Brid O'Conaill and David Frohlich
Hewlett Packard Labs.
Bristol, UK, BS12 6QZ
boc@hplb.hpl.hp.com and dmf@hplb.hpl.hp.com
+44 272 228979
Abstract
We report findings from an observational study on the nature of
interruptions in the workplace. The results show that in most
cases, (64%), the recipient received some benefit from the
interruption. However in just over 40% of interruptions the
recipient did not resume the work they were doing prior to the
interruption. Some implications for time management and
communication technology are presented.
In contrast very little is known about what might be called
timespace in the workplace; defined as the intervals of time
into which people organise their work. Despite the proliferation
of both paper and electronic time management systems there are
very few scientific studies of time management behaviour at
work. One reason for this is the very short time window of most
workplace and human factors studies. A second reason is that
many studies rely on reports of time allocation which are
demonstrably inaccurate compared to actual measures [6].
In this paper we report some preliminary findings on the use of
shared timespace in the workplace, based on an analysis of
interruptions. Interruptions are interesting because they reveal
that the timespace of any individual is not owned and controlled
in the same way as their workspace, but can collide and merge
with that of another individual unexpectedly. Herein lies a
problem for most time management systems which tend to assume a
greater degree of control over timespace than is actually
possible and overlook the potential benefit of interruptions to
individual work. Interruptions also constitute a problem for
many remote collaboration tools such as pagers, mobile phones
and real-time conferencing products. By improving synchronous
access between remote parties they tend to benefit the
initiators with the recipients having little control over the
receipt of the interactions. Over time they may, therefore,
become self-defeating as users realise the costs of making their
own timespace more available to others.
Interruptions raise questions of both practical and theoretical
significance which we set out to address in the study,
including: How many interruptions occur at work? What proportion
of time do they consume? How many are resisted rather than
taken? Who benefits from the ensuing interactions? How
disruptive are interruptions to the prior task being carried out
prior to the interruption?
The majority, 79, of these interruptions occurred in a face to
face setting in the subject's workplace. A further 21 occurred
face to face but while the subject was away from their office.
The remaining 25 were telephone calls.
Interruptions are usually thought of as being a nuisance for the
recipient. For each interruption we analysed the content, to
determine, for whose benefit the conversation had been (see
Figure 1). Non work related interactions were taken as being
jointly beneficial. The largest number of interruptions were for
the benefit of both initiator and receiver, 43.2%. The initiator
benefited solely from 32.8% and the recipient from 20.8%. Thus,
in 64% of the interruptions the recipient received some benefit
from the interaction having taken place. In 2.4% of the cases
the interruption was initiated for the benefit of a third party.
FIGURE 1: Beneficiary of Interruption
We also measured what activities took place after the
interruption. Did the recipient return to the prior activity or
was there a disruption in the flow of work? In just over 55% of
the cases the recipient returned to their original activity (see
Figure 2). Although from the data it was not possible to say if
their performance in carrying out that task had been affected.
In 4% of cases the recipient was not engaged in a measurable
work activity, (having coffee etc.) In the remaining cases the
recipient failed to return to their prior activity because they
were interrupted again, 14.4%, they proceeded to work on the
interrupt, 15.2% or they decided to work on another task.
10.4%. Note there had been no prior indication that the task
had been finished.
Only two attempts to dissuade interruptions were observed. In
one case, the initiator was given a time limit in a joking
fashion. In another, a request was made to the subject's
secretary to hold calls. This was done when a tight deadline
was approaching for a piece of work to be completed.
The complexity of these findings on shared timespace is not well
accomodated by existing technology. Some time management
systems recommend that users make themselves unavailable to
interruptions for specific periods of the day or week so as to
concentrate on their own objectives. This blanket approach
ignores the benefit that recipients receive from being
interrupted, and the service that individuals may be contracted
to perform for others. On the other hand, using the latest
communication technology to make oneself always open to
interruption reduces the length of time people have to
continuously perform the same activity.
Clearly some kind of filtering of interruptions would be
desirable, to determine if the current interruption warranted
disruption of the prior activity. Indeed, this function is
often performed in conjunction with receptionists or secretaries
to good effect. Given the brief nature of many interruptions
such a filtering mechanism needs to be very light weight since
the filtering process itself could be as disruptive as an
interruption. Video diary technology might be useful in two
ways here [1]. First, if an interruption is allowed to proceed,
recipients might benefit from reviewing a brief audiovisual
record of the end of the interrupted activity to reset its
context, particularly if that was itself an interaction. Second,
if an interruption was resisted, a record of the context for the
interruption could be saved for later reference and action.
Another approach would be to move interruptions to asynchronous
communication media. Previous work suggests that it is not the
duration of the interruption that makes it disruptive but rather
the complexity of the new task [4]. Thus, asynchronous media
should be designed to allow the storing and transfer of complex
messages and data. To this end combined audio and written
messages appear to provide benefit in terms of fewer turns and
hence fewer interruptions [3].
Keywords:
Interruptions, workplace communication, CSCW,
ethnography
Introduction
A great deal is known about how people use workspace in the
workplace, and there are well established techniques in both HCI
and CSCW for supporting these activities. For example, the
'desktop metaphor' is an attempt to recreate the kind of messy
desks that are so useful for personal information management,
[5], while 'shared workspaces' constitute an extension of the
metaphor to support interactive document use and interpersonal
information management [2].
METHOD
To answer these questions we conducted a new analysis of data
from an observational study of workplace communication in which
two subjects were shadowed with a video camera for a full
working week. The participants were mobile professionals for
whom communication formed a central part of their job. Each
exhibited a form of local area roaming; in building for one and
out of building/metropolitan for the other. Further details of
the shadowing method can be found in Whittaker, Frohlich &
Daly-Jones [7]
RESULTS
Beginning with 29 hours of video data we extracted 125 naturally
occuring interruptions. We defined an interruption to be a
synchronous interaction which was not initiated by the subject,
was unscheduled and resulted in the recipient discontinuing
their current activity. Thus, silent document delivery while the
subject was working on their PC was not counted as an
interruption, but a request for a signature while the subject
was in the course of a telephone conversation was included. On
average the subjects were being interrupted just over 4 times
every hour. The average duration of an interruption was 2
minutes 11 seconds. Approximately 10 minutes in every hour was
being spent engaged in an interruption.
DISCUSSION
Our analysis suggests that interruptions are a frequent and time
consuming feature of office life which are seldom resisted by
recipients. Furthermore many interruptions (41%) result in the
discontinuing of the interrupted task beyond the duration of the
interruption itself. However, the analysis also shows that
recipients often derive personal benefit from the interruption,
sometimes at the expense of the initiator!
