From smart selling to smart shelling

If you have ever bought anything on Amazon, you’ll be familiar with the website’s automated recommendations that, based on your previous purchases, suggest other things you might like. Similarly, Tesco club cards enable the retailer to build a complete picture of your purchases, analyse your behaviour, and tailor the promotions it sends you.

The same principle applies to smart advertising – and broadcaster Sky has just launched a technology which, using data from its set-top boxes, makes judgements about households’ income, age, class and location before deciding which commercials appear. The charity Freedom From Torture has just become the first advertiser to apply the technology.

Analytics – the techniques for analysing data sets to identify patterns and characteristics, informing decision-making – has traditionally a job for analysts rooting through vast quantities of data. But advances in technology have meant that this work can now be automated. And the technology isn’t just applicable to marketing campaigns: big data and analytics create the potential to shape the future of how Britain wages war.

We know, of course, that the intelligence services use analytics to root through massed communications data – and that they want greater freedom to do so. But in the military field, by contrast, vast amounts of information is currently collected without being subjected to analytics, according to a report by the Royal United Services Institute (RUSI).

The paper, ‘Big Data for Defence and Security’, was published last month and noted, for example, that thousands of unmanned aerial systems (UAS) are now being used for surveillance. One model, the MQ-9 Reaper, collects the equivalent of up to 20 laptops’ worth of data per sortie, but 95% of battlefield video data is “never viewed by analysts”, it says.

Elizabeth Quintana, director of RUSI’s military sciences department, who commissioned the report, says government is currently at a stage where it is looking at “where it might use [the technology] and what it could be used for”.

Analytics are being used more widely within the Ministry of Defence’s (MoD’s) human resources department. In order to determine which skills it needs to retain and attract, and how to deploy them, it uses analytics techniques provided by private analytics company SAS. The firm’s central government director Simon Dennis tells CSW that this process takes into consideration a wide range of information, such as historical manpower records, a skills database and staff satisfaction surveys. This, he says, helps the department make sure it has the right people, in the right places.

While this is, according to Dennis, “quite a complicated” process, it will be still more challenging to use ‘big data’ – a term defined by RUSI as “datasets too awkward to work with using traditional, hands-on database management tools” – to improve our defence capabilities. Dennis says that a military officer deployed in Afghanistan told him earlier this year that in order to carry out his job, he spends a lot of time going through large chunks of data and pulling out the parts he deems useful. “If he had that [analytics] technology,” Dennis says, “he could spend more time adding value and insight to the data.”

The advocates of analytics argue that the techniques could be of huge benefit to our armed forces on the frontline. And the government already has much of the raw material: James Petter, managing director for information management company EMC in the UK and Ireland, sees opportunities in the “significant amount of data the government is sitting on already.” It might, for example, help officers to “make a decision on whether to be in contact with the enemy”. A former infantry officer himself, he says that he would have “wanted to know where all my assets were, what the lay of the land was, what sort of thing the enemy would be doing and how they [were likely to] behave”. The technology, SAS’s Dennis adds, could also help eliminate ‘friendly fire’.

Petter envisages soldiers using the latest gadgets: tank commanders could use iPads, and infantry might wear pairs of the new Google glasses, which “deliver pictures directly to your eyes”. These troops, Dennis suggests, will “find ways of using the data and come up with ideas: it is in their interest if it keeps them alive!”

Other techniques which could be transferred into the military context include a technology that already exists for cars, called telemetry. Petter believes it could be applied to the MoD’s life-cycle maintenance of equipment: one of the “biggest costs the military has”. The method involves sensors attached to equipment to inform users when different parts need attention, improving efficiency and avoiding malfunctions.

The military can also draw lessons from preventative policing, says Dennis. According to US police intelligence produced from crime statistics and weather information, car crime in multi-storey car parks increases when it rains. “Thieves,” Dennis says, “don’t like to get wet”. This may seem obvious, he says, but nobody had “actually noticed it and pulled together those facts”. Knowing that increasing patrols in car parks when it’s raining improves the “chance of an arrest, and crime prevention” could change the way a commander operates. “And you can see how that would play out in a battlefield – by predicting [actions] using historical information combined with real time information,” Dennis says, the enemy’s movements become more predictable.

While some of this technology is being used in defence already, there is still some way to go, Dennis says. And Quintana explains that while analytics tools are currently being used on data sets in each of the MoD’s three forces, the challenge lies in allowing software to jump across different databases which have historically been separated for security reasons. This, she adds, “needs to be addressed”.

Many restrictions limiting the sharing of information, Dennis responds, are in place for good reason. In Afghanistan, for instance, where the UK military is working alongside Afghan police and local military, most of these forces are “on our side”, but you “don’t want to give away anything to the ones who are really the Taliban in disguise”. It’s best, therefore, if the military can communicate its intelligence – findings drawn from analysing raw data – whilst masking the origins of that data.

Analytics, Dennis says, can achieve exactly that: applying the technique produces insights that wouldn’t be obvious even to those originally supplying the information. It’s comparable to the process of credit-rating, he says: “A bank might be able to share with another bank a customer’s credit score, but it wouldn’t be able to pass on the raw data: information about every single transaction by this customer”. What’s more, when different data sets are combined the origins of intelligence become still more obscure.

So it’s pretty easy to list the potential benefits of analytics in defence; realising them is another matter. EMC’s Patter says the major difficulty lies in shifting government’s focus from “managing and securing [its] current data” to investing in new, innovative techniques. He is, however, optimistic that over time this will be achievable. Certainly, the MoD says that it’s “aware of [RUSI’s] report, which highlights some of the key opportunities and challenges the department will face within the defence information domain”, and that it “will consider its findings closely to better understand how big data solutions could benefit defence”.

The growing government focus on ICT over the last three years, Petter says, is a promising start; he thinks that in five years, “we are going to see some major progress.” The opportunities in defence are “very exciting”, Dennis says, welcoming the MoD’s response. RUSI’s report notes that some commentators observe “the MoD doesn’t have big data; Google and Amazon do”. In time, though, the MoD may just catch up.