Canberra's Duplicate Image Problem: The Key Decisions That Will Shape What Comes Next

Thousands of duplicate images are sitting inside the digital archives of ACT government agencies, and the decisions about how to handle them — delete, merge, or migrate — are now pressing enough that at least two Canberra institutions have begun formal review processes in the first half of 2026. The question is no longer whether something needs to be done. It is who has authority to do it, and at what cost.

The issue has sharpened because several agencies are mid-cycle on major digital transformation projects. The ACT Digital Strategy, which runs through to 2027, committed the territory government to consolidating its data infrastructure. But consolidation tends to surface duplicates — identical or near-identical image files stored across separate systems, sometimes held by different directorates, sometimes the result of repeated bulk uploads over years of poor file governance. When those duplicates involve official records — urban planning photographs, heritage documentation, public health imagery — the stakes climb quickly.

Where the Pressure Is Being Felt

The Australian National University's Scholarly Information Services division and the ACT Heritage Library on Mildura Street in Griffith are both understood to be dealing with versions of this problem, though the specifics differ considerably. ANU manages hundreds of thousands of digital assets across its research collections, and any deduplication exercise at that scale requires significant staff time and a clear decision framework before a single file is touched. The Heritage Library, which holds photographic collections documenting Canberra's development from the early twentieth century forward, faces a different challenge: many of its duplicates exist because digitisation was done in multiple passes, sometimes by volunteers, and metadata standards were inconsistent across those projects.

The ACT Environment, Planning and Sustainable Development Directorate is also relevant here. Planning records — including site photographs tied to development applications across growth suburbs like Gungahlin and Belconnen — have been flagged internally as an area where duplicate image holdings create genuine administrative risk. If two versions of the same photograph carry different metadata, and one is attached to a planning decision, the integrity of that record becomes a legal question, not just a housekeeping one.

The Decisions That Cannot Be Deferred

Three choices will define how this plays out. First, who arbitrates when two copies of an image conflict in their attached metadata? The ACT Territory Records Act 2002 governs public records, but it does not offer granular guidance on digital deduplication. That gap will need filling, either through updated guidance from the Office of the Australian Information Commissioner or through territory-specific policy developed by the Archives ACT team based in Mitchell.

Second, what tools are used? Automated deduplication software can process large volumes quickly, but it cannot make judgment calls about which version of a photograph constitutes the authoritative record. Human review remains necessary at the decision layer, and in a public service workforce already stretched across competing digital projects, resourcing that review is a genuine budget question. Software licensing for enterprise-grade deduplication tools can run to tens of thousands of dollars annually for mid-sized agencies.

Third, there is the question of retention. Deleting a government image file — even a duplicate — carries risk if the deletion is not documented properly. Best practice under the territory's records framework requires that disposal decisions be authorised and logged. Agencies that skip that step to save time are creating future liability.

Practical guidance from the National Archives of Australia, published in its digital preservation guidance materials, recommends that agencies complete a full asset audit before any deduplication tool is deployed. That audit process alone, for a mid-sized directorate, can take three to six months. Agencies beginning that work now, in July 2026, are looking at completion timelines that push well into 2027 — squarely inside the territory election cycle, which adds a layer of political sensitivity to decisions about resourcing.

For institutions like ANU and the Heritage Library, the path forward is more technical than political, but no less consequential. Getting the framework right before the deletions start is the only approach that avoids compounding a records problem into a records disaster.