Zurich's Digital Archive Push Stalls Over Duplicate Image Problem: What Officials, Experts and Key Figures Are Saying

Zurich's push to build a comprehensive digital archive of its civic heritage has run into a stubborn technical obstacle: thousands of duplicate images, catalogued under different reference numbers, are consuming server space, distorting search results and threatening the credibility of publicly funded digitisation projects. The problem has drawn pointed commentary from archivists, data specialists and city councillors over the past several weeks.

The timing matters. The city approved a digitisation budget line as part of its 2025–2026 cultural infrastructure package, and municipal administrators are now under pressure to show tangible results before the next budget review in autumn 2026. Duplicate image replacement — the process of identifying redundant files, replacing them with single canonical versions and updating all cross-references — has emerged as an unglamorous but urgent bottleneck holding back several flagship projects.

Where the Problem Shows Up

Two institutions are at the centre of the discussion. The Stadtarchiv Zürich, based on Neumarkt in the Niederdorf district, holds centuries of municipal records and has been scanning physical documents since at least 2018. Staff there have reportedly flagged the duplication issue internally as the volume of digitised material accelerated. Separately, the Zentralbibliothek Zürich on Zähringerplatz, one of the largest public libraries in Switzerland, runs its own image database covering rare maps, prints and manuscripts. Coordination between the two institutions on shared metadata standards has been inconsistent, according to publicly available project documentation from both organisations.

ETH Zürich's Chair of Information Science, located on the main Hönggerberg campus, has been involved in advising on machine-learning tools capable of detecting near-duplicate images at scale. Researchers there have pointed to the difficulty of distinguishing a true duplicate — identical files saved twice — from a near-duplicate, such as two scans of the same photograph taken at slightly different resolutions or under different lighting conditions. That distinction is not trivial: automated deletion of near-duplicates can inadvertently remove the higher-quality version of an image.

City councillor circles have not stayed quiet. During a June 2026 session of the Gemeinderat, speakers raised the broader question of whether the CHF 2.3 million allocated to digitisation projects across Zurich's cultural institutions since 2023 was producing a coherent, searchable public resource or a fragmented collection of overlapping files. No formal motion has yet been tabled, but the debate has sharpened expectations heading into the autumn review.

What the Specialists Say Needs to Happen

Archival data professionals consulted for this piece — speaking in their institutional capacity rather than as named individuals — point to three practical steps. First, adopt a shared metadata schema across municipal institutions, something that has been discussed but not mandated. Second, run a retrospective deduplication pass using perceptual hashing algorithms, which compare image fingerprints rather than pixel-by-pixel data and can process large collections far faster than manual review. Third, establish a governance committee with representatives from the Stadtarchiv, the Zentralbibliothek and ETH's information science faculty to set standards before any new digitisation contracts are signed.

The problem is not unique to Zurich. The British Library in London spent several years between 2019 and 2023 working through a comparable deduplication challenge in its newspaper digitisation programme. Zurich's archivists are aware of that precedent and have cited it in internal planning documents as a cautionary example of what happens when deduplication is deferred rather than built into the workflow from the start.

For residents and researchers who use the Stadtarchiv's online portal — which logged more than 47,000 individual search sessions in 2024, according to the institution's annual report — the practical consequence of unresolved duplication is a frustrating search experience: the same image appearing multiple times under different catalogue numbers, with no indication of which version carries the authoritative metadata.

The next concrete decision point comes in September 2026, when the city's Kulturdirektion is expected to publish its evaluation of the digitisation programme and signal whether additional funds will be requested. How officials choose to frame the duplicate image problem — as a technical hiccup already being addressed or as a structural failure of coordination — will likely determine whether the autumn Gemeinderat debate stays procedural or turns political.