Since Afghanistan’s first presidential election in 2004, massive election fraud and malpractices have been the rule. Confidence in the electoral process among the population is correspondingly low. As this Peace Brief explains, a new biometric voter registration system has recently been suggested by the Independent Election Commission to radically reduce voter registration risks. If it is to be effective, however, training of registration officials in using the envisaged technology is essential, as is ensuring acceptance of it among stakeholders. The timeline is short: elections are scheduled for 2018. The risk of catastrophic failure is high.
- Since Afghanistan’s first presidential election in 2004, duplicate and fake voter registration have contributed to massive election fraud. Confidence in the electoral process among the voting-age population is correspondingly low.
- Meanwhile, the country’s Independent Election Commission (IEC) has neither an accurate count of voters and their places of residence nor any effective controls against multiple registration. Elections are thus still subject to ballot stuffing.
- The IEC recently initiated a new, high-tech biometric system that uses fingerprints and digital photos to eliminate duplicate and false voter registration. If this process is to be effective, the IEC needs to focus on equipment and training.
- The IEC also needs to conduct extensive consultations with electoral stakeholders to ensure buy-in and acceptance of the new voter registration methodology.
- The timeline is short, however: parliamentary elections are (re)scheduled for 2018.
- The conundrum is that if implementation is rushed, the new system may well not work properly. The risk of catastrophic failure due to a botched rollout is high.
The first Afghan presidential poll in 2004 was conducted during a period of great optimism following the fall of the Taliban, albeit using a hastily arranged voter registration and having only scant knowledge of the population. Eleven million people ultimately registered. The voter roll established in 2004 soon became a significant problem for subsequent elections because the registration system included no real barriers against multiple registration and the Independent Election Commission (IEC) was unable to reliably identify duplicate voter ID cards for deletion. As a result, each subsequent election saw the addition of millions of new names to the voter rolls but no process to remove old or duplicate registrations. After the 2014 registration, the number of voter registration cards issued stood at twenty-three million. Yet it is estimated that only half of Afghanistan’s thirty million people are of voting age, meaning that up to eight million of the voter cards in circulation are either duplicates or obsolete.
Another problem is that registration does not link voters to individual polling stations, which is common practice in established democracies. Without an accurate count of how many voters live where, the IEC must oversupply every polling station with extra ballot papers because it has no way of knowing which polling station a voter will go to on Election Day. These additional ballot papers, along with inaccurate voter lists and the lack of credible ID cards, have enabled fraudulent ballot box stuffing. This has badly damaged the integrity and the credibility of the electoral process.
Afghan election experts have long recognized the problems with the voter lists, but resolving them has been both expensive and politically difficult. In the aftermath of the controversial 2014 presidential election result, both President Ashraf Ghani and CEO Abdullah Abdullah agreed on the need for comprehensive voter registration reform. This was codified in the National Unity Government Agreement that formed the current government. It cites the need for significant improvements to the electoral process, and established a Special Electoral Reform Commission to recommend specific reforms. One of the Special Electoral Reform Commission’s main recommendations was an overhaul of the voter registration system, which President Ghani approved in a subsequent decree in September 2015. The election law passed in September 2016 then mandated polling station-specific lists.
After the 2014 registration, the voter roll stood at twenty-three million. Yet only half of Afghanistan’s thirty million people are of voting age.
After years of delay, Afghan authorities have recently taken several important steps to correct the badly broken system. Following discussions with national and international partners, the IEC has settled on a biometric voter registration (BVR) methodology and has issued a limited international tender for voter registration equipment. According to the current plan, the IEC will collect two sets of biometric data for each applicant: a full set of digital fingerprints and a facial photograph. The personal data recorded will correspond with the requirement for the new national ID card. Because Afghanistan still does not have a residential address system, voters will be required to register where they will later vote, which is a common way to allocate voters to specific polling stations. All of the personal and biometric information will be entered into a central database, which will enable accurate removal of duplicate registrations and more reliable updating in future elections.
Assessing Biometric Solutions
An effective voter registration system must be capable of identifying duplicates from a large data set. It must also be user friendly, first, to enable temporary registration staff to collect accurate biometric data from legitimate voters so that, second, the de-duplication process can be undertaken with a high degree of confidence. Such a system would be complex and could come at the expense of citizen trust, because voters are less likely to accept the end product if they do not understand the process. Finally, the registration system must allow only eligible registered citizens to cast a vote in their respective polling stations. This can be achieved in a number of ways, the most common of which is the use of indelible ink. In past Afghan elections, however, it has been possible to wash ink off with bleach. Voter verification equipment is increasingly being used in high-fraud risk environments.
Digital Fingerprints and Digital Photos
The standard way to ensure accurate voter registration today is to collect fingerprints and photos of the applicant. Both are taken electronically and stored in a single national database, after which individual voter data is compared against all other registered citizens to identify and remove any duplicates.
Experience has shown that the quality of digital photographs is often not enough to enable the election authorities to rely solely on facial recognition software to identify and confirm duplicates. In most cases, the BVR system relies on electronic fingerprint comparisons, with photographs being used only for manual comparisons. If potential duplicates are in the thousands this approach is feasible, but if in the tens of thousands is extremely time consuming and could lead to significant delays. Given the fairly high level of double registration in recent Afghanistan elections, this scenario is plausible.
The digital fingerprinting-photo combination has another potential drawback: registration officers must be sufficiently proficient in two technologies. Examples are numerous, such as Kenya and Nigeria, of registration staff not being proficient in operating the equipment, leading to poor quality in both photos and biometric fingerprints, in turn resulting in an unacceptably high number of duplicates remaining on the voter rolls.
On Election Day, the IEC will rely on polling station-specific voter lists, photographs, and indelible ink to guard against multiple voting or impersonation. Indelible ink has a poor track record in Afghanistan, as noted, and therefore instills little confidence in the electorate and politicians alike. Photographs, meanwhile, will be missing in most female polling stations. Hence, even if the voter register is largely free from duplicates, the current approach is prone to serious abuse if it is not accompanied by some form of biometric voter identification on Election Day, such as fingerprints.
Pilot Iris Scans
The IEC could also consider additional measures to enhance quality and confidence in the process. A fingerprint-iris combination aims to collect two sets of biometric data and thereby increase the likelihood of identifying voters who register more than once. Using iris images for de-duplication has been done in some smaller countries in the Gulf (the United Arab Emirates, for example) for border security and multinational organizations for refugee registration (such as the Office of the UN High Commissioner for Refugees).
The only application of iris scanning strictly for voter registration was in Somaliland in 2016. That exercise exposed several challenges. The operators of the iris scanners must be properly trained to capture images of sufficient quality for comparison, and various corneal conditions or visual impairments among voters undermine de-duplication efforts. The existence and extent of these conditions among the Afghan population is currently unknown. It is therefore difficult to assess how many potential duplicates will not be detected by the iris system. Furthermore, training tens of thousands of ad hoc registration officers on the proper use of two types of biometric equipment remains a challenge. In addition, explaining to an electorate with less formal education how such an abstract system actually functions is a significant challenge.
Contingency Plans to Guard against Illegible Fingerprints
Fingerprinting is the oldest and most established form of biometric and is used by most security agencies around the world as the prime identifier. Still, if the fingerprints collected are not of sufficient quality, the Automatic Fingerprint Identification System cannot detect duplicates.
The main challenge to securing the necessary fingerprint quality in Afghanistan is the relatively large number of Afghans performing hard manual labor, which degrades the quality of fingerprints. Procuring high-quality fingerprint scanners, testing them thoroughly before a full roll out of registration, and paying special attention to the training of registration officials charged with capturing fingerprints could significantly reduce this risk. Doing so is possible only if the IEC considers the digital photo component a complement to fingerprinting. Using scarce resources to procure and deploy a sophisticated digital photo solution would distract from the effort to obtain quality fingerprints. Digital photos will play an integrity-strengthening function in Afghanistan but should be treated as one of several complements to the strategic Automatic Fingerprint Identification System solution.
Verification on Polling Day
Removing duplicates and phantom individuals from the voter roll is indeed a critical first step to combating electoral fraud and the voter registration process. But, if fraud is to be avoided, the system must also ensure that voters are verified on Election Day. If they are not, the risk of ballot box stuffing will remain high: without a verifiable check of voter identity at the polls, corrupt or intimidated polling staff can simply cross out names on voter lists and cast that number of ballots in areas that are not well observed.
Although the current IEC registration system has no effective verification process, risks on Election Day can be mitigated by requiring that fingerprints be collected at the polling location and verified against the voter list before a ballot is given to the prospective voter. Each polling station’s voters list would be preloaded onto small handheld devices, requiring voters to verify their identity by scanning their fingers before casting their ballots. This approach would drastically reduce impersonation and counterattempts by rogue polling staff to stuff ballot boxes because the number of ballots cast must match the number of verified voters recorded on the device.
Introducing BVR and voter verification systems is highly complex and time consuming. Frequently, first attempts fail because not enough time was allowed for planning, testing, and implementation. If rushed, specifications for equipment and software might not fully match what is required, training sessions for registration staff charged with collecting vital biometric data become substandard, and key stakeholders are not properly consulted before decisions are made, adversely impacting their willingness to accept the final voter register.
Because parliamentary elections are (re)scheduled for July 8, 2018, the IEC has little room for error. In the event of any further setbacks or delays, it will face the unpalatable proposition of conducting national elections in which part of the country uses state-of-the-art registration and other parts use old systems that inspire little public confidence and offer at best minimal protections against fraud. The IEC should therefore focus on getting the new BVR process right rather than rushing and causing dramatic fraud problems on Election Day once again.
About this Brief
This Peace Brief outlines the pros and cons of potential voter registration options using modern technologies and identifies how Afghanistan could obtain its first credible national voter register and reduce risk of ballot stuffing on Election Day. Mats Staffan Darnolf is a senior global electoral adviser with the International Foundation for Electoral Systems and Afghanistan senior expert in residence at the United States Institute of Peace.