Several reasons exist in the literature for giving team members free access to the rule construction process. Previous studies suggest that free access is important for team stability and productivity and that there is a need for knowledge to be generated in a grounded fashion, to allow for issues of unpredictability.
The literature also suggests that the interaction rule module created as part of the research described in this thesis, Phreda, must cater for emergent knowledge. The team’s understanding of appropriate interaction rules will change with
understand each other better, immediately changing the context in which rules apply. There are two instances when a member perceives another to be reneging on a behavioural contract – when the second member is deliberately doing so and when the members share differing expectations (Piccoli & Ives, 2003). Making behavioural expectations explicit helps to eliminate this second, perceived breach of contract.
The system should allow for self-management, since in some teams the members themselves are expected to control their own interaction. In many situations self- managed teams are more productive than those where hierarchic authorities dictate behaviour (Yeatts & Hyten, 1998). Studies of local communities show that when community members make their own decisions over the use of resources, sustainable solutions are more likely to be found than if solutions are imposed from outside the community (Ostrom, 1990). The decision to build open access to rule management is reinforced by drawing an analogy between a local community and the smaller complex social system of the team. Such a structure should help the sustainability of the team as a unit.
Economics is a discipline where decisions need to be made in the context of a complex social system. Determining which theoretical model to use is not always possible. Knowledge, and rules based on the knowledge, can be both predictable (ergodic) and unpredictable (non-ergodic) (dePaula & Fischer, 2005; Dow, 2004). The Bank of England chooses an existing model for determining interest rates, dependent upon the immediate context and the “instincts” of the decision-makers. That model informs the rule announcing which rate to charge (Dow, 2004). The mechanism for deciding upon interaction rules should allow situated, local input and not rely on pre-existing knowledge alone. This approach is also in keeping with the functioning of a practitioner in a community of practice. A practitioner uses any pertinent knowledge, regardless of whether it was also applicable in other instances (dePaula & Fischer, 2005). The epistemology of Layder addresses these differences between positivists and grounded theorists. He sees existing theory as a guide, and direct experiential engagement as a modifier or shaper of knowledge (Layder, 1998).
It is critical from philosophical and practical standpoints that the software be able to capture grounded knowledge from the participants.
The literature provides guidelines for how to construct a software system that supports the creation of interaction rules and manages them in a manner akin to a human moderator. If one is to create a module that fulfils the role of moderator, it will have to monitor aspects of member interaction, and provide consequences if a monitored behaviour is recognised. The interaction rule module must allow team members to select meaningful measures of behaviour, and allow symbols to be associated with these measures of behaviour as well as with the consequences that ensue should the behaviour occur. This would enable norms of behaviour to be explicitly stated as rules, providing opportunities for team knowledge to be acquired and learning to occur. If team managers or leaders were team members, there would be a mechanism for democratic local rule making, as well as for the inclusion of already established knowledge. A facility for creating contractual relations defining acceptable behaviour could replace the need for trust to some degree. It is also possible that the process of creating the rules might generate sufficient understanding amongst members to make the resultant rules redundant. Finally, the rule module would need to be fully editable by all participants. With the ability to disable, modify and create rules as desired, a dynamic and complex team would be able to develop evolutionary structures; guidelines reflecting current patterns of behaviour and also be sensitive to team maturation. This software description fits nicely into the definition of a “User Adaptive System”. The fundamental components of such a system were recently described by deVrieze (2006). A user adaptive system contains a user model, which de Vrieze defined as “a model of relevant characteristics of a user that is or can be used to personalise the behaviour or presentation of a system.” Such a model can obtain its characteristics adaptively (for example from an “intelligent agent”, mentioned earlier), or can be explicitly given them by the user.
One can replace the term ‘User’ with ‘Team’ in de Vrieze’s calculus without altering his software model. Replacing the term ‘User’ with “Complex System”
extends the applicability of the model still further. The diagram used by de Vrieze captures the components of all adaptive systems.
Figure 3. The fundamental components of a user adaptive system. (de Vrieze, 2006).
The adaptation component (Figure 3) is activated by user events. It personalises both the user interface (via the interface handler) and the behaviour of the system (via the action handler). The question handler provides answers about the user needed for the modification of the system. The question handler has the ability to act on the performance of the system, affecting the action properties and its appearance, and hence affecting the interface properties.
The design for providing software support for interaction, learning and knowledge and dealing with teams as complex systems used in this research, extends de Vrieze’s work on adaptive personalisation (section 3.1.3).