**3.1 System reconfiguration**

Systems and models are intimately related. Modeling is a set of abstract artifacts representing systems (Dori, 2003). Therefore, we can develop a meta-model while designing a system that will abstract out the system specifications by carrying over the underlying adjustable parameters. This mechanism may impose additional tasks at the developers end, and will incur further expenses; however it will open up a wide door for any future

Dynamics of System Evolution 29

The central concept of system dynamics is the idea of two-way causation or feedback (Meadows and Robinson, 2002); it is assumed that social or individual decisions are made on the basis of the *flow* of information about the system state or environment surrounding the decision makers. The decisions lead to actions that are intended to change the state of the system. New information about the system state then produces further decisions and changes. Each such closed chain of causal relationships forms a *feedback loop*. System

In other words, system dynamics include a set of concepts, representational techniques, and beliefs that make it into a definite modeling paradigm. It emphasizes a *continuous view* (Richardson, 1999) and looks for *causality* that underlies the longer-term patterns of change in systems. Here, the focus is on the general dynamic tendencies; under what conditions the system as a whole is stable or unstable, oscillating, growing, declining, self-correcting, or in equilibrium state (Meadows and Robinson, 2002). Properties of dynamic problems contain quantities that vary over time; such variability is described by the *causality*, which influences a closed system with *feedback loops*. Thus, dynamic systems are characterized by interdependence, mutual interaction, information feedback and circular causality. Every system experiences a number of changes over time and only in keeping detailed trackrecords of these behavioral changes can visualize the system dynamics. Therefore, the core objective is to tool the wrapper system so that we can capture dynamic system behavior, track any change history and discover correlations, as well as analyze and act upon it for

As mentioned earlier, each dynamic system should consist of one or more loops running over a period of time as a part of its lifecycle. The items that affect other items in the system but are not themselves affected by anything in the system are called exogenous items. These exogenous disturbances are seen as the triggers of system behavior. *Positive loops* tend to amplify any disturbance and to produce exponential growth: if the cause increases, the effect increases and if the cause decreases, the effect decreases. *Negative loops* tend to counteract any disturbance and to move the system toward an equilibrium point: if the cause increases, the effect decreases and vice versa. Stable conditions will exist when negative loops dominate positive loops. Non-linear relationships can cause feedback loops to vary in strength, depending on the state of the rest of the system. The dominating loop might also shift over time; linked non-linear feedback loops form the patterns of shifting loop dominance. Under some conditions one part of the system is very active, and under other conditions, another set of relationships takes control and shifts the entire system behavior; i.e. one particular loop is always responsible for the overall behavior of that

The timing of system behavior depends on the presence of system elements that create inertia or delays. These inertial elements are referred to as state variables (see section 4.2). Each state is an accumulation (*stock*) of information. System elements representing the decision, action, or change in a state variable are indicated by a *flow* of information to/from a state variable. Also, there could be time-delays in information flows, and we need to look

dynamics models are made up of such loops linked together (as shown in figure 1).

better service for end users and thereby extend the system lifecycle.

**4.1 Background of dynamic system behavior**

for any lagged relationships in the system.

system.

**4. System dynamics** 

adjustments without the need for refactoring or overhauling the system. Thus, it is a valuable investment to the more contolled software maintenance activities. Such a wrapper system provides direct interactions among the system users, surrounding environment, and the service providers. This link (or communication path) enables the environmental responses on new changes that need to be applied as feedback into the system. Figure 1 shows an example of such placement of a wrapper system. A set of automated surveyagents and probing stations (Mubin & Luo, 2010a) are responsible for capturing various usage factors from the system itself (such as, access rates, feature ranks, utilization factors, path traversals), from its operational environment (such as, system users, work habits, workstations, workspaces, user feedbacks, new system requirements, user satisfaction index), and from the external environment (with influential factors).

Fig. 1. Setup of a Real-Time Meta-Structure for a System to become Evolvable

#### **3.2 Real time feedback**

The feedback concept is at the heart of the system dynamics approach (System Dynamics Society); it emphasizes a continuous view that strives to look beyond regular services to see the underlying dynamic patterns. In our setup, there are four feedback loops in the proposed operational environment through the meta-structure: (1) system usage activities are observed, and quantified into utilization factors, satisfaction indices and system state; and the knowledge base identifies whether there are any noticeable changes, and instructs the application of this change into the system, if necessary. Similarly, any change or update requests from the (2) system (3) operational environment, and (4) external influences are also applied through the service queue. Thus, over a period of time, outcomes from these activities will result in the corresponding *loop dominance*, which are then traced back into the knowledge base. The responses of such a meta-structure not only depend on logical inferences, but also on the physical instant from which the outputs are produced.
