Contemporary Operating Environment

Moneyball and Small Wars

By Gareth Rice May 14, 2019

Have we been thinking about warfare the wrong way all along?

One of my favourite book/movie combinations is Moneyball by Michael Lewis. In it, Lewis tells the story of Billy Beane, the General Manager of the low-budget American Baseball team Oakland Athletics. In 2002 Beane seeks to turn the tables on the baseball industry by using statistical analysis (or Sabermetrics) to give his team a competitive advantage. In doing so, he was able to counter his smaller budget with record-breaking results. My interest in this new way of looking at an old problem drew me to the book Small Wars, Big Data: The Information Revolution in Modern Conflict which uses similar approaches to challenge the way we understand small wars, particularly counter-insurgency (COIN).

As military practitioners, I find that we all tend to have strongly held beliefs about the right way to fight an insurgency. For many, this is based on personal experience, often in Iraq and Afghanistan. If we are lucky, this is also supported by some independent research on the topic. For my own part, I found myself putting pen to paper in 2013 after a deployment to Afghanistan to describe my own thoughts on the problem. As in American Baseball in 2002 though, experience and intuition does not always provide the correct answer to the challenges that we face. Consider then the research of Berman, Felter & Shapiro in Small Wars, Big Data which seeks to provide objective analysis to how we fought these wars.

Understanding violence

Fighting any war amongst the people is fundamentally concerned with seeking a decline in violence between key actors that can ultimately precipitate a political solution. Therefore, what should concern us as practitioners are the factors that cause that violence to continue and what will cause it to reduce. Without any data to inform these questions though, we often find ourselves making important decisions based on intuition rather than evidence. This is further compounded when you consider that we rarely appreciate the effect of our decisions past a six-month deployment cycle, and may never return to find out. Should consecutive military commanders at any level have differing views on many of these challenges, the result would be at best an inconsistent operational approach.

Small Wars, Big Data seeks to examine many of those notions that we have long held dear. Does a lack of employment actually drive young males to fight for an insurgency? Does aid and development actually win hearts and minds? Does winning a village eventually translate into winning a war? I have certainly held strong views on each of these questions but never bothered to wonder where or how I came to these conclusions. When I really reflect on them, they were feelings. Of course creating jobs would undermine the insurgency I would hear myself say. Who would want to fight when you can earn more money in a safer job? Such logic is hard to deny until you consider the impact of part-time fighters, foreign fighters and the realisation that an insurgency actually doesn’t need that many fighters to be effective in the first place. Yet, without data to challenge these notions, how would we expect to know any better?

How much money and effort have we potentially spent on these operations without really understanding the war that we are engaged in? Worse yet, could we have made the problem worse without even realising? Surely the provision of aid for instance, is a no-brainer. Aid helps the population and reinforces the legitimacy of both the host nation and the counter-insurgent. But what if it actually increased violence? After all, aid can easily become another commodity that needs to be controlled and undermined by the insurgent, as the authors highlight in a case study from Mogadishu in 1991-92. The challenge for the counter-insurgent is that the correlation between the rise in violence and the provision of aid might not be easily visible or the link clearly established.

Is data the answer?

These challenges are further compounded by the inevitable operational security that compartmentalises the data that we do have. Such secrecy ultimately makes it harder for the academic world to assist us in finding solutions to these problems. A further challenge is the unfortunate legacy of Robert McNamara’s obsession with data in the Vietnam War that led to the continuation of a failed strategy. There is certainly a balance to be achieved here, although the failure in Vietnam was not so much the collection of data but how the data was interpreted, whereby body counts became a key metric of progress.

The Moneyball phenomenon has already spread to some circles in the US military and there are certainly examples of the ADF using data effectively on operations; however, its use is far from widespread. If we want to be better at fighting these wars we will have to accept that we don’t have all of the answers and we will have to be willing to be proved wrong. When I reflect on my own time in Afghanistan, I can’t help but be confronted by the mistakes I may have made by providing solutions based purely on intuition. I’m willing to be proved wrong again. Are you?  



Gareth Rice

Major Gareth Rice conducts future warfare analysis for Force Design Division. Please feel free to continue the discussion with him on Twitter @RiceGareth

The views expressed in this article are those of the author and do not necessarily reflect the position of the Australian Army, the Department of Defence or the Australian Government.


is Baseball a good analogue for military conflict? Data analytics are great when the problem lends itself not only to the collection of good, relevant data, but also when you have the tight control over the resources that you are deploying (i.e. pitchers or batters). Can we expect to get the quality and depth of data for a conflict problem that we can for a pitcher or a batter? This is a data collection and data quality problem, not so much an analytics problem. Then, even if we could get great data, do we have the ability to pull the pitcher at the end of the 6th because we know his stats fall off after 63 pitches? This is the problem of what you can do with/how you can exploit your data analysis. There will inevitably be more and more data coming from the future conflict scenarios, but arguably this will mean that we will be relying more and more on commander's experience and "intuition" to see the wheat form the chaff, and to make decisions based on uncertainty and ambiguity not the vision of clarity that some data technologists would have us believe. Let's use technology and data as best we can, and figure out what and how to measure to provide insight on the battlefield, but let's not forget to develop people who can think and learn from experience in complex environments, and not be slaves to the data scientists... developing skilled intuition and decision making under uncertainty/complexity in an information dense/rich environment is the bigger challenge (whilst we become lured and distracted by the promises of machine learning, AI and data analytics)... discuss...

Hi Rob, Thanks for the response. I hope my baseball analogy didn't distract you too much, I was more intrigued by finding new ways to look at old problems. Having said that, I completely agree with your point about developing our people who ultimately have to make decisions, with or without data. I don't think this field of study would ever remove the uncertainty of war, but it probably has the possibility to provide better measures of effectiveness and performance. Cheers, Gareth

Hi Gareth, Great article. In the commercial sector metrics such as CPC(cost per click), CAC(customer acquisition cost), and LTV(Life time value) are central to success. The same goes for the concept of Value Propostion in the Hacking 4 Defence modification of the Lean StartUp Methodology, However, as you mentioned in your article, Robert McNamara’s RAND Corp “whiz kids” leveraging computing, economic analysis, and game theory for defence application had very mixed results. I’m thinking along similar lines as you, as long as we learn from the mistakes of the “whiz kids” past. In short, the idea of spending $25,000 an hour for an F15E Strike Eagle to drop a $500,000 precision weapon on an insurgent paid $500 to install an IED with materials cost of $25 is simply uneconomic and a path to national insolvency. Spending 1000X+, in comparison to an adversary spend of X, to disrupt adversaries is often a very, very poor value proposition. And an opportunity for an adversary to ruthlessly exploit. For small wars and counter insurgencies, I would argue that measurements such as citizen/business sentiment, corruption indexing, and perhaps even Comparative Net Promoter Scores(NPS) for government and insurgency would have more value than body counts and their nebulous 1st, 2nd, 3rd order effects.

Hi Chris, Since we're already chatting on Twitter, I won't respond again but thanks for the insights. You've given me a new angle to look at for this study. Cheers, Gareth

Add new comment

Cove App


Fast access to The Cove anywhere, anytime. Additional feature of receiving notifications for new content.

Reflective Journal


Record your reflections in a structured way to improve your performance.