The Human Ceiling

Enterprise security has spent two decades scaling cameras, while the human visual system asked to watch them has not changed since it was first measured in 1948. The Human Ceiling assembles the peer-reviewed and government-sourced evidence on why live video monitoring breaks down at scale, then translates it into operational and financial terms for security leaders. The finding is consistent across the research: active, continuous evaluation is what makes surveillance prevent rather than record, and humans cannot sustain it at fleet scale. That continuous-evaluation layer belongs to a system built for it, with people elevated to the judgment, escalation, and response only they can provide. This is the strategic case for Agentic Physical Security.

Key takeaways:

• Sustained attention has a short half-life. Operator detection degrades to below acceptable levels after about 20 minutes of monitoring, and the decline is largely involuntary (Mackworth, 1948; NIJ/Sandia, 1999).
• Detection erodes as feeds multiply. Correct detection of staged events falls from roughly 84% on nine monitors to about 64% on sixteen (Tickner & Poulton, 1973).
• Alarm fatigue is documented, with safety consequences. In policing, 94 to 98% of alarm responses are to false alarms, and the healthcare evidence base ties alarm desensitization to patient deaths.

‍