By the Numbers: Paris's Hidden Crisis of Duplicate and Ghost Images Clogging the City's Digital Archives

Paris city hall is sitting on a growing pile of redundant photographs. Across the Direction de la Communication and affiliated municipal services, internal audits completed in the first quarter of 2026 found that duplicate or near-duplicate images account for a significant share of archived digital assets — creating real costs at a moment when the city is trying to leverage its post-Olympic media library to promote tourism and attract investment.

The problem is not unique to Paris, but the scale here has been amplified by a specific chain of events. The Paris 2024 Olympics generated an extraordinary volume of photography — event images, venue construction shots, legacy documentation — spread across multiple agencies, arrondissement offices, and partner organisations including Apur, the Paris urban planning agency. When those files were merged into shared servers in 2025 as part of the city's post-Games digital consolidation, deduplication was either incomplete or not systematically applied. The result is a bloated archive that costs more to store, slower to search, and harder for communications teams to actually use.

What the Numbers Actually Show

The scale of the problem emerges clearly from the data. Industry benchmarks from digital asset management consultancies operating in the European public sector suggest that unmanaged municipal photo archives routinely carry between 30 and 45 percent duplicate content. Applied to Paris's reported archive size of roughly 2.8 million catalogued images across city services — a figure cited in a 2025 budget annex presented to the Conseil de Paris — that would put the number of redundant files anywhere between 840,000 and 1.26 million assets.

Cloud storage for public bodies in France typically runs at commercial rates near €0.023 per gigabyte per month under state-negotiated contracts through the UGAP procurement platform. High-resolution event photography from a venue like the Stade de France or the Trocadéro fan zone can run to 25 megabytes per raw file. At that file size, even 500,000 duplicates represent roughly 12.5 terabytes of redundant data — and a recurring annual storage cost running into tens of thousands of euros before personnel and licensing are factored in.

The problem compounds because duplicate images do not simply sit inert. They surface in search results, confuse cataloguing staff at institutions like the Bibliothèque de l'Hôtel de Ville on rue de Rivoli, and create legal ambiguity around rights clearance when the same image exists under multiple metadata records with different attribution fields. Paris Musées, the federation overseeing the city's fourteen municipal museums, flagged rights-management discrepancies in digitised collections as a standing concern in its 2024 annual report.

What the City Is — and Isn't — Doing

The Grand Paris Express project, which is generating thousands of new site-documentation photographs weekly across the Île-de-France construction corridor, risks compounding the archive problem further. Société du Grand Paris, the public body managing the metro extension, works with dozens of separate contractors who submit visual records independently, often without centralised deduplication checks.

Paris city hall launched a Digital Transformation Plan in 2023 with a four-year mandate and a stated budget envelope of €45 million. Automated deduplication tools — perceptual hashing technology that can identify visually similar images even if file names or metadata differ — are included in the plan's Phase 2 deliverables, scheduled for completion by the end of 2026. Whether that timetable holds given National Assembly budget pressure on municipal transfers from the state is a live question that will be tested in the autumn finance round.

For communications officers, archivists, and the agencies managing the Seine-Saint-Denis regeneration corridor that grew out of the Olympics legacy programme, the practical advice from digital asset specialists is not to wait for the city-wide system. Applying open-source tools such as the Python-based imagededup library, or commercial platforms already in use at major European municipalities, to departmental archives now can reduce storage overhead immediately and produce cleaner datasets ahead of the central platform's arrival. The window is narrow. Every week of new image ingestion — from construction sites, from cultural programming at the Grande Halle de la Villette, from Seine riverbank regeneration events — makes the eventual cleanup more expensive.