Modelling real-world behaviours
As discussed in section \ref{487159}, a significant challenge is presented in trying to reproduce real-world behaviours in computer-generated models and simulations. One specific topic addressed in this study in this regard is the representation of technological anomalies, occurring as a result of functional-failure. Whilst the technological failure definition provided in chapter 2 is intended to enable the systematic detection and comparison of such events, in reality it is very difficult to build up a comprehensive picture of the frequency of such anomalies. This is due to the fact that no standardised catalogue of historical substitutions exists that includes accompanying details of the performance development trends required to be able to identify these regions of temporary functional-failure, or at least not to the extent necessary to be able to determine with any confidence the frequency of anomaly occurrence for a particular industry. This may be because the performance evidence is either commercially sensitive, or that it has not been compiled chronologically for a range of competing products into a single consistent time series. Typically, many divergent narratives may be found for the progress related to any given technology, and performance improvements often appear in an ad-hoc fashion in historical accounts. In any case, this make it difficult to build a picture of the frequency of such events for such a wide spread of technologies as would be necessary to generate a more tailored model of functional-failure anomaly and behaviours. As such, the best that can be done in such circumstances is to assume a more generalised distribution of these events. Such models commonly include the use of binomial, Gaussian, normal, and Poisson distributions amongst others, but of these Poisson is potentially the most relevant in this instance as this is suited to predicting discrete counts of events (non-negative) in a given time period \cite{quora,distributions,uses}. Consequently, these functional-failure anomalies can be treated in an analogous manner to the modelling of more conventional failure events in the maintenance and reliability analysis of products and systems by using a Poisson distribution to predict the haphazard nature of occurrence. Equally, Poisson distributions are commonly used to represent the random accumulation of citations over time, as noted by Mingers \cite{Mingers_2015}. As such, representations of functional-failure could adopt a similar strategy, whilst being calibrated to the technological case studies considered using conventional Poisson distribution control parameters (i.e. by defining the mean expected value, skewness, and translational shifts applied to the basic distribution).
Considering the behavioural challenges presented by technology substitution in more general terms, some of the more promising simulation techniques that have been developed to improve representations of real-world dynamics, briefly discussed in section \ref{174036}, are now introduced in more detail.
Agent-Based Modelling
Excluding environmental phenomena, the emergent complexity frequently observed in real-world dynamics often stems from the autonomous nature of the individuals, communities, and organisational entities that make up human societies. These distinct entities, whether individual or group-based, operate guided by their own
values, beliefs, and decision-making capabilities, with varying levels of independence from any centralised governing authorities present. Conventionally, many models of society assume that individuals and organisational entities behave in a greedy fashion \cite{medema2009hesitant}. However, there are many examples where, despite the lack of any centralised authority, real-world entities remain largely regulated
in their behaviour (such as may be seen from the cooperation observed in the global Air Transportation System), contrary to what might be expected when
combining multiple greedy entities. Such examples make it apparent that in reality individual or group-based entities demonstrate reactive, pro-active, cooperative, and social traits. Building on these observed behaviours,
Agent-Based Modelling (ABM) is one of the few methods available to social studies for exploring this level of dynamic
and stochastic complexity, by directly considering the interaction of human-centric traits within a modelled group of entities. Several fundamental features are commonly agreed on as characteristics of agents, as summarised in Table \ref{table:agent_characteristics} \cite{Macal_2006}. Of these, the most critical condition is that any agent should be capable of making independent decisions, ensuring active, rather than passive, behavioural characteristics.