Seven Seconds
A notification costs seven measured seconds of cognitive processing. The bell has been engineering attention for nine centuries. Someone finally counted.
The number arrived quietly, in a lab report on smartphone interruptions: a single notification disrupts cognitive processing for roughly seven seconds. Pupil dilation confirmed the arousal spike. The researchers found something better than the number, though. Notification frequency predicted disruption more reliably than total screen time. The architecture of the interrupt matters more than the duration of exposure.
Read that finding twice, because most of the attention discourse is built on the weaker variable. Screen-time dashboards, digital sabbaths, greyscale modes: all of it measures exposure. The study says the damage rides on the interrupt schedule. A phone that lights up forty times a day for two hours costs you more than a phone you read like a novel for four.
The bell was always an attention machine
Interrupt architecture is nine hundred years old. The monastic horarium divided the day by bells, and every bell was a designed interruption: it pulled the monk out of one attentional state and installed another. Matins, lauds, prime. The schedule was engineered by people who thought of attention as something you build, the way you build a wall.
The difference between the monastery bell and the notification is ownership. The monastery rang its own bell, on a schedule set by the community the bell interrupted. Your phone rings a bell set by an engagement team in another jurisdiction, tuned by a model whose objective function you will never read. Seven seconds per ring, forty rings a day, and the schedule belongs to someone else.
Stiegler would have filed the whole study under pharmacology without changing a word of it. The same week the seven-seconds paper landed, another group proposed an ML framework they call Algorithmic Quiet: a system that learns your focus windows and withholds notifications inside them. The poison pole, measured; the remedy pole, designed. One instrument, two dosages.
The pharmacological reading cuts off the easy conclusion. The notification is neither an evil to be abolished nor a convenience to be defended. It is a dose, and dosage is a design decision, and design decisions have owners. That is the diagnostic chain worth keeping: seven seconds per interrupt, frequency over duration, ownership over frequency.
So the useful audit of a tool asks one thing first: who sets the bell schedule, you or the drum? Count your interrupts for one day. Then count how many of their schedules you wrote.