Complexity beyond the battlefield

“Defence is a truly complex, multifaceted business,” says Nick Colosimo, an engineer and technologist at the international defence company BAE Systems. “There’s complexity in the physical equipment and digital systems we make, in the supply chain that helps make them and we how provide support. Rarely does a ship or aircraft operate on its own. They always operate as part of a vast and complicated system.”

An important challenge in conducting NATO operations on land, sea and air, as well as in space and cyberspace, is to get all the many parts of these systems working together. And as security becomes more multifaceted, this also involves government departments, intelligence services, industry, research and development organisations, academia and universities. “We need all that to achieve something greater than the sum of the parts,” says Colosimo.

In NATO terminology this is called “multidomain operations”. In making it work, NATO faces a challenge well known to modern science: that of understanding and managing so-called complex systems.

These are systems that naturally evolve and adapt to the environment around them, examples include ecosystems, economies or living organisms. The collective behaviour of such systems cannot typically be deduced by understanding the parts because they emerge from the delicate interactions between these components. As a result, complex systems are difficult to predict and prone to sudden, dramatic shifts from one regime of behaviour to another. They are also often surprisingly resilient to disturbances.

A major theme in the study of such systems is that lessons learnt about one complex system can help understand others. The study of ecosystems, for example, yields insights into economies and large-scale computer networks.

That raises the question of whether the patterns of complexity in fields such as economics, pure maths and computer science can inform how we make sense of the data challenge of multi-domain operations. So last month, a small group of experts from industry, the Ministry of Defence and complexity science came together to discuss this challenge during a New Scientist workshop sponsored by BAE systems.

For Gordon Niven, a specialist in systems thinking at the Defence Science and Technology Laboratory at Porton Down, a key obstacle in adapting defence systems to rising complexity lies in the military’s relatively fixed and structured relationships and ways of working. This has facilitated large-scale industrial warfare against a well-defined enemy for over a century.

But the nature of warfare is changing. The military now confronts more diverse adversaries, in groups of many sizes and motivations, often not associated with nation states.

So-called grey zone warfare can also be carried out through alternative means, for example via cyberspace, through the spread of propaganda and even by the weaponisation of human migrations. Add to this the explosive growth of activity in space, under the sea and the emergence of autonomous vehicles and other agents.

In this new high-complexity setting, leaders need to digest vast volumes of information from numerous operational domains, some of which may be untrustworthy. The military must also cooperate with other military and non- military organizations with different traditions and practices.

“The military has not evolved as rapidly as its operating environment,” says Niven. Military headquarters have become larger to deal with the increasing diversity of tasks. “But this makes them less efficient, less effective and less mobile. Also more vulnerable,” he says.

But neither can the military easily shift itself to more complex settings, says Niven. “The military has actually got to work across the entire spectrum of operations because they still have to do industrial warfare,” he says. “But they also have to be capable of operating in complex situations.”

That requires new ways of thinking. Henrik Jeldtoff Jensen, a mathematical physicist at Imperial College London, points out that a key feature of complex systems is that they are indivisible wholes. The operation of any subsystem only makes sense in the context of its connections to the rest. An economy, for example, cannot be analysed without understanding its components, its relations to other economies, as well as its dependence on the underlying Earth ecosystem.

This suggests that global thinking will be crucial in producing insightful pathways to multi-domain operations. Even NATO is part of a larger network, which includes not only its many allies, but also its adversaries.

One complication for NATO is the sheer scale of the system in question and the vast amount of information associated with it. The military is learning how to gather data on increasingly larger scales. But more is not always better. “What they need is the right sort of information about the right things at the right level of abstraction,” says Niven.

An important question is how machines and artificial intelligence can help. Weisi Guo is an expert on human machine intelligence at Cranfield University. He and colleagues have processed carefully chosen data on a global scale to model and predict the outbreak of peace and conflict one year in advance with over 90 per cent accuracy.

Guo and colleagues used a simple model in which a specific region is either at peace or in conflict. They then used machine learning to study the factors influencing the likelihood of a change in this status. In general, the team found that the absence of long-term trading relations with neighbouring regions is a good predictor of the emergence of violent conflict in the near future. This may be particularly important as the military increasingly plays diverse roles not only in direct combat, but in peacekeeping, natural disaster relief and in interventions to avoid conflicts escalating.

Gathering better data on human behaviour and interactions will help, says Kerstin Hötte, a researcher in finance and economics at The Alan Turing Institute. This would accelerate our ability to run more accurate simulations and to explore multiple ideas about more flexible and adaptive system organisation at low cost, without putting anything at risk.

Other complex systems can also provide valuable lessons. Jeremy Pitt, a specialist in intelligent and self-organising systems at Imperial College London, pointed to longstanding research on social dilemmas, and the diverse mechanisms humans have invented to engineer and support cooperation. Some of these could be adapted to apply in the complex military systems setting in which autonomous machines interact.

Yet, as Niven concluded, achieving better management of complexity in NATO operations will still be difficult. Real world complex systems generally emerge through an evolutionary process of learned adaptations to past events. A similar approach for the military – experimenting with new practices to see which work and which do not – is not an option.

Small-scale exercises can help but are generally expensive. An alternative is to run tests through large-scale computational simulations meant to mimic military operations, where AI can model increasingly realistic scenarios. The problem, as many participants agreed, is that it is unclear just how accurate such simulations can be.

Somehow, NATO must find a way to test and validate new ideas and that will be hard. However, by sharing ideas around how other academic disciplines model complexity and by keeping military experts closely involved, there could well be progress in future.

For more information visit baesystems.com