Imagine being able to predict enemy patrol and aerial surveillance routes with a high degree of certainty, or an increasing vehicle availability and reliability by conducting less maintenance, or predicting battlefield weather effects instantaneously using distributed sensors already inherent in military vehicles and aircraft. These capabilities are within the realm of possibility.
Military practitioners are experts in instinctive decision making with limited information; after all the fog of war is all-pervasive. However, in his landmark book ‘Thinking, Fast and Slow’ Daniel Kahneman postulates that the ‘fast thinking’ process in the human mind is terrible at statistics and finding patterns in large amounts of information. Heuristics (sub-conscious mental shortcuts) are systematically biased to make sense of complex situations by attempting to find similarities to one’s own experience that inevitably introduces cognitive biases. These mental shortcuts are great for battlefield decision-making, but not so great for deliberate decision making in operational and strategic issues. While an effective decision making process is critical for the conduct of military operations, effective appreciation of all relevant information is also essential. Data-driven decision making now allows us to not only appreciate the perceived, relevant information, but all available data. Despite commercial advances in data analytics and processing power, military uptake has lagged.
Today engineers and data scientists are able to analyse unbelievable amounts of information (think web-scale datasets) to reveal previously unknown insights including predicting crimes to within a city block and seeing inflation in real time. In addition to the ability to analyse these massive data sets, is the coming age of machine learning, or artificial intelligence. By harnessing the available data, computers have been able to learn how to translate text, discern faces and photographs, recognise speech in any language and drive cars in unfamiliar territory. This is achieved by finding statistical correlations among billions of data points. But the sheer size of the data sets renders traditional analytical techniques uneconomical and time consuming to the point of irrelevance. Enter big data.
Big data analysis involves vast amounts of data, often from multiple sources and has quickly been embraced in the business world. Traditional data analysis requires the input information be the same format, complete and accurate. In big data, accuracy is derived from the vast sum of information. Data can come from multiple programs, databases or formats and need not be complete. Army could feasibly analyse the entirety of the human resource system, financial systems, logistic systems along with communications (email and signals) to uncover trends in workforce dynamics, force generation cycle efficiency and materiel reliability.
The potential of big data has not been understated or missed by Army, it is referenced in the Army Modernisation Plan in three of the six Army Modernisation Lines of Effort including Human Performance, C3 and Situational Awareness. Here Army is scoping concepts to exploit advanced computing, information and decision support systems and tools to enable better decision-making.
Future Logistics Concept 2035 states:
‘The single most important trend likely to affect the delivery of Defence logistics in 2035 is the growing abundance of data and the need for interconnectedness of information management systems to support analysis of that data’.
Reliability and Maintainability (RAM) analysis of Health and Usage Monitoring Systems (HUMS) databases, logistic and engineering databases could yield vast efficiencies through increased availability, reduced maintenance and more economical operating costs. Army personnel and equipment are the source of this information and is well advised to start setting the foundations to enable future data-driven decision support systems.
How do we address this shortfall and take advantage of big data opportunities? Firstly, we must gain a basic understanding of data analytics and big data capabilities in order to frame potential projects and to champion their implementation. Secondly, the establishment of skilled specialists (data scientists and computational statisticians) warrants investigation. At the very least, these specialists could be engaged to advise Army on management, storage and portability of existing data sets.
Finally, and most importantly, by recognising the fundamental value of data and its undefined potential in gaining insights into the future, we must collect and store as much data as is reasonably practicable. Including everything from on-board vehicle sensors to search queries in IT systems. This could be extended to purchasing external data such as design and testing data for capital equipment. The techniques and tools of data-driven decision-making are exciting, but the underlying value is in the data itself, this is certainly a case of more is better. Army is well placed now to implement fleet wide data capture capabilities for modernisation projects. The ability to analyse this data may not currently exist within the military, but the data must come first.
The potential of data-driven decision-making should not overshadow the importance of the human element; it will not replace or reduce the need for decisive operational and tactical leaders. To the contrary, it compliments them by reducing cognitive bias and allowing vast arrays of complicated factors to be distilled into manageable insights, hence allowing the commander to focus on the art of warfare.
About the author
Tim Byrne is a Royal Australian Electrical and Mechanical Engineer’s officer who holds a bachelors degree in aeronautical engineering and a masters in project management.