report

New report by AlgorithmWatch: Identity-management and citizen scoring in Ghana, Rwanda, Tunisia, Uganda, Zimbabwe and China

A review of identity-management practices in five African countries shows that much of the continent is well on its way towards comprehensive biometric registration. It could enable comprehensive citizen scoring or automated surveillance in the near future.

The report Identity-management and citizen scoring in Ghana, Rwanda, Tunisia, Uganda, Zimbabwe and China was commissioned to AlgorithmWatch by a public-sector organization, which asked not to be cited, last May. We recently obtained permission to publish it here.

In many African countries, the obligation to issue biometric passports in the early 2000s, which the United States and, later, members of the European Union demanded, opened the door to the biometric registration of whole populations. An industry was set up to provide fingerprints readers, facial recognition technology and a vast array of software to process this newly-acquired data.

Faster border crossing

Biometric identification holds many promises for the state and for its citizens. In countries were most of the economy happens under the administration’s radar, the unambiguous identification of people part to a commercial transaction would let the state collect value-added tax more effectively and greatly expand its tax base.

The concrete effects of biometric identification can already be felt at the border crossing between the Democratic Republic of Congo and Rwanda. There, automated gates process travelers more than three times faster than humans, which reduced queuing times at the border.

Other benefits are less clear. The identity industry claims that biometric voter registration helps ensure free and fair elections but, of all the elections reviewed in the report, only one was uncontested, the Tunisian general election of 2014. It was also the only one of the sample which was not preceded by the biometric registration of voters.

Foreign involvement

Every year since 2015, the conference ID4Africa brings together identity-management professionals from African countries and beyond. An analysis of the conference archive showed that the vast majority of non-African speakers were not independent experts, but sales representative from foreign companies. Of the 93 exhibitors at the Abuja conference in 2018, only 11 were based in Africa.

 

This reliance on foreign technology causes specific problems. In Zimbabwe, it was reported that the personnel in charge of registering the population prior to a vote told people that the fingerprint readers would be able to see for whom a person would vote in the future, in effect pressuring them into voting for a specific party.

International aid can also be used to acquire technology from foreign vendors, leading to suspicions of rigging public tenders. It has been reported that the French government granted Ghana 30 million dollar in the early 2000s on the condition that it buy equipment from Sagem, a French company now part of Safran-Morpho.

Testing grounds

Foreign companies can ruffle feathers in other ways. China’s CloudWalk was awarded a contract to build a comprehensive database of faces and bodily characteristics of the Zimbabwean population. Local journalists interpreted the deal as a way for Chinese firms to improve their facial recognition capabilities on darker skins in an attempt to outpace their counterparts in the United States.

While some fear that Chinese involvement may lead to local despots’ deploying mass surveillance systems akin to the one targeting Uyghurs in Xinjiang, other risks emerge from local practices.

Rwanda, for instance, introduced a system of citizen scoring in 2004, called Ubudehe. All Rwandans are assigned to one of six categories by villagers and village elders. The category is used in particular to distribute social benefits. Students from families classified as “wealthy”, for instance, were until recently not eligible for scholarships.

Some fear that ubiquitous identification, linked to programs such as Ubudehe, will make any error in the data inescapable and reinforce existing inequalities.

 

Published: October 21, 2019
Category: report
Author: Nicolas Kayser-Bril

Leave a Reply

Your email address will not be published. Required fields are marked *

AlgorithmWatch is supported by