By Lorand Bodo
The UK think-tank Policy Exchange recently published a new report on the struggle against online jihadist extremism, or what its authors call “the New Netwar”. The report argues that we are currently struggling to find appropriate ways to combat online jihadist extremism and therefore losing the war online against the so-called ‘Islamic State’ (ISIS). To support a series of recommendations for tighter regulation of online content, the report presents an analysis of online jihadist activities. The report concludes with the findings of a specially-commissioned opinion poll, which asked respondents to express views about, inter alia, Internet companies’ efforts to clamp down on extremist content.
The report was a collaboration between Policy Exchange and a company called Human Cognition. The report’s first section was written by Human Cognition, with data collected using a commercial tool called BlackLight, which was developed at Human Cognition by one of the report’s authors. The data were compiled over eighteen months — much of the data in Arabic — from 500 to 1000 active groups/channels each week on Telegram, Torrents and other Social Media that were run by ISIS, AQ and other jihadist sympathisers (p.33). The focus of the report is clearly the online activities of ISIS, but this is contextualised, not always as sharply as it might have been, within the wider ecosystem of online jihadist activities.
The report’s findings made a splash in the media, with coverage in the HuffingtonPost, BBC, Independent and the The Guardian. The title of the BBC report, “UK ‘biggest audience’ in Europe for jihadist web content”, made us curious to investigate the nature of this claim and explore any discrepancies between the way it was reported in the media and the form in which it actually appeared in the Policy Exchange report.
The claim and how it was reported
The Policy Exchange report stated that more Internet users in the UK than any other European country had clicked on a particular sample of shortened unique resource location (URL) links to jihadist-related content elsewhere on the Internet.
This claim wasn’t reported altogether precisely by every media outlet. For example, the BBC stated simply that ‘Britain is the fifth-biggest audience in the world for extremist content after Turkey, the US, Saudi Arabia and Iraq’. Similarly, the Independent also reported this finding without the accompanying caveats and context:
‘Its analysis found that jihadi content was accessed more frequently in the UK than anywhere else in Europe, with the country in fifth place globally behind Turkey, the US, Saudi Arabia and Iraq.’
In contrast, the Guardian’s coverage of this claim was more careful to insert caveats, describing the method as merely a ‘rough’ indicator, but still not explaining the potential unreliability of this metric:
‘On one measure, the number of clicks on a shortlink such as goo.gl or bit.ly, which gives a rough view of which country a link was clicked from, shows that between mid-February and early May 2017, Turkey had the most clicks overall, with 10,810. That was followed by the US, with 10,388; Saudi Arabia, 10,239; Iraq, 8,138; the UK, 6,107; and Egypt, 5,410. More than 40% of the clicks on ISIS material were referred through Twitter.’
The Policy Exchange report itself was more careful in how it conveyed this finding:
‘Where are Users Accessing Content? Links to content are sometimes shared using ‘shortlinks’. Shortlinks are provided by services such as goo.gl or bit.ly, that take a long URL and create a shortened version to make sharing easier. One of the services which shortlinks providers offer is aggregated data on the countries from which users click on the link. This provides, with some caveats, a rough view of how many times a link was clicked in each country…[I]n the period between mid-February and early May 2017, Turkey was the location with the most clicks overall, followed by the US, Saudi Arabia and Iraq. The UK was the fifth most frequent source of clicks, and the most frequently identified source of clicks for a European country.’ (p.51)
This is an important difference between the original report and the BBC and Independent coverage: the Policy Exchange report clarifies that this is a specific, time-bounded claim, referring to data collected between mid-February and early May 2017, and that the claim is restricted to the way in which a certain sample of jihadist-related content was consumed by Internet users.
The authors also added an important footnote, caveating some limitations of this analysis: ‘Some users will have adopted methods of obscuring their location, but many will not.’ (p.51) Note the presumption in the way this caveat is framed: try switching ‘some’ and ‘many’ as they appear in the sentence, and you will see what we mean. Clearly, in presenting the caveat this way and choosing to include the claim in the main body of the report, the authors feel it is sufficiently robust to withstand scrutiny, although we feel they could have done more to explain their reasons for this confidence and to persuade readers to share it.
Three further observations ought, however, to be made. First, without knowing what the shortened links were, and which services were used to create and analyse their usage patterns, it is difficult to know whether discrepancies between shortlink-services’ data analytics have affected this analysis. Second, the availability of virtual private networks (VPNs) is so widespread, and jihadist materials — created by dedicated information security channels and groups — so often instruct people to take these precautions to hide their identities and locations when browsing the Internet, that it is a little too emollient to try to reassure readers that whilst ‘some’ may have used these services, ‘many’ will not have done so. After all, this section of the report repeatedly highlights the well-known switch of pro-ISIS activities from Twitter to Telegram, expressly because of the perceived superiority of the latter from a security persective, so the report’s authors themselves implicitly place the consumers of these shortlinks in a category of heightened communications security consistent with a higher than average use of tools like VPNs.
Third, given that a VPN not only suppresses a user’s original location but actively presents that user as being in a completely different location, the use of VPNs could have had a severely negative impact on the reliability of this data. Just how big a pinch of salt to take with this claim therefore depends on your estimate of the proportion of users taking active measures to restrict the accuracy of location-tracking when they consume jihadist-related media online. There appears to be no more rigorous an attempt in the report to control for this than the indication that ‘some users will, many users won’t,’ so we prescribe a fairly large pinch of salt, certainly larger than the BBC.
This is an important report by an institution committed to ‘evidence-led policy;’ it pursues the worthwhile aim to generate public understanding of the threat posed by online jihadist activities, and moves into the political space with a series of reform proposals regarding the system for regulating and removing jihadist content from the Internet. In this context, the reporting of this particular claim in some media fell below the standards of clarity and accuracy that should be expected. It is unfair to hold the report’s authors responsible for how some journalists have described the report’s findings, but it should be noted that the Policy Exchange itself highlights this finding, placing it in the prominent Key Findings at the top of the report. This obviously fits with the thrust of the report’s recommendations for stronger UK regulatory measures to clamp down on jihadist materials online, but the prominence afforded to this claim, and the attendant media spash generated by it, seems out of proportion with its substantive contribution to the report.
One final thought: this report provided quantitative analysis to support the claim that ISIS and other jihadist groups still pose a sizeable and significant online threat to the UK and its allies. Another way of looking at this issue would have been to explore the relationship between quantity of jihadist content online and the impact of this content as a tool for radicalisation or incitement to violence. This is especially important in the context of ISIS’s fading fortunes on the battlefields of Iraq and Syria. Arguably, ISIS’s initial success in taking and holding territory in Iraq and Syria was a part of what made the spread of ISIS’s content online so appealing — at least to those who left Britain and other countries to join its ranks, or remained at home resolved to commit violent acts in the West. Images of successful conquest, triumphant ‘fighters’ and a parallel series of images and reports about ISIS’s ‘governance’ of its new territories, are all likely to have exerted an influential impact on the suggestible individuals who were seduced by these images. This kind of impact was connected to, but distinct from, the wider theological dimensions of much of ISIS and other jihadist propaganda materials that are well covered in the Policy Exchange report. What is distinct here is the qualitative impact of certain kinds of content, which is a separate variable to the size of ISIS’s and other jihadist groups’ online propaganda.
It would have been interesting to read the authors’ views on the possible impact of ISIS’s diminishing territory and battlefield fortunes, not only on the quantity of its online propaganda, which the report covers in depth, but also the potentially variable impact over time of this content on its target audiences. Put differently, would it be so important to reform regulation to accelerate efforts to reduce the total number of pieces of original jihadist content online, if each piece of jihadist content were becoming progressively less influential over time in the incitement of further violence? This is clearly a policy-relevant dimension of the problem: is the impact of jihadists’ online medium undermined by the real-world failures of the messengers?
Lorand Bodo is a Researcher at Ridgeway, working on the use of open source intelligence (OSINT) to support national and international security. His research focuses on online extremism, particularly ISIS’s use of the Internet. You can find him on Twitter @LorandBodo.
This article was originally published on medium.com on 22 September, 2017. Republished here with permission.