**1. Introduction**

In July 2014, the entire world watched in shock and disbelief Abu Bakr al-Baghdadi declare the restoration of the 'Caliphate'. Today, al-Baghdadi is dead, and the Caliphate has been reduced to a (imaginary) shell. However, despite its current status, the 'Islamic State' (IS) merits a deeper exploration, for its impact and repercussions have been profound and far-reaching. IS's violent entrance in the global scene has shuttered much of our understanding of violent non-state actors and the salafi-jihadi movement in particular. For years the common wisdom was that the jihadi movement (and particularly al-Qaeda) had undergone an organizational transformation in the wake of the 'war on terrorism', evolving from a relatively hierarchical group to a deterritorialized network that preferred attacking the West than controlling territory. Yet, in mid-2014, IS demonstrated that jihadi groups do not mind controlling territory, if given the opportunity.

However, IS's impact has been larger than simply invalidating the main premise of the 'new wars' paradigm. As this article shows, the declaration of the Caliphate and the extent of IS's territorial control raise important questions in the study of international politics. Was IS, at the height of its activity in 2014–2016, a terrorist group, a guerrilla movement or a revolutionary state? The answer to this question will help us understand the nature of IS's challenge to the global order. If IS (and possible future copycats) are able to socialize into the international system's norms, as other revolutionary states in the past did, then the international community might consider a more conciliatory response. On the contrary, if IS's ideology and strategic considerations preclude this type of groups to become

'normal' members of the international community, conciliatory gestures will have a limited impact, no matter how generous they are.

The meteoric rise of IS did not rattle only the international system, but it had a profound impact on the salafi-jihadi movement as well. The salafi-jihadi (sub)system, as the article shows, underwent an intense period of competition and alliance-building. This competition primarily involved IS and al-Qaeda, which competed for hegemony in the jihadi movement. Drawing on Watts's (2016) distinction between 'destructive' and 'escalating competition', the article explores the phases of IS–al-Qaeda rivalry. However, contrary to Watts, the article posits that destructive competition, in terms of direct conflict between al-Qaeda and ISIS has been in reality limited, even where they have been the closest, Syria. Instead, as the article shows, the rivalry has been played out in three domains: propaganda, attacks on western targets and a 'war of bay'as', which is the main focus of this article. Prominent ideologues and jihadi groups around the world were called to take a stance on the al-Qaeda–IS conflict. Many felt the need to situate themselves on the new jihadi map, either reiterating their loyalty to al-Qaeda or siding with the new-comer. This resulted in a flurry of competitive pledges of oath coming from all the corners of the jihadi movement. This war of bay'as has ushered a period of turmoil in the jihadi subsystem. The process of system recalibration took the form of international alliance-building and local fragmentation, which a ffected most local jihadi scenes and conflicts.

The article explores the challenge IS posed to the international system and the jihadi subsystem as follows. The first section provides a brief theoretical discussion of fragmentation and alliance-building in violent non-state groups. The theoretical propositions discussed in this section are then used to inform the examination of IS and its competition with al-Qaeda. The exploration of IS's challenge is expounded in three parts. The first of the three examines the nature of IS's challenge to the international order by asking what type of actor IS (or the Caliphate) was at the height of its activity, what was its stance towards the international system, and conversely, what was the stance of the international community towards IS. The second part explores the repercussions of IS's claim to hegemony within the jihadi movement by tracing the roots and mechanisms of al-Qaeda–IS competition and identifying the main conduits of intra-jihadi rivalry at the highest level. The last part opens the focus to include IS-inspired intra-movement competition and fragmentation at the local level, highlighting the motivations behind (splinter) groups' decision to pledge allegiance (bay'a) to Abu Bakr al-Baghdadi.

### **2. Friends or Foes? Alliance-Building and Fragmentation in Non-State Armed Groups**

Militant groups ally and feud like every other political actor. Of the two, intra-group competition and factionalism is easier to pinpoint. Internal fragmentation or 'fratricide' (Hafez 2017; Staniland 2012) occurs when groups that belong to the same movement 'start targeting each other through words and/or actions' (Hamming 2020, p. 14). Competition between groups has been linked to higher lethality, as the e ffort to 'outbid' rival groups leads to more violence (Bakke et al. 2012; Conrad and Greene 2015) and suicide terrorism (Bloom 2005), especially in religious and nationalist movements (Nemeth 2014).

Alliances, on the other hand, are harder to define because not all alliances are equal. The most concise definition of alliances draws from the International Relations (IR) theory and defines alliances as 'formal or informal collaborative arrangements between two or more actors who employ terrorist tactics in the pursuit of joint interests' (Moghadam 2017, p. 7). In reality, cooperation might take various forms: ideological, logistical and operational (Jadoon 2018; Karmon 2005; Phillips 2019). In the context of the IS–al-Qaeda rivalry, ideological cooperation might take the form of a public pledge of allegiance. Logistical cooperation involves sharing material and/or non-material resources, such as weapons, funds, training, consultations and propaganda assistance. Operational cooperation refers to cooperation in the preparation or execution of attacks.

Groups might forge collaborative arrangements in one or more fields. The more cross-sectional and intensive the cooperation the deeper the alliance. Alliances can take four (ideal) forms: mergers, strategic alliances, tactical cooperation, and transactional cooperation (Jadoon 2018; Moghadam 2015). At the lower end of cooperation, groups enter into a marriage of convenience to fulfil short-term

and narrow goals. Limited in nature and scope, this cooperation is driven by practical rather than ideological goals. High-end cooperation, on the other hand, involves cooperation in more than one domain. These alliances, which range from strategic alliances to mergers and full absorption of the weaker group, constitute the highest form of alliance building between militant groups.

The scholarship on alliance-formation between violent non-state groups agrees on one point: alliances are rare (Asal et al. 2016, pp. 6–7; Byman 2014, p. 438; Oots 1989, p. 144). If anything, fragmentation is the norm. The rarity of alliances 'does not make them uninteresting' (Oots 1989, p. 145), though. Alliances have been associated with higher lethality (Asal and Rethemeyer 2008; Oots 1989), the adoption of suicide bombing tactics (Asal et al. 2016) and the prospect of pursuing CBRN weapons (Asal et al. 2012). Given their rarity but high impact, there are two overriding questions: what makes non-state alliances so hard in the first place and what motivations make groups overcome these obstacles and forge alliances against 'common sense'. Answers to these questions can be grouped under two broad categories. The first group includes studies that rest on material consideration and organizational imperatives to explain alliances (and fragmentation) in armed non-state groups. The second category contains studies that, to a varying extent, acknowledge the importance of power considerations but, at the same time, place more emphasis on ideology and shared framing in the formation and deepening of alliances.

The study of alliance formation in non-state conflicts has built on various theoretical traditions. The most daring drew from IR theory and particularly balance of power or balance of threat theories. Karmon (2005), for instance, has demonstrated the relevance of neorealist concepts and balance-of-threat alliance-building theories in the study of terrorist groups' (international) cooperation. In the same vein, Christia (2012) has demonstrated the centrality of power considerations in civil war alliances, in tandem with realist expectations as the author admits. Indeed, as some scholars have noted, non-state groups often face similar conditions of anarchy as states in the international system (Adamson 2005, pp. 43–44; Christia 2012, p. 51; Fearon 1995; Posen 1993). The lack of institutions to mitigate uncertainty and suspicion attenuates commitment issues in non-state alliance formation (Bacon 2014, p. 5; Christia 2012, p. 6). The inability of non-state groups to enforce commitment, due to the nature of their operations, renders power considerations central to their outcome. Although state sponsors can on occasions act as external enforcers, increasing overall trust and predictability (Bapat and Bond 2012), in general, the lower the degree of institutionalization within movements, the higher the possibility of fragmentation and infighting if there are changes in the distribution of power (Bakke et al. 2012).

Organizational issues and concerns have been the focus of numerous studies which, instead of IR theory, draw insights from businesses cooperation or organizational theory (Bacon 2018; Byman 2014; Horowitz and Potter 2014). Along the lines of realist-inspired literature, these studies downplay ideology in favor of resources-related considerations. Militant groups align to procure tangible and intangible resources to increase their overall capabilities. By forming alliances, groups might expect direct assistance in the form of funds and weapon transfers or training opportunities, especially if the ally holds territory. Groups with transnational aspirations might seek alliances to gain access to local resources, people and local terrain's knowledge (Byman 2014, p. 431; Jadoon 2018, p. 31). Intangible resources, on the other hand, might take the form of 'best practices' sharing (Cragin et al. 2007), as groups might benefit from the transfer of new information, technology and tactics (Asal et al. 2016; Horowitz 2010). Some groups might also want to associate themselves with an established group to improve their credibility, legitimacy and prestige or to exploit the opportunity to 're-invent their image' (Bacon 2014, pp. 4, 8). This is most evident in groups with acute organizational learning or adaptation needs that for various reasons cannot undertake self-reform to address their problems (Bacon 2014, p. 16). The notion of weakness as a driver for alliance-building is contested (Phillips 2019). Some studies consider (e ffective) state counterinsurgency and military losses as an important motivation for groups to seek allies (Kalyvas and Kocher 2007). Well-placed allies can o ffer the struggling group territory and networks to weather the storm. Al-Qaeda's reliance on Taliban's safe houses and trusted networks after US intervention in Afghanistan is a case in point (Bacon 2014, p. 8). On the contrary,

Bapat and Bond (2012) claim that weak groups might seek alliances but their inability to withstand counterterrorism pressure turns them into undesirable partners.

In this turmoil of competing arguments, ideology-based studies pose a simple counter-argument: groups ally (or fragment) for ideological reasons. Drawing from scholarship on state alliances (Barnett 1996; Holsti et al. 1973), these studies posit that a shared enemy and shared ideology a ffect partner selection (Asal et al. 2016; Gade et al. 2019). Given that groups operate in an anarchic and unpredictable environment, ideological homophily might improve the levels of trust among groups and institute commonly accepted norms and preferences. Agreement on the basics, such as conflict framing, conception of the ideal polity, and territorial aspirations (Gade et al. 2019), might smoothen the path to alignment.

Walt (1987) has demonstrated that some ideologies are extremely divisive to allow alliance formation. Therefore, ideology is most likely to play an important role in alliance formation when this ideology is unifying, not divisive. Drawing from Walt, Karmon (2005) identifies nationalistic ideologies as the least accommodating to inter-group cooperation and anarcho-communist ideologies as the most prone to forge alliances. Nationalistic ideologies are by nature exclusionary and territorially bounded, characteristics that diminish the attractiveness of alliances. On the opposite side, the main characteristic that encouraged the formation of coalitions between groups that ascribed to anarcho-communism was their fluid and decentralized ideological base (Karmon 2005, p. 42). Religious ideologies, which Karmon touches only briefly, can o ffer similar encompassing and decentralized ideological bases. Indeed, relevant research has shown that groups that share religious a ffiliation are particularly likely to form alliances (Asal et al. 2016; Phillips 2019).

The article sustains that the main di fference of religious ideologies is that the latter can be divisive locally but unifying globally. Religious groups often adopt exclusionary views towards competitive forces in their immediate circle of operations. The inherent centralization imperative in religious ideologies, such as salafi-jihadism, and competition over local resources and recruits fuel intra-field rivalries and fragmentation. At the same time, religious groups often ascribe to a transnational and transcendental understanding of community. Therefore, they aspire to speak in the name of a global community of faithful that recognizes no borders. Often lacking the capacity to reach these global communities, religious groups choose to forge alliances with local same-minded forces. International alliances of this kind are possible because the two parties do not 'compete in the same primary political and resource market'(Bacon 2014) and they have fewer things to worry about.
