ABSTRACT
A significant challenge, particularly for engineering organisations, is the ‘productservice shift’, whereby organisations move from delivering a product to the provision of through-life, sustained capability support (Oliva and Kallenberg 2003; MOD 2005; Anon 2008). This affects companies in their internal organisation and requires changes to the processes that they employ; particularly for knowledge and information management. Furthermore, the problems will get steadily worse as systems evolve from stand-
alone status to components of systems-of-systems, with all the issues associated with this shift. While this happens, the making of complex decisions will remain a human and organisational issue. This is the context of this paper. The paper outlines the background of decision-making systems (DMS) and com-
plexity before outlining the development and subsequent validation of a DMF (Decision Making system Framework) and supporting process. It builds on work initially presented at theErgonomicsAnnualConference 2008 (Molloy et al., 2008).
While there has been a considerable literature on Decision Support Systems, there appears to be a dearth on the characterisation and evaluation of DMSs that may use DSSs; Simon (1997) is one example.This is a pity; from an ergonomics perspective;
as Rasmussen (1997; 2000) has said, “A closer look at the recent major accidents show that they are not caused by a stochastic coincidence of failures and errors, but by a systemic migration of organisational behaviour towards the boundaries of safe operation. Major accidents are the side-effect of decisions made by several decision-makers, in different organisations, and at different points in time, all doing their best to be effective locally.” One of the underlying causes of this is complexity: “A system exhibits complexity
if it is composed of many integrated entities of heterogeneous parts, which act in a coordinated way and whose behaviour is typically nonlinear.” (Rycroft and Kash 1999). Some organisational characteristics which give rise to this are (Gregg 1996):
• many agents • many kinds of agents • operating in parallel • with a degree of behavioural autonomy • non-linear interactions • multiple steady states for the system • multiple steady states for human agents As Gregg says, not many of these are needed for the system to exhibit emergent
behaviour. A personalised view of this (Vaill 1998) is:
