How Stockholm's Digital Archives Ended Up Full of the Same Image Twice — and What's Being Done About It

Stockholm's municipal digital archive holds hundreds of thousands of images — photographs of planning applications, construction sites, public spaces, and heritage buildings stretching back to the mid-1990s. A significant share of them, according to internal reviews discussed by Stadsarkivet, the city's official archive authority, are duplicates: the same photograph uploaded multiple times under different file names, different case numbers, or different dates.

The problem did not arrive overnight. It accumulated across three decades of digitisation drives, departmental software migrations, and a persistent lack of unified standards for how city departments submit visual documentation. Understanding how Stockholm got here matters now because the city is midway through a 2024–2027 digital infrastructure overhaul, and duplicate image records are directly obstructing the automation tools meant to speed up planning permit reviews in neighbourhoods like Hagastaden and Järva.

A Problem Built Layer by Layer

The roots of the duplicate image problem trace to the late 1990s, when individual förvaltningar — the administrative units that run everything from road maintenance to school buildings — began scanning physical documents into their own siloed databases. Stadsbyggnadskontoret, the city planning office on Fleminggatan, used one system. Trafikkontoret used another. Exploateringskontoret, responsible for land development, ran a third. None of them talked to each other in any automated way.

When the city began consolidating these systems in the early 2010s under the Stockholm e-förvaltningsprojekt framework, engineers essentially merged the databases without a deduplication pass first. An image of a building facade on Götgatan that had been filed by three separate departments showed up three times in the merged archive. Multiply that across tens of thousands of planning cases handled over fifteen years, and the arithmetic becomes uncomfortable.

A 2022 internal audit — details of which were later referenced in a Kommunfullmäktige budget document — found that roughly 18 percent of image files in the primary planning archive were either exact duplicates or near-identical copies differing only in compression or watermark. That figure, applied to a repository then holding around 340,000 image records, means approximately 61,000 files were redundant. Storage costs for the city's primary data centre in Kista have risen accordingly, though the archive authority has not published a precise figure for what the duplication specifically costs per year.

Why the Fix Has Taken So Long

The delay has not been purely technical. Stadsarkivet operates under Swedish archival law — Arkivlagen — which imposes strict obligations about what can be deleted from a public record, even when the record is a redundant copy. Deleting an image file requires demonstrating that it carries no independent legal or evidential value. When two files are not pixel-perfect matches — one might be a slightly higher-resolution scan from a later batch digitisation project — archivists cannot automatically delete either without a case-by-case review.

That review process is manual. With tens of thousands of files in question, the backlog built up fast. The city brought in Riksarkivet, the national archives authority based in Marieberg, to help draft a legal framework for safe deletion of confirmed duplicates, a process that began formally in March 2025 and is expected to produce binding guidance before the end of this year.

Meanwhile, the 2024–2027 infrastructure project has introduced AI-assisted image comparison tools, piloted first on the archive's Södermalm planning district files. The tools flag probable duplicates for human sign-off rather than deleting automatically — a workaround that satisfies Arkivlagen but still requires staff time.

For residents and developers navigating the planning system, the practical impact shows up in processing times. A permit application in Östermalm that references photographs already in the archive can still trigger a duplicate-file flag that stalls the automated workflow, pushing the case to a manual queue. The city's own target of processing standard permit applications within ten weeks has proved difficult to meet consistently in districts where archive duplication rates are highest.

The Riksarkivet guidance, expected in late 2026, should give Stadsarkivet the legal clarity to begin systematic deletion. Until then, the advice from Stadsbyggnadskontoret for anyone submitting planning documentation is straightforward: use the reference numbers provided at submission to verify that images are not already on file, and avoid re-uploading photographs from previous applications on the same property.