How San Francisco's City Websites Ended Up Buried in Duplicate Images — and Why Fixing It Took Years

San Francisco's municipal web infrastructure is carrying tens of thousands of duplicate image files across more than a dozen city department websites, a problem that accumulated quietly over nearly a decade and is now forcing a reckoning with how the city manages its digital assets. The Department of Technology, which oversees the city's enterprise systems from its offices on Golden Gate Avenue, confirmed earlier this year that duplicate image replacement has become a formal line item in its 2025-2026 fiscal cleanup initiative.

The issue matters now because the city is simultaneously pushing a sweeping digital-services modernization effort. SF Digital Services, the office that manages sf.gov and coordinates web standards for agencies ranging from the Municipal Transportation Agency to the Department of Public Health, cannot migrate legacy content to its unified platform without first resolving redundant media libraries that slow load times, inflate cloud storage costs, and break accessibility compliance under California's ADA-related web standards.

How the Duplication Problem Built Up

The roots go back to roughly 2014, when multiple city departments began independently spinning up WordPress and Drupal installations without centralized governance. The Department of Homelessness and Supportive Housing, created in 2016 under then-Mayor Ed Lee, launched its own web presence and imported image assets in formats that overlapped with existing Department of Public Health photography. The same pattern repeated across the Planning Department's Civic Center-area offices and the Office of Economic and Workforce Development, which manages programs across neighborhoods from the Tenderloin to Bayview-Hunters Point.

When COVID-19 hit in March 2020, city communications staff uploaded vast quantities of public-health graphics, multilingual flyers, and emergency-resource images under severe time pressure and with no coordinated naming convention. A 2023 internal audit by SF Digital Services found that some image files existed in six or more copies across different departmental content management systems, each with slightly different filenames, dimensions, or metadata. That audit, referenced in the department's published modernization roadmap, estimated that redundant media storage was contributing meaningfully to cloud expenditure at a time when the city's Department of Technology budget was already under pressure.

The sf.gov platform, built on the Gatsby and Contentful stack and formally launched in phases beginning around 2020, was designed to prevent exactly this kind of sprawl. But migration from the old sites has been slower than planned. As of the department's most recent public progress report, roughly 40 percent of city department web content had been migrated to sf.gov, leaving the majority of legacy systems — and their duplicate image libraries — still active.

The Technical and Political Dimensions

Cleaning up duplicate images is not simply a matter of running a deduplication script. Each image may be linked from multiple pages, embedded in PDF documents, or referenced in outreach materials printed by neighborhood organizations in the Mission District or the Excelsior. Removing or replacing a file without auditing every link risks breaking pages that residents depend on for services information. SF Digital Services has had to build a custom toolchain to map image references before any file is touched.

Budget politics have complicated the timeline. The Department of Technology faced scrutiny during the 2025 budget cycle, when the Board of Supervisors debated cuts across city agencies. Technology modernization projects, which rarely produce the visible constituent benefits of a repaved street or a new shelter bed on Division Street, are perennial targets for deferral.

The practical path forward, according to the modernization roadmap, involves automated perceptual hashing — software that identifies visually identical or near-identical images regardless of filename — combined with a mandatory review period before any file is deprecated. Departments have been asked to designate a digital content lead by September 2026. For residents, the near-term effect should be faster-loading city web pages, particularly on mobile devices, where sf.gov's performance metrics have lagged behind the city's own targets. For city staff, it means, finally, a single image library they can actually search.