Zurich's Digital Archives Push Forward on Duplicate Image Purge — Here's What Changed This Week

Zurich's two largest public digital image repositories moved this week to formalise a shared protocol for eliminating duplicate photographs — a technical housekeeping issue that has quietly ballooned into a governance headache touching everything from historical preservation to data-storage costs. ETH Zurich's library directorate and the Stadtarchiv Zürich, based on Alfred-Escher-Strasse, confirmed they are piloting a joint deduplication framework that went into active testing on Monday, July 1.

The timing is not accidental. Swiss federal data-retention rules updated in January 2026 require cantonal and university institutions to demonstrate leaner digital asset management by the end of the third quarter. For organisations still running parallel image catalogues — some inherited from legacy systems dating back to the late 1990s — the compliance clock is now ticking loudly.

What the Problem Actually Looks Like

Duplicate images accumulate in ways that sound mundane but compound fast. A photograph of Paradeplatz taken during the UBS-Credit Suisse merger proceedings in 2023, for instance, might exist in four separate archival buckets: the Stadtarchiv's press collection, ETH's economics documentation project, a cantonal government folder and a digitised newspaper scan from the Neue Zürcher Zeitung's historical partnership with the Swiss National Library. Each copy carries slightly different metadata, different file-naming conventions and, crucially, different access permissions.

The Swiss Federal Archives in Bern estimated in a 2025 audit — covering 14 cantonal institutions — that between 18 and 23 percent of stored digital images across Swiss public repositories were functional duplicates consuming storage that cost an average of CHF 0.04 per gigabyte per month in managed cloud infrastructure. For a mid-sized city archive holding tens of millions of files, that adds up to a recurring line item that elected officials on Zurich's Gemeinderat have started asking questions about.

ETH Zurich's library, located on Rämistrasse 101, began deploying a perceptual-hashing algorithm in its image stacks in May. The tool generates a short numerical fingerprint for each photograph and flags near-identical matches — catching not just exact copies but also slightly cropped or recompressed versions of the same image. By the end of June, the system had processed roughly 2.4 million files in ETH's open-access science image collection and flagged approximately 310,000 candidates for human review.

What Happens to Flagged Images — and Who Decides

Flagging is not the same as deleting. Both the Stadtarchiv and ETH have confirmed that no image is removed without a human archivist signing off, a procedural safeguard that reflects broader Swiss caution about automated decision-making in public administration. The review queue at the Stadtarchiv, which holds the city's visual record stretching back to the 19th century, is being handled by a team working out of the reading room on Alfred-Escher-Strasse, with a target clearance rate of 5,000 reviewed files per week.

For journalists and researchers accessing the Bildarchiv — Zurich's centralised municipal photo database — the practical change arriving this week is a new status tag visible in search results. Duplicate candidates now carry a yellow flag icon alongside their catalogue entry, alerting users that the file may be superseded by a master copy elsewhere. The tag went live on July 2.

Housing researchers using archival imagery to document the Wohnungsnot crisis — particularly visual records of Zürich-West and the Langstrasse corridor where densification has transformed the streetscape over the past decade — will find that consolidating duplicates also means cleaner metadata on building permits and demolition records attached to images. That linkage between photographs and administrative data is what makes the deduplication project more than a storage exercise.

The pilot is scheduled to run through September 30, after which the two institutions plan to publish a joint methodology report that smaller Swiss municipal archives in Winterthur and Baden could adopt. Anyone needing access to specific historical image files held at the Stadtarchiv should contact the reading room directly before August, when a scheduled system migration will temporarily restrict remote access to the Bildarchiv's oldest digitised collections.