Sociotechnical Systems and the Future of Human-Model Interactions

Embracing Complexity in the Future of Work
Involved in the systems engineering of multiple domains and disciplines, Dr. Donna Rhodes says the path forward in engineering is truly model-centric.

Read the full article written by Daniel de Wolff here.

For more than 20 years Donna Rhodes has played a leading role in the advancement of sociotechnical systems theory and practices. As our world grows increasingly complex and interconnected, these engineered systems, which are technological in nature and inextricably tied to delivering value to society (e.g, multi-modal transportation or energy distribution systems), are becoming increasingly prevalent given the challenges faced in today’s world. “The interconnectivity of things is driving how we architect these systems and how we evolve them over time,” says Rhodes. Her work includes case study investigations of many types of sociotechnical systems. Her recent undertaking in the realm of maritime border security is a perfect example: The system itself consists of aircrafts, drones, ground stations, radars, ships and all other manner of assets used to protect the security of a coastal border. It’s a complicated system that requires fitting many moving parts together seamlessly while taking into consideration the fact that this system will almost certainly change as it evolves over time. “Systems constantly change and improve,” says Rhodes, “but one can only re-architect a system at specific periods over time, so in our research at MIT, one of the things we are exploring is how to determine the appropriate time to undertake a major re-architecting activity.” This means not only examining the properties of these systems (are they adaptable, evolvable?) but also determining what new approaches and design strategies could be used to design systems so that there is more predictability of both when and how they are best evolved over time.

Rhodes points out that while modeling and simulation has long been used, engineering is rapidly becoming a truly model-centric discipline. “We’ve always developed models when we perform any kind of engineering, but now the power that we have from a computational perspective combined with the needs that we have as far as our interconnectivity is concerned are driving us to extensive, more powerful use of models.”

Systems are designed using state-of-the-art model-based methods that employ sophisticated software toolsets, specifications are replaced by digital artifacts, and designs (and design changes) are evaluated using digital prototypes in simulated virtual environments. As a result, an emerging area of sociotechnical systems has been labeled digital twin technology. A true digital twin is not just an exact digital copy of a product or system. Rather, a true digital twin receives sensor data from the actual operational system. The sensors gather real time information about the environment and the operations of the actual system in order to contribute to the model. The model or digital twin is useful in terms of analysis and understanding how the system is performing, while also allowing for some level of predictability. GE’s work on digital wind farms, for example, is a true representation of digital twins. Every wind turbine has sensors that collect data and feed it back to the model. With this approach, or this collection of digital twins, one can examine the differences in wear-and-tear of individual turbines, predict the necessity for end-of-life part replacement, and examine the performance of the entire wind farm as a whole.

Digital twins in their full realization are few and far between right now. “However, we are rapidly approaching a world where there will be an inversion of the value of intellectual property,” says Rhodes. “For example, today your products might be your valuable assets. In the future, the model itself will not only possess the same characteristics as your product but will also be armed with historical information of its performance and the capacity to project that into the future.” Imagine a world where your business portfolio doesn’t consist of product technologies but rather the digital models of these technologies. As we transition into this reality of models as primary focus and product or component systems as secondary, the question of trust in these models becomes essential.

Which is why Rhodes and her research group are investigating how humans interact with models, the implications for these interactions, and how to use these findings to create more effective human-model interaction. “As we move into this world of the digital twin, where models are the prominent entity, the need to understand the relationship between humans and models is further accentuated,” says Rhodes. The study has revealed a preference for human knowledge as opposed to the encoded knowledge that models represent. With this in mind, transparency becomes a key issue. Some engineers may want total transparency in the model, whereas senior decision makers may not want or need this level of transparency. “The question is: How do we create these types of model-centric environments so engineers who want to see into the code can get the details they need, while senior decision makers can interact with a model to retrieve meaningful information without needing to know how to operate modeling toolsets?” While sociotechnical system architects like Rhodes can design technology that augments or automates intelligence in a variety of ways, she believes we are not yet at a point where humans can be completely removed from the equation. Rather, she is interested in the critical points where humans need to be directly engaged, as opposed to where we can rely on automation to make our decisions for us.

Rhodes is particularly interested in how to take an architecting approach to enterprise transformation of sociotechnical systems. As companies attempt to transform themselves to undertake a model-based approach to engineering, they often start with rewriting engineering practices or simply inserting modeling toolsets into the current system. But in reality, it has to go much deeper than that, according to Rhodes. For example, does the organizational structure have to change? And if a model-centric approach is adopted, the imperative for model trust implies a need for a cultural as well as an infrastructure shift within the organization. Another aspect of the research focuses on the new leadership roles and responsibilities that are required as enterprises transform under the digital paradigm.

Sociotechnical systems are of interest to industry, government, and academia alike. As a result, Rhodes sees powerful partnerships continuing to develop in order to better understand and benefit from these systems. “Industry is in the thick of having to make changes to these systems,” she says. “Academia can provide a means for them to observe what is going on and distinguish patterns in aspects they may be too close to, to objectively observe. In an academic setting, we can work on general frameworks and strategies that others can use to apply to the architecting of sociotechnical systems.” And the Industrial Liaison Program has become an important part of how she does her work as a researcher: “It’s impossible to put your finger on the pulse of what is going on in the world without looking outside of your own environment. As a researcher, I’ve found tremendous benefit working with ILP. My involvement in various capacities, from briefings and discussions with executives to speaking at conferences, have provided me with extremely valuable insights and knowledge. It’s an important cycle: we get to better understand industry needs and they allow us to push our research and new ideas to get their feedback. It’s a very powerful tool for everyone involved.”