How San Francisco's City Agencies Got Buried in Duplicate Images — and What They're Doing About It

San Francisco's Department of Technology has been quietly wrestling with a problem that sounds mundane until you see the numbers: city agencies collectively maintain tens of thousands of duplicate digital images spread across legacy servers, shared drives, and public-facing websites — many of them uploaded multiple times over the past decade with no consistent naming convention, no version control, and no central registry. The redundancy has compounded storage costs, slowed website migrations, and created accessibility compliance headaches that the city is now scrambling to fix.

The issue matters right now because the city is in the middle of a sweeping digital infrastructure overhaul, part of Mayor Daniel Lurie's push to streamline government services following a bruising audit period that dogged the final years of the London Breed administration. Several departments — including the Planning Department on Mission Street and the Office of Economic and Workforce Development at City Hall — have been directed to consolidate their web presences onto a unified content management platform by the end of fiscal year 2026-27. That migration exposed, in full, just how chaotic years of decentralized uploading had become.

The roots of the problem trace back to roughly 2014, when departments began building out their own microsites independently, often contracting with separate vendors who handed off asset libraries with no standardization. By 2019, the San Francisco Municipal Transportation Agency alone was managing image repositories across at least four separate platforms simultaneously, according to internal documentation reviewed during a city controller's performance audit released in early 2024. The SFMTA's Digital Services team has since collapsed those repositories into two, but the deduplication work — identifying which files are true duplicates versus slightly resized or reformatted versions — proved far more time-consuming than anticipated.

The Compounding Costs of Neglect

Digital storage is cheap in isolation. In aggregate, it adds up. City technology staff estimate that redundant image files account for a disproportionate share of the storage burden on servers hosted through the Department of Technology's data center on Turk Street in the Tenderloin. When the city renegotiated its cloud storage contracts in late 2024, analysts flagged that bloated asset libraries were inflating costs and complicating the migration to Amazon Web Services infrastructure that the Department of Technology has been phasing in since 2023.

The accessibility angle has been particularly pointed. Under Section 508 of the federal Rehabilitation Act and California Government Code Section 11135, public agencies must ensure digital content — including images — carries proper alt-text and metadata. Duplicate images uploaded at different times frequently carried inconsistent or missing alt-text, meaning the same photograph of, say, a Caltrain connection at 4th and King Street might have full accessibility tagging in one instance and none in another. Complaints filed through the Mayor's Office on Disability, which operates out of Van Ness Avenue, flagged this pattern repeatedly between 2022 and 2025.

What the Fix Looks Like

The Department of Technology is now piloting an automated deduplication tool — procured through a contract finalized in March 2026 — that cross-references file hashes to identify exact and near-exact duplicates before a human reviewer makes a final call. The tool is being tested first on the Planning Department's image library, which spans permit application photos, neighborhood character studies, and marketing materials from community outreach campaigns in districts including the Sunset and the Bayview.

The broader lesson city IT managers are drawing is structural. Several peer cities, including New York and Chicago, moved earlier toward centralized digital asset management systems — a step San Francisco debated but deferred twice, in 2017 and again in 2020, partly because departmental autonomy was treated as a political given. The cost of that deference is now measurable in staff hours spent on cleanup rather than service delivery.

For San Francisco residents and the businesses and nonprofits that interact with city web portals daily, the practical payoff should be faster-loading pages, more consistent imagery, and websites that actually pass accessibility audits. The Department of Technology has set a target of completing the initial deduplication sweep across the five highest-volume departments by December 2026. Whether the remaining 50-plus city agencies follow on the same timeline is the next question officials will have to answer.