Question
This question was posted on behalf of Ingo Boltz, member of the ACE Practitioners’ Network

I am currently working with UNDP on a project in Zambia to design a new national registration system.
The idea is to collect an infant/child fingerprint (using live capture) as close as possible to birth, but still get an image good enough to verify that same print at age 16, when the young citizen applies for his adult ID card. A positive match would be required to get the adult ID. I am trying to establish a chain of authenticity from infant-hood upwards to adulthood.

What is the best age for this? Age limits have been considered in other contexts as well? If so, what are the conclusions?

Summary of responses
The consensus is that taking the fingerprint of an infant, for the purpose of then later using the information for the basis of an adult ID card is unreliable. This is due to the changing nature of fingerprints in young children and the uncertainty of whether they can be matched to fingerprints in adulthood. A range of different ages are put forward for when it is best to first take fingerprints for the purpose of ID cards, with the European Union criterion of 12-years-old being the earliest.

The obstacles are compounded by the extremely high quality images required to adequately capture a child’s fingerprints. And in the opinion of one expert, additional logistical and operational challenges posed in Zambia call into question the feasibility of such a scheme.

By way of comparison, in Argentina a series of fingerprints are taken throughout an individual’s life, starting when a child first enters school, while in Morocco a fingerprint is first taken at the age of 16 when an individual applies for an ID card.

Examples of related ACE Articles and Resources
Encyclopaedia:
• Voter Registration Methodologies

• Voter Registration Technologies and Materials

Support and Advice:
• Consolidated reply: EMBs using Automatic Finger Identification Systems (AFIS)  

External Resources
• Carsten Gottschlich, et al. ‘Modeling the growth of fingerprints improves matching for adolescents’

• Council of the European Union, ‘Setting of the minimum age for recording and storing facial images and fingerprints in the chip of a passport’
• FIDIS, Study on ID documents: Biometrics

Names of contributors
1. Horacio Boneo
2. Ronan McDermott
3. Alexandre Michaud
4. Ahmed Jazouli
5. Khalid Waheed
6. Jagoda Petrovic Ukaj
7. Peter Wolf
8. Ingo Boltz

Hi Ingo,

In my country (Argentina) fingerprints have a long tradition that began with Vucetic, who created one of the first classification systems.  Identification begins at birth, and follows the individuals through the years.  However, I understand that in the case of newborns what is taken is the print of the foot plant rather than the fingerprints, as they do not provide a reliable identification and may change in the first years.  The first official fingerprints are taken when the child enters school and given his first ID document, which is then replaced at certain times through his/her adult life. 

However, dactiloscopy (a number of other denominations are also used for this forensic specialization) is a rather complicated subject that I feel is somewhat out of the scope of the standard electoral expert.  It might be better to get some specialized support before providing advice – it should not be difficult to get it from some Police department in a developed country, they may even have cooperation agreements with the Zambian police.

Hello Ingo,

Fingerprinting children is exceptionally difficult as the minutae are, well, minute! The image quality requirements are much greater so you can expect a higher failure to enrol rate, unless you lower the quality threshold. That compromises your error rates.

Moving forward 16 years - the relatively high failure to aquire/enrol rates observed in countries such as Zambia (I was there for three months recently as a member of the EU-EEM team looking at VR and we observed high rates of failure to aquire, especially in rural areas) combined with the previously mentioned infant/child enrollment challenges mean that the likely percentage of 16 year olds whose prints would be falsely rejected by such a system would, in my opinion, be unacceptably high. This calls into question the value of the proposed system.

I'm struggling to find literature on the comparison of the aging of biometrics from infancy to adolescence that would point to, perhaps, the use of iris over fingerprints. Certainly, iris recognition advocates claim that the iris is stable from about 1 year old onwards.

From the US NSTC's Subcomittee on Biometric's Biofoundations Catalog FAQ (see this link )

"Q: Can children’s fingerprints be collected?
Yes, in most cases, a child’s fingerprints can be collected after the age of one year or so, but the prints may not have the clarity of adult prints. It is not clear whether fingerprints taken from children can be automatically matched to those same individuals later as adults."

[emphasis added]

This suggests that newborns cannot be effectively fingerprinted. It also reflects the uncertainty with comparisons between infant and adult fingerprints of the same individual.

Any AFIS used to match templates taken from 16 year olds with those taken from infants or children would have to be specially programmed to cater for the growth. (see this link)

It does appear that you won't get a cost-effective biometric from a newborn or tiny infant. By that I mean, you can take DNA, but not a fingerprint or iris.

A challenge greater than the technical one you rightly pose this forum will be the necessary operational and logistical resources to deliver such a system to the remote villages where births (and deaths) frequently go unregistered. This has proved the undoing of both civil and voter registration in Zambia. Putting a live capture kit at District HQ will simply not work - people cannot afford the time and cost necessary to travel to register births - the project will have to get to communities and do so annually. The costs of this will be significant and should be factored in to overall project budgets.

I look forward to other submissions on this question.

Regards,

Ronan

Hi all (Thank you Mr. Boneo; I corrected my profile so that I can at last see questions):

Elections Canada does not do electoral biometrics, nor any provincial or municipal levels. 

That said, we are wondering about privacy issues, the maintenance and protection of such bank of data over long periods of time, and restrictions over the use of the data for other purposes than electoral identification (e.g. from other public and private entities whose purposes are not electoral e.g. police, secret services, etc)?

I understand this is a side question to the original question.  Please do prioritize the first question from Ingo Boltz.  Thank you.

Regards,

Alexandre

In Morocco, fingerprint of the child is taken at 16 when she/he applies for ID card.

The registration of the child is through the book civil status. To protect the chid the period between bith and registration should be close and limited by law.

Best

Ahmed Jazouli

Rabat

Helo,

This is fact that finger prints are changed from childhood to adult age.It would therefore, be

in the fitness of things that fingerprints be taken and saved after attaining adulthood.

In most of the countries 16 years age is considered to be the age for registration in national database but still many  countries register the citizens at the age of 18.

In Pakistan the qualifying age for registration is 18.

KHALID WAHEED

The following links contain some background on why fingerprints for European Union passports are only captured from the age of 12.

Main reason are high error rates, especially for the under 6 year old. After the age of 6 fingerprints still change as the child grows and special recognition alogrithms are needed to accommodate for those changes.

http://www.statewatch.org/news/2006/jul/9403-rev1-06.pdf
http://www.fidis.net/resources/deliverables/hightechid/int-d36000/doc/10/

Hi everyone,

sorry for the delay in catching up, I was on the road last week.

@Horacio:

I'm actually based in Buenos Aires, but all of my contracts ironically have been outside of the country, so I am embarrassingly ignorant of the situation in my host country. Do you know if the foot print is taken at birth, and whether it is electronically stored somewhere for later use? Perhaps we could meet for coffee one of these days if you are in town?

@Ronan:

I actually read your report, very good work! Thanks for the links provided, especially the Goettingen paper is very helpful. I have also been in touch with some Brazilian researchers who are working on the same issue. They have developed a infant scanner prototype that is promising (Hi-res photo camera to capture) but way too expensive for use in the field.

It seems to be consensus that nothing taken from infants is really of much use, including iris (also very hard to capture since infants have their eyes closes most of the time). I also doubt it would be cost effective in Zambia.

My current idea is to capture a print at one, another at 6 (school enrollment), perhaps an intermediate at 10 (no idea yet about the "event" to tie it to) and then the adult one at 16.

I completely agree that provisioning in the field is the bigger challenge though. You've been down in the provinces a lot, what would your take be at involving the Health Facilities (clinics, rural health posts, mobile medical teams) to get to the more remote people? I know Health just went through a funding scandal and they are understaffed so maybe not ideal but they DO reach 80-90% of children through immunization... Getting them to use kits would give good reach, but I'm not sure its feasible.

@all:

There are basically two issues here. 

One is to establish "ground biometric truth" from as early in the child's life as possible, which would make identity theft later much more difficult. Admittedly perhaps not the main priority in Africa but still something to keep an eye on when planning.

The other is avoiding benefits fraud. One of the issues in Zambia is plain motivation to get a child registered. There are little benefits to be had for it, and considering that, its excessively difficult to register. Hence only 6% do register in rural areas. Only by providing incentives to register can this be changed, and in most cases that will mean tying benefits to registrations (like South Africa does with its Child Rearing Grant). However, if this results in massive benefits fraud (multiple registrations of children to obtain multiple benefits) then the backlash could shut the program down (and of course its costly). So having an early print could cut down on that, even if it cannot be extrapolated to a later age for matching. 

I doubt that privacy concerns would play a major role in Africa, although I may be wrong.

Hi Ingo,

I am now in Torino, but I will be back in Buenos Aires in a few days and I will call you.

I think that what they took is a physical imprint, but the main purpose is to avoid child theft or confusions, not identity in the long-run. 

Hi Horacio,

Happy new year! Would be happy to meet up, perhaps my private email is best to organize.

Looking forward to meeting!

Ingo