Threads of Peace: Leadership and Conflict Resolution in Nested Negotiation Networks Open Access

Which variables have influenced the resolution of protracted, asymmetric, and ethnonational (PAE) conflicts in Latin America, Africa, Asia, Europe, and the Middle East between 1960 and 2021? How did specific bargaining and leadership strategies contribute to the peaceful settlement of such armed conflicts?

To answer these questions, this study utilizes negotiation system analysis to compare eleven historical cases to analyze how power, leadership, and relationships lead to the settlement of PAE conflicts. A negotiation system is “a network of actors connected through relationships that are used to influence agreements” (Bell and Wolf 2023: 86). Negotiation system analysis is a methodology used to map such systems, which involve many actors and types of actors, overlapping conflicts and coalitions, and links between local, regional, and global politics.

The specific approach is what George and Bennett call “heuristic” case study, in which induction is used to identify new variables, hypotheses, and causal paths (George and Bennett 2005). It employs a diverse set of cases in order to include numerous variables and processes. The objective is to describe a causal model that explains how power, leadership, and relationships can lead to the negotiated settlement of PAE conflicts.

This study employs a modified version of a methodology that has been used to analyze complex conflict systems in Central Asia and the Middle East (Bell 2015; Babbitt et al. 2017) and that has guided several Negotiation Task Force research projects.¹ The methodology views the relationships between parties as central to the behavior of a conflict system (Bell and Wolf 2023).

The applicability of the methodology to the cases rests on the “nested negotiation network” phenomenon. Three global trends created the “nested network” phenomenon, which the methodology takes as its core object of study. These trends are: 1) stakeholders with influence over decision-making have multiplied, 2) distinctions between global, regional, and local environments are breaking down, and 3) political leaders have found that their conflict management tool kits are unsuited to the conflicts they face (Bell and Mandell 2018).

Negotiation system analysis provides insights into relationships and links that cannot be inferred if individual actors are observed as isolated entities (Bell and Mandell 2018). For instance, an analysis of the relevant actors’ interconnectedness may reveal their degrees of influence over negotiated agreements and their capacity to leverage their positionality in the network into power and authority (Bell and Wolf 2023). In turn, “this interconnectedness has important implications for the effects of individual and collective action, and for the acquisition of skills needed to anticipate and manage these kinds of effects” (Bell and Mandell 2018: 42). An in-depth analysis of these networks can empower mediators with crucial information regarding the conflict’s actors and allow the conflicting parties to develop more effective negotiation tools and approaches, which may unveil otherwise unforeseen avenues of conflict resolution or management (Bell and Wolf 2023).

The framework uses the following definitions of terms:

Key Stakeholder: An actor that exerted meaningful influence over the peace process; provided financial, social, or political support to the negotiating stakeholders; or was a signatory party to the final peace accord.

Actors: Individuals or state/non-state organizations in the network of relationships; the most significant entities involved in the peace process (Bell and Nehrbass 2022).

Key Interests: The actors’ most significant desires, demands, and objectives that directly or indirectly relate to the conflict (Bell 2015).

Bridging or Broker Positions: An actor’s ability, in a leadership capacity, to link multiple nodes (instances of alignments between two parties) (Bell and Nehrbass 2022) or coalitions that would otherwise have remained unconnected.

Closeness Centrality: Measures how close a node/actor is to all other nodes in the network; showcases the access efficiency of each node/actor to the other actors in the network (Golbeck 2013).

Betweenness Centrality: Displays how important each node/actor is in providing a “bridge” of information between different parts of the network; highlights the nodes whose removal would cause the network to fall apart (Hansen et al. 2011).

Eigenvector Centrality: Measures a node/actor’s importance in the network stemming from their connections with other very well-connected actors; is sometimes used to measure how much influence a node exerts in the network (Golbeck 2013).

High-Intensity Relationships: Allies, adversaries, or active armed opponents.

Low-Intensity Relationships: Partners or rivals.

Cluster Identification: Within the network of positive relationships (allies+partners), subgroups of the most closely connected actors, which are identified through algorithmic cluster detection (Hansen et al. 2011). (In this study’s network diagrams, clusters are color-coded to guide the reader.)

Clusters: More densely connected subnetworks in networks.

Subnetworks: Connected actors in a broader network; not each subnetwork is a cluster.

Disconnected Subnetworks: Actors that are mutually connected, but none of them are connected to the overall network of relationships.

Spoiler: Individual Authority Figure (IAF) that hinders a peace process.

Bridge-builder: IAF that serves as a cross-coalition connector still anchored with its party.

Unifier: IAF that acts as an intra-coalition organizer.

Facilitator: Independent, third-party connector between parties.

The negotiation networks approach draws on research and experience integrating conflict system analysis and negotiation analysis. A nested negotiation network is a set of relationships between parties with an interest in or influence over a conflict within a larger network of relationships constituting a conflict system. It resists coalition analysis—a common framework—because parties in a nested negotiation network are simultaneously part of several alliances with overlapping ties and contradictory goals (Bell and Mandell 2018). A nested negotiation network is distinguished from a mere set of coalitions by the fact that parties that are part of opposing coalitions may simultaneously be allies in a coalition aligned toward a different goal. This feature renders nested negotiation networks unstable and complex, leading to the nonlinear and emergent properties that characterize other complex systems (Bell and Mandell 2018).

Power is a contested concept—there is no consensus on what it means or how to measure it (Nye 2011). This study follows Nye (and most dictionaries) in employing a minimalist, intuitive definition: power is an actor’s ability to take actions that achieve desired outcomes (Nye 2011). This definition encompasses three “faces” of power and three approaches to measuring it.

The first face of power, identified by Robert Dahl, is the ability of an actor to influence another to take an action they would not otherwise take (Dahl 1957). This is the face most people intuitively associate with power. It is power as the ability to command, to coerce, and to force. The second face of power, identified by Peter Bachrach and Morton S. Baratz, is the ability of an actor to influence agendas and discourse so that their desired outcomes are achieved (Bachrach and Baratz 1962). This face of power is structural. It may incorporate command power, as in the case of the veto-wielding members of the UN Security Council, but grows less from the proverbial barrel of a gun than an actor’s understanding of political and social institutions. The third face of power, identified by Steven Lukes, is the ability of an actor to influence other actors’ perceptions of their own interests in order to achieve desired outcomes (Lukes 2004). This is power at its most subtle. Those who are influenced by this type of power may not know they have been influenced in this way. They believe they are acting according to their own interests, potentially unaware of why they see their interests in the way they do.

Jeffrey Hart also identified a debate within the international relations literature over whether power is best measured as control over resources, control over actors, or control over outcomes (Hart 1976). This study does not take a stance on which measure is “best,” nor does it attempt to measure power in a granular way. It does, however, incorporate the central insight unifying the models, which is that control over resources and control over actors can be converted into a degree of control over outcomes, which is what matters to actors (Nye 2011).

The final set of concepts necessary to analyze power is the most basic and applies to all three faces of power and all three methods of measuring power. According to Dahl, power has a base, a means, a scope, and a magnitude. An actor’s power base is the set of resources it draws on to exert power, the means of power is the set of methods it uses to convert its power base into outcomes, the scope of an actor’s power is the range of responses it is capable of eliciting, and the magnitude is the likelihood that the exercise of power will have the desired result (Dahl 1957).

The systematic assessment of each key stakeholder conducted for this study focuses on how each actor wielded their network of relationships to their strategic advantage, and to maximize their social, political, or economic gains at the negotiating table. Therefore, it assesses factors including, but not limited to: an actor’s influence over other actors, an actor’s influence over political agendas and social discourse, an actor’s main sources of leverage (which may include resources, military strength, powerful allies, and international reputation), resources under an actor’s control, and an actor’s ability to combine these sources of power to achieve desired outcomes (Babbitt et al. 2017). The analysis of the actors’ power and leverage will remain grounded in the relationship dynamics exhibited through their negotiated networks. Descriptive analysis will categorize parties’ exercises of power, and exploratory analysis will develop a model connecting these exercises of power to a sustainable resolution of the conflicts in question.

Modern theories of leadership are largely unified by their focus on the relationship between leaders and followers (Ahlquist and Levi 2011; Khan 2016). The most minimalist definition of leader-follower leadership in the literature is the “capacity to direct the actions of others” (Ahlquist and Levi 2011: 3). However, that definition fails to incorporate the fact that a “leader does not supply leadership in the abstract but provides a particular ‘product’—a particular set of services designed to achieve some particular purpose.” (Underdal 1994: 183). People only consent to being led as long as they believe the leader is taking them in the right direction in the right manner.

Leadership occupies an important role in conflict resolution. One key role is derived from the “two-level” nature of international diplomatic negotiations (Putnam 1988). A key concern for negotiators in negotiations that involve both external and internal negotiations is the risk of involuntary defection—a group’s failure to uphold the deal its negotiator reached (Putnam 1988). The threat of violent defection plays a major role in preventing lasting peace, and the management of the risk of involuntary defection is therefore important to consider in any model of conflict resolution (Walter 2002).

Leaders play a role in minimizing the risk of involuntary defection. Putnam’s theory of two-level games notes that leaders are able to influence the opinions of various domestic constituencies (Putnam 1988). Several experimental studies also suggest that leaders are able to shape the opinions of group members (Ahlquist and Levi 2011). Leaders can, therefore, change their followers’ attitudes or directly coerce them into upholding a deal, thereby preventing involuntary defection.

But followers are not mere objects for a leader to act upon. The replacement of a leader can be considered a type of “involuntary defection” that can occur in two-level games. Removing a leader whom the group believes will not meet the group’s interests resembles a preemptive “failed ratification” of the deal. A leader attempting to prevent involuntary defection may therefore have to walk a fine line, changing followers’ attitudes while preventing them from choosing a new leader.

Leadership change can also be a potential trigger for negotiation breakthroughs, rather than a form of involuntary defection, possibly because new leaders face different political incentives than established ones and because particular leaders may have nonrational constraints that prevent conflict resolution (Lieverfeld 2016).

A second key role for leadership analysis in conflict resolution is found in the negotiation itself. Some research suggests that the ability of some negotiators to direct the actions of others in service of reaching a deal is important—or even necessary—to reach international agreements requiring collective action, especially as situations grow more complex (Underdal 1994). Negotiation leaders guide the negotiating group as a whole toward its common purpose (Underdal 1994; Eckersley 2020). Although “at the table” behaviors of negotiation leaders are beyond the scope of network analysis, negotiation leadership depends both on the behaviors of individual negotiators and the position of the parties they represent in the larger conflict system (Underdal 1994).

The framework for analysis incorporates eight core concepts from the leadership literature: structural and informational theories of leadership; and transactional, transformative, unilateral, adaptive, coercive, and instrumental forms of leadership.

In structural theories of leadership, “the leader is someone who occupies a particularly important position in some predefined institutional structure” that gives them control over various levers of decision-making (like agenda control) or a particularly potent ability to advocate for a specific outcome (Ahlquist and Levi 2011: 6). In informational theories, the leader is someone who helps groups solve coordination problems by virtue of their visibility to all group members or access to privileged information (Ahlquist and Levi 2011). In transactional leadership, leadership is based on a series of explicit or tacit agreements between leaders and followers, in which leaders provide some sort of reward for followers exhibiting desired behaviors (Khan 2016). In transformative leadership, leaders motivate followers to strive toward “higher order” needs by changing their values (Khan 2016). Unilateral leadership is essentially leadership by example, in which leaders strive to solve a problem via their own actions, motivating their followers to do the same (Underdal 1994). Informational leadership models have shown that unilateral action can serve as a useful signal to other group members that allows them to better coordinate and achieve better outcomes (Ahlquist and Levi 2011). Adaptive leadership is “the practice of mobilizing people to tackle tough challenges and thrive” (Heifetz et al. 2009: 14). This interpretation stresses the importance of distinguishing between formal authority, informal authority, and leadership. According to adaptive leadership theory, acts of leadership define a leader, but an individual’s position in a formal authority structure does not. Coercive leadership is leadership by incentives, in which leaders reward desired behavior and punish undesired behavior (Underdal 1994). Coercive leadership is therefore a form of structural and transactional leadership. Instrumental leadership is based on specialized knowledge or skills (or followers’ belief in such), in which followers obey the leader’s directives based on their belief that the leader is uniquely capable of solving particular problems due to their access to privileged information or individual skills (Underdal 1994).

The methodology of this study is centered on assessing relationships between parties to a conflict and therefore has an established set of definitions that are used in classifying these relationships. For this study, relationships are classified as allies (++), partners (+), rivals (-), adversaries (–), and active armed opponents (x).

Allies (++): Stephen M. Walt (2009: 86) defines an alliance as “a formal (or informal) commitment for security cooperation between two or more states, intended to augment each member’s power, security, and/or influence.” He identifies the essential element in an alliance as a commitment to support each other against some external actor(s). We add that the relationship among allies includes “stable and extensive arrangements for security cooperation with a long history, as well as significant material or financial support, often directly related to a party’s armed struggle against an armed opponent” (Babbitt et al. 2017).

Partners (+): In other scholarly works, the words “allies” and “partners” usually have been used interchangeably. There have been attempts to carry out a “more rigorous categorization of alliances” in order to facilitate a better understanding of what brings allies closer together or drives them apart (Ghez 2011). In this article, we use both categories—allies and partners—to show the level or type of commitment these relationships entail. We define relationships between partners as characterized by “material, financial, or ideological support, which may or may not be linked to an armed struggle.” For example, parties that maintain productive and effective “channels of cooperation and assistance” are listed under this category (Babbitt et al. 2017).

Rivals (-): A large quantity of research on rivalry is concerned with analyzing “pairs of states repeatedly experiencing conflict” (Lemke and Reed 2001). Early studies about interstate rivalry “identify rivals by ‘dispute density,’ a minimum number of disputes within a specified period,” and the emphasis placed on phases suggests that rival relations can potentially change depending on different phases (Lemke and Reed 2001: 457). As with alliances, some scholars differentiate between various “types” of rivalry based on the extent to which rival relations may endure or vary (Goertz and Diehl 1993). Thus, based on such characteristics identified with the concept of rivalry, we define rivals as “parties with whom the party is in a state of political, ideological, or economic rivalry [competition] but with whom there is no direct armed combat. Rivalry does not exclude the possibility of cooperation on specific issues, and it captures a wide range of competitive behavior including conflict over contested territory as long as there is no ongoing armed combat” (Babbitt et al. 2017).

Adversaries (–): Amitai Etzioni (2011), in his discussion of how China is being framed as an adversary in the United States, uses the words “threatening,” “force,” and “enemy” in conjunction with the term “adversary”. As implied by Etzioni (2011), adversaries are regarded as a threat to a party’s interests and security, although parties in this relationship do not rule out the possibility of socializing with one another. As such, we define adversaries as parties with whom the party is in a “state of open, most likely mutually acknowledged, hostility, yet not in a state of direct, active, armed combat” (Babbitt et. al. 2017). The distinction between rivals and adversaries is less delineated than that of other relationship networks. The degree of cooperation was the primary indicator of their separation. Additionally, diplomatic recognition may be missing between actors in an adversarial dyad. Although the divide is blurry, it is necessary, as a dyad’s position on this continuum can have drastic repercussions on the relationship.

Active armed opponents (x): A number of writings on regional conflict refer to armed opponents in regard to a war or military conflict (e.g., Hartzell 2009). While the term armed opponents often is used interchangeably with armed factions, war adversaries, and rivals, armed opponents are considered rival groups that have chosen to engage in military conflict (Hartzell 2009). In an attempt to differentiate the concept from adversaries, we define active armed opponents as parties with whom the party is currently in a “state of direct, armed combat” (Babbitt et. al. 2017). (All emphases in original.) (Bell and Wolf 2023).

As was the case with previous negotiation network analysis efforts, the model was constructed using structured, focused comparison (SFC) (Bell and Wolf 2023). SFC uses a standardized set of general questions to structure investigation and focuses analysis on its research objectives.

The model was constructed in two stages. The first was a descriptive phase in which researchers identified possible explanatory variables and causal pathways by analyzing individual cases. The second was an exploratory phase in which researchers constructed a causal model by comparing cases.

The SFC uses the Bridging Insights source collection² as its initial material source, although further research was necessary to answer specific questions for specific cases. The source collection is used to identify the boundaries of each conflict system.

In terms of its scope, this article draws its insights from a comprehensive report containing eleven 6,000-word case studies focusing on conflicts around the world. The network analysis, on average, looked at the relationships between twenty-eight actors per case, including but not limited to states, paramilitary groups, international institutions, Indigenous organizations, and individual authority figures. Each case study entailed an in-depth analysis of how the network of relationships between key stakeholders, each actor’s positionality on the network diagrams, their leadership potential, and resulting leverage interacted to impact the outcomes of the peace process. In every case study, all parties that were a part of the formal negotiations that resolved the PAE conflict are included in our analysis. In the two unresolved cases, we consider parties that were part of the formal negotiations before these talks collapsed. In all cases, we then expanded the scope of our analysis to include certain other actors. These other actors are included if they meet one of two criteria: 1) they are connected with the formal negotiation parties through high-intensity relationships (see definitions), or 2) they otherwise exert significant influence on the peace process during its final year.

We define actors that exerted significant influence as a) actors that provided extensive financial, social, or political support to one or more formal negotiating parties, or b) actors who are considered as relevant in previous, aggregated scholarly work about the respective conflict; actors were identified as such if they are frequently mentioned in at least one third of the scholarly source materials that we reviewed about that conflict.³

Researchers then assess each actor’s interests, goals, sources of leverage, internal conflicts, and leadership dynamics by analyzing the existing literature using a framework adapted from previous efforts using this methodology (Bell and Wolf 2023). They then conduct an in-depth analysis of each key stakeholder’s distinct network of relationships and classify it under the following categories, on a positive-to-negative spectrum: allies, partners, rivals, adversaries, or active armed opponents. These stakeholder assessments ultimately lead to the creation of negotiation network diagrams, which explore the complex relationship dynamics presented by each conflict through the lens of power and leadership. We also rely heavily on “the premise that in a complex regional conflict system, there is a strong connection between the relationships between the main actors and their ability to negotiate” (Bell and Wolf 2023: 84). As a result, “we argue that evaluating the actors’ relationships on a scale ranging from extremely negative (active armed opponents) to extremely positive (allies) indicates the potential for negotiation or mediation to take place” (Bell and Wolf 2023: 85).

In the narrative for each case study, we closely inspect how the distinct network of relationships impacted the overall scope of the conflict and its manifestation of power and leadership through network analysis, in which we measure values such as closeness centrality, betweenness centrality, eigenvector centrality, and clusters, and their impact on the conflict. We also provide an overview of the key conflict dynamics, such as a comparison between the number of alliances vs. adversarial relationships, the overall ratio of the positive vs. negative relationships, polarized coalitions, and well-connected leaders that exerted the most influence on the peace process. By analyzing these metrics, we shed light on sources of power and leverage, internal factions, leader–follower relationships, mediator/broker positions, and the impact of these variables on the actors in conflict who relate to, lead, and influence each other (Babbitt et al. 2017).

The network diagrams explicitly take all parties that acted in a leadership capacity (including individual authority figures, state actors, organizations, etc.) into account. The position of each leader on the network diagram, and hence the scope of their relationships with other parties to the conflict, interact with the leaders’ ability to leverage their leadership potential, the level of power they possessed at the bargaining table, and the impact of their leadership capacity on the overall peace process. Researchers also examine the network analysis data to describe the sub-conflicts that make up the system, the involvement of each actor in each conflict, the network of relationships, and the emergent properties of the system. Researchers then use this information to develop hypotheses answering the identified research questions in service of the driving purpose: to determine why these conflicts reached a durable settlement. A member of the research team that had not previously reviewed the case studies assessed all eleven case studies in conjunction upon their completion to ensure adherence to the overall structural framework.

Our comparative framework relies on three key categories: (1) Relationship Dynamics, (2) Power Dynamics, and (3) Leadership Dynamics. Within the scope of these categories, this study’s approach focuses on the leadership and power elements that emerged from an in-depth exploration of negotiated relationships between the key stakeholders and their respective associations, which we explored through the negotiation network diagrams and the subsequent network analysis that highlights the relationship patterns within each conflict.

The inter-case analysis provides a case overview that systematically compares the six core cases, the three special cases, and the two unresolved cases. Based on the case overview and the patterns that emerged from our engagement with the relationship, power, and leader dynamics presented by the eleven case studies, we encapsulated our qualitative and qualitative findings through a list of nine observations. The focal point of these observations illustrates, but is not limited to: the role of informal negotiations; the involvement, and exclusion, of civil society actors and Indigenous groups from the negotiating table; the distinction between low- intensity vs. high-intensity relationships; the significance of mediators as bridge-builders; and the multifaceted exercises in informal and formal leadership. Researchers take an iterative approach to define values of interest due to the risks of defining these a priori, given substantial disagreement over relevant concepts in power and leadership analysis and the relative lack of theory to guide conflict network analysis (George and Bennett 2005). Following this analysis, researchers then developed a descriptive conflict resolution model that relies heavily on the patterns and observations that emerged from conducting the comparative analysis and that may serve as a prescriptive starting point for actors within a PAE conflict environment aiming to reach a negotiated settlement. The core focus of the article lies on the comparative insights derived from the inter-case analysis.

On the negotiation network diagrams, each “node” represents an actor, and the connecting lines between the nodes display each instance of an alignment between them (a linkage between two nodes in the network) (Bell and Nehrbass 2022). Each alignment on the diagrams conveys the same intensity, and the length of the lines that create the alignment between two nodes does not carry any structural significance. The negotiation network analysis considers only the relationships that existed in the year that led to the signing of a peace agreement, or after its immediate fallout in the case of unresolved conflicts. Within the analytical framework, all alignments between these nodes/actors are presented as dyads, formed by non-directed relationships (A–B). For the analysis of the “all positive relationships” (allies+partners) in the context of each conflict, color-coded clusters (using the Clauset-Newman-Moore algorithm)⁴ are defined as subgroups of the most closely connected actors.

Our methodology does not include a full mapping of all formal/informal relationships that constitute the conflict environment. We are not claiming to provide the reader with an exhaustive list of all players in a conflict, and neither do we claim to analyze all conflicts from vastly competing angles (social, political, economic, etc.). Instead, we are applying a relationship-centered framework, particularly with an emphasis on power and leadership, which places the parties at the center of the conflict, at the center of our analysis, and then expands outward to map the relationships of these key stakeholders that exercised significant leverage or leadership potential. This methodology allowed us to identify crucial power and leadership dynamics and assess the output of these variables on the conflict. Further expanding the list of actors would have led us to lose analytical focus, and risk bringing individual, normative preferences in terms of inclusion/exclusion of said actors, which we seek to avoid for the sake of generating valid insights regarding PAE conflicts. If notable informal actors, especially non-state actors, Indigenous groups, or civil society actors exerted significant influence, we included their contributions in the analysis. Otherwise, the consequences of their exclusion from the peace talks are captured in either the respective study or in the comparative analysis as well as in the conflict resolution model. While we attempted to take every actor that exerted significant influence on the peace processes into account, due to the wide breadth of the actors and limited information available through existing literature, we cannot fully control for omitted variable bias (Bell and Nehrbass 2022). Although the protracted conflicts lasted multiple decades, our dataset, narratives, and negotiation network diagrams only account for the relationship dynamics that existed in the year that the key stakeholders signed a peace accord, or in the case of the unresolved conflict, in the year immediately following the collapse of the peace process. The historically evolving nature of these relationships could hence not be fully captured by this study.

PAE conflicts and their subsequent resolution involve several complex drivers. In order to understand what factors lead to the peaceful resolution of PAE conflict, we have systematically evaluated the variables, conditions, and overarching observations that emerged from our analysis of the cases under investigation. This comparative analysis provides an overview of the six core cases, three special cases,⁵ and two unresolved cases.

We encapsulated our quantitative and qualitative findings through a list of nine observations. This comparative analysis seeks to capture how the network of relationships presented by these case studies aligns with and deviates from the interplay of power and leadership, to explore “why it worked.”

1. Informal negotiations heavily influence the setup of formal talks before peace negotiations begin.

Our observations match previous research findings that the architecture of a negotiation that happens “away from the table” is a crucial component of successful bargaining (Lax and Sebenius 2006). When formal negotiations commence —i.e., representatives of the conflict parties physically sit down at the table—important process choices such as determining the scope of the talks or selecting the parties to be formally included in the talks already have been made. This can either happen organically (with no party thinking strategically about these setup questions) or by design (with one or several parties influencing the pre-negotiation environment to get to a more favorable setup). In our resolved cases, it is notable how much informal groundwork for formal negotiations has already happened by the time formal peace negotiations commence.

For instance, the Northern Ireland–Provisional Irish Republican Army (PIRA) conflict demonstrates that prior to the commencement of the official peace negotiations, in the lead-up to the signing of the Good Friday Agreement (1998), Trimble and Hume engaged in a series of back-channel talks and displayed unilateral leadership in an attempt to build a common platform between their political parties.

In the Sudan–Sudan People’s Liberation Movement/Army (SPLM/A) conflict, under the guidance and third-party mediation of the Intergovernmental Authority on Development (IGAD), the SPLM/A and the Sudanese government agreed upon the Machakos Protocol in 2002 prior to signing the Comprehensive Peace Agreement (CPA) in 2005. This protocol enabled them to create mutually agreed-upon principles as a prerequisite for further discussions (BBC 2019). These “pre-negotiations” also became apparent through General Lazaro Sumbeiywo’s active mediation. General Sumbeiywo had served as the official head of the IGAD-led peace process in Sudan during the late 1990s, which contributed to his understanding of the conflict’s context. It also allowed him to obtain visibility and credibility in the eyes of the conflicting parties, which allowed him to demonstrate instrumental leadership. Initially, the meeting at Machakos appeared unlikely to produce any meaningful results. As the meeting neared an unsuccessful close, Sumbeiywo locked two delegates from the National Congress Party (NCP) and the SPLM/A in a room together on the night of July 19, laying down a “now or never” ultimatum to both parties (Rolandsen 2011: 556) in an effort to encourage them to negotiate informally, which led to the signing of the Machakos protocol the next day. In the case of the Philippines–Moro National Liberation Front (MNLF) conflict, Fidel Ramos, then President of the Philippines, pursued a policy of inclusive and constructive engagement with the MNLF, which helped build trust and confidence between the parties before the commencement of the official peace process that led to the Jakarta Accord (1996).

In the case of South Africa, at a time when political violence between the African National Congress (ANC) supporters and Inkatha Freedom Party (IFP) supporters was becoming rampant, Archbishop Desmond Tutu (affiliated with the South African Council of Churches) and Bishop Stanley Mogoba arranged a June 1993 meeting between Mandela and Buthelezi. This meeting led to the creation of an acceptable compromise on the sticking points surrounding power distribution between the national government and the regional governments (Camay and Gordon 2004).

These “away-from-the-table negotiations” through back channels, bilateral negotiations, or meetings convened by third actors are crucial. It is during this phase that parties exercise effective leadership to influence negotiation preconditions, ensure the inclusion of a more favorable composition of parties in formal talks, strengthen their bargaining position, solidify their negotiation coalition, and influence the no-deal options of other parties. Focusing exclusively on the analysis of formal peace negotiations is hence missing the point when analyzing the successful exercise of power and leadership to end PAE conflicts. There is the groundwork that successful negotiators do first. These activities reconfigure the network of relationships and ensure a “negotiation architecture” that is more conducive to successful peace negotiations.

2. A reduction in major acts of violence precedes the negotiated ending of protracted, asymmetric, ethnonational conflicts.

In all nine resolved cases, either formal ceasefires or informal arrangements to halt all major acts of violence between the key parties preceded the final peace agreement. In the case of South Africa, political violence occurred between the ANC and IFP supporters, but there was no formal, “all-out” conflict between the ANC and IFP. In the two unresolved cases, there has been no ceasefire or partial ceasefire at best (e.g., the Afghan Taliban and the United States agreed on a mutual suspension of hostilities while the Taliban did not commit also to ending its attacks on the Afghan government, which was excluded from the Doha talks). There has been a long debate among peace researchers about the importance of a ceasefire for ending armed conflict, with some arguing that a ceasefire plays a vital role in the effectiveness of peace talks, since parties may rely upon the ceasefire to build trust and create a more inclusive environment that is conducive for negotiations while also signaling their willingness to cooperate (Clayton and Sticher 2021; Bara and Clayton 2022). On the contrary, others argue that since peace processes do not exist on a linear plane and frequently ebb and flow, a ceasefire is not always necessary as a prerequisite for a peace agreement. The peace process between the Colombian government and the Revolutionary Armed Forces of Colombia People’s Army (FARC) highlights how a peace process can exist without a ceasefire, although in the case of Colombia, both parties ultimately secured a bilateral ceasefire with the signing of the Agreement on Bilateral and Final Ceasefire in 2016 (Forster 2019). Nonetheless, it is worth noting that both sides had informally halted their acts of active violence before agreeing on the formal ceasefire (Segura and Mechoulan 2017).

Our observations suggest that putting an end to widespread violence first, before engaging in more substantial political negotiations, may be an important sequencing factor in the resolution of PAE conflicts. For example, in the case of Sudan–SPLM/A following the escalation of the Darfur conflict, negotiations facilitated by Chad in N’Djamena resulted in the signing of the Humanitarian Ceasefire Agreement between the Sudanese government and the SPLM/A in 2004. During the Northern Ireland–PIRA conflict, in the year prior to the signing of the Good Friday Agreement in 1998, the PIRA announced a ceasefire, which allowed Sinn Féin to be included in the peace discussions (Hancock 2008). In the case of Sudan, the third-party mediation of Chad as a broker between the government and the SPLM/A led to the ceasefire becoming a possibility.⁶ These examples illustrate the role of power and leadership in brokering a ceasefire.

In the case of Northern Ireland, the gatekeeping authority of the British and Irish governments acted as their base of power, providing them with the means to set a ceasefire requirement as a prerequisite to gaining access to the bargaining table. Here, the British and Irish governments embodied the second face of power. This, in addition to the absence of any alternative peace processes, pushed all parties to comply with the ceasefire requirement.

3. A dense network of positive relationships between parties in a conflict system does not necessarily correlate with an inclusive design of formal negotiations.

At first glance, it appears that our findings contradict established research findings with regard to the importance of inclusion in peace negotiations. Civil society groups can generate greater public participation and representation in the peace process (Wanis-St. John and Kew 2008). Participation of civil society leads to additional legitimacy for the peace negotiations and also correlates with the sustainability of the peace agreement (Pfaffenholz et al. 2006). Nonetheless, it is important to note that civil society actors are almost never allocated a seat at the formal negotiation table (Wanis-St. John and Kew 2008). Particularly in Track I negotiations, the involvement of civil society actors is hindered by political elites, in part because attempting to foster cooperation and coordinating with a large number of diverse civil society actors further complicates the likelihood of arriving at a peace agreement (Pfaffenholz et al. 2006).

In most of the case studies under analysis, civil society groups and Indigenous communities, while playing an important advocacy role and providing the groundwork for the negotiations, were excluded from the formal peace process. For example, during the Guatemala–Guatemalan National Revolutionary Unity (URNG) conflict, Mayan leaders demanded direct representation at the bargaining table pertaining to Indigenous rights. They vehemently objected to the classification of 60 percent of the population as merely a sector within civil society, represented under the umbrella of the Assembly of Civil Society (Preti 2002). In response, the Indigenous sector formed the Coordination of Organizations of the Mayan People of Guatemala (COPMAGUA), which consisted of over 200 different groups. Despite the COPMAGUA’s advocacy, their impact on the peace process was limited by both sides of the negotiation table (Alvarez 2002). In 1994, the ASC attempted to hold its ground by meeting with the URNG and government negotiators to lobby for its positions, initiating regular informational meetings with the UN team and representatives of the Group of Friends countries (Krznaric 1999). Nevertheless, the ASC was not allocated a seat at the final bargaining table by the government or the URNG.

In the case of the Bangladesh–Parbatya Chattagram Jana-Samhati Samiti/Shanti Bahini (JSS/SB) conflict, the Indigenous Pahari people of the Chittagong Hill Tracts (CHT) region were excluded from formal representation during negotiations. Civil society groups such as the Hill People’s Council, the Hill Students’ Council, and the Hill Women’s Federation were excluded from the negotiation process (Saha 1999).

However, based on our negotiation network analysis, we argue that there are two different kinds of exclusion from formal peace negotiations: mitigated exclusion, which we define as exclusion from formal peace talks mitigated by positive, informal ties between at least one formally included party and the excluded party, and unmitigated exclusion, which we define as the congruence of a party’s exclusion from formal talks and its marginalization in informal negotiations because of a lack of positive, informal ties with any of the parties at the table.

In the case of unmitigated exclusion, certain parties to a conflict strike a deal without taking the interests of the excluded party into account. An example is one of our unresolved cases, Afghanistan. In Afghanistan, the 2020 Doha Deal between the United States government and the Taliban resulted in the withdrawal of U.S.-led foreign military forces from Afghanistan. In return, the Taliban agreed not to allow extremist groups, such as ISIL al-Qaeda, to operate on Afghan soil. However, the U.S.–Taliban deal completely excluded the Afghan government and did not take its interests into account. Many experts in Afghanistan and also the international community perceive the deal as significantly flawed and largely skewed in favor of the Taliban (Azizi 2021).

In the case of mitigated exclusion, formal negotiations may exclude one or more parties; however, the formally excluded party can compensate for this lack of formal representation by leveraging its positive relationship with at least one party at the table. For example, the Angola–Union for the Total Independence of Angola (UNITA) conflict demonstrates how even though the Indigenous population did not gain direct representation during formal negotiations—and hence is not formally represented on the network diagrams (Figures 1 and 2)—the People’s Movement for the Liberation of Angola (MPLA) represented the interests of the Mbundu and Mestiqos communities. The MPLA possessed the highest eigenvector centrality, illustrating how within the network of positive relationships, it wielded the highest influence and authority due to its connections with other very well-connected actors. On the other hand, the UNITA, which also emerged as a well-connected actor due to its positive associations with other key stakeholders, advocated for the interests of the Ovimbundu people (Paul et al. 2013).

This suggests that looking only at “who sits at the table” and drawing conclusions about whose interests will or will not be met by the final deal misses an important underlying dynamic of PAE conflicts. If a formally excluded party’s interests are represented indirectly by a formally included ally or partner—for example, Indigenous groups or civil society are excluded from the peace process but their interests are represented through political factions or guerrilla groups—a dense positive relationship network de facto compensates for a certain degree of formal exclusion and may facilitate the resolution of the conflict.

To be clear, this does not suggest that marginalized groups—or conflict parties more generally—should not care about securing a seat at the table. It also does not mean that actors do not need to worry if they are not included in formal talks. However, this observation suggests that there is no perfect congruence between “included/excluded” and “influential/not influential.” Instead, the ability of parties to leverage their network of relationships to influence even those negotiations in which they are not directly involved is an overlooked source of power in PAE conflicts.

4. The ability of civil society groups and Indigenous groups to influence peace negotiations depends less on external recognition bestowed on these groups by state actors, and more on these groups’ ability to organize, often under the leadership of skilled unifiers and bridgebuilders.

In the absence of formal or informal leadership that can serve to unify and represent the various civil society factions and the Indigenous communities, their power diminishes or they are excluded from the peace talks. We see this in the Guatemala case after Quezada Toruño’s resignation. He was the leader of the civil society umbrella group, the ASC. As a result of his resignation, embodied structural dimensions of leadership rapidly disintegrated, thereby diminishing the ASC’s influence on the peace process due to internal divisions (and the elites’ tendency to exclude them), even though the ASC had done the groundwork to set up these negotiations.

The second possibility that can emerge is a rebel group/organization claiming to represent the interests of civil society actors/the Indigenous population may co-opt their agenda in an attempt to further its own sociopolitical ends (which can easily backfire). For instance, in the case of Bangladesh (Figures 3 and 4), the key rebel group, Parbatya Chattagram Jana-Samhati Samiti (PCJSS), was acting as a representative of all the Indigenous Pahari Chittagong Hill Tracts communities, forming alliances and negotiating on behalf of them (the Indigenous community). This was inconsistent with the preference of many Indigenous Pahari community groups, who sought to advocate for and represent their own interests (Rizal and Yokota 2006). This disagreement created factions within the PCJSS and as a result, the exclusion of civil society and the Indigenous Pahari community ultimately weakened its ability to form working partnerships, as well as weakening its negotiating position with the government of Bangladesh.

5. In protracted, asymmetric, ethnonational conflicts, the military often acts as a distinct political actor with idiosyncratic interests and a unique network of relationships.

Our analysis sheds light on the unique role that a state’s military can play in PAE peace negotiations. Simplistic interpretations of the role of the military in civil wars may lean toward one of two extremes. The first holds that the military is the first and foremost tool of oppression and violence used by the government to strengthen its position at the bargaining table or to crack down on marginalized groups, such as Indigenous people. This interpretation of the military aligns with the first face of power. According to the second simplistic interpretation, the military is an apolitical actor that simply implements and executes government policy, and as long as the government makes peace, the military will comply. We find that in PAE conflicts, the military can best be understood as a unique political actor with its own web of negotiated relationships and idiosyncratic interests. These may well overlap with the interests of its political leaders, but this is not always the case.

For example, in the Guatemala–URNG conflict (Figures 5 and 6), while the military was not in favor of the peace process, it tolerated developments in the process because it sought to transform its negative, violent image resulting from its hostile relationship with the URNG and guerilla groups, and also to end its isolationist tendencies stemming from the military’s aggressive history (Alvarez 2002). The military formed its own network of alliances and partnerships with key stakeholders, independent of the government’s network of relationships, to serve its objective of acquiring additional power and leverage. Furthermore, when the coalition of various domestic and external stakeholders promoted the Peace Accords that concluded Guatemala’s conflict, it required the military to agree on diminishing its size, budget, and scope (Ruhl 2005). Although the military had to make concessions, in line with its vested interests, it successfully secured the guarantee of extensive amnesty for the various human rights violations that it committed during confrontations with the URNG and guerilla groups (Preti 2002).

The Turkey–Kurdistan Workers’ Party (PKK) conflict also shows how although the Justice and Development Party (AKP) utilized the Turkish Security Forces as an apparatus for violence, the Turkish Security Forces were also a distinct actor with their own network of relationships—often in an adversarial or active armed opponent capacity—and a resolve to protect their own political and economic interests (Exum 2017). Activities that signify the Turkish Security Forces’ role in an adversarial or active armed opponent capacity may include their oppression of Kurdish civil society and Kurdish activists, along with their internal divisions and hostile relationship with the factions that confronted them, including the Ergenekon faction and the Gülen Movement. In fact, the PKK’s grievances regarding a buildup in the Turkish Security apparatus contributed to the breakdown of the peace process (MacDonald 2015).

This observation reinforces established findings about the role of militaries in civil wars. The military’s power of coercion provides it with the ability to enforce its will on the state that created it, leaving it with the choice between complying with the orders of the political elites or using its considerable coercive leverage to pursue independent interests (Feaver 1999).

6. Mediators bridge structural holes and connect otherwise unconnected clusters of actors.

Conflict research literature has yielded insights about the important role of mediators in the resolution of civil wars. Existing literature indicates that mediators’ success can be traced to their ability to act as a credible bridge of information, which can help alleviate the negotiating parties’ uncertainty and mutual mistrust (Kydd 2003; Savun 2008; Svensson and Wallensteen 2010). Through their socioeconomic leverage, mediators can utilize the third face of power to impact the key stakeholders’ decision-making calculus by dangling carrots and sticks in the form of the costs and benefits of reaching a settlement (Beardsley et al. 2019).

Our case studies indicate that the United Nations (UN) may be particularly well situated to act in a mediator capacity in PAE cases. The UN presence brings a measure of credibility that may be acknowledged (even respected) by most conflict parties due to the UN’s international status, historical reputation, and conflict resolution expertise. Additionally, the UN’s involvement in the peace process may translate into economic resources, bring international attention to the conflict, and create opportunities for multiparty dialogue. While the UN need not necessarily serve as a mediator in order for its participation to be constructive, it may well act as a third-party mediator after formal negotiations commence. Even the UN’s peripheral presence can facilitate the process of arriving at objective standards of behavior for the conflicting parties, which may prevent the peace talks from falling apart down the line. The UN was involved in eight of the nine resolved cases. In seven of those cases, it acted in a mediator capacity, while its involvement with the other two cases was on the periphery, as part of the relationship network but not as a mediator. Thus, our research findings do not imply that UN mediation is a prerequisite for resolving PAE conflicts, but it is extremely conducive to the process.

The UN did act as a mediator in the Philippines–MNLF conflict, working closely with the Organization of Islamic Cooperation (OIC) to facilitate peace talks by offering political and diplomatic backing for the process. While the OIC acted in a quasi-leader capacity for the Muslim population, which allowed it to bring both the MNLF and the government of the Republic of the Philippines (GRP) to the negotiating table, the UN provided political, financial, and technical support for the parties, enabling them to successfully reach a peace agreement.

While the UN’s involvement as a mediator in the resolution of PAE conflicts remains frequent, not all successful mediators are UN organizations or UN-affiliated individuals. For example, international governments, domestic organizations, or individual authority figures can also serve as mediators, and their mediation strategy may differ accordingly. For instance, in the case of the Colombia–FARC conflict, Venezuela and Cuba played a key role by conducting back-channel negotiations with the Colombian government and the FARC in order to facilitate more formal negotiations. In particular, Cuban President Fidel Castro and Venezuelan President Hugo Chavez were critically important in the process by bringing the Colombian government and the FARC to the negotiation table. During these talks, Chavez encouraged Colombian President Juan Santos to negotiate with the FARC and convinced him that a negotiated solution was the ideal conclusion to the conflict (Beaumont 2012).

If there is one common theme to note, it may be that mediators play an important role in bridging structural holes in negotiation networks and connecting otherwise unconnected clusters of actors. Individual authority figures who are able to leverage their cross-cluster ties seem to be in an especially good position to serve as mediators in a PAE conflict. The Sudan–SPLM/A conflict (Figure 7) serves as an example by highlighting how Lazaro Sumbeiywo (lead IGAD mediator) successfully acted as a viable partner by leveraging his bridging position between the ruling NCP; the SPLM/A; John Garang, the leader of the SPLM/A; Ali Osman Taha, the First Vice President of Sudan and chief NCP representative during the talks; and international actors, such as the United States, the United Kingdom, and Norway.

However, it is important to note that actors who are well positioned within the network to act as mediators must also possess the will to exercise and leverage their broker positions. In the Afghanistan conflict, Turkey played a significant role in terms of setting up various regional dialogue initiatives that demonstrated Ankara’s convening power (Dostyar and Farahi 2022). In fact, Turkey possessed the highest eigenvector centrality, indicating its high level of connectivity with other well-connected actors in the network. However, Turkey did not leverage its broker position to facilitate an inclusive peace process for Afghanistan.

7. Groups of aligned actors form clusters in negotiation networks, pointing to the importance of coalition building in multiparty peace negotiations.

Within the network of positive relationships (allies+partners), clusters are defined as subgroups of the most closely connected actors (Hansen et al. 2011) that rely on intensive “collaboration and communication” in their response to a conflict (Karali et al. 2020: 1281). We utilized the Clauset-Newman-Moore cluster algorithm to detect clusters in the positive relationship networks of all resolved and unresolved cases. These clusters—color-coded in Figures 8 and 9—are generated automatically by the algorithm, with no manual input from the researchers. Quantitative findings from this cluster detection are in line with our qualitative research findings regarding the coalition building that occurred between mediators and politically aligned actors during peace processes.

For instance, in the Sudan–SPLM/A conflict, our cluster algorithm picked up on three distinct clusters. The green cluster includes the United States, the United Kingdom, and Norway, three actors who coordinated extensively to promote their shared objective of pushing the peace process forward by supporting the interests of the SPLM/A and pressuring the NCP to come to the negotiating table. This triumvirate coordinated so closely and purposefully that they formed an official grouping—dubbed the “Troika”—to work collectively on all matters related to Sudan. The Troika countries mostly supported the SPLM/A, but occasionally worked together with the NCP, one of the SPLM/A’s adversaries, especially through Norway’s support by leveraging its bridging position as a partner figure to both opposing coalitions (Rolandsen 2011), as illustrated by Figure 8. The algorithm also identified the red cluster, which includes the SPLM/A, IGAD, Kenya, Ethiopia, Eritrea, and Uganda, emphasizing the group of East African nations that supported the SPLM/A. Finally, the blue cluster, composed mostly of individual authority figures, brokers, and mediators, illustrates the importance of individual-level leadership during the Sudan peace process.

Similar clusters can also be identified for unresolved protracted, asymmetrical, and ethnonational conflicts, such as the Turkey–PKK case, in which the three clusters indicate clear divisions between the primarily Turkish coalition (orange), the regional Kurdish coalition (green), and the Kurdish coalition based in Turkey (blue). These clusters indicate how strong ties of communication and cooperation existed within the Turkish government’s coalition, pan-regional Kurds, and the Turkish Kurds, even though there were very few positive connections across these three blocs. Hence, although the absence of third-party mediators and the historical hostilities that existed between the key stakeholders prevented inter-coalition bridge-building, which ultimately contributed to the unresolved and protracted disposition of the conflict, intra-coalition political alignment in this multiparty negotiation remained evident (Savran 2020).

8. In most negotiation networks, prior to the resolution of protracted, asymmetric, ethnonational conflicts, low-intensity relationships outnumber high-intensity relationships.

Although in all but one of the resolved protracted, asymmetric, ethnonational conflicts, low-intensity relationships outnumbered high-intensity relationships,⁷ this was especially marked in the Angola–UNITA and Philippines–MPLA conflicts, in which there were more than double the number of partnerships and rivalries than alliances and adversarial relationships. This phenomenon highlights that in resolved PAE conflicts, competition and working associations are much more common than outright hostilities and long-standing, historically positive ties. This may point to how a peace process is more likely to lead to fruitful results in the absence of direct hostility (adversarial relations) between the key negotiating stakeholders. On the other hand, long-standing alliances are not a prerequisite for successful negotiations, as long as key actors are willing to work with each other.

However, it is important to note an outlier; namely, that in the Bangladesh–JSS/SB conflict, high-intensity relationships (adversaries+allies) significantly outnumbered low-intensity relationships (rivals+partners). This discrepancy may be a result of the close collaboration between the internal actors in the network and their external partners and the heightened hostilities between the two subnetworks (the PCJSS/SB vs. the government of Bangladesh/military), or the limited number of scholarly sources available (in English) may have restricted the researchers’ ability to analyze the negotiated relationships surrounding the Bangladeshi peace process.

The negotiation network diagrams also highlight that while disconnected subnetworks of allies are common (occurring in all eleven cases), partnerships can be utilized to bridge structural holes and establish working relationships that even in the absence of long-standing ties can lead to the resolution of the conflict in multiparty negotiations. For example, the Northern Ireland–PIRA network (Figures 10 and 11) illustrates that while the Loyalist, Republican, and international subnetworks did not form cross-coalition alliances, they did form cross-coalition partnerships, especially through the mediator and bridge-builder positions occupied by individual authority figures, including John Hume, David Trimble, George Mitchell, and Bertie Ahern.

Similarly, the France–FLN (Algeria) network (Figure 12) highlights how the National Liberation Front (FLN), which was the key rebel group; the French government; and the Secret Army Organization (OAS), a fringe militant group, formed three separate subnetworks of alliances. However, through Ahmed Ben Bella’s bridge-builder capacity and structural position as a political leader, and the United Nation’s position as a facilitator throughout the peace process, these three subnetworks formed cross-coalition partnerships (Figure 13) that ultimately resulted in the resolution of the conflict.

9. Individuals and organizations exercise leadership in vastly different ways, which leads to vastly different consequences for peace processes.

The activities that individuals and organizations undertake in order to influence the negotiation environment to their advantage are not easily defined. Even if we apply a simple definition of leadership to individuals—and for the purpose of this study define leaders as “individual authority figures”—we must recognize that the activities in which they engage, including building momentum, facilitating, unifying, and brokering, are diverse and wide-ranging. They may also engage in activities that derail the peace talks, acting in a spoiler capacity to safeguard their own, narrow interests.⁸

For example, in the Angola–UNITA conflict, Jonas Savimbi—who founded and led the UNITA movement starting in 1976—acted in a spoiler capacity, exhibiting coercive leadership that sabotaged the peace process during its initial stages. In 1991, UNITA and MPLA began peace negotiations, starting with the Bicesse Agreement and a commitment to a ceasefire (Hurst 2009). However, fighting resumed once José Eduardo dos Santos and Savimbi challenged one another in the first multiparty election in Angola since its independence, which resulted in dos Santos’ victory and Savimbi’s violent contestation of the election results. After the assassination of Savimbi, the civil war began to transition to peace talks in 2002, leading to another ceasefire, followed by formal negotiations between UNITA and MPLA.

Similarly, in the case of Turkey–PKK, President Recip Erdog˘an’s unstable, often erratic approach to the peace process, political competition with the Democratic Peoples’ Party (HDP), hostility toward the PKK, and continuous dismissal of Kurdish political and social identity hindered negotiations (Savran 2020).

In contrast, through Nelson Mandela’s adaptive leadership, the South African conflict shows how individual authority figures can exercise leadership in a broker and momentum-builder capacity by building positive relationships with other stakeholders in the conflict. As highlighted by Figure 14, Mandela effectively built a large coalition of aligned stakeholders and mobilized this coalition for conflict transformation. The success of the United Democratic Front (UDF) in terms of incorporating diverse actors, from the South African Communist Party (SACP) to the Congress of South African Trade Unions (COSATU), also showcased his skill. Mandela also effectively de-escalated high-tension situations throughout the early 1990s while fostering partnership ties with international actors.

In the Northern Ireland–PIRA conflict, David Trimble of the Ulster Unionist Party (UUP) acted as a bridge-builder between his own party, which was part of the Loyalist coalition, and SDLP (Social Democratic and Labour Party), which was part of the Republican coalition. He displayed unilateral leadership while also attempting to embody the role of a unifier between the various factions and fringe groups of the Loyalist coalition. As a result, Trimble had to walk a fine line between defending his position of authority and credibility within his own party, against the Loyalist coalition members such as the Orange Order, who opposed the Good Friday Agreement (Tonge and Evans 2004), while also maintaining his position as a mediator and negotiator with the Republican coalition, and with both the British and Irish governments, who were also key actors in the conflict resolution process.

No consistent “type” of actor demonstrated the highest eigenvector centrality across all the cases, which indicates their high level of connectivity with other well-connected actors in the network. We do notice that a number of these well-connected actors are external mediator figures, but political parties, rebel groups, domestic individual authority figures, and domestic and international organizations are all capable of displaying high eigenvector centrality values. In line with the principles of adaptive leadership, this may indicate that the type of organization or the formal position granted to an individual does not limit their ability to maneuver themselves into critical positions within the relation network that would allow them to exert their leadership potential.

This study has uncovered variables that have influenced the resolution of protracted, asymmetric, and ethnonational (PAE) conflicts in Latin America, Africa, Asia, Europe, and the Middle East between 1960 and 2021. In particular, it has analyzed how specific bargaining and leadership strategies have contributed to the peaceful settlement of such armed conflicts.

Methodologically, the study integrates conflict system analysis and negotiation analysis. It is based on insights regarding the “nested negotiation network” phenomenon, which are sets of relationships between parties within a larger conflict system. In complex conflict systems, state and non-state actors can simultaneously belong to different, indirectly connected subnetworks of relationships. This has important implications for the ways in which conflict parties are able to exercise power and leadership to influence a peace process.

This study focuses on the role of negotiated relationships, with an emphasis on power and leadership dynamics in resolving protracted, asymmetric, and ethnonational conflicts. We systematically analyzed eleven cases, including six PAE conflicts that were resolved through negotiations (Northern Ireland–PIRA 1998, Guatemala–URNG 1994, Sudan–SPLM/A 2005, Philippines–MNLF 1996, Bangladesh–JSS/SB 1997, and Angola–UNITA 2002), two unresolved PAE cases in which negotiations failed (Turkey–PKK 2015 and Afghanistan–Taliban 2021), and three resolved cases that did not meet the definition of a PAE conflict but displayed significant similarities with them (South Africa 1994, France–FLN 1962, and Colombia–FARC 2016).

We have identified the following nine patterns in regard to the resolution of PAE conflicts:

1. Informal negotiations heavily influence the setup of formal talks before peace negotiations begin.

2. A reduction in major acts of violence precedes the negotiated ending of PAE conflicts.

3. A dense network of positive relationships between parties in a conflict system does not necessarily correlate with an inclusive design of formal negotiations.

4. The ability of civil society groups and Indigenous groups to influence peace negotiations depends less on external recognition bestowed on these groups by state actors and more on these groups’ ability to organize, often under the leadership of skilled unifiers and bridge-builders.

5. The military often acts as a distinct political actor with idiosyncratic interests and a unique network of relationships.

6. Mediators bridge structural holes and connect otherwise unconnected clusters of actors.

7. Groups of aligned actors form clusters in negotiation networks, pointing to the importance of coalition building in multiparty peace negotiations.

8. In most negotiation networks prior to the resolution of PAE conflicts, low-intensity relationships outnumber high-intensity relationships.

9. Individuals and organizations exercise leadership in vastly different ways, which leads to vastly different consequences for the peace process.

Our study applied a relationship-centered framework through the lens of power and leadership, which places the key conflict actors at the center of the analysis, along with their formal and informal relationships with other actors that were either connected with them through high-intensity relationships or that exerted significant influence on the peace process. As a result, we unveiled crucial power and leadership dynamics, along with their impact on the respective conflicts.

Our findings uncovered variables that can generate an environment conducive to the resolution of PAE conflicts. We discuss how an actor’s strategic position in the relationship network impacts their leadership potential and their influence on the formal peace process, and how individual authority figures and organizations can leverage their inter-coalition and intra-coalition connections to move conflicts closer to a successful resolution, while safeguarding their own vested interests at the bargaining table. We also highlight the significance of mediators that serve as bridges between coalitions in a leadership capacity, and show the complex interplay between leadership and power that actors can wield to hinder or advance a peace process.

A few caveats are in order. Negotiated relationships in PAE conflicts are in flux. Our quantitative analysis is based on a snapshot right before the conclusion of negotiations. While our qualitative analysis considers key developments leading up to a peace process, a detailed analysis of how exactly relationships evolve over time and the implications of said evolution on the peace process is beyond the scope of this study.

This study uses a relationship-driven negotiation analysis framework. This is not to deny the importance of factors such as interests, perceptions, culture, values, and religion in PAE conflicts. However, we focus here on the negotiated connections and on their interplay with power and leadership to address the research question.

The framework of our analysis is restricted by the fact that only a limited number of cases fit the criteria defined by the original call for research proposals issued by the project of which our study is a part. As a result, we cannot draw statistically significant, generalized conclusions about peace negotiations on the basis of the cases under observation. An additional limitation of our findings is that the eleven cases deviate on the three PAE variables. Also, the comparison between the resolved and the unresolved cases is limited because only two unresolved cases were selected from a much larger number of unresolved PAE conflicts.

Since relationships in PAE conflicts are often in flux, our model does not fully capture the historically evolving nature of said relationships and their impact on the conflict resolution process. Considering the wide breadth of variables and nuances involved in each conflict, future studies would benefit from evaluating the case studies from different frameworks of analysis, such as social, political, cultural, and economic, and discussing findings in a comparative perspective. Finally, the task of covering such a breadth of complex cases means that the analysis had to be simplified in significant ways. For future iterations of similar cases, consultations with regional subject matter experts would be conducive to the process of unveiling “why it worked.”

In terms of avenues for future research, we see several promising starting points. Future studies would benefit from expanding this study’s analytical framework to include a wider variety and number of cases, which may contribute toward the generation of statistically significant findings. We propose to optimize the data-gathering process, unifying the data analysis scheme and analyzing connections between changes in negotiation networks and conflict escalation patterns more thoroughly. We also see potential for a separate study that analyzes the role that external economic incentives play in the context of domestic power and leadership dynamics. While our findings demonstrate that in PAE conflicts, economic incentives can be a viable tool for pushing a peace process across the finish line, the exact mechanism by which this happens is not fully understood. Finally, future research could focus on analyzing the causes and consequences of the formation of clusters, coalitions, and subnetworks of actors in conflict systems, and how these different types of groupings influence a negotiation process