Zurich Discovers Duplicate Images Clogging Digital Archives, Plans Major Fix

Zurich's municipal digital archive contains hundreds of thousands of duplicate image files. That is the central finding driving a remediation effort now underway across several city departments, including the Stadtarchiv Zürich on Alfred-Escher-Strasse and the digital services unit housed within Stadt Zürich's IT-Gsetz programme. The duplication problem is not new — but the cost of ignoring it has finally grown large enough that administrators can no longer look the other way.

The issue matters now for a straightforward reason: the city is mid-way through a broader digitisation push that has accelerated since 2022, when Zurich committed to expanding e-government services under the national E-Government Strategie Schweiz framework. Migrating legacy image databases into new content management systems exposed what years of siloed departmental uploading had quietly produced — the same scanned photographs, planning documents, and heritage images stored multiple times under different filenames, sometimes across three or four separate repositories.

A Problem Baked In by Bureaucratic Structure

The roots of the duplication crisis trace back to the early 2000s, when individual Zurich departments began digitising records independently, without a unified metadata standard. The Baugeschichtliches Archiv, located in the Predigerkirche complex in Niederdorf, digitised thousands of historical urban-planning photographs. Simultaneously, the Stadtarchiv was scanning its own holdings. Neither institution had a shared deduplication protocol. When the city later attempted to consolidate records into the central GEVER document management system — the Swiss federal standard — identical files arrived from multiple sources carrying different identifiers.

The problem was compounded by normal administrative turnover. Staff who understood the original filing logic left, and replacements, lacking documentation, uploaded fresh copies of images already in the system rather than risk losing material. By the time ETH Zürich's chair for information science conducted an informal audit of three city departments in 2024, preliminary findings suggested duplication rates in some image collections exceeded 30 percent. That figure, while not yet officially published for the full municipal system, aligned with what infrastructure teams had been telling internal project managers for months.

Storage costs alone make the situation untenable. Enterprise-grade archival storage in Swiss data centres — the city uses facilities that meet the strict Swiss Federal Archives guidelines on sovereignty and security — runs at a significant premium compared to commercial cloud solutions. Maintaining redundant copies of tens of thousands of high-resolution image files is not a trivial expense for a city already managing a housing shortage that has pushed one-bedroom rents in Zürich Kreis 4 past CHF 2,000 per month, squeezing every line of the public budget.

What Remediation Actually Looks Like

The current remediation project, which city IT project documents indicate formally launched in the first quarter of 2026, involves a combination of perceptual hashing algorithms — software tools that can identify visually identical images even when file names or formats differ — and manual curatorial review for edge cases. The Stadtarchiv and the Baugeschichtliches Archiv are the first two institutions brought into the programme. Cantonal guidelines require that no original file be deleted without a secondary sign-off, which slows the process but protects against accidental data loss.

For residents and researchers, the immediate practical consequence is limited — public-facing portals like the e-Pics image database remain accessible throughout. But genealogists, urban historians, and architects who regularly consult the Baugeschichtliches Archiv have noticed intermittent slowdowns in search response times over the past eighteen months, a direct symptom of bloated databases struggling with redundant indexing.

The longer lesson is institutional. Switzerland's direct-democracy model means Zurich voters periodically approve or reject technology budgets at the ballot box. The next IT infrastructure credit package is expected to come before city parliament by late 2026. How clearly officials can explain what went wrong — and demonstrate that the deduplication project is fixing it — will shape how comfortable Zürich Stadtrat members are recommending a yes vote to their constituents. Getting the archive clean is, in that sense, not just a technical task. It is a political one.