Urban Siege in Paris: A Spectrum of Armed Assault

Urban Siege in Paris: A Spectrum of Armed Assault

John P. Sullivan and Adam Elkus

In 2009, we laid out a conceptual model of terrorist “urban siege” based on the Mumbai attacks.  [i] As noted by several observers, the recent terrorist attack in Paris on the Charlie Hebdo offices may have succeeded due to the unfortunate fact that security officials expected other attack modes (such as airline bombs), not a run and gun in the heart of an urban center. [ii]

While it would be tempting to posit Paris as another bloody data point explained by our conceptual schema, Paris is in fact cause for broadening and expanding it. Unfortunately, the world faces urban security threats that span a spectrum of organization and lethality.  Future threats may look like Mumbai (as has been seen in the Mumbai-like operation against the Westgate shopping mall in Nairobi) or they may resemble Paris. [iii] And there is a large spectrum of threats that occupy the threat envelope in between.

Here we review the timeline of the attacks, analyze continuities and complications with urban siege schemas and relevant incidents, review relevant analysis that could inform a more robust analysis of urban siege, and close with a set of our own questions for researchers and practitioners about what assumptions we need to make in planning for, training to stop, and red-teaming urban siege scenarios.

Urban Siege in Paris

In France, the new year opened with a horrific urban siege. This latest installment of urban guerilla action involved armed assault and massacre, execution of police, a massive manhunt and two hostage-barricade situations. Three days of terror saw the deployment of 88,000 personnel from the French Interior Ministry – ranging from community police to specialized gendarmerie, augmented by a large military contingent. [iv]

On Wednesday 7 January, a car pulled up in front of the offices of Charlie Hebdo, a satirist magazine in Paris’ 11^th Arrondissement.  Two men – the Kouachi brothers, Cherif and Said – got out of the car; they were dressed in black and carried automatic weapons.  After making inquiries the made way to the office and opened fire, killing one.  When they arrived at the office they again opened fire killing the editor and 9 others including a police officer guarding the editor, as well as a police officer Ahmed Merabat, the first responding officer to the scene. [v]

After the attack the self-styled Mujahideen fled the scene. The next day (8 January) they robbed a gas station near Villers-Cotterets. The same day a female police officer, Clarissa Jean-Phillpe was killed in Montrouge, a Paris suburb by Amedy Coulibaly (who has been linked to the Kouachi brothers.  On 9 January, the Kouachi brothers robbed a car in northeast Paris.  A few minutes later the Kouachi brothers took a hostage and the suspects were chased by helicopters in a massive manhunt.  Area schools and businesses were put under lockdown and large numbers of officers from the National Police and Gendarmerie (including GIGN and RAID respectively) were deployed with support from the Army.  [vi] The suspects settled in for a hostage-barricade situation (the second siege) at a printers suite in Dammartin-en-Goele. The suspects boasted they would become martyrs.

A third incident at the Hyper Cacher, a kosher grocery, was conducted by Amedy Couliby to support the Kouachi brothers.  [vii] This siege, which resulted in the death of Couliby and four hostages, was terminated by a police counter assault.  The counter assaults were coordinated, simultaneous actions with the taketown of the Kouachi brothers to limit risk to the hostages at the grocery since Couliby said he would execute them if police assaulted the Kouachis’ location.  

The aftermath of the assaults includes questions about intelligence failure, fear of follow-on attacks conducted by activated sleeper cells, threats to police, and the threat of attacks in The United Kingdom and United States.[viii] Finally, the conflict has both physical and virtual dimensions as seen by a wave of attacks against French websites. Hackers, responding to the French public’s defiance in the face of terror, hit 19,000 French websites with denial-of-service attacks. [ix] A group of pro-Syrian regime hackers briefly commandeered the French newspaper Le Monde’s Twitter account, tweeting a message mocking the post-attack hashtag #JeSuisCharlie. [x]

Analyzing the Paris Urban Siege: Continuity and Complication  

On one end of a spectrum of urban assault lethality and sophistication is the Mumbai attacks. The attackers belonged to a cohesive and organized terrorist organization and received guidance, direction, and real-time information support from an offsite handler. On the lowest end is a garden variety “active shooter” more akin to the Columbine school shooters – no training, no guidance, no resources, no contacts but nonetheless possessing an willingness to kill and die. The organized terrorist group type of attacker is obviously capable of waging urban siege. Operating in small squads, they can challenge police command and control and on-site response through dispersion, firepower, and entrenchment.

As David Kilcullen observed in an application of our work to the Westgate mall attack, the Mumbai attack also was terror by “remote control” – the attackers utilized Skype, cellphones, and satellite phones to connect to an offsite operations team in Pakistan monitoring social media and news reporting concerning the ongoing attack. The Nairobi attack exhibited similar characteristics.[xi]  Six suspects affiliated with al-Shabaab executed a hostage-barricade assault against the mall complex. [xii] Multiple squads executed a coordinated attack and successfully entrenched within the mall complex for four days. 72 people died before Kenyan security forces could retake the mall.

As with Mumbai, the Kenya attackers prepared for an entrenchment scenario, fused various weapons and teams, and thwarted a disorganized and bureaucratically disjointed security force response long enough to exact a gruesome toll. The attack, though novel in its ferocity, sophistication, and toll, was preceded by a drumbeat of urban terrorist attacks in which al-Shabaab demonstrated urban assault capabilities.  The catastrophic impact of poor command and control cannot be overstated. An extensive Guardian report suggested that disputes over police and military command and control delayed response. Not only were attackers able to entrench and kill more victims, but a friendly fire incident also occurred and militants were able to foil a first joint police-army counterattack with sniper fire. [xiii]

Nor has Kenya been the sole instance of a suicide commando assault since Mumbai. In December of last year, the Pakistani Taliban launched a gruesome attack on a school for children of Pakistan army officers. [xiv] In a repeat of previous urban siege patterns, attackers provisioned for a long attack quickly pushed into the school. However, unlike in Beslan there would be no entrenchment and hostage situation. Pakistani forces responded within 15 minutes and killed all of the attackers. However, the security forces were too late to save the 132 children and 10 school staff slaughtered by the terrorists during the initial attack. Pakistani Taliban, Afghan Taliban, and al-Qaeda urban operations continue within urban centers in South Asia, a site C. Christine Fair has noted is one large “urban battlefield.” [xv]

While we do not suggest that older hostage and armed attack scenarios were simple, the operational challenges associated with these types of attacks dwarf the typical single-site, hostage-barricade assumptions seen in terrorist operations such as the 1970s Munich Olympics incident or the spate of aircraft hijackings seen during the wave of terror that preceded the current wave of radical Islamist terror. As J. Paul D. Taillon noted in his study of hijacking and hostages, successful counterterrorist operations involved forward base access, cooperation, and specialized units capable of dislodging attackers. [xvi]  In contrast, modern urban sieges will require first responders to meet attackers head-on, regardless of sophistication and armament.  Such direct police action is necessary to stop the ‘kinetic momentum’ and minimize casualties. [xvii]

For example, in June 2014 a team of heavily armed Pakistani Taliban militants assaulted the Jinnah International Airport in Karachi. [xviii] Attackers disguised themselves as airport personnel and were successfully held off by security personnel and finished off by military reinforcements. They were provisioned for a long siege, but failed to survive long enough to inflict major damage. [xix]  Whatever damage (material, human, and symbolic) they inflicted, it could have been far worse had airport security officers not immediately responded to the incident. 

For police, responding to simple, single site attacks requires a high degree of tactical proficiency.  Larger, more complex, area-wide simultaneous assaults require a high degree of coordination and the employment of operational art.  Urban operational art for the police demands integration of patrol, special operations (tactical response including SWAT, bomb squad, riot/crowd control, media/public information, detectives and investigation, and intelligence, as well as synchronization with the fire service, emergency medical services (EMS), emergency management, civil authorities, and potentially the military.  Such coordination may be needed at multiple locations in a single jurisdiction or among authorities spread across multiple jurisdictions.

Our ability to comment on the Paris attacks is limited and based on details currently known in the open source. However, we can observe several important similarities and distinctions to the urban siege model we have outlined in prior work.

Some aspects of the Paris attacks had at least superficially to other observed urban terrorist attacks. While the main actual attack itself was relatively brief, the attackers themselves hid out in the Paris metropolitan area, lengthening the period of terror and fear. The incident reached a bloody climax when the assailants – seeking martyrdom and desiring a fight to the death – holed up in a small warehouse with a hostage and subsequently died at the hands of French law tactical responders. [xx]  Both gunmen received tactical training related to basic weapons usage, and one gunman may have visited Yemen to receive further instruction and financing. [xxi]

The distribution of the siege is also relevant. The Charlie Hebdo incident must be understood as an integrated whole, with the opening assault against the newspaper offices just one (high-profile) component.  For three days, attackers went on a killing spree, distributing their attacks in time and space around Paris and its environs.  During this time, the French security authorities were forced to deploy an enormous force to find, fix in place, and neutralize the suspects before they could accumulate a larger kill count. The attacks were synchronized to achieve maximum impact, and police faced enormous difficulties handling both situations simultaneously.

The Hyper Cacher hostage taker, for example, demanded that authorities cease their pursuit of the Kouachi brothers. While the police raid that broke the siege at the Hyper Cacher may seem improvised and amateurish to some observers, it was also conducted under extremely unfavorable conditions. [xxii] Both police raids had to be synchronized for hostages to survive.  Moreover, coordinating a massive interagency response is complex and should be considered an operational success for the French security authorities given that difficulties in interagency coordination are an impediment to many operational responses and notably operational response to the Nairobi mall attack in 2013. [xxiii]

While these elements may be familiar to those that respond to, cooand, or analyze urban siege, other elements of the attack were more novel. Analyzing the Paris attacks, Clint Watts argued that the future of jihad was “inspired, networked, and directed:”

The jihadi movement may have finally become what its original luminaries always wanted it to be – and in Paris of all places. The amorphous connections between the Charlie Hebdo attackers, the Kouachi brothers – who attributed their actions to “al Qaeda in Yemen” – and kosher market attacker Amedy Coulibali – who pledged allegiance to the Islamic State in a recently released online video – may reflect exactly what some early jihadi strategists intended: broad based jihad via a loose social movement.  …. Years ago, Bruce Hoffman rightly proposed a spectrum approach to understanding al Qaeda comprising of a core, affiliates, and locals. His framework was appropriate but now needs some updates with the rise of the Islamic State. With two competing poles and a spectrum of adherents littered throughout at least five continents, jihadi plots and their perpetrators might best be examined through the blending of three overlapping categories: ‘directed’, ‘networked’ and ‘inspired’. These three labels should not be seen as discrete categories but instead as phases across a spectrum – some plots and their perpetrators will bleed over these boundaries.[xxiv]

Counterterrorism analysts have often argued over whether the future of jihad lies with centralized, hierarchal (if not completely top-down) groups capable of organized and lethal attacks or small groupings of alienated, mostly self-directed local attackers. [xxv] Watts suggests that this dichotomous understanding is ultimately misleading – it may be possible for an attack to feature such strange incongruities as terrorists belonging to two rival organizations (the Islamic State in Iraq and al-Qaeda) cooperating together.

Indeed, Islamic State supporter Coulibaly (with logistical support from other men that French authorities have detained) operated alongside al-Qaeda-identifying gunmen. [xxvi] All three were part of a known network of French domestic extremists that orbited around a charismatic yet amateur and unofficial religious figure. [xxvii] We leave discussion of what this means for the global terrorist threat landscape to counterterrorism specialists who will be informed by additional data. However, these debates, typologies, and considerations have practical meaning for operational authorities tasked with preparing for and countering urban attacks.

The three-day Paris siege complicates the assumptions of the conceptual schemas we and others have laid out regarding urban siege and urban terrorism. [xxviii]Attackers did not belong to a single group – they were part of a common network that somehow received inspiration and possible direction from two ideologically opposed terrorist organizations. Investigators are still hunting for possible leads, but it is safe to say that the attacks were a “tangled” mess that involve uncertain connections between the attackers, local terrorist connections in Europe and external organizations in the Middle East. [xxix]

The threat of simultaneous attacks, follow-on attacks, and the tangled web of influence this situation involves complicates operational response. Police must assume from now on that attackers might derive logistical support, inspiration, funding, and/or direction from a diverse combination of local, regional, and extra-regional sources. Moreover, they cannot also assume that one large attack by an attacker group is all they must contend with – synchronized attacks may occur designed to augment the execution and impact of one attack mission. Campaigns containing multiple simultaneous (or near-simultaneous) and/or sequential attacks (including attacks or engagements during exfiltration and escape) must be accounted for and demand the development and employment of operational art for urban battle. [xxx]

While much of the urban sieges since Mumbai demonstrate continuity, complications and change suggest the need for new thinking, including full-spectrum policing, operational art, including operations-intelligence integration to support command.

Diagnosing Urban Siege: Towards A Spectrum of Armed Assault

Deriving problem classes of urban siege requires a look at both the organizational dimension of the attack and the actual means of operational preparation, planning, and execution.

As per Watts’ typology of terrorist organization and influence, we believe that the organizational dimension of the attack matters a great deal in creating reasonable assumptions for training, response planning, and wargames/red-teaming efforts.  We summarize his typology below. While we make no claims that Watts’ typology is the only or necessarily the most accurate template for analyzing jihadist organization, we believe it at least illustrates many of the analytical challenges involved.

First, there are obviously the most traditional kind of attack organization. “Directed” attacks, Watts notes, assume a large degree of central organization by an external group and high lethality and capability. These attacks have become seemingly less likely as improved Western law enforcement, intelligence, and military efforts have made it difficult for attacks to be organized from the top-down. However, as Gartenstein-Ross and Leah Farall have noted, one should not count these attacks out. [xxxi] Moreover, hierarchal organization does not necessarily assume a rigid, military-style command and control structure, and Gartenstein-Ross has noted that our understanding of the global jihad remains too fragmented and incomplete to make sweeping judgements about the likelihood of directed attacks.

“Networked” attacks will assume fighters with some degree of training (perhaps derived from overseas conflicts) and some degree of connection to overseas terrorist organizations or communities of terrorist practice. But, in contrast to elaborately planned directed attacks, Watts notes, networked attackers will constitute a “swarm” that brings together operatives, resources, and perpetrators as needed. Key variables in networked attacks include the local strength of foreign fighter networks, availability of weaponry, and the Western security environment that jihadists must contend with. Watts has suggested elsewhere that the chain of foreign fighters and radicals that being funneled to and from Western states to foreign battlefields may be modeled with the collective intelligence optimization technique known as ant colony optimization. [xxxii]

Finally, “inspired” attacks feature “bungled plots and random violence” by “jihadi wannabees.” While directed attacks and networked attacks demand a complex interagency operational response, “inspired” attacks may not typically not fit the urban siege conceptual schema. Competent law enforcement should be able to handle it, as “inspired” but often incompetent jihadists are frequently just as much of a danger to themselves as they are to their targets. However, one cannot rule out that directed or network attacks may spawn copycat inspired attacks, complicating security response, intelligence, and investigation before, during, or after an urban siege scenario. It is possible that future “inspired” cells may develop sophisticated capacity on their own or through interaction with other cells over time (although it is expected that this is difficult to achieve).

Next, we summarize Gartenstein-Ross and Daniel Trombly’s October 2012 report on the use of small arms by terrorists. [xxxiii] Gartenstein-Ross and Trombly note that the use of small arms figures highly in terrorist strategic thought and must be analyzed as a function of a larger jihadist war of attrition. Al-Qaeda documents outline a strategy for a war of attrition rooted in a combination of complex, multi-member operations and smaller attacks. Complex and large-scale missions force the target to expend significant resources to prevent future attacks of that type, while smaller operations create a constant threat stream and foster an atmosphere of fear and paranoia while driving up costs gradually. [xxxiv] A vehicle for this is the use of firearms and armed assault:

For both large-scale and small-scale attacks, firearms figure prominently in al Qaeda’s strategy. A considerable corpus of written works underlies the significant role given to small arms. For years, al Qaeda and other jihadi organizations have published documents on the value of these weapons. In Abd al Aziz al Muqrin’s A Practical Course for Guerrilla War, a book based on writings that first appeared in al Qaeda’s online journal Mu’askar al Battar, multiple chapters describe tactical and operational planning for urban warfare. Techniques covered include assassination, hostage taking, attacking motorcades, assaulting and clearing fixed targets, and setting up ambush positions. Additional volumes cover the acquisition and maintenance of small arms.[xxxv]

Having outlined the strategic aim behind al-Qaeda contemplation of armed assault, Gartenstein-Ross and Trombly create a typology of urban assault types. Assasination attacks involve terrorist targeting of a high-profile individual. Single-shooter attacks aim for a symbolic target or location of importance to the enemy. Two-shooter teams allow terrorists to conduct more sophisticated attacks over extended periods of time. Mass attacks and frontal assaults denote terrorist operations against fixed targets. Finally, complex urban warfare attacks include multi-man teams and hybrids of the aforementioned attacks. Terrorists may also mix hostage taking, robberies, and defensive siege combat with any one of these attack types. [xxxvi]

Both the Watts and Gartenstein-Ross and Trombly typologies address essential aspects of an urban siege scenario. The organizational capacity and style in one of the categories Watts outlines may dictate the nature of the small arms attack drawn from Gartenstein-Ross and Trombly’s study. Moreover, as we have previously suggested, an attack of one Watts type may lead to follow-on and/or concurrent attacks featuring another Watts organizational attack type and multiple possible Gartenstein-Ross and Trombly small arms attack types.

Murky and ill-structured incidents like the Paris incident suggest the need for greater integration between the levels of analysis in both surveys. Both cover core elements of the problem – operational direction and mechanism of attack respectively – but understanding organizational capacity and the causation of attacks may help explain the overlaps between attack execution types that Gartenstein-Ross and Trombly note at the conclusion of their report. “With firearms attackers have great flexibility,” Gartenstein-Ross and Trombly rightly note. Once an attack has begun, they can select new targets and counter law enforcement.” [xxxvii]

This, when coupled with the potential for a more unpredictable attacker set composition, suggests that conceptual integration is of more than just academic or high-level policy relevance. It matters very much for operational preparation for countering armed assaults. In order to train, prepare, red-team, plan, and allocate resources properly before the attack, police and other security agencies need to have scripts, scenarios, and models of how an attack is organized, rehearsed, and executed.

Questions about Future Urban Siege

While we do not propose our own typology, the Paris attack and newer research and analysis by Watts, Garteinstein-Ross and Trombly, and others suggest some pressing questions for both researchers and operational responders to consider when pondering urban siege post-Hebdo. In pondering these questions, we hope that researchers and operational responders can grope towards some conceptual synthesis between the levels of analysis that Watts and Gartenstein-Ross and Trombly cover in their analyses. We list them below:

1. What kind of organizational assumptions should we utilize when building urban terrorism scenarios?

Both the Mumbai and Paris urban sieges had similar results – prolonged mayhem by multiple groups of attackers. However, in one operation (Mumbai) the terrorists belonged to one group distributed into multiple teams. During the Paris attack the terrorists belonged to a loose common network but only loosely coordinated their synchronized operations. This created two different kinds of command problems for the first responders. Of course, the police may not and likely won’t know which type of adversary they are facing during the initial course of an actual attack sequence and must rely upon real-time intelligence and operational reports to develop situational assessments.

During the Mumbai operation first responders struggled to handle the command and control problem of countering a distributed operation. But in the Paris attack state capacity was high and this was not as grave of a challenge. Rather, the primary challenge was locating the perpetrators, connecting disparate incidents, and later during the hostage situations dealing with the new problem of an attacker that synchronized an attack to coincide with a main operation. This demanded synchronized police response, which was achieved.

The command implications most relevant for an urban siege problem will depend greatly on what kind of organizational assumptions we make about the connection between attacks in the urban terror scenario. But this question also pertains very much to the prevention of attacks before they happen.

A Mumbai-like scenario requires extensive preparation, planning, a forward base, and attack vector that might expose planners and operatives to vulnerability. The terrorist “kill chain” in this case may be amenable to detection and penetration. In contrast, a Paris-like scenario does not have to be analyzed by the familiar recourse to “intelligence failure” explanations – it is perfectly possible that an attack like Paris might occur absent the systemic intelligence-sharing and indications and warnings flaws observed after Mumbai. [xxxviii] Here the terrorist “kill chain” may be more obscure and difficult to penetrate.

2. How should we weight maximum casualties and maximum disruption in the assumptions we make about terrorist mission planning?

One issue that Gartenstein-Ross and Trombly implicitly raise is the dichotomy between disruption and casualties as objectives in an urban siege scenario. Certainly, killing a lot of people can induce disruption and disrupting a key site or system can lead to a substantial amount of casualties. But they ought not to be regarded as interchangeable. One can induce a substantial amount of disruption and fear without Mumbai or Peshawar kill counts – the toll of the Paris attack sequence was small compared to those incidents yet it also induced a massive mobilization of French security forces and led to fear, suspicion, and a backlash that may complicate future counter-terrorism efforts. [xxxix]

This raises some core questions about urban siege scenarios from the point of view of the attacker. Is there a tradeoff between casualties and disruption? How many casualties are necessary for disruption? Is one kind of objective easier to achieve than the other in an urban siege? Do attacker objectives change dynamically during the middle of an incident in response to new information? It should be noted that attackers themselves also may not see a distinction between casualties and disruption or heavily consider it in their planning. Enough casualties automatically suggest disruption, and disruption may be a primary objective with casualty count as a side effect.  

All of these questions bear heavily on organizational assumptions and choice of attack tactics and weapons. As Gartenstein-Ross and Trombly note, firearms allow tremendous flexibility both prior to and during the prosecution of an urban siege. Police will be better able to model, red-team, and train for urban siege scenarios if they have a greater idea of how terrorists themselves view success and failure conditions for urban sieges.

3. How should we think about social media information and operational security during urban sieges?

We devote the most space to this issue due to the fact that operational security (OPSEC) in response has become more acute due to changes in the social media landscape since we last wrote on urban siege.

The issue of social media and OPSEC is by no means new. The Mumbai attacks were one of Twitter’s first real-time crises, with both locals and foreigners giving contradictory and confused play-by-play as the event unfolded. However, the increasing saturation of social media platforms and the ubiquity of Twitter and other social media platforms are increasingly bringing uncertainties about social media and OPSEC to the fore. 

In this attack sequence, social media played a significant role for police, the media, the community, and terrorist organizations alike.  Jihadists and their supporters used social media to praise the attacks, and the #JeSuisCharlie meme went viral extremely quickly. [xl] Additionally, Twitter became a tool for tracking terrorists and developing situational awareness (for all actors), and social media became a key operational security concern as the tactics, techniques, and procedures of security forces are now broadcast in real time by both new and conventional media and terrorist can track that presence as seen in the warning for police to keep a low profile on social media minimize the potential for terrorist ambush. [xli]

Both everyday citizens and major news media organizations maintain social media presences. Social media increasingly drives news during crises, and fusing social media information has grown easier over time due to the increasing maturation of third party client applications.  It has become easier for perpetrators of incidents to monitor feeds, as long as they have manpower to spare or are suitably entrenched in a manner that allows them to monitor feeds unimpeded.

While it is important to remember that attackers (if they successfully infiltrate) begin with the advantage of surprise and responders face an uphill challenge in sorting through contradictory information, so do attackers as well.  More research and assessment needs to be done about cognitive and organizational limitations on how attackers receive, process, and utilize social media information during crisis scenarios.

The human factors and emergency response literature is replete with analysis about how the incident commander’s situational awareness challenges, but we know comparatively little about that of the attacker group. [xlii] It is plausible that information fusion and processing difficulties may be negated by external support and planning (like the Mumbai attack’s handlers), but it also may just add yet another information channel to process as an extra burden. It is only by modeling the information processing challenges attackers face (and how technology may help or worsen them) that law enforcement organizations can gain an realistic idea of OPSEC considerations in future crisis scenarios and justify them to external audiences.

Conclusion: Are We Charlie?

The first question routinely asked after every major terrorist attack is “can it happen here?” Until more information is available about the Charlie Hebdo attacks, it is hard if not impossible to even offer informed speculation about the answer. Our own work is based on news reports and others’ analyses and we will eagerly monitor how well they hold up as more detailed information continues to emerge about the attacks.

Our purpose in writing this piece, however, is not to argue about the potentials for urban siege. We know that armed assault and urban siege is likely to remain a dangerous threat in both the developed West and the developing world. [xliii] However, noting that the possibility for urban siege exists is no longer sufficient or useful. What increasingly matters is how the attack will occur, and we hope that our analysis and questions will spur others to move forward in intensely studying and wargaming the variations and permutations of urban siege.

We titled our first piece on urban siege: “Postcard from Mumbai: Modern Urban Siege.” It is our sincere hope that, whatever the tangled aftermath of Paris, we do not see too many more lethal “postcards” of urban siege from any more cities. [xliv]

End Notes