Zurich's Duplicate Image Crisis: The Key Decisions Ahead for the City's Digital Archive

Thousands of duplicate images sit inside Zurich's municipal digital archive, and the question of what to do with them is no longer theoretical. The Stadtarchiv Zürich, headquartered on Alfred-Escher-Strasse, has spent the better part of three years digitising historical photographic collections spanning the city's postwar reconstruction, the Kreis 4 and Kreis 5 urban renewal campaigns of the 1980s, and the SBB rail corridor projects along Langstrasse. The problem is that bulk scanning, compressed timelines, and shifting vendor contracts have left the archive with a significant proportion of records filed in duplicate or near-duplicate form, consuming server capacity and complicating public access through the online portal.

Why does this matter now? The Stadtarchiv's digital collection feeds directly into the broader Zurich Open Government Data initiative, which the city has committed to expanding through 2027 under its Smart City Zürich programme. Duplicate records inflate apparent collection size, degrade search accuracy, and slow retrieval speeds for researchers, journalists, and the growing number of residents who use the portal to trace property histories during the ongoing Wohnungsnot housing crisis—when a tenant disputes a landlord's renovation claims, for example, archived building permits and photographs become legal reference material. Errors in that archive carry real consequences.

The immediate decision point falls to the archive's leadership and to the city's Department of Digitisation and Information Management, which operates partly out of the Stadthaus on Stadthausquai. By September 2026, administrators are expected to present a preferred methodology to the city council's committee on cultural infrastructure. Three broad options are on the table. The first involves deploying a commercial AI-based deduplication tool—several Swiss IT vendors active in the Zurich market have already submitted preliminary proposals. The second is a manual curation process, assigning additional specialist staff to work through the backlog collection by collection. The third is a phased hybrid: run an automated first pass to flag likely duplicates, then route flagged items to trained archivists for final decisions.

The Cost and Capacity Question

Budget is the central constraint. The Stadtarchiv's annual digitisation allocation for 2026 is understood to run in the low-to-mid single-digit millions of Swiss francs, and any new deduplication programme would compete with commitments already made to digitise the ETH Zürich-adjacent Bildarchiv collection and the Zentralbibliothek Zürich's map holdings on Zähringerplatz. Staff expansion is not straightforward either: qualified digital archivists with experience in Swiss-German municipal records are a narrow hiring pool, and salary benchmarks in the canton make recruitment from outside Zurich expensive. The city's HR frameworks under the Personalamt set pay grades that limit how quickly new specialist roles can be stood up.

There is precedent elsewhere in Switzerland. The Staatsarchiv Basel-Stadt completed a comparable deduplication exercise in 2024, having worked through roughly 180,000 flagged image records over eighteen months using a hybrid model. The Basel experience suggests that fully automated tools, while fast, produce an unacceptably high rate of false positives when applied to historical photographic material where two nearly identical shots from the same camera session should sometimes both be retained for scholarly reasons. The lesson for Zurich's administrators is that speed and accuracy are not the same target.

What Happens Next

The September presentation to the city council committee will not be the final word. Council members are likely to request a pilot phase—possibly confined to one discrete sub-collection, such as the Limmatquai streetscape photographs from the 1960s—before committing to a citywide approach. That pilot, if approved in autumn, would run through the first quarter of 2027, with results feeding into the broader Smart City Zürich progress report scheduled for spring of that year.

For researchers and residents who rely on the archive, the practical advice is straightforward: when retrieving records through the online portal today, cross-reference any single image against at least two catalogue entries before treating it as unique documentation. The Stadtarchiv's reading room on Alfred-Escher-Strasse accepts in-person appointments on weekday mornings, where staff can help verify whether a suspect duplicate actually represents a distinct record. Until the city settles on its methodology and works through the backlog, that human check remains the most reliable safeguard available.