This article was written as a collaboration between Jack Mulcaire and editor-in-chief Jonah Bennett, and draws on Jack’s experience covering media narratives during the Syrian conflict.
We’ve heard a lot lately about social media filter bubbles. Supposedly, social media is allowing people to self-segregate into echo chambers where they spread misinformation, fail to encounter opposing views, and become radicalized to the point of destroying our democracy. This is generally seen as a problem that especially affects populists. Columbia Journalism Review has declared that the “pro-Trump media sphere” during the 2016 election was characterized by “polarized outlets” peddling “disinformation: the…construction of true or partly true bits of information into a message that is, at its core, misleading.” As the story goes, these media spheres prey on the low education of the populist base and lead them into information traps.
Social media has indeed shifted the balance of epistemic power in favor of defense, that is, the ability of even casually organized groups to maintain epistemic sovereignty for their own worldview against the mainstream, whether that worldview is sound or delusional. From the perspective of the mainstream, which disagrees with these people by definition, this is purely negative; someone is now able to be quite wrong, where previously they would have been set right. Further, a shift towards epistemic defense reduces mainstream information-based political power. We can’t be at all surprised then to see the negative responses from the mainstream. But we also can’t take it too seriously; information content is proportional to surprise.
What we can take more seriously is that group epistemology is hard, and the amateur filter bubbles do indeed fall into groupthink and badly reasoned ideas. The empowerment of defense is often just the empowerment of ignorant people to mislead each other and persist in their misinformed worldviews. Worse, it increases the power of charlatans, demagogues, and hostile actors to spread disinformation to unwitting members of the public.
But this is not supposed to happen to the epistemic elite. They are supposed to do a better job at maintaining higher epistemic standards, and for the most part, they do. As such, both the official journal of the Council on Foreign Relations and the Harvard Kennedy School have stated that the solution to the crisis of social media disinformation and resultant right-wing populism is to rely on journalists and credentialed experts to discover and disseminate the truth. That is, to shift the balance of power back in favor of a central mainstream with high epistemic standards.
Again predictable, but hard to disagree with in principle; not everyone should be maintaining their own worldview without some benevolent aid. A society needs high quality official sources of information, it needs to educate the population in the view produced by these sources, and it needs to protect itself against the trolls, extremists, special interests, and foreign influences who would subvert that media ecology. This is especially important in a democracy, where a breakdown of the systems of popular education will lead to loss of political control.
But there’s a problem: the experts are also susceptible to the epistemic perils of the Internet.
The media cycle moves fast. The cycle of event and reaction that takes place online and on television ensures that everything is forgotten and irrelevant within a few days. We use events for political point-scoring and personal attention-getting, like the Indians used every part of the buffalo, before discarding the carcass and moving on to the next event on the assembly line.
The frantic way in which our culture uses and discards events means that there are no consequences for being wrong. If you did or said something foolish, you can just wait a few days and it will be completely forgotten, especially if it’s something a bit too complicated to be easily encapsulated in a headline. Virtually no one is keeping score. No one’s looking back at your track record to see if your takes and your predictions ended up right or wrong. Even if they do, it usually doesn’t matter. It’s all about what’s happening now, and what’s happening next.
This leads to a lack of accountability for the experts and opinion-shapers who hold functionally important positions in our media ecology. The rush to write on questionable reports and apply favorable narratives as quickly as possible leads to degrading epistemic standards, and error. Experts at the top of the media pyramid, who are supposed to have high standards, fall into the same failure modes of social media epistemology as everyone else.
This is amply demonstrated by Mark Ames’ recent piece on the relationship between Twitter journalists and the infamous Islamic State Twitter propagandist ShamiWitness.
Ames’ piece is a good first step towards a reckoning with the now-forgotten analytical and journalistic mistakes of the Syrian War. ShamiWitness was a key account for following the Syrian conflict on Twitter from 2011 to 2014. He served as a content aggregator, an advocate for the rebel cause, and an important source of information for journalists and analysts. This was problematic, because by 2013 he was openly supporting the Islamic State and helping them recruit foreign fighters.
I’ve long called for a reckoning with the sins and mistakes of the ShamiWitness era of Syria Twitter, and for an examination of the lessons that this forgotten social media history can teach us about how media coverage of conflicts is generated. Ames’ article on the cozy relationship that Western Syria experts until recently had with ShamiWitness is important for this. It is an attempt to take a step back and look at the performance of the opinion-shaping class with the benefit of hindsight.
Ames’ basic thesis is that Western think-tankers and journalists were too friendly to ShamiWitness and that they bear responsibility for boosting his influence as an Islamic State recruiter. Ames gets some things right—the relationship between ShamiWitness and Western opinion-shapers got too close. But Ames’ narrative is also too simplistic. Think-tankers and journalists didn’t merely gravitate to ShamiWitness because (according to Ames) they were bloodthirsty, evil neo-cons, and he was a bloodthirsty, evil Islamic State supporter. Rather, the reliance on ShamiWitness and others like him was a classic case of a social media filter bubble.
Anonymous pro-rebel accounts enabled foreign policy journalists in their self-deception and confirmation bias, and the public’s understanding of the Syrian conflict suffered as a result. The think-tank class proved just as susceptible to the pernicious effects of internet epistemology as Trump-supporting, Facebook-browsing Boomers. The saga of ShamiWitness and his interlocutors is a story of how social media can cause pernicious feedback loops in the media environment.
I got caught up in it, too. I made a Twitter account at the start of the Arab Spring because I wanted to follow the wars that had broken out around the Middle East. I was generally supportive of these uprisings, and hoped that the turmoil could lead to a more peaceful and democratic Arab World. My search for information and commentary on the Syrian conflict, which was just breaking out, led me to ShamiWitness’ Twitter account. I instantly thought of it as a useful resource. At that time, he wasn’t yet “the most prominent Islamic State fanboy on Twitter,” to quote jihadism expert Aymenn Jawad al Tamimi:
His perspective was clearly that of an Islamist but—undoubtedly through prior tracking of social media—he seemed to have a broad knowledge of Syria’s Sunni insurgency with a particular focus on Salafi and jihadi groups.
Twitter is vital to understanding the Syrian conflict. Twitter is where primary source information—photos, videos, public statements—was released to the world. Twitter is where partisans of the warring factions commented on the conflict in real time and where interested observers constructed maps that made sense of the conflict. ShamiWitness was at the center of it all. His account was identified by researchers at King’s College London as the single most important and influential jihadist disseminator account on social media, followed by 65% of the Twitter accounts in a dataset of foreign fighters and their supporters. Network analysis of his account by researchers at the University of Wollogong in Australia demonstrated the breadth of his influence and is worth quoting at length:
[Our research] clearly indicates four distinct populations (or networks) of Twitter users [interacting with ShamiWitness and the content he posts]: (1) international media, (2) regional Arabic media, (3) IS sympathizers, and (4) IS fighters….This network also includes journalists, such as jenanmoussa, brown_moses (citizen journalist), policy analysts, such as charles_lister and hxhassan, and other international entities, which are not directly related to IS…[These groups] used the network service @shamiwitness has provided for a fairly long period of time.
In other words, ShamiWitness sat squarely at the intersection of international journalists and think-tankers with anonymous online supporters of the Syrian rebellion and the Islamic State. In this network, support of the rebellion was the norm. Unfortunately, this inclination to support the rebellion led to some extreme cases of wishful thinking. Even in 2011, headlines like “Syrian opposition: united, peaceful and not Islamist” by Michael Weiss should have been met with skepticism. By the summer of 2012, Eliot Higgins was declaring that the Assad government was on the “brink of collapse.” In 2013, The Economist was declaring that the rebels were winning, with the support of quotes from Charles Lister, and WSJ’s Elizabeth O’Bagy portrayed the conflict as characterized by neatly delineated zones of “moderate” and “extremist” control, with “moderates” comprising 80-85% of the rebellion. The Daily Beast proclaimed that Hezbollah’s entry into the war was not only a sure sign of Syrian President Bashar al-Assad’s imminent fall, but also likely to result in the disbandment of Hezbollah’s armed wing. Even as late as 2015, Lister was publishing articles explaining “Why Assad Is Losing.”
For most of this time, I was right there with these experts. I was reading the same sources they were, seeing the same combat videos, battlefield maps, kill counts, and article headlines they were, distributed across the pro-rebel Twitter network in which ShamiWitness, and other popular, anonymous pro-rebel accounts played such a large role. I believed much the same narrative. But we were missing the forest for the trees.
At every step of the way, the good traits and successes of the rebels were viewed as having the greatest possible significance, while their failures and flaws were minimized. There was a lot to minimize. This was quintessential social media groupthink at work. As early as 2012, Western foreign policy journalists, located outside of Syria and relying mainly on pro-rebel activists for information on events inside the country, had made up their minds that Assad was doomed. However, those actually on the ground in the country were not so certain, and tended to regard the outcome as “impossible to forecast.” By January 2014, the rebellion had descended into vicious infighting between the Islamic State and other jihadist factions, and the entirety of rebel-held Eastern Syria was falling into the hands of the Islamic State. Al-Qaeda-linked factions controlled much of what remained of rebel-held Syria. The intra-rebel civil war and jihadist ascendancy put the final nail in the coffin for any hope of major international support for the rebellion or broad-based popular support for it in Syria. Ugly incidents like the infamous video of Abu Saqqar the heart-eater didn’t help, either.
The PR battle had decisively turned, and the actual battle was turning decisively, too. Militarily, most rebel gains had been met with worse reverses. Rebel forces near Damascus had been encircled, as key neighborhoods made separate truces with the government. Rebels were quickly swept from the Qalamoun region. Homs, the former “capital of the rebellion,” was almost completely taken by government forces, and the Aleppo front was in total stalemate. Iranian and Russian support for Assad was only increasing, and Hezbollah’s forces were having decisive impacts on the battlefield around Damascus. Overwhelming popular support for the rebellion had failed to materialize: the army remained loyal (something Joshua Landis correctly understood in 2012), and large numbers of urban Sunnis continued to support the Assad government, or at least refused to support the rebellion. The commentariat’s projections of imminent rebel victory continually failed to obtain in large part because they failed to account for the support that Assad’s allies and domestic base were willing and able to give.
“We underestimated the price [Iran] and Russia were willing to pay to keep Assad in power—which made the Obama administration’s messages about Assad needing to go even more empty,” John Allen Gay, executive director of the John Quincy Adams Society and a noted expert on Iranian foreign policy, told Palladium in an interview.
The turning point had come. That should have been clear at the time. The Syria Twitter community couldn’t see it because that narrative didn’t come through in the dedicated pro-rebel sources most journalists relied on, ShamiWitness foremost among them. The pro-rebel commentariat continued to get the big picture wrong. A great example of this phenomenon is Lister’s 2013 breakdown on the rebellion’s composition for Business Insider. Lister’s analysis of the numbers and apparently political affiliations of the spectrum of rebel groups was more or less correct. However, his takeaway—that further Western support for the FSA and “moderate Islamists” would hurt both Assad and the irreconcilable jihadists—was still false. As Sam Heller has shown, no part of the “rebel spectrum” was ever going to be an adequate anti-jihadist force, and the jihadists always had decisive advantages in unity and motivation. This was demonstrated by the ease with which al-Nusra and its allies destroyed the Western-backed Syrian Revolutionary Front in 2014 and the U.S.-armed Hazem Movement in 2015. That outcome should have been foreseeable, but the Syria commentariat didn’t want to see it. Lister proposed substantively the same plan—escalating support for chosen opposition factions—in 2016.
So, where does ShamiWitness fit into all this? He, and accounts like him, enabled self-deception. People turned to him for good enough reasons, at first. His account was a legitimately useful source of information. People wanted access to photos and videos of rebel actions, and information about what was going on politically inside the rebellion. ShamiWitness could provide that; as Michael Weiss stated, ShamiWitness “posted useful videos.” I don’t disagree. However, many of these videos were faked or posted without vital context. Smart people could ferret out most of the outright fakes, but falsehood still leaked through. Secondly, he posted in English. Most journalists and experts simply aren’t good enough at Arabic to go deep into local social media and find relevant primary source material, so they rely on dedicated, anonymous aggregators like ShamiWitness to provide them with selected content in English. Thirdly, he was ardently pro-rebel. Lister, Higgins, Weiss, and many of their allies had already made up their mind that the rebels were the good guys and that they were winning.
ShamiWitness’ transformation from generic young Islamist and rebel supporter into rabid ISIS propagandist makes the self-deception of the Syrian Twitter community incredibly clear. As their source transformed before their eyes, in a way that poetically mirrored the changes happening in the opposition as a whole, journalists didn’t adequately question the conception of the conflict they had formed in cooperation with him and other anonymous accounts like him. The red flags mounted, both about ShamiWitness as a person and about the conflict as a whole. Those flags were ignored.
Social media, especially a section of social media that is hyperfocused on analysis of a bloody war, doesn’t lend itself to this kind of necessary reflection.
I spoke with Joshua Landis, head of the Center for Middle East Studies at the University of Oklahoma—one of the few people to be more or less correct in his analysis of the conflict throughout—about the role of social media in the misreading of the Syrian conflict.
“The problem with social media is that it creates bubbles. So much of social media is cheerleading. Outside observers, such as ShamiWitness and many other opposition supporters, created false expectations of a unified ideology [among the rebels],” Landis said. “Not only did revolutionary fervor create a [social media] bubble but it caused putative liberals in the West to root for regime-change so hard that they embraced the jihadists.”
Social media interlocutors played a key role in creating bad analysis.
“I think the main point of the Syria controversy is external observers projected a civil, secular Western democratic façade onto what was really an uprising of conservative Islamist rural Sunnis simply because their interlocutors tended to be from the former camp,” Alex Mello, security analyst at energy consultancy Horizon Client Access, said in an interview with Palladium. “On social media…but in general also, especially on foreign engagement circuit—the optics-friendly Westernized guy in the suit doing the rounds at the Turkish hotel conferences wasn’t really representative of the broader rebellion.”
Weiss is correct that he and others properly distanced themselves from ShamiWitness once it became clear that he had crossed the final line into being an operative of the Islamic State, something Ames does not adequately acknowledge. Pro-rebel sympathies were quite natural. Assad’s forces were indeed responsible for the majority of civilian casualties, and had instituted a vast system of enforced disappearance and torture even early on in the conflict. Assad’s supporters were usually quite open about their intent to commit war crimes. More relevant to what we’re discussing here is the fact that pro-Assad content aggregators (the pro-Assad versions of ShamiWitness, people like Leith Abu Fadel) were often not only engaged in bloodthirsty rhetoric but also obviously mendacious. The same behavior—minimization of their own side’s flaws, maximization of all their successes, endless spin—were in ample evidence on the pro-Assad side. Disinformation from Syrian State TV and pro-Assad Twitter accounts made it easier to ignore the lies of omission committed by the likes of ShamiWitness.
Maybe it shouldn’t be surprising that so many people, myself included, failed at the time to see what with hindsight should have been clear. Some of these mistakes, however, were avoidable. As I indicated earlier, Prof. Joshua Landis came off better than others did; a look at what sets him apart can reveal why his analysis was more successful. Landis had deep familiarity with Syria prior to the onset of the conflict; he had a chance to form impressions about Syria before the war, rather than becoming an instant expert after the conflict had begun. He also had pre-existing personal relationships to his Syrian relatives, giving him access to non-activist Syrians physically located in country. Even as early as 2011, Landis’ access to a “second opinion” in the form of actual Syrian civilians rather than interested partisans aided his critical analysis of the claims and videos circulated on social media by the likes of ShamiWitness, which other outlets swallowed. Finally, Landis was an academic; his role was merely to produce accurate analysis. Most other professional Syria observers, including those discussed here, had either committed themselves to advocating for one side or the other, or worked for organizations that were engaged in advocating for particular policies in Syria.
Others didn’t fare so well. Social media gives us historically unprecedented access to primary source material. However, in Syria that didn’t result in correct predictions about the course of the conflict or in better policy prescriptions for how to handle it. The story of ShamiWitness and those who relied on him as a source is a cautionary tale about how social media allows smart people to construct factually convincing narratives that are nevertheless so rife with confirmation and selection bias that they are worthless for understanding the big picture.
But despite my focus here on the epistemic pitfalls, we can’t assume that the foreign policy commentariat is beset only by problems with epistemic hygiene. It’s impossible to talk about the foreign policy process without mentioning the role domestic and foreign intelligence agencies, U.S. public relations firms, and lobbyists play in muddying the waters on international issues. The information environment is seeded by actors who have various diverging interests.
Syria was no different. Ketchum, a New York-based public relations firm, scored a major coup by securing an opportunity for Vladimir Putin to publish an op-ed about Syria in the New York Times. Another PR firm, Brown Lloyd James, set up an infamous Vogue Magazine profile of Syria’s pretty first lady, Asma al Assad, shortly before the outbreak of the conflict. The Saudi-employed PR firm Qorvis “ran the Twitter account for the Syrian Opposition coalition.” The family of Omran Daqneesh, the famous “Dust Boy” who appeared in an viral photo during the later stages of the battle of Aleppo, were used by both rebel media activists and Syrian State TV. And both sides have used even cruder methods since the start of the war. The point is: when information about the war appeared on social media or traditional media, it was often put there on purpose by highly interested actors.
Sometimes, PR firms and other interested parties to conflicts are not content to control the information environment by flooding the op-ed sections of major newspapers. In the case of the First Gulf War, Kuwait hired the firm Hill & Knowlton to organize the now-infamous Nayirah testimony: the daughter of the Kuwaiti ambassador to the U.S. pretended to be a regular Kuwaiti girl and told the Congressional Human Rights Caucus she had personally seen Iraqis charge into hospitals and rip babies out of incubators, leaving them to die. Amnesty International corroborated her report. Everyone heard the testimony through breathless media coverage. It was a stunning success for Kuwait. President George H. W. Bush cited the testimony as justification for intervention. The only problem, of course, is that it was later discovered she was the daughter of Saud Al-Sabah, a Kuwaiti ambassador. And worse, the testimony of atrocities was fabricated. Some 28 years later, Hill & Knowlton remains a successful PR firm in D.C. This was an extreme example, but it’s illustrative. Wholly fabricated stories have been present throughout the Syrian conflict, although it’s not clear which ones if any were deliberately created by professionals.
Journalists and bloggers fall prey to these schemes because structural issues exerting pressure on the media make journalists more susceptible to manipulation. Trolls who like to dupe the media are usually quite competent, but PR firms, lobbyists, and intelligence agencies subvert the media for a living. They are highly aware of vulnerabilities, and very good at exploiting them.
These vulnerabilities have been magnified by structural issues stemming largely from the declining revenues and quickened pace of the digital environment:
- Salaries and benefits have fallen relative to other careers, driving talent, experience, and expertise away. In practice, this means fewer foreign bureaus, overall reduced staff, and a staff that is much younger and willing to work for subsistence wages at New Media companies because the title of their position is ‘editor’ or ‘journalist’ as opposed to the more accurate descriptor of ‘content farmer’ or ‘content aggregator.’
- The lower talent and frantic pace means fewer in-depth pieces, and more focus on content farming. Sometimes in-depth pieces perform really well, but it’s virtually impossible to maintain that level of quality consistently, especially when some pieces take six months to write. It’s much easier to post low effort content like cat videos and hot takes, so that’s what happens.
- The faster paced news cycle means more brutal deadlines. News doesn’t stop at 5 p.m. It just keeps coming. There are more content sources to draw from than ever. Hitting deadlines harder is necessary to make sure someone doesn’t ruin your scoop by posting it on Twitter. This means less time and effort can go to fact-checking, analysis, and contextualization.
- Expense account budgets for reporting have gone way down. In-depth pieces often require costs: hotels, flights, insurance, etc. Lower expense budgets means fewer original investigative pieces.
- More roles for journalists to fill. The modern journalist often is responsible for writing multiple pieces a day, promoting those pieces on all forms of social media, doing TV/radio hits, attending events and parties, and constantly texting and talking on the phone with sources. This schedule never ends. This again means less focus on the core business of producing high quality information.
All of these factors boost the epistemic problems we see in the modern media, to say nothing of impure motives.
One way to look at the bigger picture here is that journalistic integrity, and an information ecosystem with high epistemic standards, are public goods. The market, which is to say the average viewer of advertisements, cares as much about cat videos and hot takes as they do about careful reporting and analysis, and so cannot support the larger expense required for the latter.
Careful journalism with high standards, which could avoid the failure modes visible in the ShamiWitness case and others, can thus only be done by a charitably or publicly supported institution, or a monopoly or oligopoly, because only then would it have the freedom to ignore market pressures pushing in the wrong direction. So as the market pressures have forced media to become more competitive, the result is low information journalism that delivers the appearance of careful reporting, or even just entertainment, without the substance.
But even a public monopoly journalistic institution, like the CBC in Canada, or other state-run media, has only the freedom to do the right thing, not necessarily the inclination. The hard problem at the core of all this is how to build an institution inclined to produce truthful reporting, and how to incentivize it properly. The quality of our solutions to these kinds of problems, which can’t simply be solved by the market or a quick budget adjustment, is a key measure of the health of our civilization.
In this case, the ShamiWitness era of Syrian civil war reporting, that measure came back negative. We saw even the more official sources getting mired in bad epistemology, and manipulated by propagandists. This isn’t supposed to happen, but it did. We’re supposed to have an epistemically sound system of experts and high-standard journalists at the top of our media ecology. Instead, at least in this case, we have just another echo chamber without the long-term attention span to do careful analysis in proper context, but with pretensions to authority.
With Joshua Landis and others, we see that there is good analysis, and good structure, somewhere in the system. Perhaps that is where we should be looking for solutions to the crisis of public epistemology. But this sound and careful analysis is drowned out and overshadowed by the cacophony of low-information journalism, or technically correct but nonetheless misleading professional analysis.
These kinds of failures of our information systems may be behind the general loss of trust in the mainstream that underlies the demand for filter bubbles, but unplugging from the mainstream because we can’t trust it anymore, facilitated by social media, does not solve the problem. It’s just a symptom of the problem. We still need an intellectual authority with high epistemic standards. This is a crucial piece of infrastructure for a civilization to be able to make good collective decisions.
This, the problem of public epistemology, is one of the key questions for our project of institutional re-engineering: how should sound epistemic authority work, and how can its output be prioritized over that of special-interest propagandists and low-information journalists?