Zurich's Digital Archive Drive Hits a Wall: Officials and Experts Sound Off on Duplicate Image Problem

Zurich's bid to consolidate its sprawling municipal image archives into a single searchable platform has run into a concrete obstacle: tens of thousands of duplicate photographs clogging the system, slowing access times and inflating storage costs at a moment when the city is under pressure to trim its digital infrastructure budget.

The problem surfaced publicly this spring, when Stadt Zürich's digitisation team flagged the issue in an internal review presented to the Stadtrat. The archive, which draws from sources including the Stadtarchiv Zürich on Neumarkt and the photographic collections held at Museum Haus Konstruktiv on Selnaustrasse, had accumulated what administrators described as a structurally significant proportion of redundant files — multiple copies of the same image, often ingested separately by different departments over years of uncoordinated uploads.

What the Experts Are Saying

Specialists in digital preservation argue the problem is neither unique to Zurich nor easy to solve. Researchers at ETH Zürich's Chair of Information Science have been examining automated deduplication algorithms as part of broader work on large-scale image dataset management. The core difficulty, according to that research community, is that standard hash-based matching — which flags files as identical based on binary content — fails to catch near-duplicate images: photographs of the same subject taken seconds apart, or slightly re-cropped versions of the same scan. Those require more computationally expensive perceptual hashing or machine-learning classifiers, which introduce their own error rates.

For public archives, the stakes of a false positive are high. Accidentally deleting a photograph that appears to be a duplicate but is actually a distinct document constitutes an irreversible loss under Swiss federal archiving law. The Schweizerische Nationalbibliothek in Bern has dealt with comparable issues in its own digitisation programmes and has cautioned publicly against aggressive automated deletion without human review layers. That caution is echoed by archivists at the Stadtarchiv on Neumarkt, who are understood to be urging a slower, more methodical approach than some city technology officers prefer.

The tension maps onto a familiar institutional fault line. IT departments, facing pressure on server costs — commercial cloud storage rates in Switzerland have climbed sharply since 2024 — want a faster resolution. Archivists and historians push back, citing the irreversibility of deletion and the legal obligations encoded in the Bundesgesetz über die Archivierung. Neither side is wrong, but the gap between them is where backlogs grow.

Practical Implications for Zurich's Digital Infrastructure

The timing matters. Zurich's city administration committed in its 2024–2028 digital strategy to making municipal historical records searchable to the public through a single portal by the end of 2027. That deadline now looks tight. The Stadtarchiv currently holds more than 1.2 million indexed items across its collections, and the image deduplication review is one of several technical prerequisites for the portal launch.

Vendors pitching deduplication tools have been circling. At least two European firms specialising in cultural heritage digitisation presented their products at a closed-door session held at the Rathaus on Limmatquai in May. Neither contract has been awarded publicly, and the procurement process sits under standard Submissionsrecht rules, meaning the city is not required to disclose bids until a decision is made.

What comes next depends largely on which approach the Stadtrat endorses. The two realistic paths are a phased human-assisted review, which is slower and more expensive in staff hours but legally safer, or a hybrid automated-plus-audit system that uses algorithmic flagging to shortlist candidates for human sign-off. The second option is what most technical advisers appear to favour, though no formal recommendation has been issued.

For Zurich residents hoping to access historical photographs of neighbourhoods like Aussersihl or the Langstrasse quarter through the eventual public portal, the practical message is straightforward: the launch timeline is under pressure, and the city's digital ambitions will only move as fast as this unglamorous but consequential data-cleaning work allows.