When Real-Time Intel Still Isn’t Fast Enough
We now live in a world where both man and machine can access data on almost any topic at any moment. Documentation of our world happens in real time, through a constant, autonomous torrent of ones and zeroes — and research and recall of that information have been reduced to mere mouse clicks. With all data available at all times, opportunities — and adversaries — can also move in real time. So we should ask ourselves, “How do we move faster?” This is the domain of predictive analytics — a concept that isn’t new, but the potential of which, in a world not limited by data or processing power, is expanding rapidly.
I’m at Lockheed Martin where we focus relentlessly on expanding and improving the technology and tradecraft to remain ahead of adversaries. Our investments in predictive analytics primarily serve the goal of anticipating threats emerging from dynamic environments, and being able to do so faster than others. (it would be naïve to think that our adversaries are not finding opposing uses for these technologies). From predicting the locations of roadside bombs to pinpointing the next government collapse, exploiting available data requires high-performance collection and rapid, thorough, and transparent analysis.
It is fascinating, however, that the solutions we’ve developed have also turned out to be effective in fighting other threats to safety and wellbeing – among them, criminal networks and bacterial infection.
Granted, the political and military turmoil right now in Syria and Iraq is a more typical focus of the analysts using something called LM Wisdom, the solution we developed to automate the collection of data and subject it to advanced processing and analysis. LM Wisdom is being used to monitor events in real-time, and correlate, aggregate, and index massive sets of multi-language data. By using processing steps and filters, analysts can collect information and integrate everything from locations to tonality of messages to modes by which people are communicating with one another. Once a model of what’s happening right now is created, correlation algorithms tailored to specific problem sets enable the prediction what might possibly happen next. Information processed through LM Wisdom augments traditional intelligence gathering, so decision makers can understand various threats and what they could rapidly become.
For over half a century, the aerospace and defense industry has been at the forefront of defining advanced analytic techniques, because we needed them to address highly complex engineering challenges. Some of these challenges are well known, such as propelling a man faster than the speed of sound and safely landing a man on the moon. Countless others may never fully be appreciated by the public at large. Perhaps the most impressive aspect of these early solutions was that they all relied on a multi-disciplinary approach, combining mathematics, engineering, computer science, and physical sciences.
What has become abundantly clear across the decades is that any application of analytics to a complex problem relies on three essential components. Analysts need to acquire meaningful and abundant data sets, often from multiple sources internal and external to the organization. Algorithms are then needed to weed out the noise from high-value information and “connect the dots” across the information. Lastly, analysts must rely on their tradecraft – the investigative skills required to ask the right questions of big data.
What is new, however, is that we are no longer limited by data or processing power. Data is enormous and available in real-time — we are now, as many have observed, in the era of Big Data. Processing power, meanwhile, is now so immense that we can capitalize on this abundance. It might seem that more data would increase the unlikelihood of finding the proverbial needle in the haystack, but this challenge is largely overcome by the sheer processing power available in modern computing platforms. The true value of expansive data is in the enablement of analytic prospecting — quickly identifying and recognizing patterns and connections within the data. We can look beyond finding the needle to finding patterns that might indicate the presence of a needle. We can truly start going faster than real-time.
Moreover, the same multi-disciplinary approach and computational ideas used to simulate airflows of fighter jets or predict missile trajectories can now be applied to harness data and unearth actionable intelligence in previously intractable areas. For example, we have employed data analytics to assist in the discovery and identification of criminal networks responsible for producing and distributing counterfeit drugs. Using essentially the same tools we use to make sense of political and military turmoil, we were able to discover the true identities and aliases of key players as well as the flow of money through the illicit network.
It turns out, too, that the same toolkit can be applied in medical settings. We found that the signals that human bodies constantly emit can be tracked just like a missile or satellite. For example, our team developed an algorithm that detects sepsis, a potentially fatal blood condition, in patients’ bloodstreams up to 14 hours faster than currently employed techniques. Our bodies give off signals like temperature, blood pressure and white cell count, and using these signals, the algorithm can help health care systems and providers deliver more personalized medicine with higher likelihood of improved outcomes.
The power and applications will only continue to grow and spread. Big data will only get bigger. The more computing devices we connect to the internet of things and the more areas to which we apply complex algorithms will only expand the information we have prior to making decisions. As data and processing power cease to be a limiting factor such analysis will revolutionize the way we interact with the world and measure the risks of our decisions.
Of course, these growing capabilities are also available to people who mean to cause us harm. Meeting the challenges that they will present will always be a matter of staying ahead. In a world not limited by data or processing power, real-time awareness will not be sufficient. We will need to be faster.
For more expert insights on the power of predictive analytics, see HBR’s Insight Center, Predictive Analytics in Practice.



Marina Gorbis's Blog
- Marina Gorbis's profile
- 3 followers
