Paris's Digital Archive Problem: The Hidden Scale of Duplicate Images Clogging the City's Public Records

Paris city hall is sitting on a digital storage crisis it has only recently begun to quantify. Municipal technicians auditing the Direction de l'Urbanisme's document management systems found that duplicate image files — scanned planning permits, heritage photographs, and infrastructure diagrams — account for an estimated 34 percent of total stored data across the primary archival servers, according to internal assessment work completed in the first quarter of 2026. The redundancy problem is not unique to Paris, but the scale here is amplified by years of parallel digitisation efforts run by separate arrondissement administrations that never shared a common deduplication standard.

The timing matters. The Grand Paris Express expansion, now pushing through civil works at stations including Villejuif-Institut Gustave Roussy and Le Bourget RER, generates enormous volumes of technical imagery — survey scans, subsurface mapping, progress documentation — feeding into repositories managed by Société du Grand Paris. When those image banks overlap with the Direction de la Voirie et des Déplacements' own records, the duplication compounds fast. Officials overseeing the Seine urban regeneration corridor, where redevelopment from Bercy to the Boulogne-Billancourt waterfront is generating fresh rounds of planning submissions, face the same problem in real time.

What the Numbers Actually Show

Storage costs are the most immediate pressure. A terabyte of archival-grade cloud storage on contracts typical for French public bodies ran to roughly €22 per month per terabyte in 2025 procurement rounds, according to publicly available framework pricing from the UGAP — the Union des Groupements d'Achats Publics, the central purchasing body for French public institutions. When duplicate images inflate a dataset by a third, the budget waste is direct and calculable. For a department holding 900 terabytes of imagery, a 34 percent redundancy rate means paying for roughly 306 terabytes of files that deliver no additional information.

The problem is not merely fiscal. Planners working on housing delivery in Seine-Saint-Denis — where the banlieue municipalities of Saint-Denis and Aubervilliers are pressing applications linked to post-Olympics legacy development — rely on quick retrieval of baseline site photographs and previous permit imagery. When search results return multiple near-identical versions of the same scan, staff must manually verify which version is authoritative. The Atelier Parisien d'Urbanisme, known as APUR, flagged retrieval-efficiency losses in a working note circulated to partner organisations in late 2025, identifying image deduplication as one of three priority actions for shared database improvement.

The Paris municipal budget for digital infrastructure — grouped under the Direction des Systèmes et Technologies de l'Information — stood at approximately €47 million for 2025, a figure drawn from the city's published budget documents. Deduplication tooling, which can automate the identification and archiving of redundant files using perceptual hashing algorithms, typically costs between €15,000 and €80,000 for an enterprise-scale deployment, depending on whether the city uses open-source frameworks or licensed platforms. By that measure, a one-time remediation project would pay back its cost in storage savings within months.

Where the Fix Has to Start

The practical path forward runs through data governance before it runs through technology. The Bibliothèque historique de la Ville de Paris on Rue Pavée, which holds digitised photographic archives going back to the Haussmann era, completed its own deduplication pass in 2023 and reduced its active image inventory by 28 percent without losing a single unique document. That project used a phased approach: automated flagging first, then human curatorial review for images where pixel similarity scores fell in an ambiguous middle range.

Housing and planning applications in the 13th and 18th arrondissements — both currently active zones for rental-market enforcement and new-build permitting — are among the document streams most likely to contain layered duplicates, because applicants often resubmit modified versions of earlier plans without a formal version-control step. City IT teams are now piloting a mandatory file-fingerprinting step at the point of upload for planning submissions, starting with the 13th arrondissement's local instruction unit. If the pilot reduces duplication rates to below ten percent by the end of 2026, the approach is expected to be extended city-wide through the next municipal digital strategy cycle, running from 2027 to 2031.