Stockholm's municipal digital archive holds hundreds of thousands of images, street scenes from Södermalm, construction records from the Slussen redevelopment, heritage photographs from Gamla Stan. A significant portion of those files, archivists now acknowledge, are duplicates: the same image stored under different file names, in different folders, sometimes at different resolutions, uploaded by different departments that never talked to each other.
The problem did not appear overnight. It accumulated across roughly three decades of fragmented digitisation work, each wave of scanning and uploading governed by whichever department happened to be running a project at the time, with no single city-wide standard enforced until recently.
A Long History of Siloed Systems
The roots of the duplication crisis trace back to the mid-1990s, when Stockholm Stadsarkivet, the city's official record office on Kungsklippan, began its first systematic push to digitise photographic collections. The work was painstaking and expensive, and different city departments handled their own holdings independently. Stockholms Stadsmuseum on Ryssgården digitised its collection separately. The urban planning office, Stadsbyggnadskontoret, ran its own scanning programme. So did the traffic authority, the parks department, and several district councils.
By the 2010s, Stockholm had at least four major internal image repositories running in parallel, none of them fully synchronised. When the city migrated toward a unified content management platform around 2017, staff began importing files from all those legacy systems at once. Deduplication logic was not built in from the start. The result: the same photograph of, say, a 1970s tram on Vasagatan might appear in the archive under three different reference numbers.
A 2023 internal review by Stadsarkivet found that in one sample set of 40,000 digitised images, roughly 18 percent were near-identical duplicates, different file sizes or slight colour corrections, but functionally the same picture. That figure alarmed city information managers, both for the storage costs it implied and for the confusion it created when journalists, researchers, or municipal departments tried to retrieve images for official use.
The Clean-Up Campaign
Since January 2026, Stadsarkivet has been running a structured duplicate-identification programme using perceptual hashing software, a technology that compares images based on visual content rather than file names or metadata. The tool flags pairs or clusters of images that are visually identical or near-identical, which archivists then review manually before any deletion takes place. No file is removed without a human sign-off, a safeguard the office built in after a smaller 2021 pilot accidentally flagged a set of legitimately distinct before-and-after construction photographs as duplicates.
The current programme covers the archive's post-1945 photographic holdings first, roughly 210,000 images, with pre-war and early-20th-century material scheduled for a second phase beginning in early 2027. Storage alone for the full municipal image library costs the city a recurring annual budget line; reducing the duplicate count is expected to trim that overhead, though Stadsarkivet has not published a specific projected saving.
For ordinary Stockholmers, the practical effects are indirect but real. Journalists requesting archive images through the city's public access portal, journalists from publications including this one, have long encountered the frustration of search results returning four near-identical versions of the same photograph, with no clear indication of which is the definitive or highest-quality copy. Once the cleanup is complete, Stadsarkivet plans to implement a canonical-file tagging system, so each unique image has a single authoritative record regardless of how many legacy copies existed before.
The project also raises a broader civic question about how Stockholm manages its visual history going forward. The city's new digital governance framework, adopted by Kommunstyrelsen in March 2026, sets binding metadata standards for any image uploaded to municipal systems, standards that should, in theory, prevent the next generation of archivists from inheriting the same mess. Whether those standards hold across every department will depend on enforcement, not just policy documents.