A Framework of Human Systems Engineering. Группа авторов
measures of sociotechnical alignment such as iterative surveys, natural language processing of communications, and explorative modeling to proactively identify and quantify sociotechnical risk.
Historically, SE frameworks have been shown to produce, on average, better development results than efforts lacking discipline. This is not to say that viable products cannot be developed outside of an established framework, it merely suggests that well‐defined frameworks reduce risk and provide a basis for communicating intent. Thus, by expanding the framework to explicitly examine sociotechnical measures and calculate additional areas of risk, Epoch 4 provides the foundation for addressing these risks and raising the likelihood of a successful development effort.
The importance of defining risk in sociotechnical systems has been previously noted by other researchers. For example, Greenwood and Sommerville (2011) demonstrated an analysis approach for the identification for sociotechnical risks associated with coalitions of systems, and Johansen and Rausand (2014) posited a framework for defining complexity for risk assessment. Epoch 4 integrates this body of work into the main development environment.
3.3.3.1 Sociotechnical Network Models
Higher‐fidelity models of system development environment of the hardware and software typically ignore the necessity for the explicit recognition and modeling of the sociotechnical network. Luna‐Reyes et al. (2005) noted that many information system development activities fail to deliver because of social and organizational factors. A concurrent modeling and analysis environment that recognizes the effects of stakeholder interactions would address this shortcoming.
While modeling of sociotechnical networks is not new (Hu et al., 2010), Epoch 4 fully integrates both the modeling and subsequent exploration and experimentation of the sociotechnical network into the development cycle and uses tools such as systemigrams (Mehler et al., 2010), agent‐based models, and system dynamics models to visualize and model the influence between stakeholders. This development of a representative sociotechnical network model provides the ability to monitor the trends and alignment between stakeholders and potentially identify existent and emerging risks.
This concept of modeling of the sociotechnical network during system development has been demonstrated by the authors who used changes in information entropy (IE) to assess whether the stakeholders were moving closer or diverging in their alignment of key belief areas (El Saddik, 2018). The temporal belief and alignment measures discussed in Section 3.3.1 indicated risk to system development. Epoch 4 explicitly recognizes both the source of sociotechnical risks and the importance and necessity to ameliorate them.
3.3.3.2 Digital Twins
Epoch 4 recognizes that the acceptance of any development effort is dependent upon successfully meeting the preferences of myriad stakeholders. System development environments that advocate frequent stakeholder interactions attempt to create an environment where there are few surprises and disagreements are worked out in near real time. In most cases this represents an improvement from the staged formal interactions that may occur asynchronously (such as the V model) and are document based. However, frequent interactions with stakeholders can be costly and, in some cases, may not be feasible. By augmenting the development environment with digital twins, Epoch 4 provides a mechanism to reduce the burden on the full set of stakeholders as well as allowing real‐time experimentation during all phases of development.
There are a number of definitions of digital twins, but the most useful for the discussion here is from El Saddik who defines a digital twin as: “A digital replica of a living or non‐living physical entity. By bridging the physical and the virtual world, data is transmitted seamlessly, allowing the virtual entity to exist simultaneously with the physical entity” (Rosen et al., 2015). Developing digital twins is used extensively in industry (Tao et al., 2018) and in healthcare (Baillargeon, 2014; Bruynseels et al., 2018). While there has been work in modeling stakeholder in various domains with an array of preferences (Le Pira et al., 2015; Tracy et al., 2018), quantitatively modeling the preferences of stakeholders in the development space per the aforementioned set of sociotechnical measures is not nearly as prevalent.
Epoch 4 advocates the creation of digital twins to model the preferences of the stakeholders as they relate to the development activity and the associated effects that the implementation of the development may have. The models of the digital twins can be straightforward. For example, Bayesian networks have been used to model stakeholders. The key is that a digital model of each of the agents is created with the interactions between them quantitatively described and how the introduction of new evidence affects preferences.
3.4 The Artificial Intelligence Opportunity for Building Sociotechnical Systems
Consider the definition of AI offered by Poole et al. (1998): “The study of intelligent agents: any device that perceives its environment and takes actions that maximizes its chance of successfully achieving its goals.” As the fourth epoch of SE matures, there is significant opportunity to employ various AI techniques to enhance quality and reduce risk during capability development. AI can be seen as computational agents that work with the development team throughout the development life cycle, essentially human–machine teaming from a sociotechnical perspective.
Figure 3.3 provides a high‐level taxonomy of AI. For the application discussed here, techniques discussed above can be viewed as approaches to enable the identification of risk events in the sociotechnical space and a forward projection of what will happen if the risks are ignored or mitigation actions are specifically taken. For example, in the context of risk identification, it is straightforward to examine specific applications of AI. Conditional decision trees can be developed to identify risks based upon temporal sociotechnical measures. Natural language processing is useful to assess the tone, and possible misalignment, by examining electronic communications. Bayesian reasoning can be employed to predict the possible consequences of the misalignment. Artificial neural networks can be used to learn from previous cases as well as dynamic data.
Figure 3.3 A taxonomy of artificial intelligence.
Initially, it is anticipated that AI will be used primarily for the development of complex sociotechnical systems that will be long lived and are likely to evolve over their lifespan. These systems have diverse stakeholders, and the stakeholder population will evolve over time. Understanding the perspectives of the stakeholders who may not be available or even exist during development is a necessary requirement to field a system that meets the requirements of the diverse stakeholder population. Further, the evolution of the deployed system as well as the sociotechnical ecosystem provides additional challenges. Modeling stakeholders' behaviors and evolution with AI provides an approach for prospective analysis of risks.
3.5 Using AI to Track and Interpret Temporal Sociotechnical Measures
For truly effective HSE in system development, it is essential to measure alignment between stakeholders and their belief that the project will succeed along numerous axes to rapidly identify existing and emerging risks. In the previous work cited by the authors (Barry and Doskey, 2019; Doskey and Barry, 2018), the approach taken to ascertain this information was via survey or forensic assessment. While this can work effectively on small development efforts, this approach does not scale when the number of stakeholders begins to increase into the tens or more.
Using AI there is an opportunity to conduct real‐time and near‐real‐time analysis of sociotechnical measures using natural language processing and sentiment analysis. Sentiment analysis is well established in social media (Ortigosa et al., 2014), and APIs exist, such as Google Natural Language API (Putra, 2019), which allow individuals to quickly create