Estonia's biometric identification information system will be completed two years behind schedule and in reduced volume. The face recognition component will be dropped for lack of funding.
Interior minister at the time Andres Anvelt (SDE) presented an ambitious plan during a cabinet meeting in the summer of 2017. The state hoped to complete an automatic biometric personal identification system in less than three years. Machine-readable fingerprints and face recognition images or faceprints were meant to be assembled in a single database that could be used for identification by state agencies and companies by May of 2020. The latter would include hospitals, weapons stores, ISPs and banks.
The government gave the €20 million project the green light. Based on initial plans, the project should be completed in a little over six months. However, Tarmo Olgo from the Ministry of Internal Affairs' audit division said that the project is only nearing the end of the initial analysis phase.
"It was believed that because the idea was there, we just needed to work out a few details and could move on to the tender process in three months. Actually, such major IT solutions require a lot of analysis before procurements can be made," Olgo explained.
Margit Ratnik, head of the Police and Border Guard Board's (PPA) identity and status bureau, said too much time was spent on arguing over whether the new information system will hold all or just biometric data. The initial memorandum presented to the government was more ambitious.
"Next to biometric data, the memorandum also treated with biography. The scope of the memorandum concerned a central personal information database that was perhaps a little too ambitious and not very rational in terms of personal data protection. Discussing and arguing over this logic has taken up too much time," Ratnik explained.
Necessary but long-drawn-out analysis has left the project two and a half years behind schedule, meaning that the biometric identification database should be completed in 2022.
€27 million instead of €20 million
Other difficult decisions have been made, primarily as concerns funding. Nearly half of project funding was initially meant to come from the EU.
"Foreign funding was misplanned again. They asked for a lot, while conditions showed some components would be eligible and some wouldn't," Ratnik said.
To make matters worse, the Ministry of Internal Affairs carried out a market survey toward the beginning of the year and concluded the development would cost €27 million instead of the initially planned €20 million. The PPA asked for additional funding in the volume of €9.2 million euros in spring but only received a little over €1 million. This means that the PPA now has €17 million for the project, EU subsidies included, and the initial project will have to be dialed back.
"This sum is enough for a solution to allow authorized institutions to request biometric identification data. The funding does not include facial recognition capacity."
In addition to face recognition, an important part of the work to render fingerprints usable also requires additional financing. Existing fingerprint databases are operated by different agencies and sport rather uneven quality.
"For example, fingerprints accessible through the forensic system go back to the year 1960 and are kept on paper today. All of them need to be digitized somehow if we deem it sensible and necessary and entered into a central database. The data that is already digital needs to be examined in terms of quality and a decision mace of whether it needs to be digitized again. These decisions will have to be made by someone, and it costs quite a lot of money," Ratnik admitted.
Small pilot project
Ratnik said the PPA will start looking for additional funding as the first stage of the project is launched. The first tenders will hopefully be declared this year. That said, the system will not be developed all at once. Tarmo Olgo said that a small-scale pilot project will be put together first.
"A small copy of the future central system will be procured for modest cost where all major work principles can be tested using less data, also whether interfaces between systems work. Provided it will work, it will be simple to expand to everyone who could benefit from it. If it won't work, we will not have spent too much money and will quickly learn what exactly is broken," Olgo said.
The state should have an overview by next fall, which is when the project council will meet at the interior ministry and decide how to proceed. Olgo said that the ministry's audit division also recommends taking the project back to the government.
"We know today that it is unlikely we will be able to build a system like the one invisaged in 2017 for the money we have. Rather, the ministry will have to determine what can and cannot be done and go back to the government to ask whether there is still interest in the idea and whether we're prepared to contribute more resources. Perhaps we should learn from various past megaprojects."
Editor: Marcus Turovski