Wednesday, 2 October 2013
The spectrum of control architectures (Figure 1) goes from centralized to decentralized structures. Movement in any direction implies a trade-off between oversight and efficiency, on the one hand, and local autonomy and operational flexibility, on the other. The following are the most common control architectures.
Figure 1: The spectrum of control architectures
Centralized – Has the advantage of relatively simple control; however, if the situation for which it is configured changes, then a massive effort is required to reconfigure it. Is characterized by a single point of failure, resulting in a lack of general robustness.
Heterarchical (or fully decentralized) – Relatively robust because there is very little to break, but this lack of structure also makes it very difficult to control and makes undesirable chaotic behaviour possible.
Hierarchical – Consists in a top-down decomposition of control. Is efficient in that it forces an expected behavior; but is inflexible and branches can become uncontrollable if an intermediate element is incapacitated. Autonomy of nodes is quite limited.
Federated – A compromise between hierarchical and heterarchical structures. Like the latter, the nodes have a high degree of autonomy but communicate through specialized middle nodes. Has improved robustness and flexibility over the other architectures, but does not allow for dynamic restructuring.
Holarchical (or holonic) – A hybrid, recursive and hierarchical structure which is able to generate dynamic linkages to form an impromptu control structure. Couples the flexibility of fully decentralized architecture with the stability and predictability of a centralized and hierarchical control.
The control structure defines the performance of an organization to the extent that it defines its authority structure; it affects task and resource allocation (MacMillan et al. 2002); it indirectly affects performance by influencing collaborative processes; and it defines the external fit of the organization, that is, its adaptability to changing situations (Hollenbeck 2000). Given that they involve several organizations, MOS preclude control structures that assume a single chain of command, such as centralized, hierarchical, or holarchical architectures. In fact, MOS are constituted of autonomous and sovereign organizations that try to exploit a networked environment, and this entails the use of a decentralized architecture, where generally an administration board is used as means of governance. However, when a new entity is created, the latter, as a new organization, can have a centralized or hierarchical command, which can sometimes become problematic. Control structure in MOS must be rather defined in terms of the intensity of control (partial or full) of partner organizations on their representatives in the collaboration space, endowing it with more or less autonomy as a new organization. Another issue is the power relationships between the partner organizations, which can give one party more decision making authority and thus more global control over the mission of the newly created MOS.