The number emerged from Mark's methodology of direct field observation rather than self-report. Workers shadowed by her research team consistently described themselves as focused for long stretches, occasionally distracted. The measurement showed something else: fragmentation at a cadence the workers themselves could not perceive. The gap between believed work and actual work is, in Mark's framework, not a matter of self-deception but a structural limitation of introspection.
The decline from 150 seconds to 47 seconds did not proceed smoothly. It accelerated in measurable steps, each correlated with a new layer of the digital environment — persistent email, smartphones, notification streams, always-on messaging, algorithmic feeds. Each layer lowered the activation energy required to switch. Each layer raised the gravitational pull of the next available engagement. The worker's cognitive system, responsive to environmental cues as it evolved to be, followed the gradient toward shorter engagements and more frequent switches.
Forty-seven seconds is the number Edo Segal identifies in the foreword as the measurement that unsettled him. It is not a productivity figure. It does not appear on any dashboard. It describes the cognitive substrate on which every dashboard depends. You On AI celebrates capability expansion; Mark's number describes what is happening to the attentional architecture of the people wielding the capability.
The AI moment, in Mark's framework, does not invent fragmentation. It intensifies a trajectory already well underway. Claude Code shortens the interval between question and answer to seconds. The imagination-to-artifact ratio Segal celebrates compresses what were multi-minute or multi-hour intervals into the span of a single prompt cycle. The result is a workflow in which 47 seconds is no longer the floor but the ceiling of sustained engagement on any single cognitive mode.
Mark arrived at the measurement through the continuation of a research program she began in the early 2000s at UC Irvine. Her team's methodology required ethics approval, consent from participating organizations, and the willingness of workers to be observed continuously during their workdays. The longitudinal structure of the research — with successive studies conducted across two decades — allowed Mark to track the trajectory rather than a single point. The numbers are not simulated or extrapolated; they are measured.
The measurement is empirical. Not a philosophical claim, not a rhetorical figure — a number produced by direct observation of workers in real environments, replicated across multiple studies.
The trajectory predates AI. The decline from 150 to 47 seconds occurred before the arrival of large language models, driven by the accumulating layers of the digital environment.
The floor may drop further. AI tools that respond in seconds eliminate the micro-intervals that previously bounded attentional fragmentation, compressing the temporal architecture of cognitive work to a degree that has no empirical precedent.
Subjective experience cannot perceive it. Workers whose attention switches every 47 seconds report the sensation of sustained focus. The measurement describes what the introspective report misses.
The number is diagnostic, not prescriptive. Mark's framework does not argue that 47 seconds is catastrophic per se. It argues that 47 seconds is a measurement, and measurements require accounting.