PVM
- Performance Variability Management
Distinctions
between different types of performance are usually based on an evaluation of the
outcome. It is thus common to distinguish between accidents, incidents, near
misses, and normal performance. In the cases where performance has failed in
some way, it is furthermore common to refer to the concept of error-as-cause (in
distinction to error-as-event) and specifically to embrace a notion of the
"human error" as a definite category of causes. Although this approach
has been a de facto standard for a good many years, it has a number of
major shortcomings that gradually are becoming clear.
One
is the failure to distinguish clearly between the phenotypes
(manifestations) and genotypes (possible causes) of events with unacceptable
outcomes.
Another
is the failure to acknowledge that finding the causes of an event or an
outcome is a case of reconstruction rather than identification, hence a
social judgement made in hindsight.
Yet
another is that the approach completely disregards the positive aspects of
failures, namely that they provide an opportunity to learn.
A
final shortcoming is that the common approach only poorly describes
performance failures that can be explained as the result of actions at the
organisational and management levels (distal rather than proximal causes).
A clear
alternative to the common approach is provided by the so-called ecological view
on error. It is, however, preferable to avoid the use of the term
"error" completely, since it is ill-defined. A more precise term would
be performance deviations, corresponding to the view of cognitive
systems engineering. This change in perspective has two consequences
with important theoretical and practical implications. The first is that
performance deviations can be considered for all levels of a system, including
the system as a whole, rather than for individual actors only. The second is
that the focus is shifted from explaining the negative consequences of
performance failures, to understanding the nature of performance deviations.
It is a
characteristic of all dynamic systems, simple as well as complex, that they are
able in some way to adapt their performance and retain control
of the situation. This goes for simple biological systems, human beings, complex
technological artefacts, and organisations – possibly including even
societies. One prerequisite for this is the ability to detect performance
deviations and compensate for them in a timely and efficient fashion. This means
that the issues of performance deviation management become central to
understanding what formerly was called accidents and errors. Specifically, it
becomes necessary to formulate clearly the principles for performance deviation
management that can serve as the basis for practical techniques on the shop
floor, in the control room, in the cockpit, and in the boardrooms.
Without
going into details here, the basic principle of PDM is to focus on how detection
and correction can take place. Detection is clearly a
prerequisite for correction, and systems should be designed to make detection
easier and correction possible. The former addresses issues of how easy it is to
detect and evaluate changes in system status – or even precursors to changes.
The latter is concerned with the details of system design, specifically how it
can be ensured that sufficient time and resources are available to enact the
corrections and achieve a recovery. For performance deviations that either lead
to very significant negative consequences, such as a loss of mission or loss of
life, or that are frequently recurring, the issues of detection and correction
are normally complemented by issues of prevention. The latter
involves the ways in which effective barrier systems and barrier functions can
be designed, implemented and maintained.
© Erik Hollnagel, 2005