Taking a nap is risky business. During WWI, a man named Jack Dunn was sentenced to death for sleeping at his post. According to the British Empire’s Manual of Military Law, if you did nod off, it could be off with your head.
“A sentinel found asleep or drunk at his post while on active service would, if the character and the circumstances of the offence were sufficiently grave, be liable to suffer death.”1
‘Keeping watch’ is a task most people are familiar with. The context may vary – bodyguards on the lookout for assassins, detectives on a stakeout, or Homer Simpson minding the controls to the Springfield Nuclear Plant. Keeping watch is the primary task of lifeguards, quality controllers, security guards, and air traffic controllers.
These jobs have a reputation for being sleepy. For a brief time, I was a page at a legislative assembly. My job was to sit next to the speaker of the house and fetch water or documents for the ministers upon request. To do that, I needed to monitor the assembly for people trying to visually grab my attention. I would always become drowsy after about twenty minutes. It was a surprise to me – and especially my dad – that I didn’t get fired. But as it turns out, this wasn’t entirely a ‘me’ problem.
Lisanne Bainbridge, in Ironies of Automation, points to a body of literature that suggests humans have about a 30-minute attention span when it comes to visually monitoring activity (a.k.a. keeping watch). We’re not equipped to stay alert for that long. She says it is “impossible even for a highly motivated human being” to consistently spot abnormalities if they have been monitoring for a “long period”.2 The example given in the text is of an operator watching a screen. In the remainder of the paper, Bainbridge makes the case that the monitoring problem cannot easily be automated away, and she was right.
In March 2018, a semi-autonomous Uber test vehicle killed a pedestrian in Tempe, Arizona.3 The backup driver was unable to react quickly enough. As the story goes, if the backup driver had been paying closer attention, the death would have been avoided. However, the backup driver had been operating the car for 43 minutes.
Lifeguards have well-established protocols for sustaining high levels of operator attention. This week, an amphibious colleague of mine shared an example from his local pool.4 There are two lifeguard chairs. A lifeguard will spend 20 minutes in one chair, then move to the other chair for 20 minutes, then take a 20-minute break.
This protocol allows the operators to reset their attention, thus maintaining a high level of attention across the monitoring system. The two chairs also provide continuity in the shift, enabling an acceptable flow of information from the incumbent lifeguard to the incoming lifeguard.
It’s possible that a single lifeguard working a several-hour shift would have more complete information (like which areas are slippery, which rascal kids are causing trouble, and what time the swim club starts). But when it comes to responding to risks, attention is just as valuable as information. Sacrificing some knowledge, by rotating lifeguards is easily compensated by the increase in the minimum level of operator attention.
Both pool facilities and traffic systems have been physically engineered to be safer. It’s also true that they have been protocolized to be safer. Attention-boosting protocols have emerged in response to the necessary, but difficult, task of monitoring. When it comes to improving performance, saying “pay attention” simply isn’t enough.
If you’re trying to maintain a high production quality or a high level of safety in your business, you need to establish a simple alertness protocol. Role rotation is a great 80/20 approach to solving that problem. Keep in mind the rough 30-minute individual attention threshold, and go from there.
1 The Punishment for Falling Asleep at One’s Post by Kelly Buchanan (here).
2 Ironies of Automation by Lisanne Bainbridge (here).
3 Death of Elaine Herzberg, Wikipedia (here).
4 Dorian’s website (here).
