SF's Digital Archive Overhaul: What Officials, Experts, and Key Figures Are Saying About Duplicate Image Replacement

San Francisco's Department of Technology is under pressure to overhaul a years-old problem that has quietly inflated storage costs and slowed public-records requests: thousands of duplicate images embedded across city databases, from Planning Department permit files to SFMTA traffic-camera archives. The push to replace those redundant files with a standardized, deduplicated system has moved from IT back rooms to budget hearings at City Hall, with officials and civic-tech advocates drawing hard lines over how fast it needs to happen—and who should pay for it.

The issue matters now because San Francisco is mid-way through a broader effort to digitize public services before the fiscal year 2026-27 budget takes effect in August. Mayor Daniel Lurie's administration has flagged digital infrastructure as a priority after the city's Controller's Office identified ballooning cloud-storage expenditures. Redundant image files—building permits with the same photograph attached dozens of times, for instance—are a concrete, fixable part of that problem, and civic-tech groups say the window to act is narrow.

Where the Pressure Is Coming From

The San Francisco Civic Technology Alliance, a nonprofit that has consulted with city departments on open-data standards, has been vocal on the issue. The group points to the Planning Department's permit portal on Eddy Street and the city's DataSF open-data platform as two systems where duplicate image proliferation is demonstrably worst. In internal presentations shared with supervisors on the Board's Government Audit and Oversight Committee, the Alliance has argued that a phased replacement program—beginning with the highest-traffic databases—could cut redundant file volume by roughly 40 percent within 18 months, based on audits conducted in late 2025.

The Department of Technology has not publicly disputed that figure. Its director-level staff briefed the committee in May 2026, outlining a proposal to use a hash-based deduplication protocol—software that compares unique image fingerprints and consolidates identical files to a single stored copy. The approach is already standard practice at large municipal governments including New York City's Department of Records & Information Services, which completed a comparable project in 2023.

Voices from the Mission District-based Code for San Francisco brigade, a volunteer civic-hacking group affiliated with Code for America, have been more skeptical of the city's timeline. Members presenting at a June 2026 open meeting at the Exploratorium's Pier 15 campus argued that without dedicated staffing, the deduplication effort risks stalling the same way a 2021 SFMTA data-cleanup initiative did—a project that was eventually folded into a broader contract that ran $2.3 million over its original budget, according to city procurement records.

What a Fix Actually Involves

The practical challenge is not purely technical. City agencies store images under different naming conventions, uploaded by different vendor systems over decades. The Department of Building Inspection, whose offices sit at 49 South Van Ness, uses a permitting platform that was last overhauled in 2018. Reconciling its image library with the newer systems adopted by the Planning Department requires manual review of edge cases—images that are visually identical but carry different metadata, or files that need to be preserved separately for legal reasons tied to active litigation.

The city's proposed contract for the replacement work, currently in a request-for-proposals phase, carries an estimated value of between $800,000 and $1.2 million depending on scope, according to the Department of Technology's published procurement notice dated June 15, 2026. Several established San Francisco-based technology contractors, including firms operating out of the SoMa corridor, are expected to bid.

Supervisors on the audit committee are expected to take up the RFP at their next hearing, tentatively scheduled for late July. Civic-tech advocates say anyone tracking the issue should monitor DataSF's public changelog—updated weekly at data.sfgov.org—for evidence that deduplication work has begun on pilot datasets before a full contract is awarded. If the RFP closes on schedule and a vendor is selected by September, the first phase of image replacement could be completed before the end of calendar year 2026. Miss that window, advocates warn, and the project is likely to slip into the next budget cycle.