Facial recognition in public space: three years of case files

We reconstruct implementation in four cities from court files and information requests. None evaluated the system before deployment.

Technological capability is neither bought nor decreed: it accumulates in organisations that outlive several governments. That is the blind spot of much innovation policy in the region, which tends to fund projects and rarely to fund institutions.

What we find

The survey shows that the sectors where the country built real capability share an unglamorous characteristic: moderate but stable funding over more than two decades, with a predictable technical career path.

By contrast, programmes with funding peaks followed by abrupt falls produced teams that dispersed and equipment left without trained operators.

Why it matters

Useful science policy looks less like a bet and more like a boring, sustained budgetary commitment, accompanied by evaluation rules the research community recognises as legitimate.

“A laboratory is destroyed in a year and rebuilt in fifteen.”

What we propose

1. 01Establish a multi-year funding floor with automatic indexation and periodic external evaluation.
2. 02Include auditability and data portability requirements in all public technology procurement.
3. 03Link postgraduate training programmes to the capability priorities defined by the system.

Methodology

The study combines budget analysis, bibliometrics and interviews with researchers and programme managers. Data and processing code are published under an open licence.

This text is part of the institute's standing editorial line. Data and replication code, where applicable, are published in the Data section.