How Berlin's Public Image Archive Ended Up Full of Duplicates — and What It's Costing the City

Berlin's Senate Chancellery confirmed last month that a systematic audit of the city's centralised digital image archive — used by everything from the BVG's marketing teams to Senatsverwaltung für Stadtentwicklung planning documents — had uncovered tens of thousands of duplicate files clogging shared servers. The audit, initiated in late 2025 under the SPD-led coalition's broader digital infrastructure reform push, found that redundant image assets across senate departments were consuming roughly 40 percent of allocated storage on the city's primary content management platform.

The problem did not appear overnight. It accumulated over nearly two decades of parallel digitisation drives, each department building its own workflow with little coordination from the centre. Understanding how Berlin arrived here matters because the fix is neither cheap nor fast — and because the same structural failures that created this mess are still shaping how the city handles data in 2026.

A Patchwork System Built Department by Department

The roots run back to the mid-2000s, when Berlin's cash-strapped administration — the city was effectively broke after the Bankgesellschaft Berlin collapse — could not afford a unified digital asset management system. Departments improvised. The Senatsverwaltung für Kultur began scanning exhibition archives independently. The Stadtentwicklung office digitised planning photos using a different software vendor. BVG, as a separately incorporated public transport company, ran its own image library entirely. By the time the city's IT consolidation body, ITDZ Berlin, started pushing for integration around 2015, there were at least seven distinct cataloguing conventions in active use across the administration.

The result was predictable. When departments eventually gained access to shared storage — hosted on servers at the ITDZ data centre on Berliner Allee in Weißensee — they uploaded existing local archives wholesale, without deduplication. A single photograph of Alexanderplatz, for instance, might exist in the system under three different filenames, two different aspect ratios, and four departmental folders. Multiply that across thousands of images and years of uploads, and the archive became structurally unreliable. Staff couldn't trust search results. Projects were delayed. Teams started maintaining shadow archives on local hard drives, which made the central system even less useful.

A 2023 internal review by ITDZ Berlin — the results of which were reported by the Tagesspiegel at the time — estimated that storage inefficiency was adding roughly €800,000 annually in avoidable infrastructure costs across the consolidated server estate. That figure does not include staff time lost to manual workarounds.

The Cleanup and What Comes Next

The current deduplication project, which the Senate Chancellery linked to its Digitalisierungsstrategie 2025–2030 roadmap, is being carried out in phases. Phase one, covering the Stadtentwicklung and Finanzen departments, was scheduled for completion by June 30, 2026. Phase two, which brings in the cultural institutions clustered around the Kulturforum on Matthäikirchplatz, is set to run through the end of the year. BVG is negotiating a separate integration agreement that would finally connect its image library to the central system, though no signed contract has been announced.

The practical consequences for ordinary Berliners are indirect but real. Delays in planning document publication — some traced to archive failures — have slowed public consultation processes in fast-developing areas like Tempelhof-Schöneberg and the Heidestraße corridor near the Hauptbahnhof. Housing campaign materials produced by the Senate's Wohnraumversorgung Berlin were, according to a parliamentary question filed in March 2026, reprinted at least twice due to incorrect image versions being pulled from the archive.

Going forward, the city's digital reform office has specified that all new image uploads to the central system must pass an automated hash-check before ingestion — a standard deduplication technique that most large organisations implemented years ago. Departments have until January 1, 2027 to comply. For a city that spent years building seven different answers to the same question, that deadline is both overdue and, on past form, optimistic.