Nubian Rights Forum & 2 others v Attorney General & 6 others; Child Welfare Society & 9 others (Interested Parties) [2020] eKLR facilitate the security of personal information stored in each electronic identity card issued to each individual citizen and registered foreign individuals resident in a country in question. He averred that centralization of identity data facilitates optimum allocation of resources by ensuring that a fraction of the limited resources are devoted to the sole security of a single database as opposed to millions of databases. Thirdly, that in any case, a decentralized biometric identity system would not dispense with biometric parameters such as fingerprints. 435. Mr. Omwenga further averred that NIIMS is neither entirely centralized nor entirely decentralized. This was because, first, the operationalization component of NIIMS is centralized in the sense that respective identities of individuals would have to be authenticated from one system as opposed to the current system where fragmented personal identities are respectively held in separate functional databases variously maintained by governmental ministries, departments and agencies (MDAs). Secondly, that NIIMS is decentralized to the extent that registrants will be issued with a card containing limited personal identity information such as unique identification number as opposed to biometric identity information hence the risk of identity theft attributable to decentralized systems is obviated. It was his averment therefore that a more accurate description of NIIMS would be an integrated identity system in which what is integrated is the identity data as opposed to functional databases variously maintained by MDAs. Accordingly, it is possible to maintain one population register in a number of digital copies at virtually minimal marginal costs and in the unlikely event of a breach on a copy of a digital population register, the other copies will remain unaffected. 436. Regarding the contents of population registers, Mr. Omwenga observed that it is impossible for individuals to prove their identities without providing sufficient personal information which include selected biometric and biographical information. That in deciding on the content of a population register, States are enjoined to balance two competing interests- the States’ legitimate interest in maintaining a vital identity mechanism (based on biometrics) against the equally important interest in respecting the individuals’ right to privacy. In his view, based on his knowledge and expertise, the processing of biometric data for purposes of establishment and operationalization of identity systems is not only justifiable under contemporary international law but is also a trite practice, world over. 437. It was his deposition that fingerprint matching has been used to identify individuals for thousands of years and across cultures. He further averred that an identity system that solely relies on the matching (deduplication) of one biometric feature is probabilistic as opposed to deterministic. However, combined with the matching of other biometric features and biographical data points (such as date of birth, facial image, parentage information, birth certificate, amongst others), it results in a multimodal identity system that is deterministic. 438. According to Mr. Omwenga, the probabilistic nature of individual biometric features justify the need for the requirement of multiple biometrics such as retina and iris patterns, hand geometry, earlobe geometry, voice waves and DNA in identity legislations, policies and systems. That the probabilistic nature of fingerprint matching (deduplication) is attributable to the fact that perfect matching of all data points in a fingerprint is not necessarily possible owing to extraneous factors. To increase the probability of a perfect fingerprint match, deduplication software is architecturally and functionally designed to keep templates of fingerprints of all available fingers of an individual. In his view, the NIIMS deduplication software is also designed as such. Accordingly, where a fingerprint is distorted, NIIMS would ideally have nine more sets of fingerprints with which to extract templates and run a match. In the unlikely event that all the fingerprints are distorted, or where an individual has none, the individual in question would not be disenfranchised in any way whatsoever because during the design of NIIMS, the probabilistic nature of fingerprint matching and the associated misgivings were anticipated and have been addressed by NIIMS in three ways. 439. He enumerated these as, first, for purposes of enrolment, in its current implementation design, NIIMS requires certain minimum data points from a potential registrant. These include the date of birth, parentage information, residence information, current national identification number for adult citizens, birth certificate in respect of http://www.kenyalaw.org - Page 64/174