Canberra's Duplicate Image Problem: How the Capital Stacks Up Against Wellington, Ottawa and Edinburgh

Canberra's public sector holds more images per capita than almost any other city its size in the world. That distinction is becoming expensive. Across ACT government agencies, the Australian National University's digital repository, and the National Archives of Australia's Parkes Place facility, duplicate image files — identical or near-identical photographs stored multiple times across separate systems — are consuming server space, distorting search results, and inflating licensing costs for images agencies already own.

The issue has sharpened in 2026 for a specific reason: the ACT Government's Digital Infrastructure Consolidation Program, which entered its second phase in March, is forcing agencies to migrate legacy storage systems onto a unified cloud environment. That migration is surfacing thousands of duplicate assets that had been silently accumulating since at least 2014, when earlier digitisation drives swept through institutions including the National Library of Australia on Parkes Place and the ScreenACT archive at ACTEW House.

What Canberra Is Actually Doing About It

The ACT's Digital, Data and Technology Solutions directorate confirmed in its March 2026 quarterly report that deduplication tooling has been integrated into the government's Microsoft Azure migration pathway. The approach uses perceptual hashing — a technique that identifies visually similar images even when file names or metadata differ — rather than the simpler checksum matching that only catches exact binary copies. It is a meaningful upgrade. Checksum tools miss resized, recompressed or slightly cropped versions of the same photograph, which are precisely the kind of duplicates that accumulate when communications teams across Civic Square, London Circuit and Dickson all grab the same stock image from different subscriptions.

ANU's Scholarly Communications team has been running a parallel effort through the university's institutional repository, ANU Open Research, targeting roughly 340,000 image assets uploaded since 2018. A university spokesperson's written guidance to depositors, updated in April 2026, now explicitly requires researchers to check the repository's deduplication index before uploading datasets. University College London ran a comparable program in 2023 and reported removing around 18 percent of stored image assets as confirmed duplicates — a figure that, if applied to ANU's holdings, would represent a substantial reduction in storage overhead.

Wellington City Council completed a city-wide digital asset deduplication project in late 2024 under its Smart Capital Initiative, contracting the job to a New Zealand-based records management firm. Ottawa, a federal capital with a public-sector workforce profile similar to Canberra's, embedded deduplication requirements into its Directive on Service and Digital in 2023, making it mandatory for all new federal image uploads rather than a retrospective clean-up. Edinburgh's City of Edinburgh Council went a different route again, prioritising human review over automated tools after a 2022 algorithm incorrectly flagged archivally significant variant photographs as duplicates — a cautionary outcome that ACT archivists have cited in internal guidance notes.

The Cost Question No One Has Fully Answered

Precise dollar figures for Canberra's duplicate image overhead remain elusive. Cloud storage costs under the ACT's whole-of-government contract are not publicly itemised by file type. However, the broader Digital Infrastructure Consolidation Program was budgeted at $34 million over three years in the 2024-25 ACT Budget, and deduplication is listed as one of several efficiency mechanisms expected to contribute to projected savings across that period.

For institutions like the National Film and Sound Archive at McCoy Circuit, the stakes go beyond server bills. Duplicate records in a cultural heritage context can create provenance confusion — two identical catalogue entries for the same image can generate competing rights claims or obscure which copy carries the authoritative metadata. The NFSA has been working with the Australasian Performing Right Association and a network of state archives since January 2026 on a shared metadata standard designed partly to prevent cross-institutional duplication.

What happens next will depend on how consistently the ACT government enforces its new deduplication requirements once the Azure migration completes, currently scheduled for the fourth quarter of 2026. Agencies that have already migrated — including Transport Canberra and the Environment, Planning and Sustainable Development Directorate — will serve as the test cases. If Ottawa's mandatory-at-upload model proves easier to sustain than Wellington's retrospective clean-up approach, expect pressure on the ACT to tighten its intake controls before the next wave of government communications digitisation arrives.