Paris Archives Tackle Thousands of Duplicate Images Clogging City Databases

Paris's municipal digital archive system contains hundreds of thousands of duplicate image files, a sprawling cataloguing failure that has quietly accumulated across more than a decade of uncoordinated digitisation drives and now threatens to undermine the city's broader effort to make its cultural heritage freely accessible online. The problem, long acknowledged in internal reviews, has finally forced a reckoning as the city presses forward with infrastructure investments tied to the Paris 2024 Olympic legacy programme.

The timing matters. Since the Games closed in August 2024, Paris City Hall has been pushing hard to activate what it calls the legacy phase — channelling post-Olympic momentum into urban regeneration, tourism infrastructure, and the digital promotion of the city's public spaces. That promotion effort depends heavily on image libraries. When those libraries are clogged with near-identical files catalogued under different metadata strings, the downstream costs — in staff hours, storage, and licensing errors — compound fast.

A Fragmented System Built Over Fifteen Years

The origins of the problem trace back to the early 2010s, when several Paris institutions launched independent digitisation campaigns without a shared technical standard. The Bibliothèque historique de la Ville de Paris on Rue de Rivoli, the Musée Carnavalet in the Marais, and the Direction des Affaires Culturelles each built their own image repositories. Files migrated between systems during server upgrades, and the same photograph — of, say, the Canal de l'Ourcq or the covered passages of the 2nd arrondissement — would be ingested multiple times under slightly different file names, resolutions, or rights metadata.

By the time the Grand Paris Express construction accelerated from 2019 onward, generating thousands of new documentary images of the Seine-Saint-Denis suburbs and the inner banlieue, the duplication rate in some collections had reached levels that made routine searches unreliable. Archivists working on the project found themselves manually cross-checking results, a workaround that consumed time budgeted for actual cataloguing work.

Paris Musées, the umbrella body overseeing the capital's fourteen municipal museums, opened approximately 150,000 images to the public domain in 2020 under an open-access policy — a genuinely significant step that drew attention across European heritage circles. But the open-access release also exposed the underlying chaos: users downloading from the Paris Musées collections portal quickly discovered that the same Eugène Atget photograph of a Montmartre alleyway might appear two or three times with contradictory date attributions.

What Forced the Current Review

The trigger for the current push toward systematic duplicate removal was a 2025 audit commissioned through the Direction de la Transformation et des Relations avec les Usagers, the city body responsible for digital services. That audit, whose findings were circulated to relevant cultural directorates in early 2026, reportedly identified storage redundancy as a significant and addressable cost. The city's cloud storage contract for heritage data runs to several million euros annually, and duplication is a direct contributor to that bill.

The review has also been shaped by European pressure. The EU's cultural heritage interoperability standards, developed partly through Europeana — the pan-European digital library based in The Hague — require member institutions to submit clean, deduplicated metadata if they want their collections listed on the platform. Paris institutions have lagged behind counterparts in Amsterdam and Madrid in meeting those standards, and the gap has become a reputational irritant for a city that positions itself as a global cultural capital.

The practical path forward involves a phased deduplication programme, starting with the highest-traffic image categories — architectural photography, Seine riverscape images, and historical street documentation from before 1950. Institutions involved are expected to adopt a common metadata schema aligned with the RIOS standard used by several other French national bodies. Staff retraining at the Carnavalet and at the Médiathèque de l'Architecture et du Patrimoine will be part of the rollout.

For Parisians and researchers who rely on these archives — from doctoral students at the Sorbonne to heritage consultants working on Haussmann-era renovation permits in the 10th arrondissement — the practical advice for now is straightforward: cross-reference any image pulled from a municipal database against at least one secondary source before committing to rights assumptions. The cleanup is coming, but it will take time.