by Clint Watts
Jihadis were watching YouTube, though. It was easy to see they were by monitoring the accelerating number of views for videos showcasing Zarqawi’s violent legion in Iraq. The next generation of jihad had already moved on from bin Laden, inspired by a new terror group using a whole new medium. And al-Qaeda in Iraq started its own rebranding in October 2006, changing its name to the Islamic State of Iraq and launching its own media group, al-Furqan. The rebranding and split media operations didn’t represent a fracture, at least not yet. Al-Qaeda’s al-Fajr continued distributing the newly branded Islamic State’s al-Furqan content; it had to. Without jihad’s Iraqi legions and their relentless attacks, al-Qaeda’s central propaganda outfit had little to cheer.
Forums were still the preferred method of initial content distribution from real terrorist groups and the site of high-level discussions on strategy, but YouTube opened a pathway for commoners to accelerate content sharing and discussion. Content continued bleeding over from forums to mainstream social media applications as Facebook and Twitter overtook the internet.
Forum administrators, partly out of self-preservation, warned in 2009 of the security risks of spreading jihad on social media. But their warnings didn’t carry much weight. Jihadi forums weren’t exactly safe and dependable, either. Ansar al-Mujahideen shuttered in August 2010 when Spanish authorities arrested a key administrator.11 Occasionally, many forums would go offline at once, suffering what appeared to be distributed denial-of-service (DDoS) attacks coming from counterterrorism efforts. Every single one of the forums was surely penetrated by intelligence agencies. My colleagues and I joked that there might be more counterterrorists than terrorists on some of the forums.
Even as forums joined jihadi communities together, creating dialogue, content sharing, and engagement among al-Qaeda’s core supporters, radicalization and recruitment of new members remained limited and principally in-person. Social media changed that quickly. Video content showing off al-Qaeda in Iraq’s deadly rampages went viral, and open dialogue on YouTube, Facebook, and Twitter extended the reach of terrorist groups to Muslim diaspora populations worldwide.12 Not only could terrorists upload their destructive operations on social media, but recruits, fanboys, and suicide bombers could increase their notoriety by posting video manifestos or martyrdom tapes. On some forums, contributors urged members to join Facebook, seeking to open up discussions beyond jihad’s elites. The effects of jihadi migration to mainstream social media quickly became clear even in America.
In 2008, five Americans from Alexandria, Virginia, mysteriously disappeared and then surfaced in Pakistan, where they were detained on terrorism charges. YouTube videos inspired the five friends: it was on the social media platform that they met an al-Qaeda operative named Saifullah, who helped guide their travels to Pakistan.13 Just a year later, the Somali terror group al-Shabaab’s rapid rise, fueled by recruitment of foreign fighters in diaspora communities, pointed to Facebook recruitment occurring in Minneapolis, Minnesota.14
Social media allowed rising jihadi groups and their supporters to bypass al-Qaeda’s central leadership. Now any terrorist group, large or small, could spread its propaganda, connect with its loyalists, and guide in new recruits. Al-Qaeda affiliates with lower technical capability and limited internet access could create mainstream social media pages.
After 2001, Western counterterrorism continued to improve each year, making terrorist travel to fronts in Iraq and Afghanistan more difficult by the day. Bin Laden and al-Qaeda’s central leadership suffered unending drone strikes and losses. Zarqawi’s death, in 2006, followed by two years of U.S. special operations raids in Iraq withered the newly branded Islamic State of Iraq dramatically, slowly dimming its star. As some al-Qaeda groups crumbled under relentless strikes, al-Qaeda in the Arabian Peninsula, immersed in war-torn Yemen, ascended in the jihadi hierarchy. AQAP brought attacks on the U.S. homeland through concealed bombs carried across the ocean on commercial aviation or by inspiring homegrown recruits in America to attack at home. The message from al-Qaeda strategists in Yemen shifted: If you can’t join us on the battlefield, then stay where you’re at and conduct jihad at home. U.S. counterterrorism officials’ fear of internet jihad may have been overplayed initially, but AQAP’s new strategy, the advances of social media, and the rise of a new jihadi star, American-Yemeni cleric Anwar al-Awlaki, confirmed their worst fears.
Awlaki’s connections to the 9/11 hijackers and his controversial time as an imam in northern Virginia led to his departure for Yemen in 2004. The Yemeni government arrested him in 2006, in connection with an al-Qaeda kidnapping plot. When he was released in 2007, Awlaki’s English YouTube sermons spread like wildfire. One popular sermon offered “44 Ways to Support Jihad,” encouraging supporters to perpetrate suicide operations in the West, and a separate message called for jihad against America. He operated a blog and MySpace and Facebook pages, engaging directly with jihadi fanboys around the world providing religious guidance and operational direction.15 By 2010, Awlaki’s social media persona had shifted from religious scholar to freedom fighter. Pictures showed him clad in ammunition belts and sporting a rocket-propelled grenade (RPG) launcher. Another American jihadi online propagandist, Samir Khan, ended his popular jihadi blog InshallahShaheed, which he authored from his parents’ basement in North Carolina, and moved to Yemen, where he teamed up with Awlaki in 2010.16 Together they produced and promoted AQAP’s new English online magazine Inspire, providing directions for supporters worldwide to perpetrate violent plots locally.
Inspire traveled through Twitter, Facebook, and jihadi forums at light speed. Awlaki’s virtual inspiration soon connected to attacks far and wide. U.S. Army major Nidal Hasan emailed Awlaki for guidance prior to killing fellow soldiers at his Fort Hood army base. Umar Farouk Abdulmutallab, an online recruit in Nigeria who was attracted to Awlaki’s magnetism, traveled to Yemen, where he later donned an underwear bomb before boarding a Christmas Day plane bound for the United States. The bomb failed, but the implications were dire.
Awlaki radicalized supporters via mass social media dissemination and directed them in private messages. Al-Qaeda had finally synchronized all of its online operations on social media. Radicalize, recruit, operate, finance, train, direct—everything could be done from afar on YouTube, Facebook and Twitter. Content and narratives moved around the world at a frantic pace, enticing sympathetic supporters, enraging diaspora communities, accelerating recruitment, and cloaking terrorist operations. The internet had saved al-Qaeda from demise, and social media seemed to be restoring its lethality. At least it seemed that way at first, but the social media era would soon take terrorism in an unexpected direction.
* * *
Open-source intelligence became a booming business as terrorists moved to social media. Jihadis liked to show off, feeding their egos through Twitter and Facebook posts retweeted and liked by their diaspora communities and fellow fighters. YouTube videos rapidly radicalized some new jihadi prospects, but they also provided a window into terrorist operations for all those seeking to understand and dissect al-Qaeda’s operations, particularly those in Iraq. Much like Hollywood directors, terrorist social media video producers had definitive styles they used and locations from which they like to film. Downloading a jihadi video stream allowed one to understand the size of the terror cell, their accents revealed Arabic dialects denoting their homelands, and stray frames sometimes even showed their hideouts in Iraq. Task organization charts outlining a terrorist group’s chain of command could be created from YouTube videos, and their screen names tied to email addresses.
By 2006 and 2007, cadets enrolled in the terrorism and counterterrorism classes I instructed at West Point gathered online videos and created their own real-world intelligence projects from the barracks. The U.S. government, initially, was slow to react to this new intelligence opportunity. During unclassified training sessions, U.S. government personnel often shouted down academics during presentations: “Hey, you can’t talk about that. That’s
classified information.” During breaks, academic presenters often conducted a hasty Google search and revealed to these analog counterterrorists that their precious secret intelligence sat on the World Wide Web for everyone to see. The advantages the U.S. intelligence community had once enjoyed were slipping away in the open-source world.
A few years before, in 2004, American journalist James Surowiecki had published a book called The Wisdom of Crowds, which described how the internet provided a vehicle for crowds to make smarter decisions than even the smartest person in the crowd, working alone, could make. Collective intelligence mined from the internet through crowdsourcing proved effective in three types of decisions: coordination challenges, where groups work together to determine an optimal solution, such as the best way to get to work or travel overseas; cognition calculations, where people involved in a market compete to provide the right answer, such as guessing the winner of an election; and cooperation networks, where a central system collects information and the crowd then controls behavior and enforces compliance—think Wikipedia as an example.
The internet lowered the barriers for users to contribute their data and opinions in each of these situations and significantly decreased the costs of collecting such information. The advent of the iPhone propelled this phenomenon even further with social media apps rapidly capturing user experiences. Amazon provided millions of product reviews, Yelp’s app created restaurant recommendations for any neighborhood, and Rotten Tomatoes guided people to the better movies available in theaters. As I began to further develop my blog, selectwisdom.com, in 2010, I thought crowdsourcing might be a way to harness counterterrorism analysis in the same way it had helped terrorists march forward online.
I started the blog off slow. I’d write a short post on whatever al-Qaeda affiliate had pulled off an attack that week. I’d profile those groups that I knew better than others. I’d showcase the great research of those who didn’t get enough attention for their work, predominantly those outside of Washington, D.C.’s thought bubble. I hoped to gain a sizable audience to gather a statistical sample of opinions, pushing for at least thirty contributors but often getting a bit more. When challenged on my analysis, I’d post a survey on Twitter, collecting answers from a few dozen of the world’s emerging counterterrorism superstars and occasionally a real terrorist or two, a few sympathizers participating in terrorist forums, or a terrorist fence-sitter debating travel to a far-off battlefield. They’d provide a bit of perspective, too, adding an alternative view from across the oceans. The following week, I’d publish the survey results on my blog and start a new round of dialogue. My readership was light, but slowly building.
I kicked off 2011 with a new terrorism forecast. Expert predictions litter news outlets every New Year, and since I’d been working on a new theory, I placed my bet on January 2, 2011.
“Osama bin Laden will be killed this year!” I offered, with some supporting analysis.
In the previous weeks, I had employed a prediction technique I’d previously taught in government intelligence analysis courses to estimate when al-Qaeda’s leader might be killed. Rumors around Pakistan suggested that bin Laden might be moving about, President Obama was itching to leave Afghanistan, and I checked some actuarial tables to see what percentage of people bin Laden’s age (he was fifty-three) die naturally. I also assumed that bin Laden, if he were to be killed by a drone or a raid, would meet his demise in the warmer months, when terrorists roamed more freely and special operations forces picked up their pace. The bet was a longshot, but I didn’t believe it to be crazy. Americans had hunted bin Laden for ten years; he couldn’t hide forever and still keep control of his network.
The prediction, however, was actually designed as a vehicle for crowdsourcing an important question. What would al-Qaeda and the world of terrorism be if bin Laden were no more? I used my New Year’s bin Laden prediction to provoke the audience to answer my “Post–bin Laden” survey. My attempts at crowdsourcing this survey failed miserably. Rather than yielding great wisdom or important insights from experts, the results instead returned a pattern of answers of no consequence. “Nothing will change” and “It doesn’t matter” became patent answers from the best thinkers in the field, regardless of the question.
My crowds weren’t wise; they seemed a bit lazy and dumb. Even more, on Twitter, they were hypercritical. Reviewers of my analysis couldn’t wait to tell me my prediction was wrong or that my questions were stupid. Crowdsourcing the future wasn’t working as planned.
I assumed from the outset that the crowd I was building met the preconditions for being wise: diverse, independent, and decentralized. Linked primarily on Twitter, positioned in wildly different geographies, sometimes even in conflict zones, I wrongly believed the crowd I polled had diversity in their opinions and independence in their thinking. While physically the crowd was highly decentralized, social media somehow unified their thinking, overtaking any local knowledge or diverse expertise they had previously accumulated. I went back into the research to figure out why this kept happening.
A political scientist at the University of Pennsylvania, Philip Tetlock, had performed the most expansive collective judgment research to date, in his 2005 book Expert Political Judgment. Tetlock surveyed hundreds of experts in political thought over two decades. After thousands of questions, he determined that experts, en masse, were no more successful at predicting future events than a simple coin toss. Tetlock observed two kinds of forecasters. Borrowing from a Greek saying, “The fox knows many things, but the hedgehog knows but one big thing,” he classified those good at predictions as “foxes” and poorer performers as “hedgehogs.” What differentiated the two groups’ success was how they thought, not what they thought. Tetlock’s foxes were self-critical, used no template, and acknowledged their misses. Hedgehogs, by contrast, sought to reduce every problem to a single theory, were not comfortable with complexity, were overconfident in their assessments, and placed their faith in one big idea, pushing aside alternative explanations. I saw a lot of hedgehogs in my online surveys and occasional foxes. I needed to find foxes to get insights, and to do it, I decided to employ social engineering, the same techniques I’d used in West Point prank calls. I needed to trick respondents to reveal their true tendencies as either a fox or a hedgehog. Once I found the foxes in the pack, I’d investigate their responses as the outliers in a sea of predictions.
Daniel Kahneman and Amos Tversky, two Israeli American behavioral psychologists, provided the social engineering tricks I needed to separate the foxes from the hedgehogs. Over many years of research, they identified a series of heuristics—mental rules people use to make decisions—and noted the circumstances where biases emerged that led to incorrect judgments. These two gentlemen had determined long ago the predictive missteps I observed in my polls. Status quo bias, a belief that tomorrow will most likely look like today, ruled my responses. Loss aversion, a tendency to avoid anticipated losses rather than pursue equally likely gains, filled the results of counterterrorism policy questions. Herding, the tendency of large groups of people to behave the same way and pursue groupthink, drove my social media recruits to the same set of answers.
Armed with Tetlock’s insights and Kahneman and Tversky’s heuristics and biases, I changed my approach. The next survey sought not the wisdom of the crowd, but the wisdom of outliers—those gems arising when crowds fail. Instead of asking simple yes-or-no questions, I flooded respondents with as many potential outcomes as I could think of, making it challenging for nonexperts to wade through the responses. I cornered novices and less innovative thinkers by playing to status quo bias. Every question had a “no change” response option, surrounded by responses imitating common thinking stripped from Google searches, newspaper headlines, and cable news pundits. With every question, I offered survey takers a comment box or allowed them to craft an “other” response.
The last week of April 2011, I sent out one large survey designed to ignore the majority opinion and
instead look for the outliers. I used some of the same questions from January, like “What will be the chief consequence of bin Laden’s death?” but this time I added more strategic questions, such as “Over the next 2 years, the largest portion of Persian Gulf donor contributions to extremism will go to . . . ?” After all the questions, I asked the respondents how confident they were in their answers. Finally, I needed to know why people were outliers, so, while the responses were anonymous, I gathered a bit about their education level, foreign travel, professional background, and preferred information sources. In total, more than three hundred people answered the survey. I coded and studied the results, and then it happened.
I woke to an endless string of text messages on my phone on May 2, 2011. U.S. Navy SEALs had killed Osama bin Laden in Abbottabad, Pakistan. My prediction had been correct, and my Twitter feed of only a couple hundred followers suddenly became more active than usual. For a brief Google search period, news of bin Laden’s death brought a world of visitors to my New Year’s prediction. My small blog suddenly had an audience, and I had a new opportunity to gather perspectives from a larger crowd.
I quickly updated the survey from the previous week, rapidly polling people regarding the sudden death of al-Qaeda’s leader. I gathered another 150 or so responses to this survey; some were the same people I’d queried the week before bin Laden’s death. The results of the two surveys, when compared, showed how a real-world shock suddenly shattered experts’ confidence in their assessments.
Analyzing the results from before and after bin Laden’s death gave me a select group of foxes. Hedgehogs again picked “No change” in response to bin Laden’s death and were highly confident in their assessment. All the responses were anonymous, but when I compared the highly confident “no change” responses, they routinely clumped into two hedgehog buckets. Government folks, military, and intelligence types largely shared the same answers to questions repeating what might commonly be heard in the Washington, D.C., beltway. Academics followed a similar pattern, sticking closely to the preferred classroom theories of the Ivy League. On social media, both crowds whined about how long the survey was and how it was a “waste of time.” Matching their attributes with responses, I could see the Pentagon bureaucrats and tired academics in the bunch thumbing their noses at each question. In later survey iterations, I could even see how long respondents spent answering questions. The higher their confidence, the less time they spent answering questions, and, not surprisingly, the more wrong they were.