Director of National Intelligence Lt. Gen. James R. Clapper, USAF (Ret.), once observed that one of the peculiar behaviors of the intelligence community is to erect totem poles to the latest fad, dance around them until exhaustion sets in, and then congratulate oneself on a job well done.
One of our more recent totem poles is big data. Big data is a byproduct of the wired world we now inhabit. The ability to amass and manipulate large amounts of data on computers offers, to some, tantalizing possibilities for analysis and forecasting that did not exist before. A great deal of discussion about big data has taken place, which in essence means the possibility of gaining new insights and connections from the reams of new data created every day.
Or does it?
Some interesting assumptions about big data need to be probed before we dance some more around this totem pole. A major problem is the counting rules. Eric Schmidt, the chairman of Google, has said, “We create as much information in two days now as we did from the dawn of man through 2003.” He quantifies this as five exabytes of data (5 x 1018). Schmidt admittedly counts user-generated content such as photos and tweets, for example. All of this may be generated; but is it information, and more importantly, is it intelligence?
This data clearly is information—to someone—but very little of it would qualify as intelligence. It does qualify as a very large haystack in which there are likely to be very few needles that will be of use to anyone engaged in intelligence. To cite a more relevant example, the National Security Agency (NSA) programs lately in the news went through millions of telephone metadata records, which led to 300 further inquiries. The argument can be made that without the NSA metadata program, these leads might not have existed at all; but a means-and-ends argument remains over the larger big data claims.
It is intriguing that most of the more optimistic claims about the possibilities of big data come not from intelligence analysts or their policy customers but from people in information technology. This is self-serving, to be sure, but it does raise concerns about hype. Beyond examples such as the NSA program or link analysis, there have not been many concrete examples of specific applications of big data. Some do exist: the terrorism connections so noted; the details of weapons systems; or economic data. However, no amount of data will get at some of the key questions uppermost on the minds of policy makers: intentions. What North Korea or Iran or any other nation or leader will do next is not very susceptible to data. Moreover, as one senior policy official remarked, “I do not want data, I want knowledge and insight.” This is an extremely important point: our customers want knowledge and insight, not data. And no; data in and of itself will not necessarily produce knowledge and insight, any more than crowds will produce wisdom.
Some pushback on the utility of big data has taken place. As several analysts have noted, correlation, which is relatively easy with data, does not equal causation. The Iraq weapons of mass destruction experience is a good illustration of this pitfall. It is easy to confuse large amounts of data with in-depth knowledge and expertise. To a certain degree, data are much easier to deal with as they are almost mechanical and can give the illusion of useful insights when they actually may be somewhat ephemeral. As Kenneth Cukier and Viktor Mayer-Schoenberger noted in a 2013 article in Foreign Affairs: “Big data is a resource and a tool. It is meant to inform, rather than explain … .”
The real threat is the big data hype will drag the intelligence community even further from what should be one of its main contributions to policy makers: creating and disseminating in-depth knowledge. The intelligence community used to be the fount of knowledge on certain key topics. This has faded in the last decade or so for several reasons. First, many issues have been jostling for attention at the top of the priority list, instead of one main issue such as the Soviet Union. Second, our two predominant concerns of the past decade—counter-terrorism and counter-insurgency—essentially are tactical issues. So, we have lost strategic focus and strategic capability. This becomes serious as we look at the now-dominant cadre of analysts we have hired since 2001, most of whose experience has been tactical.
Analytical managers therefore will have to be on guard to assure themselves that their analysts also continue to be “knowledge workers,” a term coined by Peter Drucker around 1959 to connote someone who works primarily with information or who develops and uses knowledge in the workplace. Creating knowledge goes back to expertise. Big data are not enough. What the data mean, in context, and being honest about what the data do not tell us, would be much more useful as an analytical guideline than the continued, unjustified hype.
Mark M. Lowenthal is the president and chief executive officer of The Intelligence and Security Academy. For another perspective on this question, see "A Longtime Tool of the Community" by Lewis Shepherd.