"Modelling the Enterprise and its Actors as Triply-Articulated Anticipatory Systems"
We present PAN, a toolset for the analysis of strategic risk in large enterprises. We take an enterprise to comprise one or more actors, an actor being an embodied individual whose anticipation of its world may be characterised by three models: how things in the world behave, what states-of-affairs the actor desires and how the former can be orchestrated to satisfy the latter. All these models are necessarily inaccurate and, as they are continually being recast by the actor in the light of its experience, they may not even be mutually consistent. Inaccuracies in the first model expose the actor to 'performance' risk, e.g. that the behaviour of a designed system differs from that predicted by its design. Inconsistencies between the first and third models expose the actor to 'composition' risk, e.g. that a group of interoperating systems collectively exhibit unanticipated outcomes, either destructive or emergent. Inconsistencies among all three models expose the actor to 'implementation' risk, e.g. that an expressly demanded state-of-affairs, although successfully established, fails to satisfy the actor's desire. The advent of 'open systems' (e.g. the 'Semantic Web') has increased the significance of composition risk. Simultaneously, consumers of systems and services have been encouraged to express their demand in terms of states-of-affairs relevant in their own models (e.g. network-enabled conflict), rather than those of suppliers. This 'asymmetric demand' increases the significance of implementation risk. The discipline of systems engineering addresses performance risk, through powerful analytical and simulation models, and covers some aspects of composition risk. Soft systems methodologies seek to address composition risk and some aspects of implementation risk, but offer no analytical power. Using PAN, the enterprise's actors may construct representations of their models of themselves and compose these models together to locate, estimate and mitigate the risks posed to the enterprise by gaps and inconsistencies among them. Further, the resulting composite models define the granularity and stratification of the ontologies operative in the enterprise and thereby entail the architectural structure of enterprise-wide data platforms and agent-based systems.