Zurich's Archive Institutions Race to Tackle Duplicate Image Problem Exposed This Week

A city-wide review of digital image holdings, completed this week, has confirmed what archivists at several Zurich institutions had long suspected: tens of thousands of duplicate photograph files have been silently consuming server space and distorting public search results for years. The audit, coordinated across repositories including the Stadtarchiv Zürich on Alfred-Escher-Strasse and the Zentralbibliothek Zürich on Zähringerplatz, identified redundant files running into the hundreds of gigabytes across shared cataloguing systems.

The timing matters. Zurich's major cultural institutions have spent the past three years migrating analogue holdings into unified digital platforms, partly funded through a 2023 cantonal programme aimed at making historical records searchable by the public. Duplicate images — the same photograph ingested multiple times under different metadata tags — quietly degrade that investment. A researcher querying the holdings for, say, a 1940s image of Bahnhofstrasse can now receive the same file returned four or five times in a single results page, a problem librarians describe as both an efficiency drain and a credibility issue for digitisation work.

How the Problem Accumulated

The root cause is institutional. When Credit Suisse's collapse in 2023 triggered budget reassessments across cantonal departments, several digitisation projects were handed off between teams or merged mid-stream. Files ingested by one contractor were later re-ingested by a successor without deduplication checks. The Stadtarchiv Zürich alone holds more than 1.2 million digitised items, according to its published catalogue figures, and even a small percentage of duplication compounds quickly at that scale.

ETH Zürich's library, based on Rämistrasse, runs a separate digitisation pipeline but participates in the Swiss national aggregator platform, which feeds into Europeana. Cross-institutional uploads mean that a duplicate originating in one Zurich collection can propagate into the national index and then into the European one — a cascade that takes considerably more effort to unwind than the initial error required to create. The Swiss Federal Archives in Bern flagged this propagation risk in a technical note circulated to cantonal partners in April 2026, though the specific scale of cross-border contamination from Zurich holdings has not been publicly quantified.

Practically, the costs are not trivial. Commercial cloud storage for large cultural institutions in Switzerland runs at roughly CHF 0.022 per gigabyte per month on standard enterprise contracts. For a repository carrying several hundred gigabytes of redundant data, the annual waste runs to several thousand francs — modest individually, but significant when multiplied across a canton with dozens of smaller municipal collections from districts like Oerlikon, Wiedikon, and Schwamendingen that feed into the central system.

What Institutions Are Doing Now

The Zentralbibliothek has confirmed it will run a phased deduplication process through August, using perceptual hashing software that compares image content rather than just filename or file size — a method more reliable for catching items re-scanned from the same physical original at different resolutions. The Stadtarchiv is understood to be evaluating the same toolset, though its timeline has not been announced publicly.

For members of the public who rely on these archives — historians, journalists, architects researching Zurich's built environment — the near-term effect will be temporary gaps in the online catalogue as files are reviewed, merged, or removed. The Zentralbibliothek's digital portal is expected to show reduced availability for parts of its photographic holdings during the remediation window in July and early August.

Longer term, the audit is pushing a conversation that archivists across the German-speaking world have been deferring: whether individual institutional ingestion pipelines should be replaced by a shared ingest layer that runs deduplication before files ever enter a permanent repository. That kind of infrastructure requires both political will and capital investment. Zurich's experience this week gives that argument fresh, locally documented evidence to draw on.