In an era where artificial intelligence can generate human-like text and smartphones possess more computing power than the systems that guided Apollo 11, one might assume that urban policy challenges would be met with equally sophisticated solutions. Yet, as 2026 unfolds, San Francisco is preparing to deploy a method for counting its homeless population that would have felt familiar to scribes documenting the Domesday Book nearly a thousand years ago.
The city's next homeless point-in-time count is scheduled for January 29, representing a biennial ritual that has become increasingly disconnected from modern data capabilities. This exercise, while necessary for securing federal funding, relies on an army of volunteers armed with clipboards wandering city streets during specific hours, manually tallying individuals who appear to be experiencing homelessness. The inherent subjectivity and limitations of this approach have drawn sharp criticism from advocates and policy experts who question whether such an archaic system can possibly capture the complexity of contemporary homelessness.
Federal Mandates and Outdated Methodology
The primary justification for maintaining this practice is straightforward: cities must conduct these counts to qualify for federal homeless assistance funds. This requirement has locked municipalities into a methodological framework that prioritizes consistency over accuracy. San Francisco's approach, until recently, involved sending counters out between 8 p.m. and midnight—a timeframe that created significant ambiguity about who was homeless versus who was simply out for evening activities.
The city has announced modest improvements for the upcoming count. Counters will now operate between 5 a.m. and 10 a.m., theoretically reducing confusion between homeless individuals and nighttime revelers. More significantly, volunteers will now engage in conversations with the people they're tallying, rather than making visual assessments from a distance. This change addresses a fundamental flaw: without direct interaction, a wealthy tech executive in casual attire could be mistaken for someone living on the streets.
The Paradox of Progress
These improvements, while welcome, illuminate a deeper systemic problem. San Francisco delayed implementing these common-sense changes for years due to a perverse methodological constraint: maintaining consistency with past counts. The logic was that changing the approach would break the ability to compare year-over-year data, creating a discontinuity in the historical record. This created a situation where the city intentionally preserved known flaws to maintain data comparability—a textbook example of how rigid adherence to tradition can undermine genuine progress.
The comparison to medieval record-keeping is not hyperbole. The Domesday Book, compiled in 1086, represented a comprehensive property survey of England conducted through direct observation and manual documentation. While revolutionary for its time, its methods bear striking resemblance to San Francisco's current approach: physical presence, visual confirmation, and handwritten tallies. The fact that this methodology persists in an age of real-time data analytics and integrated service databases reveals a troubling gap between technological capability and policy implementation.
Subjectivity and Accuracy Challenges
The Jack Dorsey example highlighted in discussions about the count's flaws—where a billionaire might be mistaken for a homeless person based on appearance—underscores the method's reliance on subjective judgment. This subjectivity extends beyond individual misidentification to fundamental questions about who qualifies as homeless. The distinction between someone sleeping on the street and someone in temporary shelter, or between a person in supportive housing who spends time outdoors and someone truly unsheltered, requires nuanced understanding that a brief visual assessment cannot provide.
The Tenderloin neighborhood exemplifies these challenges. With one of the city's highest concentrations of both homelessness and supportive housing, counters must differentiate between residents of single-room occupancy hotels who step outside for fresh air and those without any shelter at all. Without conversation and context, these distinctions become nearly impossible to make accurately, potentially inflating or deflating counts in ways that directly affect resource allocation.
Existing Alternatives Already in Use
Here's the most frustrating aspect of this situation: San Francisco already possesses far more sophisticated data collection systems. The city continuously compiles extensive information about homelessness through service utilization records, shelter occupancy data, and coordinated entry systems. This information, submitted regularly to federal agencies, provides a more dynamic and accurate picture of homelessness than any single-night count could ever achieve.
These integrated databases track real-time interactions with homeless services, medical care, and social support systems. They can identify patterns, measure outcomes, and provide longitudinal data that a point-in-time snapshot simply cannot capture. The existence of these systems makes the continued reliance on clipboard counts not just anachronistic but actively redundant.
The Broader National Context
San Francisco is hardly alone in this predicament. Cities across the nation conduct similar counts using comparable methods, creating a patchwork of data quality and reliability. Some municipalities have moved to more appropriate counting hours years ago, but the fundamental limitations persist. The federal requirement, while well-intentioned in creating accountability, has inadvertently created a lowest-common-denominator approach that discourages innovation.
The biennial nature of the count compounds these problems. Homelessness is a dynamic, fluid phenomenon that can change dramatically over months or even weeks. A count conducted every two years provides snapshots so infrequent that they miss seasonal variations, economic shocks, and policy impacts that occur between measurements. It's akin to measuring a river's flow by checking it once every two years at a single point—technically a measurement, but hardly representative.
Toward a Modernized Approach
What would a truly modern homeless counting system look like? It would integrate real-time data from service providers, use predictive analytics to identify at-risk populations, and employ technology like anonymized mobile data (with appropriate privacy protections) to understand population movements. It would prioritize longitudinal tracking over single-night snapshots and focus on outcomes rather than raw numbers.
Such a system would require rethinking federal requirements and investing in data infrastructure. But the payoff would be substantial: more accurate resource allocation, better policy evaluation, and a clearer understanding of what interventions actually work. The technology exists. The data exists. What's missing is the political will to transition from a medieval census to a modern measurement system.
Conclusion
San Francisco's modest methodological improvements for the January 29 count represent progress, but they are incremental changes to a fundamentally flawed system. The city—and the nation—must confront the uncomfortable truth that we are using 11th-century methods to address 21st-century challenges. While federal funding requirements create legitimate constraints, they should not prevent us from building the better systems we already know how to create.
The clipboard-toting volunteers heading out on January 29 deserve respect for their dedication. But their efforts would be better supported by a system that values accuracy over tradition, integration over isolation, and continuous measurement over biennial rituals. Until then, San Francisco will continue counting homeless individuals with methods that would make a Domesday Book scribe feel right at home—a testament not to our capabilities, but to our failure to apply them where they matter most.