Zurich's Duplicate Image Problem: The Numbers Exposing a Hidden Digital Waste Crisis

Zurich's public institutions collectively store an estimated 40 percent of their digital image assets as duplicates, according to analysis circulating among IT procurement officers at the city's Stadthaus on Stadthausquai ahead of a planned infrastructure review this autumn. That single figure, if it holds up under formal audit, points to a problem that is simultaneously mundane and expensive: organisations are paying to store the same photograph twice, sometimes dozens of times, across fragmented server environments.

The issue has gained urgency because of timing. The aftermath of the UBS-Credit Suisse merger has pushed large financial institutions based in Zurich's Paradeplatz district to consolidate data infrastructure at speed, exposing legacy archive systems that were never properly deduplicated. At the same time, the canton of Zurich's ongoing push to centralise e-government services under the ch.ch digital platform means city departments are migrating image libraries from siloed local drives into unified repositories — and finding redundancy rates that strain both budgets and processing capacity.

What the Data Actually Shows

Storage is not cheap. Commercial cloud archiving in Switzerland, where data residency laws under the Federal Act on Data Protection require local hosting for sensitive public records, runs at roughly CHF 0.025 per gigabyte per month for cold-tier storage at Swiss data centre operators. That sounds trivial until you multiply it across tens of millions of image files. A mid-sized cantonal department managing urban planning photography — think aerial surveys of the Limmattal valley or construction documentation from the ongoing Rosengartentunnel project — can accumulate several terabytes of raw imagery annually. If 40 percent of that volume is redundant, the wasted spend compounds year on year.

ETH Zurich's Computer Vision Laboratory on Rämistrasse has published research showing that perceptual hashing algorithms — software tools that generate a compact numerical fingerprint of each image — can identify near-duplicate files with accuracy rates above 97 percent, even when images have been resized, recompressed or lightly edited. The same research indicates that running such a deduplication pass across a one-terabyte archive typically completes in under four hours on commodity hardware, making the technical barrier to action very low.

The Zurich Cantonal Archives, the Staatsarchiv on Winkelriedstrasse, began a pilot deduplication programme in March 2025 covering its digitised photograph collection. While the archive has not yet published final results, the project scope document — publicly accessible under cantonal transparency rules — identifies the elimination of redundant derivatives as a primary target, noting that scanned prints were routinely saved in multiple resolution variants without systematic naming conventions to distinguish them.

Practical Steps and What Comes Next

For private organisations, the calculation is more straightforward. Marketing departments at firms headquartered around Zurich's Kreis 5 tech cluster and along Binzmühlestrasse in Oerlikon have increasingly adopted digital asset management platforms — tools such as Bynder or Canto — that flag duplicate uploads at the point of ingest rather than relying on retrospective cleanup. The upfront licensing cost for such platforms typically runs between CHF 12,000 and CHF 50,000 annually for a mid-sized team, a figure that needs to be weighed against ongoing storage and labour costs.

The city's IT coordination body, Zentrale Informatik, is expected to publish revised storage governance guidelines before the end of 2026 as part of the broader Smart City Zurich programme. Those guidelines are anticipated to recommend mandatory deduplication checks before any image collection migrates onto shared infrastructure.

For individuals and small businesses, the practical advice is simpler. Free tools including digiKam — open-source software maintained by the KDE community — perform perceptual-hash duplicate searches on local drives without uploading data to external servers, an important consideration given Swiss privacy norms. Running such a scan on a 100-gigabyte personal photo library typically takes under 30 minutes on a standard laptop and frequently recovers 15 to 20 percent of occupied space. The numbers are not glamorous. But in a city where even a one-room flat in Wiedikon now costs upwards of CHF 1,800 a month and every square centimetre of physical space is rationed, the idea that digital space deserves the same discipline is gaining converts.