Hong Kong's Digital Archive Push Faces a Reckoning Over Duplicate Images: What Happens Next

Hong Kong's public and institutional archives are sitting on a growing pile of redundant digital files — and the decisions about how to clean them up will shape how residents, researchers and businesses access historical and civic records for years to come. The trigger is a familiar one: years of rapid, often uncoordinated digitisation have left major repositories holding multiple copies of the same images, sometimes scanned at different resolutions, sometimes duplicated across siloed databases with inconsistent metadata tags.

The timing matters because the territory is mid-way through a broader push to modernise its data infrastructure under the Smart City Blueprint for Hong Kong 2.0, which covers the period to 2025 and beyond. With that framework now under review and Greater Bay Area data-sharing protocols growing more demanding, administrators can no longer defer the question of which image records are authoritative, which are redundant and who gets to decide.

Where the Problem Is Most Acute

The Hong Kong Public Libraries system, which operates 74 branches across the territory including its central hub at Causeway Bay's Hong Kong Central Library on Gloucester Road, has been quietly flagging the problem internally for more than two years. The library's digital collections span historical photographs, maps and cultural materials, many of which were scanned in separate batches by different contractors. The result is that a single archival photograph may now exist in three or four versions inside the same system, each with slightly different file names and incomplete provenance records.

The situation is similarly complicated at the Hong Kong Film Archive in Sai Wan Ho, which holds one of the most significant moving-image collections in East Asia. The archive has digitised thousands of film stills and promotional materials over the past decade. Without a unified deduplication protocol, staff managing access requests must sometimes manually verify whether two apparently identical image files are genuinely the same asset or represent different print generations — a process that adds hours to what should be routine requests.

The government's own records office, the Government Records Service, which operates under the Innovation, Technology and Industry Bureau, has begun consulting with counterparts in Singapore's National Archives on how that city-state handled a comparable consolidation exercise between 2019 and 2022. Singapore reduced its digital storage overhead for public records by consolidating duplicate assets — a process that took roughly 18 months and required new cross-departmental data governance rules before a single file was deleted.

The Key Decisions Ahead

Three choices now sit on the table for Hong Kong's institutional custodians. First, who has authority to designate one image file as the master copy and retire the others? This is not purely a technical question. Under current government ICT guidelines last updated in March 2024, individual bureaux retain custody of their own digital assets, meaning there is no single body with the power to issue a territory-wide deduplication mandate. A policy change would require sign-off at bureau level at minimum.

Second, what happens to the retired duplicates? Permanently deleting files carries archival risk — a lower-resolution duplicate may, in specific research contexts, be the only surviving evidence of how an original item was processed at a given point in time. The Hong Kong Museum of History at Chatham Road South in Tsim Sha Tsui, which manages a photographic collection running to hundreds of thousands of items, has argued internally for a tiered approach: retire duplicates to cold storage rather than delete them outright, at least for a defined review period.

Third, how does this interact with public access commitments? The territory's Data Protection Principles under the Personal Data (Privacy) Ordinance apply even to historical images if they depict identifiable individuals. Any bulk deduplication exercise touching those files must include a privacy review, which adds both cost and calendar time to what administrators might prefer to treat as a purely technical housekeeping exercise.

Practically speaking, institutions that move first on this will be better placed when the next round of Greater Bay Area data-sharing negotiations opens — likely in late 2026 or early 2027. Cross-border data infrastructure agreements increasingly require partner institutions to certify the integrity and uniqueness of shared assets. For Hong Kong's archives, that certification is currently impossible to provide with confidence. Getting there means making the hard calls now, not after the next government IT audit cycle forces the issue.