How Zurich's Digital Archives Ended Up Full of Duplicate Images — and What the City Is Doing About It

Zurich's public institutions are sitting on a sprawling, overlapping mess of digital images. The Stadt Zürich's administrative departments, together with cultural bodies such as the Zentralbibliothek Zürich on Zähringerplatz and the Stadtarchiv on Neumarkt, have collectively accumulated digital photo collections estimated in the hundreds of thousands of files — with internal reviews suggesting that in some repositories, duplicate or near-duplicate images account for a significant share of stored assets. The exact figures vary by institution, but the problem is structural, not accidental.

Understanding how this happened requires going back roughly fifteen years. Between 2010 and 2018, Swiss municipal bodies, universities and cultural institutions each launched their own digitisation drives, often funded through separate Federal Office of Culture grants and canton-level IT budgets. There was no unified metadata standard, no shared deduplication protocol and no common platform. ETH Zürich's library, the Stadtarchiv and cantonal agencies each built their own systems. When staff transferred files between departments — or when contractors delivered digitisation batches — the same image frequently entered multiple databases under different file names, different timestamps and occasionally different resolutions.

Smartphones and the Second Wave

The second major inflection point came after 2015, when smartphone cameras became standard equipment for city staff, event photographers and institutional communicators. The Stadtarchiv, which holds records dating back to the medieval period alongside modern photographic collections, began receiving unfiltered bulk uploads from departments across the city. The same scene from a Rathaus press conference or a Langstrasse neighbourhood event might arrive in four versions — shot by four different phones, uploaded by four different staff members, none of them aware the others had done the same.

Cloud storage made the problem cheaper to ignore. Storage costs dropped steadily through the early 2020s, and the administrative incentive to deduplicate — which requires staff time, software licensing and editorial decisions about which version of an image to keep — simply did not exist at the budget level. By 2023, the Zentralbibliothek Zürich alone had flagged the duplicate image problem in an internal digitisation review, though the institution has not published specific figures publicly.

The UBS-Credit Suisse merger of 2023 added an indirect pressure. Several Zurich-based financial communications teams, rationalising their own digital asset management after the consolidation, turned to vendors offering AI-assisted deduplication tools. That market activity helped legitimise the technology for public-sector buyers watching from across the Limmat. By mid-2025, at least two cantonal IT procurement tenders referenced automated image-comparison functionality as a required feature — a first for Swiss municipal procurement language.

The Path to a Fix

Deduplication is not simply a matter of deleting identical files. Near-duplicate detection — identifying images that are visually similar but differ in crop, brightness or compression — requires either manual editorial review or machine-learning classifiers trained on the specific content types in a given archive. For a body like the Stadtarchiv, which must weigh archival completeness against storage efficiency, deleting what appears to be a redundant image carries legal and historical risk. A higher-resolution version of a 1970s Zürich-West construction photograph, for instance, may carry different evidential value than a lower-resolution copy filed under the same event.

The Swiss Federal Archives in Bern published guidance in January 2025 on managing digital duplicates in public collections, recommending a phased approach: automated flagging first, human review for items predating 2000, and bulk deletion protocols only for assets created after 2010 and held in more than three identical copies. Zurich's institutions are broadly aligning with that framework, though implementation timelines differ.

For organisations outside the public sector — the many Zurich-area pharmaceutical companies along the Glatttal corridor, for instance, or private cultural foundations in Zürich-Seefeld — the practical advice from digital asset management consultants is to audit before procuring. Running a deduplication pass on an existing library before migrating to a new platform can reduce storage costs and, more importantly, prevent the problem from embedding itself in whatever system comes next. The window to do that cheaply is now, before the next generation of AI-assisted workflows bakes the redundancy in deeper.