What the Pen Reveals About the Sword: Rhetoric-Based Mapping of Insurgency Factional Structure in Iraq

What the Pen Reveals About the Sword: Rhetoric-Based Mapping of Insurgency Factional Structure in Iraq Michael Gabbay Information Systems Laboratories, Inc. 200 W. Mercer St., Ste 410 Seattle, WA 98119 mgabbay@islinc.com Paper presented at the Annual Meeting of the International Studies Association March 28, 2008 Note: A condensed version of this paper appeared as Mapping the Structure of the Sunni Insurgency in Iraq, CTC Sentinel, 1(4):10 12, March 2008

1 Introduction The Iraq conflict is a complex fusion of a rebellion against foreign occupation and an internal civil war a conflict that seems intent on exploring almost all axes of violence between its participants. Divisions within the Sunni insurgency in Iraq have critically influenced the evolution of the conflict and will no doubt bear critically upon its subsequent course and resolution. Major points of contention between nationalist-leaning insurgents and caliphate-minded, pan-islamic jihadists have been, among others, Sunni participation in elections, the indiscriminate targeting of Shiite civilians, and the nature of the threat posed by United States-backed Sunni militias, known as awakening councils. Understanding the divisions between insurgents at the level of specific insurgent groups is key to devising effective counterinsurgency and conflict resolution strategies. In Iraq, this task is complicated by the proliferation of insurgent groups, most of whom claim an Islamist mantle, and the murky nature of their origins, composition, and leadership. In this paper, we describe a quantitative methodology for constructing diagrams that characterize and clarify insurgency factional structure using insurgent rhetoric as data. These factional map diagrams can shed insight into insurgent factional dynamics involving cooperation, rivalries, decision making, and organizational cohesion. Our results suggest that the coarse-graining of the Sunni insurgency into a nationalist-leaning camp on one side and Al Qaida-inspired jihadists on the other needs to be further resolved to serve as a guide for US counterinsurgency policy, especially in an era where security gains hinge on the cooperation of the Sunni awakening councils. The zeal and apparent dexterity with which contemporary Islamist insurgent and terrorist groups convey their messages over the internet and satellite television channels like Al-Jazirah has usually been cast as part of a public relations struggle with the US and the West more broadly over the hearts and minds of the Islamic masses. In Iraq, it is apparent however that these media also increasingly serve as forums in which insurgent groups compete with each other for the loyalties, not of a more-or-less neutral population-at-large, but rather of those who already support the insurgency, including insurgent fighters themselves. Accordingly, an analysis of insurgent rhetoric can provide a window into the factional structure and dynamics within the insurgency. This window is all the more valuable given that much of the information on insurgent groups possessed by counterinsurgent forces is by necessity classified and even at that level there is likely little knowledge of the leadership deliberations within and between insurgent groups which 1

ultimately underlie their decisions and rhetoric. 2 Factional Mapping Methodology Our notion of factional structure involves the integration of measures of: (i) insurgent group ideological or strategic differences; (ii) cooperative relationships between groups; and (iii) the overall influence of each group. We describe these measures below for the Iraqi context but the methodology is broadly applicable to insurgencies in which multiple autonomous insurgent groups are present, as is true for most modern cases. Our methodology uses a fusion of concepts from spatial models of politics [1, 2] and social network analysis [3, 4]; the former for its mathematical framing of political competition and voting behavior as occurring along a policy or ideology space, such as the familiar liberal-conservative dimension of US domestic politics, and the latter for its mathematical framing of group structure as a network of bilateral ties between actors and the associated analytical metrics for assessing actor influence and roles. The factional map s inclusion of an ideological axis assumes that interactions along this dimension are important. This is in accord with certain network-based models of group opinion change, in particular, Friedkin and Johnsen s social influence network theory [5] and our variant which emphasizes nonlinear dynamics [6]. The factional maps were originally developed as visualization tools for application of the political elite decision making model described in [7] to the leadership of the FARC insurgent group in Colombia [8] and Russian succession decision making under Putin [9]. Our previous work, however, derived from surveys which solicited area analysts to make direct judgments on ideological/policy issues and actor relationships whereas our Iraq study is based directly on text data. 2.1 Data The data used to construct our Iraqi factional maps includes eleven Sunni insurgent groups listed in Table 1 and spans the time from mid-2003, when Islamist insurgent groups started emerging, through early 2007, just prior to a process of alliance formation among the nationalist-leaning groups. In order to look at changes in factional structure, we divide this time span into two periods: August 2003 July 2005 and August 2005 April 2007, which we refer to as the first and second data periods respectively. The data set consists of hundreds of translated insurgent statements from jihadist websites and interviews of insurgent group officials in print and broadcast media 2

Group Symbol Islamic Army in Iraq IAI 1920 Revolution Brigades 1920RB Mujahidin Army MA Rashidin Army RA Al Qaida in Iraq AQI Ansar al-sunnah Group ASG Salah al-din al-ayyubi Brigades JAMI Fatihin Army FA Iraq s Jihadist Leagues IJL Shield of Islam Brigade SIB Just Punishment Brigades JPB Table 1: Insurgent groups included in analysis. Only the top 7 groups are included in the first data period. as provided by the US government s Open Source Center. Insurgents make specific operational claims of attacks and issue broad creed and methodology statements of ideology and doctrine as well as more focused policy communiqués. Figures 1 and 2 show the factional maps for the first and second data periods and we now describe the procedure used to generate these diagrams. 2.2 Targeting Policy For our measure of ideological and strategic differences between groups we consider the target classes US troops, Iraqi security forces, Shiite militias, government officials, civilians, etc. that are claimed by insurgent groups. In particular, we calculate the value of a targeting policy variable which scores each insurgent group by the average legitimacy of the target classes it claims operations against, where the legitimacy of each target class is the acceptability of attacking it within the set of insurgent groups as a whole, at least according to their public statements. The targeting policy is plotted along the horizontal axis of the factional maps where lower targeting policy scores indicate the presence of less acceptable, more controversial targeting claims. The motivation behind this choice is that disagreement over what types of targets are legitimate has often been the primary source of dissension within Islamist insurgencies [10], and, moreover, the pronouncement of whom one has slain is perhaps the most signal form of political rhetoric, not simply cheap talk. To calculate the targeting policy, we first assess the attitude that each group expresses regarding the acceptability of targeting a given class. Ta- 3

Attitude Meaning Value 2 Operational claims against class 1 Targeting class is legitimate but no operational claims 0 Ambivalent or no opinion expressed -1 General or mild condemnation against targeting class -2 Strong or specific condemnation against targeting class Table 2: Attitude coding scheme for rating insurgent group position regarding targeting a given target class. ble 2 summarizes the coding scheme used to assign attitude scores and Table 3 shows the matrix containing the attitudes of each group for each class for the second data period. The attitude coding scheme is as follows: if a group specifically claims attacks on a target class, then the class is assigned an attitude value of +2 for that group; 1 if a group states a target is legitimate in an interview or policy statement but does not make specific operational claims against that class, a +1 attitude value is assigned; a value of 0 is assigned if the group makes no claims and expresses no opinion or ambivalent opinions about a target class; if a group makes a general but weak condemnation of targeting a given class, an attitude of -1 is given; and a -2 attitude value is assigned if a group strongly condemns targeting a given class, by, for instance, condemning another group for targeting that class or taking specific actions to not target it, such as issuing orders not to attack civilian voters during an election. A legitimacy score for each target class can now be calculated as the average of its attitude values for the whole ensemble of insurgent groups. In Table 3, the classes are sorted in descending order of legitimacy with the most controversial targets at bottom. The value of a group s targeting policy variable is then calculated as the average legitimacy of the target classes it claims, i.e., those classes for whom the group s entries are +2 (we only use claimed targets rather than the full matrix of entries because of the relatively large number of zeros for certain classes and the more ambiguous nature of scoring attitudes other than claims). Targeting policy increases from left to right along the table corresponding to a more discriminate or selective use of violence. 1 If a group does not explicitly claim a target class but does claim specific attacks that are readily correlated with news media accounts of attacks against that class, then a +2 is also assigned. This in particular applies to AQI targeting of Shiite and Sunni civilians. 4

Target Legitimacy AQI ASG JPB IAI IJL FA SIB MA 1920 RA JAMI Class US Forces 2 2 2 2 2 2 2 2 2 2 2 2.00 Shiite Militias 2 2 2 2 2 2 2 1 2 1 1 1.73 Iraqi Govt. Forces 2 2 2 2 2 2 2 2 2 0 1 1.73 Police 2 2 2 2 2 2 2 2 1 0 0 1.55 Spies, Agents 2 2 2 2 2 0 2 0 0 0 1 1.18 Kurdish Militias 2 2 2 2 0 2 0 1 0 1 0 1.09 Iraqi Govt. (civ.) 2 2 2 2 2 0 0 0 0 0-1 0.82 Foreign Civilians 2 2 0 1 0 0 0 1 1 0 0 0.64 Oil Pipelines 1 2 0 2 2 0 0 0 0 0 0 0.64 Politicians 2 1 2-1 0 0 0 0-1 0-1 0.18 Sunni Local 2 0 0-2 0 0 0 0 1 0 0 0.06 Leaders Shiite Civilians 2-1 0-1 -1-1 0-1 -1-2 -1-0.64 Sunni Civilians 2-1 0-2 -1-1 0-1 -2-2 -1-0.82 Targeting Policy 0.80 1.26 1.28 1.34 1.38 1.62 1.64 1.76 1.82 2.00 2.00 Table 3: Matrix of insurgent group attitudes regarding target classes for the Aug. 2005 Apr. 2007 data period. The legitimacy for each target class is listed in descending order. Targeting policy for each group is listed in ascending order. 5

2.3 Cooperative Ties For the cooperative relationship measure, we look at the number of joint communiqués and claims of joint operations between groups; the strength of the cooperative relationship in the factional maps is indicated by the thickness of the links between groups. A joint communiqué is a statement signed by two or more groups and indicates the presence of communication and some level of agreement among the leadership of the groups issuing it. Furthermore, it demonstrates a willingness of the groups involved to be publicly associated with each other. 2 Joint operations are typically only claimed by one of the participants. Such operations indicate local coordination among insurgent groups, although the public disclosure of such cooperation presumably reflects a policy decision by the claimant group at least at the level of its media officials. 2.4 Prominence For assessing the overall influence of a given group within the insurgency, we look at its prominence within the rhetoric of the other insurgent groups (not in the news media where AQI would dwarf all others), under the assumption that more influential groups will be referred to more frequently by their insurgent brethren. The prominence of a group is calculated as the number of times it is referred to by other groups in the form of: (i) joint communiqués; (ii) joint operations; and (iii) simple mentions unrelated to (i) and (ii). 3 The prominence is normalized so that the average prominence of the groups is set to a value of one. Our prominence measure does not admit a simple interpretation as it conflates notions of military power, persuasive appeal, and notoriety. But, then again, neither does the notion of influence itself; two groups, an outgoing one and an insular one, may possess comparable numbers of fighters but the outgoing one that participates in joint efforts with other groups can be rightly said to be more of a leader within the insurgency than the insular one. We believe that the prominence measure is preferable to simple comparison of the volume of operational claims issued by each group as these claims are often highly inflated and, even if accurate, would only gauge a group s military power and not capture the less tangible 2 Occasionally, joint communiqués are later denied by a supposed signee, in which case, that signee is omitted for our purposes. 3 Multiple mentions of a group in a single communiqué or interview are only counted as one reference. Also, we do not restrict references to a group to only one name. For instance, references to AQI also include Zarqawi and his pre-aqi group Monotheism and Jihad (first period) and the Islamic State of Iraq (second period). 6

persuasive elements of influence and leadership. 3 Factional Structure We highlight some observations about factional structure that arise from the factional maps, focusing on the second data period shown in Fig. 2, because of its recency and the greater amount and quality of data available. Looking at the targeting policy spectrum, the groups are more evenly spread out along it than one would expect from a simple binary division into nationalist-leaning and jihadist salafist wings. AQI appears on the extreme left and the Rashidin Army and the Salah al-din al-ayyubi Brigades (JAMI) appear on the far right. In terms of prominence, AQI is seen to be very significant but nowhere near the dominant group one would expect based on its overwhelming presence in the news coverage. Its share of the total number of inter-group references is 13%, roughly in line with Summer 2007 US military estimates that AQI was responsible for about 18% of attacks attributable to Sunni insurgent groups [11]. It is the Islamic Army in Iraq that is seen to be the most prominent (it is commonly believed to be the largest insurgent group) followed by the 1920 Revolution Brigades the influence of the latter, however, has been substantially reduced since it fissioned in March 2007, with the breakaway faction taking the name, Hamas of Iraq. On the low end of the scale, the Shield of Islam Brigade is seen to have zero prominence having never been mentioned by other groups. This shows the value of our prominence measure in sorting out the influence of insurgent groups because the little-known SIB s 9% share of the total attacks claimed by insurgents in a one month survey of insurgent online statements [12] is signficantly higher than more prominent and established groups like 1920RB, JAMI, and RA, and, in fact, SIB has been only heard from sporadically since. Observing the network of links between groups, it is clear that the joint operations network displays more symmetric cross-cutting between groups than does the network of joint communiqués. This likely indicates greater operational cooperation between field units on a local level as compared with the more selective cooperation between leadership elements required for issuing policy statements. The isolation of the jihadist salafist groups, AQI and ASG, on this leadership level is particularly striking, and perhaps AQI s insularity bears responsibility for the fact that its prominence is lower than the estimate of its military power stated above. In contrast, the groups that are commonly considered to be more nationalist in orientation IAI, MA, 7

2 1.5 (a) joint communiques: Aug2003 Jul2005 IAI ASG MA 2 1.5 (b) joint operations: Aug2003 Jul2005 ASG IAI MA prominence 1 1920RB prominence 1 1920RB 0.5 AQI 0.5 AQI JAMI RA 0 0.8 1 1.2 1.4 1.6 1.8 2 targeting policy JAMI RA 0 0.8 1 1.2 1.4 1.6 1.8 2 targeting policy Figure 1: Factional map diagrams for the period Aug. 2003 Jul. 2005. Thickness of links between groups is proportional to the number of: (a) joint communiqués and (b) joint operations. 2.5 2 (a) joint communiques: Aug2005 Apr2007 IAI 1920RB 2.5 2 (b) joint operations: Aug2005 Apr2007 IAI 1920RB prominence 1.5 1 AQI ASG MA JAMI prominence 1.5 1 AQI ASG MA JAMI RA 0.5 RA FA IJL JPB 0 SIB 0.5 1 1.5 2 targeting policy 0.5 FA IJL JPB SIB 0 0.5 1 1.5 2 targeting policy Figure 2: Factional map diagrams for the period Aug. 2005 Apr. 2007: (a) joint communiqués and (b) joint operations. 8

1920RB, JAMI, RA are all inter-connected, forming a clique in social network terms. And given the substantial spread of these groups in targeting policy, the factional map suggests that it is on the level of ties among leaders, rather than ideological or policy congruence, that the lumping of these groups into a single nationalist-leaning camp is most on the mark. The lack of publicly declared leadership links can be an important indicator of the relations between groups, especially if they operate in the same geographical areas, and also of possible pathways for defection of fighters between groups. While there have been a number of claims of fighters switching allegiances between the nationalist and jihadist salafist camps and between AQI and ASG within the latter [13, 14, 15], to our knowledge, there have been no claims of such defections within the nationalist camp, where the leadership links are strong and groups likely seek to preserve good relations [16]. It seems reasonable to assume that the presence of operational links between groups that do not have declared leadership ties would further heighten the possibility of defections between those groups. A strong indication that the targeting policy dimension is indicative of broader ideological divisions is the fact that three major alliances of insurgent groups have nucleated around the most prominent group in each region of the targeting policy spectrum: AQI s Islamic State of Iraq on the left; in the middle, the Islamic Army s Jihad and Reform Front which includes the Mujahidin Army, Fatihin Army, and a splinter group of ASG, the Ansar al-sunnah Shariah Commission; and, on the right, the Jihad and Change Front which consists of the 1920 Revolution Brigades and the Rashidin Army as well as some smaller groups. These latter two alliances consist of groups that are relatively nearby in targeting policy as does another alliance of JAMI with Hamas of Iraq. A broad coalition called the Political Council for Iraqi Resistance (PCIR) has appeared recently whose members are the Jihad and Reform Front, and the JAMI/Hamas alliance. But its later appearance suggests that it is less cohesive than its component alliances. The PCIR also appears to be a bloc for negotiating with the US, a policy component that is not captured by the targeting policy variable. Although proximity in targeting policy does appear to be an important factor for alliance formation, it is not the sole determinant as network ties and other components of ideology matter as well. For instance, the Mujahidin Army has had a strong relationship with the IAI throughout the insurgency and shares a common salafist identity; factors which seem to have induced it to ally with the IAI rather than with the 1920 Revolution Brigades to whom it is closer in targeting policy. These results suggest that analytical metrics and predictive models for alliance formation and cooperation more generally 9

are feasible. Evolution of the factional structure can be observed by comparison of the factional maps of the first and second periods. Changes are seen in the three components of the maps ideology, network structure, and prominence. The targeting policy reflects the rift that has opened up between the nationalist-leaning and jihadist salafist groups. There is a greater spread in targeting policy in the second period due primarily to the discord over AQI s targeting of Shiite and then Sunni civilians; the ordering along the targeting policy dimension of the most prominent groups AQI, ASG, IAI, MA, 1920RB, JAMI is the same in Figs. 1 and 2 but the overall separation is much larger for the second period. The rift is also paralleled by a drop in the network ties between the nationalist and jihadist salafist wings, particularly at the leadership level. In the first period, the cooperative links, as gauged by joint communiqués, between the nationalist-leaning IAI and MA with the jihadist salafist ASG are relatively strong whereas they are absent in the second period. This, however, cannot be attributed to disputes over targeting since ASG and IAI are still very close in targeting policy in the second period. Given the sparser data coverage and the less mature use of the media, the prominence measure for the first period is more subject to uncertainty than the second. However, the ASG s steep drop relative to its flanking neighbors, AQI and IAI, is particularly noteworthy. This certainly reflects its isolation as noted above, but it is also likely that ASG s relative military strength fell as well, given claims by AQI that fighters from ASG joined the Islamic State of Iraq and the May 2007 defection of the ASG Shariah Commission to the IAI s Jihad and Reform Front. 4 We speculate that this can be largely attributed to ASG being located in an ideologically unstable position relative to AQI and IAI, losing market share of insurgent fighters to them after the onset of the sectarian civil war. ASG s identity rhetoric, with its aspirations for a restored caliphate, mocking of nationalism, and venomous religious derogation of Shiites, is very close to that of AQI, but its targeting policy is closer to IAI, even expressing disapproval of targeting Shiite civilians as observed in Table 3. As Sunni-Shiite violence intensified, hardcore jihadist salafists were attracted to AQI having abundantly proven that it could walk the walk in line with its extremist rhetoric by pursuing an extreme targeting policy and by its declaration of the Islamic State of 4 A force consisting of 2,000 men according to the IAI but only 2 by the ASG s account [15, 17]. The drain of ASG fighters was implicitly acknowledged by a statement of its emir in which he counsels mujahidin to reconsider shifting between groups [18]. 10

Iraq. On the other hand, those insurgents with more of a nationalistic bent would have been attracted to the IAI s expressed belief in an intact Iraq, as exemplified by the ASG Shariah Commission s defection. That the defections from the ASG had an ideological component, at least on the nationalist side, is strongly suggested by the ASG s eventual reversion to the name of its original jihadist salafist nucleus the Al Qaida (central)-linked Ansar al-islam having originally changed it to Ansar al-sunnah as part of its expansion early in the insurgency [19]. We now make some observations about the Islamic Army in Iraq that arise from the factional maps. It is striking that the IAI appears as the most prominent group and also at the center of the targeting policy spectrum. This is very suggestive of a spatial politics paradigm in which positioning oneself at the location of the median voter is advantageous [1], but where the voters in our case happen to be insurgency s pool of fighters and active supporters. The asymmetry in the structure of the IAI s joint communiqués network shown in Fig. 2(a) is also striking as one might have expected a more symmetric structure along the lines of its joint operations network in Fig. 2(b). Consideration of the structure of the leadership network alone would place IAI solidly in the nationalist-leaning camp but the targeting policy shows that the IAI is situated on the hardline edge of the nationalists, a fact which is bolstered by its joint operations network, where it alone among the most prominent nationalists has conducted joint operations with both ASG and AQI. The fact that what may be the most powerful nationalist group is also its most extreme could have serious implications for the cohesion and demands of the PCIR in any negotiations with the US. 4 Implications for Group Composition and Goals The spread of nationalist-oriented groups along the targeting policy dimension is puzzling given that their stated high-level political goals are essentially identical: a territorially intact, non-federal Iraq with a basis in Islamic law and a place for all Iraqi sects and ethnic groups. But in terms of the targeting policy spectrum, the Islamic Army in Iraq is roughly equidistant between AQI and the Rashidin Army. What then is the reason for this incongruity between the divergence of targeting claims and the seeming convergence of political goals among the nationalist-leaning groups? We believe that this narrowly posed puzzle is actually related to a larger and more manifestly important question: why have not the nationalist-leaning groups undergone greater consolidation given their similar declared political 11

goals? We argue that this mismatch between targeting policy and professed political goals reflects fundamental divergences in ideologies and commitments to unstated political goals on what is nominally the nationalist side of the insurgent spectrum, which, in turn, has inhibited unification efforts among the nationalists. In our spatial politics framework, where insurgent groups compete for the votes of insurgents and other active insurgency supporters, the IAI s location at the center of the targeting policy space implies that it draws significant support from both jihadist salafists on the left and nationalists on the right an inference about its factional composition that has crucial implications. In order to satisfy both constituencies, the IAI, for reasons of organizational cohesion and survival, must jointly pursue the goals that each constituency holds most dear; for the salafists, that is Sunni shariah rule and for the nationalists, it is an intact Iraq. Accordingly, it is our contention that the Islamic Army in Iraq as well as its allies in the middle region of the targeting policy spectrum are deeply committed to re-establishing Sunni central rule in Iraq a revanche, however, that is unmistakeably religious in nature and for which heavy sectarian bloodletting is an acceptable cost. In contrast, groups on the right side of the spectrum, like the 1920 Revolution Brigades and the Rashidin Army appear more committed to preserving the integrity of Iraq s Arab nature, within the context of an expulsion of US forces. Although they no doubt harbor some ambitions of seizing central power themselves, their strategy for doing so likely stops short of re-igniting a Sunni-Shiite civil war and these groups would be more disposed to settle for a lesser goal of Sunni parity with Shiites, in which Sunnis are given what they perceive as a fair share of power in an Iraq freed of US occupation. These conclusions may have a crucial bearing on the Sunni awakening council militias currently cooperating with the US given that they are reported to have a substantial number of former insurgents in their ranks. The differences between the nationalist groups in targeting policy are also paralleled in identity rhetoric. The IAI not only claims operations against a wider array of targets but also applies religiously derogatory terms Crusader US soldiers, the pagan Iraqi National Guard, and apostate Shiite militias to its targets, a practice much less frequently employed by groups such as 1920RB, JAMI, and RA. The IAI and the MA display a level of hostility towards Shiites in their identity rhetoric that does not rest easily upon their professed desires for a unified Iraq where all sects are welcome. Even their rhetoric about the supposed harmonious nature of sectarian relations in pre-invasion Iraq glowingly evokes a land ruled by Sunni caliphs or kings, where other sects are beneficently tolerated yet distinctly subor- 12

dinate [20, 21]. Perhaps most ominously, the Islamic Army in Iraq consistently refers to removing, not just the US occupation, but a more dangerous Iranian occupation [22, 23, 20, 24]; such a call, in all practicality, would almost surely amount to a bloody Sunni-Shiite civil war. Tellingly, officials associated with the 1920 Revolution Brigades and the Rashidin Army have dismissed the notion that the Iranian occupation is more dangerous than the US one [25, 16, 26]. Of course, the IAI cannot explicitly state its goal of seizing power, but the public presentation of a watered-down agenda, with real goals concealed, as well as the instrumental formation of broad but loose coalitions, such as the PCIR, is classic insurgent strategy [27]. 5 Conclusion The factional mapping methodology we have developed takes a rich data source insurgent rhetoric and uses it to generate a compact quantitative and visual representation of insurgency factional structure. Given the proliferation of insurgent groups in Iraq, the methodology provides a very useful way of clarifying which groups are important, where they stand in relation to each other on an ideological and strategic level, and their cooperative relationships on a political and operational level. The use of targeting policy as an ideology indicator provides finer resolution of fundamental differences between insurgent groups than simply looking at their high level political goals. Consideration of declared political goals alone leads to over-estimating the level of ideological congruence between the nationalist-leaning groups, a misimpression that would be compounded by the apparently strong network of relationships between the leaderships of the major nationalist groups. The structure of alliances that has formed within the insurgency gives credence to the use of the targeting policy spectrum as a variable which encodes meaningful differences between insurgent groups. Since the targeting policy is based on public rhetoric, it sheds light on the constituencies that insurgent groups rely on and compete over, and so, can also be used to assess the factional composition within individual insurgent groups. The Islamic Army in Iraq s position in the center of the targeting policy space suggests that it has achieved its power by appealing to both jihadist salafists and nationalists, which, in turn, implies that, for organizational cohesion, it needs to pursue an agenda that satisfies both constituencies installing Sunni religious-based rule in an intact Iraq. We also showed how the factional maps can capture the temporal evolution of insurgency factional structure in a way that generates useful insight into the dynamics that shape insurgent 13

decision making. References [1] Melvin J. Hinich and Michael C. Munger. Analytical Politics. Cambridge University Press, Cambridge, UK, 1997. [2] Keith T. Poole. Spatial Models of Parliamentary Voting. Cambridge University Press, Cambridge, UK, 2005. [3] Stanley Wasserman and Katherine Faust. Social Network Analysis: Methods and Applications. Cambridge University Press, Cambridge, UK, 1994. [4] M.E.J. Newman. The structure and function of complex networks. SIAM Review, 45(2):167 256, 2003. [5] Noah E. Friedkin and Eugene C. Johnsen. Social influence networks and opinion change. Advances in Group Processes, 16:1 29, 1999. [6] Michael Gabbay. The effects of nonlinear interactions and network structure in small group opinion dynamics. Physica A, 378:118 126, 2007. [7] Michael Gabbay. A dynamical systems model of small group decision making. In R. Avenhaus and I.W. Zartman, editors, Diplomacy Games: Formal Models and International Negotiations. Springer, 2007. [8] Michael Gabbay. Application of a formal model of group decision making to the FARC Secretariat. Technical report, Information Systems Laboratories, June 2006. [9] Michael Gabbay. Application of a social network model of elite decision making. Paper presented at the annual meeting of the International Studies Association, Chicago, IL, Feb. 2007. [10] Mohammed M. Hafez. Why Muslims Rebel. Lynne Rienner Publishers, Boulder, CO, 2003. [11] Anthony H. Cordesman. Iraq s Sunni insurgents: Looking beyond Al Qa ida. Technical report, Center for Strategic and International Studies, July 2007. 14

[12] Daniel Kimmage and Kathleen Ridolfo. Iraqi insurgent media: The war of ideas and images. Technical report, Radio Free Europe/Radio Liberty, 2007. [13] OSC. Amir of Islamic State of Iraq offers US safe withdrawal, December 22, 2006. [14] OSC. Al-Furqan Media responds to accusations of false claims by Al- Mujahidin Army, September 04, 2007. [15] OSC. Islamic Army in Iraq spokesman denounces US-Iran alliance, views jihadist rifts, December 09, 2007. [16] OSC. Jihadist forum posts interview with Al-Rashidin Army amir, August 08, 2007. [17] OSC. Ansar al-sunnah Group denies joining Jihad and Reformation Front, May 03, 2007. [18] OSC. Amir of Ansar al-sunnah Group addresses Muslims, mujahidin, enemies, October 14, 2007. [19] OSC. Ansar al-sunnah Group goes back to former name Ansar al-islam, December 7, 2007. [20] OSC. Iraqi Islamic Army spokesman views Iraq events, al-anbar, terms for talks with US, September 16, 2007. [21] OSC. Jihad and Reform Front spokesman discusses foundation, political plan, September 25, 2007. [22] OSC. Iraqi Islamic Army spokesman on resistance operations, talks with Americans, April 11, 2007. [23] OSC. Jihad and Reformation Front announces death of field commander 4 Sep, September 05, 2007. [24] OSC. Islamic Army in Iraq amir urges jihadist groups to cooperate against enemies, October 16, 2007. [25] OSC. Jazirah TV s talk show discusses expected US strike against Iran, September 23, 2007. [26] OSC. Iraqi AMS spokesman explains stand on other Islamic parties, Awakening councils, November 20, 2007. 15

[27] David Galula. Counterinsurgency Warfare: Theory and Practice. Praeger Security International, Westport, CT, 1964. 16