What Would You Say You Do Here?

Mike Judge’s 1999 satire Office Space parodies turn-of-the-century white-collar office work at a fictional software company called Initech. In an iconic scene, corporate management consultants referred to as “the Bobs” interview Initech personnel in a search for “efficiencies.” When they sit down with Tom, a middle-aged employee who works in customer management, the meeting quickly devolves into absurdity.

Tom’s job is bringing the requirements of the company’s customers to the software engineers. One of the Bobs asks, “Why can’t the customers just take their requirements directly to the software people?” Because, Tom replies, engineers aren’t good at dealing with customers, but he is. The Bobs assume this means Tom himself brings the customers’ requirements to the engineers, but in fact Tom’s secretary actually does that, or more often, the fax machine. Finally, one of the Bobs deadpans: “What would you say you do here?” To which Tom exasperatedly responds, “Well look, I already told you! I deal with the [expletive] customers, so the engineers don't have to! I have people skills! I am good at dealing with people! Can't you understand that?” 

Tom’s difficulty justifying his job illustrates one of the larger disruptive trends unleashed by the information revolution and an emerging threat to the traditional model of intelligence analysis: the disintermediation of knowledge. Disintermediation means removing the middleman. And intelligence analysts—the part of the intelligence community that “deals with the customers,”—risk ending up like Tom if they don’t successfully redefine their role in our changing world.

In the old model, an intermediary like Tom could prevent a customer from accessing information unless they went through him. In the slower-paced world of the 20th-century, where information was scarce and knowledge took a long time to accrue, this model made economic sense. If someone wanted to find the answer to a difficult question they used to have to travel across town and spend hours in a library. It was reasonable, then, to hire an expert to just answer the question for you. This is in short, how professions came about in the first place.

Consider stock brokers. Not so long ago, most people had neither the time nor inclination to master the complicated world of financial markets. Those who did—stock brokers—were able to lower transaction costs for their customers while charging them a fee. In other words, brokers had a value proposition. But when electronic stock trading applications like E-Trade came along, almost completely removing the need for intermediaries to perform the now-simple action, the halcyon days of plain old “stock brokers” were numbered.

Similarly, intelligence analysts are  the ‘hub’ of the intelligence cycle, producing intelligence in an industrial-like process, serving as an intermediary between intelligence collectors and intelligence consumers. Thousands of sharp, dedicated analysts work diligently every day to refine and repackage volumes of ‘raw’ intelligence into digestible, ‘finished’ form for American policymakers and military commanders, ostensibly to provide them with decision advantage.

Indeed, at the outset this was seen as the feature: General Hoyt Vandenburg, the second Director of Central Intelligence, explicitly stated that the then-new CIA “would, like a central assembly line, put [raw collection] and its own work together and present an overall picture in a balanced national intelligence estimate including all pertinent data.” Over time, the assembly line model became routine, its products commodified to the point where they could be dismissed as “CNN plus secrets.”

Although some experts think analysts are likely to remain the “central element in the policy-intelligence relationship,” that proposition is doubtful if the current model is retained, because as Artificial Intelligence (AI) expert Kai-Fu Lee puts it: “Much of today’s white-collar workforce is paid to take in and process information, and then make recommendations based on that information—which is precisely what AI algorithms do best.”

A Changing Paradigm – the Promise and Peril of Information Technology

During the Cold War, it was easy to justify the analyst’s role. They had exclusive access to exquisite information provided by expensive technical collection platforms that could peek behind the Iron Curtain. Armed with this unique capability to collect Soviet secrets that weren’t available from any other source, intelligence analysts were the lens through which this data was viewed and built their reputation as reliably objective truth-tellers, even in light of several widely-publicized failures. The “Cold War Consensus” lent their judgements respectability, if not always uniform acceptance. In effect this meant the IC held a monopoly.

But disintermediation has been coming for the intelligence analyst for many years. It was after all former CIA Director Robert Gates, who, back in 1989 noted that the creation of the White House Situation Room decades earlier had initiated a great change in how intelligence was provided to the President, in a way that “had yet to be fully appreciated.” Fast-forward thirty years and the cutting-edge IT that allowed President Kennedy to receive raw reporting from collectors is available to most, if not all.

The information revolution is having profound effects on this monopoly in what is now a data-driven age. Because information technology now stores and retrieves information on its own, transaction costs have been virtually eliminated. People can simply type questions on their keyboard, or increasingly, ask Alexa. As attention spans shrink and meetings expand, analysts spend less time interfacing with their customers, resulting in stacks of unread intelligence reports. Unfortunately, speed trumps quality in the information age.

Soon, savvy policymakers and military commanders will expect to have access to everything, anywhere, at any time, and to be able to recall any fact or figure almost instantaneously. The language of contemporary knowledge professions reflects this shift. Professionals in the private sector are replacing the dead metaphors of the past—verbs like produce or deliver with present participles like servicing, sharing, filtering, cognifying, and brokering—all of which imply continuous action. Intelligence can no longer be conceived of as product—instead intelligence should be recognized as a service, analysts as the providers of this service.

Resultantly, it is fashionable today to be sanguine about the dawn of practical machine learning. Algorithms, for instance, are fast enough to appear instantaneous to people. A search algorithm can scan billions of words and find correlations in immediately. But this speed leads many to  overestimate the benefits of AI, perhaps because of a  widespread confusion about how it works, combined with our inherent difficulty understanding probability. Algorithms are useful tools—but greedy, shallow, and brittle. And as Pentagon CIO Dana Deasy says, without accurate data, “AI is irrelevant.” Unfortunately, the vast majority of the world’s data is anything but.

What They Do Here: The Enduring Importance of Intelligence Analysis

Intelligence analysts who primarily perform the time-intensive and error-prone task of what one professor at the National Intelligence University calls ‘data sifting,’—think editing spreadsheets and writing situation reports—are indeed likely to be replaced by AI in the next few years. But these analysts are working at the bottom of the knowledge hierarchy, a conceptual model that visualizes how data aggregates to information, is refined into knowledge, and is (hopefully) applied with wisdom. The true utility of analytic thinking is much closer to the top, in the uniquely human realm of cognition.

What is the value proposition of intelligence analysis in an age when information flows freely, answers are only a millisecond away, and algorithms make recommendations? Intelligence does the fastest IT and smartest AI can’t: make sense of it all.

Sense-making is the process through which organizations understand the world. Organizational understanding is a gestalt that emerges from the collaboration of all parties involved—collectors, analysts, AIs, and the consumers themselves. Forward-thinking intelligence officers have urged the IC to adopt a sense-making model for years, but as we move further into the 21st-century, the risk of not making this transition becomes existential.

NGA Director Robert Cardillo hints at this when he says that analysts today should be “coherence control officers” in an incoherent world. He told me in an interview earlier this year that the IC “has to be different than Google.” While Google can retrieve answers to questions in milliseconds, it doesn’t provide you with context, background, or understanding. In a similar vein, Gregory Treverton concludes that intelligence is ultimately storytelling. Intelligence failures happen when the story doesn’t’ match reality, or if you like, when a new chapter begins but we’re still reading the last paragraph. The analyst’s challenge, then, is to tell the story of reality in a coherent manner that decision-makers understand and internalize.

To stay relevant in the information age, the IC should abandon the product-delivery metaphors of the past and make intelligence a sense-making experience, not a thing that can be packaged and delivered. Intelligence is not the white paper or the slide deck or the overhead satellite image. Intelligence is the experience of understanding something more completely than before, and the analyst’s role is to facilitate that understanding.