The Price Of A Half-Second

AMultitasking is a word borrowed from machines, and it flatters the people to whom it is applied. A processor genuinely can hold several jobs at once; a person, on every measurement so far devised, cannot. What looks like simultaneity is a rapid alternation between tasks, and every alternation has to be paid for somewhere. The scale of the alternation is not in dispute. Software that logged the keystrokes of 1,900 office employees in 2019 recorded a median of 1,200 window changes in a working day, with attention moving to something new every two minutes and forty seconds. What that movement costs, and who should be asked to carry the cost, is where the argument begins. Employers have tended to treat a fragmented day as a failure of personal discipline, correctable by training. A second view holds that fragmentation is designed into the instruments of the modern office, and that asking individuals to resist it is rather like asking them to resist the weather.

BFelix Rosenthal, a cognitive psychologist at the Hallam Centre for Attention Research, has spent much of her career putting a number on that cost. Between 2015 and 2021 her laboratory took 4,300 volunteers through paired tasks and timed the moment of changeover. The average switch cost 0.45 seconds where the change was imposed by the experimenter, against 0.38 seconds where the volunteer chose the moment; on tasks governed by a complicated rule the figure passed 1.4 seconds. Half a second, Rosenthal concedes, is nothing at all. Her argument is arithmetical rather than dramatic: an expense too small to notice, met twelve hundred times between breakfast and the end of the day, becomes one of the larger items in the accounts. She is equally plain about what her stopwatch cannot see. It counts the switch and says nothing whatever about the reason for it, and what an hour of alternation does to the person performing it is not a quantity her equipment can capture.

CThat claim invited a test outside the laboratory, and in 2022 Rosenthal's group ran one. Eleven firms were persuaded to release their internal message logs, and 640 office workers agreed to have their screens sampled for six weeks. The finding was not that interruption is expensive but that its price depends on its timing. Where a message arrived in the middle of a step, the average return to full working speed took 4 minutes 12 seconds; where it arrived after a step had been finished, the same return took 1 minute 5 seconds. The penalty was confined to interruptions falling before a natural stopping point, and those arriving at a boundary made no measurable difference at all. The laboratory figure, in other words, was visible in a real working day, but only when the day was cut in the wrong place.

DBoris Nakamura, an organisational ergonomist at the Rhodes-Lund Institute, accepts the measurements and disputes what is drawn from them. An interrupted worker, he argues, is never fairly compared with an undisturbed one, because the undisturbed one is an invention; the comparison that matters is with the colleague left waiting for an answer, whose own work stops meanwhile. Studying 58 teams that introduced protected quiet hours between 2019 and 2022, he found that the share of tasks finished by their agreed date rose from 61 per cent to 74 per cent, while the average wait for a reply from a teammate grew from 40 minutes to nearly three hours. Nakamura volunteers two limitations. The teams adopted the policy themselves, so they may have been teams already disposed to guard their working time; and he counted deadlines met, not the quality of what was handed in.

EThe argument has left the journals. Since 2023 a code of practice adopted by a federation of large employers has required signatories to record how often staff are contacted during agreed periods of protected work and to publish a yearly summary. Nakamura regards this record-keeping as the most useful part of the code, because it obliges firms to keep evidence that can be examined by somebody other than themselves. Rosenthal, whose figures campaigners quote more often, is here the more sceptical of the two. An audit, she observes, counts the switches without ever asking whether the job required them, so an employer whose work is genuinely unpredictable will look no worse than one whose culture merely rewards the instant reply. Both dislike the growing habit of publishing a fall in message volume as though it were, in itself, a gain in attention.

FWhat is emerging is a division of the question rather than an answer to it. Where a task keeps its state outside the head, in a checklist, a form half filled in or a machine that will not proceed until the previous step is complete, the switch costs little and the worker returns to a place the work has held for them. Where the state of the task is held in the head alone, as in drafting, diagnosis or design, there is nothing to return to and the cost is the whole of the reconstruction. Whether office work as a whole is moving towards externally held state or away from it is disputed, and neither researcher claims to know. What both reject is the framing that has dominated public argument: that switching is either a personal failing or a harmless habit, when in most offices it is a property of the way the work has been arranged.