Zurich's Digital Archive Crisis: The Key Decisions Ahead on Duplicate Image Replacement

Thousands of duplicate images are sitting inside Zurich's public digital archives, consuming server space, distorting search results, and costing municipal institutions money they don't have spare. The city's main cultural and administrative bodies are now under growing pressure to agree on a coordinated approach to duplicate image replacement — and the decisions made in the coming months will shape how Zurich manages its digital heritage for the next decade.

The issue has gained urgency because of a deadline. The City of Zurich's IT and digitisation strategy, adopted in 2023, set a 2026 review point at which all major public institutions were expected to demonstrate progress on digital collection hygiene — including the removal or replacement of redundant image assets. That review window is now open, and several institutions have yet to submit compliance plans to the Stadtrat.

Where the Problem Is Concentrated

The Stadtarchiv Zürich on Alfred-Escher-Strasse and the Zentralbibliothek Zürich near the Predigerplatz are carrying the heaviest loads. Both institutions digitised large volumes of photographic material between 2018 and 2022, often ingesting images from multiple sources without deduplication protocols in place. The result is collections where the same photograph, map, or illustration can appear under three or four separate catalogue entries, each carrying its own metadata — sometimes contradictory — and each occupying independent storage allocation.

At the Zentralbibliothek alone, internal assessments reported to the Zurich cantonal government indicated that duplicate or near-duplicate files accounted for a significant share of total image holdings, straining both the institution's storage budget and its public-facing search interface. The library's e-rara platform, which serves researchers across the German-speaking world, has been flagged by academic users for returning cluttered results when querying historical Swiss periodicals.

ETH Zürich's library division, based on the main Rämistrasse campus, has taken a more aggressive approach, deploying perceptual hashing algorithms since early 2025 to identify near-duplicate images in its scientific image repository. The method compares visual fingerprints rather than file metadata alone, catching duplicates that escaped earlier rule-based filters. ETH's approach is now being studied as a possible template for city-level adoption, though cost remains a sticking point: licensing comparable software for municipal use is estimated to run into six-figure annual expenditure in Swiss francs.

The Decisions That Will Define the Outcome

Three choices now sit on the table for city administrators and cultural directors. First, whether to centralise duplicate detection under a single city-wide digital infrastructure unit, or leave each institution to procure and run its own tooling. Centralisation offers economies of scale but requires political agreement across bodies that have historically guarded their operational independence.

Second, institutions must decide what replacement means in practice. Deleting a duplicate outright risks losing variant metadata attached to the secondary file — metadata that archivists argue can carry independent research value. The alternative, merging records, is labour-intensive and requires trained staff that the Stadtarchiv, operating with a core team of under 30 archivists, does not currently have in surplus.

Third, there is the question of public transparency. Zurich's direct democracy tradition creates a reasonable expectation that residents can scrutinise decisions about public collections. A proposal circulating within the cultural department would require any mass deletion of archive images above a certain volume threshold to go through a formal consultation process — slowing the technical work but giving the public a meaningful voice.

The Stadtrat is expected to receive consolidated recommendations from the relevant department heads before the summer recess ends in mid-August 2026. If agreement is reached then, pilot programmes at one or two institutions could begin in the fourth quarter of this year, with city-wide rollout targeted for 2027. If consensus breaks down — particularly over budget allocation — the process moves to a formal referendum track, which under Zurich's cantonal rules could add another 12 to 18 months before any binding decision. For archivists watching server costs rise and search quality decline, that delay is not abstract: it is measured in francs and in frustrated researchers.