How San Francisco's Digital Archives Ended Up Full of the Same Photo Twice — and What the City Is Doing About It

San Francisco's municipal image databases contain tens of thousands of duplicate photographs — the same pothole, the same permit inspection, the same shelter bed photographed and uploaded multiple times across incompatible city systems — and officials are now working through a formal remediation effort to clean up records that have ballooned in storage costs and complicated everything from insurance claims to housing code enforcement.

The problem did not appear overnight. It accumulated across roughly a decade of piecemeal technology decisions made by departments that rarely talked to one another. Understanding how the city arrived here matters because the same structural conditions — fragmented procurement, no unified data governance policy, pressure to digitize fast — are still present, and without deliberate reform they will produce the same result again.

A Decade of Siloed Digitization

The roots of the duplicate-image problem trace to the early 2010s, when departments including the Department of Building Inspection on Stevenson Street, the San Francisco Municipal Transportation Agency, and the Department of Public Works each launched independent document-management and photo-archive projects. Each signed separate vendor contracts. Each built storage systems that could not communicate with the others. Inspectors in the field, often working off mobile devices with poor connectivity, would upload a photo, lose the confirmation, and upload it again. Multiply that across thousands of field workers over ten years and the redundancy compounds rapidly.

The city's 311 service request platform, which processes complaints from neighborhoods ranging from the Tenderloin to the Excelsior, became a particular accumulation point. Residents submitting photos of encampments or damaged sidewalks had no visibility into whether their image had already been logged by a city worker hours earlier. The system accepted and stored both versions. When the Department of Homelessness and Supportive Housing began cross-referencing shelter availability data with field photography in 2021, staff found duplicate image sets attached to single case files — a problem that slowed caseworker workflow and inflated cloud storage invoices.

The San Francisco Department of Technology, headquartered on Seventh Street in SoMa, flagged the systemic issue in a 2023 internal review, though the findings were not widely publicized at the time. The review found that city departments were collectively paying for redundant cloud storage across at least four separate enterprise contracts. Efforts to consolidate those contracts under a citywide data infrastructure initiative stalled during the fiscal turbulence of 2023 and 2024, when budget shortfalls forced repeated delays to capital technology projects.

What the Remediation Effort Looks Like Now

The current cleanup effort, coordinated through the Department of Technology and touching systems at City Hall on Dr. Carlton B. Goodlett Place as well as field-facing platforms used by Public Works crews, relies on automated hashing tools that compare image files and flag duplicates for human review before deletion. The process is deliberately cautious: a photo mistakenly deleted from a building permit file can become a liability issue if a code dispute ends up in litigation.

Staff have been working through a backlog that, by the department's own internal accounting shared with the city's budget analysts this spring, runs into the hundreds of thousands of flagged image pairs. Progress is steady but slow. The work is also unglamorous — it does not carry the political visibility of, say, the city's AI deployment pilots or the ongoing BART Fare Integration project — which means it competes poorly for staffing resources.

For city departments, the practical lesson is that procurement decisions that look cheap in year one accumulate hidden costs by year five. For San Francisco residents, the immediate implication is more concrete: cleaner image records mean faster resolution of 311 complaints, more accurate housing inspection files, and fewer disputed permit histories when property owners try to refinance or sell in neighborhoods like the Mission or the Outer Sunset.

The Department of Technology has said it plans to publish updated data-governance guidelines applicable to all city departments before the end of fiscal year 2026. Whether those guidelines will carry any enforcement mechanism — or remain advisory — is the question that will determine whether San Francisco finds itself sorting through the same problem a decade from now.