asked to verify someone claiming to be a person known to the system, in which case the system only has to compare the new information with one item in the stored data, or be asked to identify a person, thus requiring the system to compare the new information with every item in the database.

Automatic Fingerprint Identification System (AFIS) refers to a computer system capable of establishing the identity of an individual through fingerprints by the use of biometrics. AFIS can be used for example in criminal investigations and in elections.

Nigeria is one country in which AFIS has been used for election purposes. The EMB identifed and removed milions of duplicates from the voters register based on decentralized datasets (using a number of local government areas rather than the whole country as one dataset) using AFIS. AFIS has also been used in the voter registration process in Democratic Republic of Congo and Venezuela and electoral rolls have in previous elections been produced by Pakistan's National Database and Registration Authority (who has systems including fingerprint biometrics) on behalf of the Elections Commission of Pakistan. It was also considered for the 2003 elections in Yemen to address problems with the current voters register. In Yemen, the issue was not to develop a new AFIS system, but rather to explore the possibility to merge the voter registration with the existing AFIS-based civil registry system. Read more about these examples below.

The main reasons for the introduction of AFIS in the electoral process has been identification and removal of duplicates in the voters register to guarantee “one person one vote” in elections, and the main considerations to be made are as with the introduction of all technology the electoral process:

• Cost – the introduction and use of AFIS are likely to have high costs attached (read more about cost projections in the case of Yemen on page 44 in the attached report) and it is important to go through a thorough procurement process to find the most appropriate and cost effective system.
• Time – a realistic understanding of the time required to introduce an accurate AFIS is necessary. It may take months or even years to develop the system and produce ID cards based on it.
• Accuracy and credibility - Achieving acceptable levels of accuracy needs thorough consideration (see examples of potential problems of AFIS on page 43 in the attached file)   and it is important that there are realistic expectations of what AFIS can deliver, both by the EMB and other stakeholders.

Replies were received, with thanks, from:

• Ronan McDermott
• Antonio Spinelli

Links to related resources:

• Automatic Fingerprint Identification System in Wikipedia
• Russia - The State Automated Elections System (Case study in ACE)
• Elections and Technology topic area in ACE
• Technology for Voting in ACE
• Voter Identification Methods in ACE
• Registration Data Collection in ACE
• Independent National Electoral Commission in Nigeria
• Election Commission of Pakistan

Quote from the ACE Encyclopaedia on Automatic Fingerpring Identification Systems:

“While this technology is not new, the electronic methods of recording and recognizing an individual finger print advanced substantially during the last decade of the 20th century. Today, identification can be achieved in a few seconds with reasonable accuracy. As a result, the use of automated fingerprint identification systems (AFIS) that record, store, search, match and identify finger prints is rapidly expanding. AFIS can be integrated in a suite of applications that work together to provide a comprehensive fingerprint and palm print identification system.”

Individual responses in full below:

Nigeria has a massive AFIS infrastructure but has never (to my knowledge) run a full analysis on any dataset larger than a number of Local Government Areas. Significantly, despite removing many millions of duplicates identified by AFIS, the Independent National Electoral Commission (INEC) never (as of late 2005) sought or brought a single prosecution for multiple-registration against any offender!

Prior to the 2002 Elections there, in one State, AFIS processing was only conducted on 5 of 23 Local Government Areas (LGAs) prior to the election. Some 130,000 duplicate registrations were identified by AFIS from just five LGAs out of a total voting population of approximately 2.5 million.

AFIS was a feature of the Digital Registration Kits used in Democratic Republic of Congo for voter registration there in 2005. However, the processing was limited to the maximum number of persons registered on that kit, approximately 4,000. I am not aware of how many duplicates were identified in this deployment of AFIS - but the benefits were immediate - the system identified the duplicate in near real time, thereby preventing duplicate registrations on that kit. I am not aware if any back-end AFIS was conducted on larger datasets.

Electoral Rolls have, in previous elections, been produced by Pakistan's NADRA (National Database and Registration Authority) on behalf of the Elections Commission of Pakistan. Since NADRA's systems include fingerprint biometrics, it is likely that some AFIS processing was conducted on some or all of this data. ECP is creating a new voter database that will not include fingerprint data.

Hi Mark!

I have run into some instances of electoral management bodies (EMBs) considering the use of Automated Fingerprint Identification Systems (AFIS) for voter registration purposes. My most recent experience and closer involvement with AFIS was in Yemen in 2001. It is already six years ago, so please consider that things might have changed/improved since then.

I have attached a report presenting the findings of a technical diagnosis mission of the voter registration system of Yemen conducted by Michael Yard and I for IFES in mid-July 2001.

At that time, in view of the crucial Parliamentary elections to be held in 2003, the Yemeni EMB faced an enormous pressure to undertake overdue reforms of a highly obsolete and unreliable voter registration system and to amend the resulting voters’ lists, which were considered to be widely flawed. Until then voter register/registration problems had gravely undermined both the credibility of the EMB and of the elections the EMB administered.

One of the three options examined by the voter registration assessment mission was the plan of the Government of Yemen to use a new database established through an already existing Civil Registry Project (piloted by the Department of Civil Registry of the Ministry of Interior) as the basis the new voter register for the 2003 elections.

So, in the case of Yemen it wasn’t so much about the EMB itself considering to develop its own AFIS-based voter registration system, rather about exploring the feasibility of merging the existing (but by then still in its very infancy) AFIS-based civil registry system developed by the Civil Registry Department with the new voter registration database that was about to be established by the EMB. In the end, luckily the EMB and the Government opted to follow the recommendations contained in our assessment.

With specific regard to the use of AFIS technology, the following chapters of the attached report might be of interest:

• “OPTION 2: Using the Civil Registry Project as Basis for a New Voter Registry” at page 23 of the report; and
• "Potential Problems with AFIS Identification Systems", page 43

It would be very interesting to find out how far the Civil Registry Project of the Ministry of Interior of Yemen has gone in these years.