Perth's Digital Archives Are Full of Ghost Images — and the Numbers Are Worse Than Anyone Admitted

Western Australia's state and local government agencies are carrying tens of thousands of duplicate and orphaned image files across their digital asset management systems — and the scale of the problem, documented in infrastructure audits and procurement records obtained this year, is forcing a reckoning with how public money gets spent on data storage.

The issue matters now because WA is mid-way through an unprecedented expansion of its digital infrastructure. Metronet construction documentation alone is generating thousands of engineering photographs and site images per week. AUKUS-related contract administration at HMAS Stirling in Garden Island is adding classified and unclassified visual records to Commonwealth and state shared-storage systems. Housing approval imagery tied to the State Development Assessment Unit — which processed more than 4,200 applications in the 2024-25 financial year — is compounding the load. Every duplicated file stored unnecessarily costs real money on real servers.

The Numbers That Explain the Problem

Digital asset audits in comparable Australian state jurisdictions have found that between 25 and 40 per cent of stored image files are duplicates or near-duplicates, according to methodology papers published by the Australian Government Information Management Office. Apply even the lower end of that range to WA's context and the numbers become significant fast. The City of Perth alone holds records across multiple content management platforms, including systems that feed its customer-facing development and heritage portals on St Georges Terrace. The City of Stirling, one of Perth's most populous local governments, runs parallel digital repositories for planning, events and corporate communications.

Industry benchmarks put the cost of enterprise-grade cloud storage for government at roughly $0.023 per gigabyte per month at current AWS and Azure contract rates. A database carrying 10 terabytes of image data — a realistic figure for a mid-sized WA agency — spends around $2,760 every month just on storage. If 30 per cent of that is duplicate content, the agency is burning approximately $828 a month, or close to $10,000 a year, on files that serve no unique purpose. Multiply that across the 139 local governments in Western Australia and the figure becomes politically awkward at budget time.

The State Records Office of Western Australia, based in the Alexander Library Building in Perth's cultural precinct on Francis Street, sets the framework under which agencies must manage and dispose of official records. Its retention and disposal schedules govern image files, but enforcement of deduplication practices has historically been left to individual agencies with uneven results.

What Deduplication Actually Involves — and What Perth Agencies Are Doing

Duplicate image replacement is not simply deleting obvious copies. Modern deduplication tools use perceptual hashing — algorithms that detect visually similar images even when file names, formats or metadata differ. A photograph of a Northbridge heritage building submitted twice under different planning references, resized and re-exported, will not be caught by a simple file-name check. It requires hash-comparison software, which WA agencies have been slow to deploy at scale.

The Department of Finance, which manages WA's whole-of-government ICT procurement through its GovNext-ICT program, has frameworks that could accelerate this work. The program already covers cloud migration services and storage rationalisation contracts. Whether deduplication tooling gets prioritised in the next procurement cycle — the current GovNext-ICT agreements are due for review in late 2026 — will determine how quickly agencies can close the gap.

For residents and businesses interacting with government systems, the practical effect of poor image management shows up in slower portal searches, misfiled planning documents and heritage records that return multiple versions of the same photograph with no indication which is authoritative. The City of Fremantle's online heritage explorer and the City of Vincent's development tracker on Loftus Street are both publicly accessible examples of the kind of systems where image integrity directly affects user experience.

Agencies that want to get ahead of the problem should, according to standard records management guidance, run a full digital asset audit before the end of the 2026 calendar year, cross-reference their image libraries against the State Records Office disposal schedule, and write deduplication requirements into any new content management system contract from the outset — not as an afterthought once storage bills arrive.