'Our History Keeps Getting Erased': Boston Communities Push Back on Duplicate Image Replacement in Digital Archives

A growing number of Boston residents, community historians, and digital archivists are raising alarms about a practice that has been quietly degrading local online records for at least two years: automated content management systems flagging authentic neighborhood photographs as duplicates and replacing them with generic stock images pulled from commercial libraries. The problem has surfaced most acutely in digitized collections tied to community organizations in Roxbury, Jamaica Plain, and East Boston, where decades of photographic documentation of immigrant neighborhoods, protest movements, and housing campaigns are at stake.

The issue has particular urgency right now because the city of Boston is in the middle of a multi-year push to digitize historical neighborhood records under the Boston City Archives digitization initiative, which received funding through Mayor Michelle Wu's FY2025 budget. That work, concentrated at the City Hall Annex on School Street and at the Boston Public Library's Kirstein Business Branch on City Hall Plaza, means more records than ever are moving through automated processing pipelines — and more are vulnerable to algorithmic misclassification.

What Communities Are Losing

The duplicate-detection algorithms used by several widely adopted content management platforms work by comparing image hashes or pixel-pattern signatures. When two images score above a similarity threshold — sometimes as low as 85 percent — the system can automatically flag one as redundant and substitute a licensed stock alternative. For photographs of, say, the corner of Blue Hill Avenue and Warren Street in Dorchester taken in 1987, there is no stock image that serves as an adequate replacement. There is simply no substitute for the original.

The Madison Park Development Corporation, which has maintained photographic records of Roxbury neighborhood development since its founding in 1970, has been among the organizations working to document the scope of the problem locally. Staff there have identified instances where photos tied to affordable housing campaigns along Dudley Street were misprocessed through a third-party document management vendor, resulting in placeholder imagery appearing in public-facing web collections. The organization has not released a full count of affected files, but the situation has prompted it to pause use of automated duplicate-detection features in its current archive software.

Community members at a Jamaica Plain Neighborhood Council meeting in June described discovering that photographs from the 1990s Green Street community garden campaigns — images that documented early environmental justice organizing in the neighborhood — had been silently replaced in a shared digital folder system used by multiple local nonprofits. For many of those residents, who have spent years fighting displacement and advocating for housing affordability, the loss of visual documentation felt like an extension of the same erasure they experience in other domains.

Systemic Gaps and Practical Steps Forward

The problem is not unique to Boston, but it lands with particular weight here given the city's dense concentration of university-linked digital humanities programs. Northeastern University's NULab for Texts, Maps, and Networks, based on Huntington Avenue, has been developing guidance on human-in-the-loop review protocols specifically to address automated misclassification risks in community archive projects. Their draft framework, circulated internally this spring, recommends that any collection containing pre-2000 photographs be exempted from fully automated duplicate resolution until a qualified archivist reviews flagged files individually.

Boston Public Library's Digital Commonwealth program, which aggregates digitized collections from more than 130 cultural institutions across Massachusetts, requires contributing organizations to retain original files regardless of what automated platforms do downstream — a policy that has shielded some collections, though not all, from permanent loss.

For community organizations that lack dedicated archival staff, the practical advice from digital preservation specialists is straightforward: disable automatic replacement features in any content management system handling historical images, maintain at least two independent backup copies in separate storage locations, and audit any collection that passed through automated processing between 2023 and the present. The cost of external hard drives sufficient to back up a mid-sized community photo collection runs roughly $80 to $150 at current retail prices — a fraction of what recovery efforts cost once images are gone.

The Boston City Archives is expected to release updated digitization guidelines before the end of the third quarter of 2026. Advocates say those guidelines need explicit language addressing duplicate-detection risk — and that community organizations shouldn't have to wait to find out what the city decides before protecting their own records.