Fake News and Data Mining: Mapping Today’s Media for Intel Analysis
Whatever your take on the 2016 U.S. presidential election, it is becoming obvious that one of the biggest losers of the campaign was the mainstream media. After generations of relying upon newspapers and network news to tell us what is going on in the world – including information about a candidate, their experience, views and policies – Americans are now finding their news sources to be increasingly inaccurate, biased, agenda-laced, and as politically motivated as the politicians themselves.
For intelligence analysts who rely on major media outlets as a form of open-source intelligence (OSINT), this level of distrust has significant and permanent implications on how we do our jobs. Alternative sources of open-source information, such as digital news, blog sites, and social media, have proven to be even more contentious for finding accurate information.
OSINT and the Media
Simply put, OSINT is intelligence information derived from open or publicly available sources. This is distinct from intelligence disciplines that require covert or special technical collection, such as signals intelligence (SIGINT) and human intelligence (HUMINT). The type of information collected by OSINT is intended to be seen by a large audience, so the information is usually not protected. Common sources include mass media, newspapers, websites, brochures, magazines, professional literature, and television broadcasts.
To the intelligence analyst, OSINT is relatively easy to collect and doesn’t involve all the handling requirements of classified information. Today, OSINT has grown to comprise most, if not all, of the information many analysts use or have access to. Advances in commercial imagery, drone technology, and the proliferation of new media outlets, bloggers and social media platforms have given us an ever-increasing amount of open-source information. The problem with all this available information is that much of it isn’t trustworthy, and neither are many of the authors or purveyors of it.
We’ve come to the point in American politics where we no longer trust traditional news outlets to provide honest information. A Gallup Poll conducted in September puts America’s trust in mass media at an all-time low of just 32%. This is an 8-point drop from a poll conducted just a year earlier. Instead of expecting our traditional news outlets to inform us in an objective manner, expose corruption and promote the common good, two-thirds of Americans now view them as a tempest of persuasion, opinion and bias. We see these sources openly favoring one candidate over another or one political spectrum over another.
To help intelligence analysts, there are many well-established critical thinking and structured analytic techniques that can help evaluate both an individual article and the source that created or broadcast it. However, what’s missing is a model to help make sense of the media landscape. Before getting into the weeds of evaluating each individual article as a potential intelligence source, analysts need to have an understanding of what an overall map of the media environment looks like.
Similar to using a geographic map of a tactical battlefield, a map of the media environment is crucial for identifying, evaluating and placing OSINT sources within general categories. Mapping is a form of intelligence preparation by visualizing the battlespace, but in this case media sources make up the terrain.
Mapping the Media
I have previously mapped out today’s Intelligence Community using a series of concentric circles to show the different types of intelligence agencies and organizations. The map shows how information technology has allowed intelligence as a business to proliferate into state and local government levels as well as in the commercial marketspace.
A few years ago when working with author Shane Harris on his book, The Watchers: The Rise of America’s Surveillance State, I talked him through the map I created to help him understand how the intelligence community has desegregated and has infiltrated almost every aspect of society. Shane, who is presently the National Security and Intelligence Correspondent for the Wall Street Journal, noted that the media is organized in a similar manner. Much like it did in the intelligence community, information technology has been able to desegregate the media environment and push information out to more diverse areas in the media.
From our discussion, I created another map showing what the media looks like today:
Each ring in the diagram is a potential intelligence source with unique characteristics that analysts need to understand. While each ring and source could be further broken down and profiled exhaustively, these layers give analysts a better understanding of the type of information they are getting. As a general rule, the further you get from the center, the more wild and unrestrained the environment gets.
At the core of the map labeled “Mainstream Media,” there are numerous editors, producers, and other corporate figures. The people in these positions are paid to exercise quality control over content, facts, placement, and other forms of presentation. However, they disappear as you move toward the outer rings. As such, we should expect stories from the center to be fact checked and polished prior to going to print, air or publication. Unfortunately, the loss of trust in our news sources lies squarely on the shoulders of these people and the authors they are supposed to be checking.
As we move farther from the center, vigorous editing becomes less important, affordable, or even desired. If you’re pulling material from the outer ring, such as from social media, you can almost certainly expect to find articles rife with rumor, exaggerations, sensationalism, and even totally bogus stories. This includes fake news sites and false advertising, also known as “clickbait.”
This is not to say that the inner ring of mainstream media, even when journalistic standards are upheld, is the gold standard for stories with true intelligence value. Many of these editorial and quality assurance controls are the primary reason for the bias that analysts encounter. Bias can influence a story through the language that is used, implied messages, the story’s placement on a page, and even the omission of relevant facts or complete stories altogether.
When using media for OSINT, analysts must take into account that bias is everywhere and in every source. While most bias is political in nature, there are also professional, technical and financial biases. A magazine could be professionally biased towards a certain subject matter if it’s the sweet spot of their audience’s interest; podcasts can be considered technically biased because they are almost exclusively audio files; and fake or sensationalized stories are financially biased because they are written as clickbait for advertisers instead of as an accurate story.
Media Bias and Data Mining Tools
Analysts use data mining tools to organize and visualize large to massive data sets, but most of these tools are not designed to identify the different types of bias in the media. If you’re using data mining tools to pull articles, web-pages and broadcasts, these tools will show you just what you asked for, including things you don’t want.
Key words and phrase searches are built on algorithms that can’t identify biases found in the timing of the article, their placement on a print or web page, or the bias of omission. From your search terms, they pull in all the inaccuracies and irrelevant articles equally. If you’re trying to find information about “Clinton in New York,” data mining tools are going to return information about the entire Clinton family; the town of Clinton, NY; the Clinton Street Baking Company in New York City; and the Avalon Clinton apartments on West 52nd Street.
To analysts using data mining tools, understanding the media map is even more important because it provides a structure to limit searches. The parameters used on data pulls can then be explained to a customer or decision-maker of your analysis.
By identifying which ring a particular media source is from, you can quickly pinpoint the desirable and undesirable qualities of that source. For example, if you’re looking to understand the decision-making process of a specific foreign leader or personality, the “Mainstream Media” and “New Media” rings may have your best sources. However, if you’re looking to pull the pulse of the conservation or buzz surrounding that same leader, “Blogs and Social Media” would be much more effective, as they will return less about facts and more about perceptions.
The changes in our media landscape are ongoing, constant and never ending. For intelligence analysts, mapping this ever shifting landscape is a critical first step in navigating it.
About the Author: Erik Kleinsmith is the Associate Vice President for Business Development in Intelligence, National & Homeland Security, and Cyber for American Military University. He is a former Army Intelligence Officer and the former portfolio manager for Intelligence & Security Training at Lockheed Martin. Erik is one of the subjects of a book entitled The Watchers by Shane Harris, which covered his work on a program called Able Danger tracking Al-Qaeda prior to 9/11. He currently resides in Virginia with his wife and two children.