Use of Artefacts 103
CONSEQUENCES OF TECHNOLOGY CHANGE
Since technology is not value neutral, it is necessary to consider in further
detail what the consequences of changes may be and how they can be
anticipated and taken into account. The introduction of new or improved
artefacts means that new tasks will appear and that old tasks will change or
even disappear. Indeed, the purpose of design is partly to accommodate such
changes either by shaping the interface, by instructions and training, etc. In
some cases the changes are palpable and immediate. In many other cases the
changes are less obvious and only occur after a while. This is because it takes
time for people to adapt to the peculiarities of a new artefact and a changed
system, such as in task tailoring. From a cybernetic perspective, humans and
machines gradually develop an equilibrium where demands and resources
match reasonably well. This equilibrium is partly the result of system
tailoring, partly of task tailoring (Cook & Woods, 1996). Technically
speaking an equilibrium is said to exist when a state and a transformation are
so related that the transformation does not cause the state to change (Ashby,
1956, p. 73-74). From this point of view technology is value neutral only if
the new artefact does not disturb the established equilibrium.
One effect of the adjustment and tailoring that makes a system work is
that some aspects come to be seen as canonical and while others are seen as
exceptional. This means that users come to expect that the system behaves in
a specific manner and works in a specific way, i.e., that a range of behaviours
are normal. Similarly, there will be a range of infrequent (but actual)
behaviours that are seen as abnormal or exceptional, but which people
nevertheless come to recognise, and which they therefore are able to handle
in some way.
When a system is changed, the canonical and exceptional performances
also change. The canonical performance comprises the intended functions
and uses, which by and large can be anticipated – indeed that is what design
is all about. But the exceptional performance is, by definition, not
anticipated, hence will be new to the user. Representative samples of
exceptional performance may furthermore take a long time appear, since by
definition they are infrequent and unusual. After a given time interval – say
six months – people will have developed a new understanding of what is
canonical and what is exceptional. The understanding of canonical
performance will by and large be correct because this has occurred frequently
enough. There is, however, no guarantee that the understanding of
exceptional performance will also be correct.
The introduction of a new artefact can be seen as a disruption of an
equilibrium that has developed over a period of time, provided that the
system has been left to itself, i.e., that no changes have been made. How long
this time is depends on a number of things such as the complexity of the
104 Joint Cognitive Systems
system, the frequency of use, the ease of use, etc. After the disruption from a
change, the system will eventually reach a new equilibrium, which however
may differ somewhat from the former. It is important to take this transition
into account, and to be able to anticipate it. Specifically, if only one part of
the system is changed (e.g., the direct functions associated to the artefact) but
not others, such as the procedures, rules or the conditions for exceptions, then
the transition will create problems since the system in a sense is not ready for
the changed conditions.
Specifically, the manifestations of failures – or failure modes – will
change. The failure modes are the systematic ways in which incorrect
functions or events manifest themselves, such as the time delays in a
response. New failure modes may also occur, although new failure types are
rare. An example is when air traffic management changes from the use of
paper flight strips to electronic flight strips. With paper flight strips one strip
may get lost, or the order (sequence) of several may inadvertently be
changed. With electronic strips, i.e., the flight strips represented on a
computer screen rather than on paper, either failure mode becomes
impossible. On the other hand, all the strips may become lost at the same
time due to a software glitch, or it may be more difficult to keep track of
them.
Finally, since the structure of the system changes, so will the failure
pathways, i.e., the ways in which unwanted combinations of influences can
occur. Failures are, according to the contemporary view, not the direct
consequence of causes as assumed by the sequential accident model, but
rather the outcome of coincidences that result from the natural variability of
human and system performance (Hollnagel, 2004). If failures are seen as the
consequence of haphazard combinations of conditions and events, it follows
that any change to a system will lead to possible new coincidences.
Traffic Safety
An excellent example of how the substitution principle fails can be found in
the domain of traffic safety. Simply put, the assumption is that traffic safety
can be improved by providing cars with better brakes. The argument is
probably that if the braking capability of a vehicle is improved, and if the
drivers continue to drive as before, then there will be fewer collisions, hence
increased safety. The false assumption is, of course, that the drivers will
continue as before. The fact of the matter is that drivers will notice the
change and that this will affect their way of driving, specifically that they will
feel safer and therefore driver faster or with less separation to the car in front.
This issue has been debated in relation to the introduction of ABS
systems (anti-blocking brakes), and has been explained or expounded in
terms of risk homeostasis (e.g., Wilde, 1982). It is, however, interesting to
Use of Artefacts 105
note that a similar discussion took place already in 1938, long before notions
of homeostasis had become common. At that time the discussion was
couched in terms of field theory (Lewin, 1936; 1951) and was presented in a
seminal paper by Gibson & Crook (1938).
Gibson & Crook proposed that driving could be characterised by means
of two different fields. The first, called the field of safe travel, consisted at
any given moment of “the field of possible paths which the car may take
unimpeded” (p. 454). In the language of Kurt Lewin’s Field Theory, the field
of safe travel had a positive valence, while objects or features with a negative
valence determined the boundaries. The field of safe travel was spatial and
ever changing relative to the movements of the car, rather than fixed in space.
Steering was accordingly defined as “a perceptually governed series of
reactions by the driver of such a sort as to keep the car headed into the middle
of the field of safe travel” (p. 456).
Gibson & Crook further assumed that within the field of safe travel there
was another field, called the minimum stopping zone, defined by the
minimum braking distance needed to bring the car to a halt. The minimum
stopping zone would therefore be determined by the speed of the car, as well
as other conditions such as road surface, weather, etc.
It was further assumed that drivers habitually would tend to maintain a
fixed relation between the field of safe travel and the minimum stopping
zone, referred to as the field-zone ratio. If the field of safe travel would be
reduced in some way, for instance by increasing traffic density, then the
minimum stopping zone would also be reduced be a deceleration of the car.
The opposite would happen if the field of safe travel was enlarged, for
instance if the car came onto an empty highway. The field-zone ratio was
also assumed to be affected by the driver’s state or priorities, e.g., it would
decrease when the driver was in a hurry.
Returning to the above example of improving the brakes of the car,
Gibson & Crook made the following cogent observation:
Except for emergencies, more efficient brakes on an automobile will
not in themselves make driving the automobile any safer (sic!). Better
brakes will reduce the absolute size of the minimum stopping zone, it
is true, but the driver soon learns this new zone and, since it is his
field-zone ratio which remains constant, he allows only the same
relative margin between field and zone as before. (Gibson & Crook,
1938, p. 458)
The net effect of better brakes is therefore that the driver increases the
speed of the car until the familiar field-zone ratio is re-established. In other
words, the effect of making a technological change is that the driver’s
behaviour changes accordingly. In modern language, the field-zone ratio
corresponds to the notion of risk homeostasis, since it expresses a constant
106 Joint Cognitive Systems
risk. This is thus a nice illustration of the general consequences of
technological change, although a very early one.
Typical User Responses to New Artefacts
As part of designing of an artefact it is important to anticipate how people
will respond to it. By this we do not mean how they will try to use it, in
accordance with the interface design or the instructions, i.e., the explicit and
intentional relations, but rather the implicit relations, those that are generic
for most kinds of artefacts rather than specific to individual artefacts. The
generic reactions are important because they often remain hidden or obscure.
These reactions are rarely considered in the design of the system, although
they ought to be.
Prime among the generic reactions is the tailoring that takes place (Cook
& Woods, 1996). One type is the system tailoring, by means of which users
accommodate new functions and structure interfaces. This can also be seen as
a reconfiguration or clarification of the interface in order to make it obvious
what should be done, to clearly identify measures and movements, and to put
controls in the right place. In many sophisticated products the possibility of
system tailoring is anticipated and allowance is made for considerable
changes. One trivial example is commercial office software, which allows
changes to menus and icons, and a general reconfiguration or restructuring of
the interface (using the desk top metaphor). Another example is the modern
car, which allows for minor adjustments of seats, mirrors, steering wheel, and
in some cases even pedals. In more advanced versions this adjustment is done
automatically, i.e., the system carries out the tailoring or rather adapts to the
user. This can be done in the case of cars because the frame of reference is
the anthropometrical characteristics of the driver. In the case of software it
would be more difficult to do since the frame of reference is the subjective
preferences or cognitive characteristics, something that is far harder to
measure.
In many other cases the system tailoring is unplanned, and therefore not
done without effort. (Note that even in the case of planned tailoring, it is only
possible to do that within the limits of the built-in flexibility of the system.
This a driver cannot change from driving in the right side to driving in the
left side of the car – although Buckminster Fuller’s Dymaxion car from 1933
allowed for that.)
The reason for unplanned tailoring is often that the artefact is incomplete
or inadequate in some ways; i.e., that it has not been designed as well as it
should have. System tailoring can compensate for minor malfunctions, bugs,
and oversights. A good example is how people label – or re-label – things. A
concrete example is the lifts found in a university building in Manchester,
UK (Figure 5.4). It is common to provide a call button by each lift, but for
Use of Artefacts 107
some reason that was not the case here. Consequently, a sign at the right-
hand lift instructed people that the call-button was by the left-hand lift, while
a sign at the left-hand lift instructed people that a bell-sound indicated that
the right-hand lift had arrived.
Another type of tailoring is task tailoring. This happens in the case where
it is not sufficient to tailor the system, or where system tailoring is
impossible. The alternative is then to tailor the tasks, i.e., to change the way
in which the system is used so that it becomes usable despite quirks of the
interface and functionality. Task tailoring amounts to adjusting the ways
things are done to make them feasible, even if it requires going beyond the
given instructions. Task tailoring does not go as far as violations, which are
not mandated by an inadequate interface but by other things, such as overly
complex rules or (from the operator’s point of view) unnecessary demands to,
e.g., safety. Task tailoring means that people change the way they go about
things so that they can achieve their goals. Task tailoring may, for instance,
involve neglecting certain error messages, which always appear and seem to
have no meaning but which cannot easily be got rid. The solution is to invoke
a common heuristic or an efficiency-thoroughness trade-off (ETTO) rule,
such as “there is no need to pay attention to that, it does not mean anything”
(Hollnagel, 2004). The risks in doing this should be obvious. Task tailoring
may also involve the development of new ‘error sensitive’ strategies, for
instance being more cautious when the system is likely to malfunction or
break down.
IF YOU HEAR THE
BELL PLEASE USE
THE RIGHT-HAND
LIFT
BOTH LIFTS ARE
OPERATED BY THE
BUTTON ON THE
LEFT
Call
button
Figure 5.4: Unusual call-button placement.
..................Content has been hidden....................
You can't read the all page of ebook, please click here login for view all page.