The fundamental question that third-party facilitators or mediators (hereafter, the mediator) should ask themselves at the outset is whether their intervention is indicated at all. However, perhaps because there is currently a distinct competitiveness or territoriality in regard to facilitating or brokering peace processes, this question is seldom asked. Best practice suggests that peace processes are more likely to succeed where the parties engage directly with one another and take ownership of the peace process and its products. But where belligerent parties are unwilling to engage with each other, collaborate in joint structures or even meet in the same room, third parties or mediators are often required. For this reason, this article attempts to identify key lessons learned about effective third-party practice during the early phases of engagement with armed groups and state parties in the context of a peace process. This article may also provide some useful insights to armed groups and state parties in terms of how to negotiate effective third-party roles or at least to appreciate the considerations of third parties.

Mediators should be cautious where it seems one or both of the parties to a conflict wish to engage in talks without any intention of reaching agreement. This phenomenon is common in a world where there is increasing international or regional pressure on parties in armed conflict to engage in negotiations. Motivations for engaging may derive only from the economic rewards of being seen to do so or the punitive conditions or consequences of failing to do so. Engagement may also be a military stratagem serving to bolster the parties' image or to secure a period of respite prior to another military manoeuvre. In these circumstances, a mediator must think carefully before engaging parties in an unworkable or potentially damaging peace process. While negotiations can build a party's commitment to political dialogue, it is also clear that the fallout of failed negotiations can prevent or delay a more serious peace process.

 Mediators should also be able to walk away from a process that has become dishonest or counter-productive. States whose international profile is closely associated with their successful participation in particular peace processes tend to make frequent third-party interventions in those processes, and it is often difficult for officials to walk away for fear of damaging their careers or their country's reputation.

An important corollary to this is that mediators have at least some independence in exercising a measure of control over the process and doing so may mean that the mediator acts more independently than the parties would initially wish. Failure to do so could lead to a chaotic process in which parties exercise unbearable pressure on the mediator. While it is crucial for a mediator to have the trust of the parties and their consent to act as a mediator, this should not be construed as an imperative towards appeasement. The parties will generally come to respect a measure of control over the process, provided it is exercised impartially and professionally. This needs to be asserted at the outset – even at the risk of 'losing' the mediation.

In the first engagements the mediator needs to establish and build confidence in his or her role, and accordingly the skills of listening, empathetic understanding and impartiality are important. It will be important for the mediator to respect the rules of confidentiality and insist that the parties do so too.

The mediator must possess knowledge of creative process options and the capacity to move from one form of negotiations to another: formal to informal, committee to sub-committee, bilateral to multilateral, one-on-one or side talks. In order to do this the mediator should, with the parties' consent, establish some responsibility for the orchestration of the negotiations including facilitating the preparation of an agenda. General Sumbeiywo managed this at the outset of the Machakos round of Sudanese peace talks. The failures to do so in the Burundi peace talks became a significant complication.

A good mediator is acquainted with the substance of the issues at the heart of the conflict as well as the nature of the possible solutions, and they have access to the necessary national, regional or international authority to secure the parties' compliance with their process requests. Preferably the mediator should also be able to guarantee (or know that others will guarantee) the resources necessary to facilitate the negotiations. Mediators may often need to rectify the imbalance in access to resources and knowledge possessed by non-state parties. In this regard, state or multilateral organisations may be more clearly indicated as mediators in certain conflicts than individuals or eminent persons.

External pressure on both state and non-state actors (i.e. threats to their resources, status, or pressure from regional powers) is critical but not decisive in inducing a state of mind conducive to participation in negotiations. What is more important than external pressures is the armed group's subjective appreciation of a negotiated settlement as the first prize – as something that can actually deliver on their bottom line demands.

Three important barriers to achieving such willingness to enter a peace process are: the legacy of previous bad faith negotiations; the belief that the adversary is unable to meet its bottom line demands; or the current imbalance in power between the two parties. Thus, the first task of a third party is to engage the armed groups (and the other parties) on the grounds that a negotiated solution is both possible and capable of generating a lasting and enduring settlement of the conflict. These negotiations are usually separate, unsolicited, preliminary, informal confidence-building exercises. The mediator, as in subsequent engagements, is required to share best and worst alternatives to a negotiated agreement, to separate positions from interests and explore possibilities.

A mediator should build the parties' understanding that their enemy is also their negotiating partner. The barriers, problems and constraints that affect one's negotiating partner are also likely to affect the success of the peace process. In this sense, recognising that both parties have some mutual obligation to assist each other in dealing with their respective constraints may encourage collaborative and constructive approaches and an understanding of each side's need to frame a solution in a form its supporters can digest.

It is critical that both parties appreciate the need for demonstrable gains so that they both can claim benefits arising out of the engagement and negotiations for their respective constituencies. Thus in South Africa the negotiations initially followed a tit-for-tat series of concessions that both sides were able to hold out to their supporters. It is important to recognise that peace is itself a 'bread and butter' benefit for both sides.

Armed groups need to turn around the slogans and battle cries which served them well during periods of mobilisation. 'No compromise' or 'victory or death' are not appropriate rallying cries for parties in a peace process. The African National Congress (ANC) needed several months to bring their cadres round to the possibility of a negotiated settlement.

There is a tendency for parties to personalise the causes of the conflict and to believe that the removal or destruction of the opposing leader will secure the conditions for peace (the so-called Savimbi option). The practice of undermining a rival leader's authority may well mean that an agreement will either not be concluded, because of accusations of bad faith (as in the so-called Karuna issue in Sri Lanka) or worse, the leader will sign a peace agreement without the support of their constituency or military commanders. It is thus important that parties respect each other's leaders and their legitimacy.

Initial engagements are not only about securing the armed group's entry, but involve the very nature and details of the peace process itself. Such negotiations involve establishing the role of the third party. Building trust in the process involves building trust in the mediator. Due to the intrigue and manoeuvering involved in establishing the Burundi talks, the Tanzanian facilitation team was on the back foot throughout the process.

Pre-negotiations: Rules and arrangements

The range of process issues that can be pre-negotiated is extensive and the more that is agreed the smaller the list of potential flash points that might threaten the process later. Such issues include: decision-making formulae at the talks; the extent and nature of representation of the parties at the talks; important logistical questions such as the venue and transport; the powers and role of the third party; the removal of pre-conditions to the talks; time frames; rules of procedure, etc. A comprehensive negotiation of these process issues with an armed group is also a form of preparation for the substantive talks to come.

Negotiating a peace process 'road map' can provide both clarity and confidence in the direction and trajectory of the peace process. In South Africa it was only after the conclusion of an agreement on such a road map (in a memorandum of understanding between the South African government and the ANC) that negotiations proper were able to commence. The failure to negotiate a road map at the outset in Sri Lanka has seen that process stranded in a state of uncertainty.

Logistical arrangements

An issue that may have disproportionate importance in the discussion of negotiation arrangements with armed groups is the equality of treatment and status of the parties. Whilst state parties are sometimes loathe to accord full and equal status to non-state parties and armed groups, the very process of engaging across a negotiating table is itself an equalising mechanism. However, sensitivities over equality may mean symbolic issues such as the number of bodyguards participants may bring, the motor vehicles they use, their hotel rooms and other issues of formal status may have more importance than might seem warranted to an outsider. There are also key issues of asymmetry between state parties and armed groups that need to be addressed, such as access to resources, finances and expertise, securing visas for travel, and any perception of inferior treatment in the process.

Pre-agreements

One reason why an armed group may be loath to enter into a peace process is the fear that it may not succeed in some critical issues in the negotiations. Certain assurances can provide confidence in the process and security to an armed group: for example, pre-agreeing some outcomes that affect neither party's bottom lines. Typically, 'binding fundamental principles' might govern the conduct of the parties during the transition, the outcome of any constitution-making process down the line, or simply guarantee the rights of minorities in a post-conflict society. Such binding principles were critical in South Africa and Namibia.

Agreeing on confidentiality, transparency and inclusivity

The mediator should ensure mutual clarity on rules of confidentiality so that the parties do not leak information about the process to secure certain off-the-table advantages. Breaches of confidentiality have frequently scuppered peace processes before they have started (for example the talks between the National Council for the Defence of Democracy and the Burundian government in Rome, 1998). Party-to-party engagement is generally best conducted in an atmosphere of candour and confidentiality. Yet failure to provide a measure of transparency on the substance of the talks can lead to disinformation and suspicion about the nature of the deals being struck between political elites. It is thus necessary for the mediator to balance the need for confidentiality within the process and transparency about it.

The conduct of the media, and the parties' use of it, can have an important effect on the overall negotiating environment: belligerent war talk, continued propaganda and disinformation can sour the atmosphere. The mediator can offer to regulate information about the peace process itself, and may raise the need to agree on rules regarding each party's approach to the media. It is also critical for the parties to be aware that there are multiple audiences on whom public statements may have quite different effects. Triumphalist reporting of negotiations to supporters may affect the constituency of the negotiating partner and undermine the capacity to keep to agreed compromises. While the participants at the table may allow some space to either side to 'talk up' the fruits of their participation, this may not be understood so strategically by the other's rank and file (as in Northern Ireland). In this regard media releases should address the sensitivities or insecurities of persons who are not at the table but who may have a critical role in destabilising the peace process.

In the interest of a durable and sustainable peace agreement the mediator should endeavour to point out the need for inclusivity especially in regard to any long-term political or constitutional implications of the agreement which affect people and groups not represented at the talks. It is not the mediator's duty to represent unrepresented parties, but for the durability and workability of the outcome of the peace process, a third party may raise the need for inclusivity in the processes and/or institutions created by the peace process. It is unclear whether the Norwegian facilitators did enough to persuade the government to adopt a bipartisan approach to the Sri Lankan talks. In any event, the absence of such an inclusive approach is the principal problem the process now faces.

An enduring pressure on the mediator from the outset is to ensure that the peace process builds and sustains a momentum. Momentum in this context means concrete achievements and deliverable benefits from the process, or an ever-broadening area of common agreement. In the initial phase, this may mean interim humanitarian access, or provisional ceasefires or even the agenda of the talks. In some situations the use of 'sunrise' and 'sunset' conditions can be used where issues of principle are agreed but implementation is delayed, or there is a phased removal of the disputed condition or circumstance. In both cases the armed group may be precluded from entering a peace process if it cannot at least demonstrate that a central underlying principle has been or will be addressed (e.g. withdrawal of troops, representation in state structures).

The most important test of a mediator's judgment is when and how to facilitate agreement by introducing compromises in these initial stages. In circumstances where it may be difficult or overly risky for a party to accept another party's proposal on issues of process and/or substance, or make one of their own, it is the role of the mediator to provide parties with 'neutral' compromise proposals or single texts from which both can negotiate. In this way, parties may be empowered to move past initial positions and onto discussing substantive issues. This is a high-risk strategy, especially if one party insists on using the mediator's agenda for the negotiations and the other party insists that it be rejected (Sudan). Moreover, the mediator stands to be accused of bias by one or both sides, as was the case in both Burundi and Sudan. However the compromise packages in both cases introduced the likely middle ground, which the parties, although unable to accept then, were later able to implement or refine.

There are several substantive issues that may arise in the initial stages of a peace process that require attention by the mediator, including ceasefires, amnesty and other important issues of international law, and implementation of agreements.

Cessation of armed hostilities

It is not uncommon for the parties to a peace process to insist on the continuation of armed conflict during the talks on the grounds that it is such armed conflict that has created the conditions and established the weight of the armed group at the table. This was the situation in Sudan until the Sudan People's Liberation Movement / Army overran the garrison town of Torit with disastrous consequences for the negotiations. In general the continued conduct of armed hostilities does not create an environment favourable to a successful peace initiative, as the capacity to negotiate can become hostage to battlefield fortunes.

The issue of whether and how to negotiate a ceasefire in the preliminary stages of engagement is a context-specific choice and difficult to discuss fully in this article. However, one critical consideration for a mediator dealing with the issue of a ceasefire at the outset of a process is that ceasefires can be notoriously difficult obligations to fulfill. Even though an armed group or state party may believe or wish to give the impression that it is capable of fulfilling the obligation, it would be most damaging for the process if one or other of the parties subsequently fails to live up to its commitments. In Sri Lanka the large and disproportionate number of ceasefire violations by the Liberation Tigers of Tamil Eelam is steadily eroding confidence in the ceasefire agreement. In this regard the third-party mediator may have some role in counselling the parties on undertaking unrealistic obligations.

Difficult issues and international law

Questions of amnesty, human rights, treatment of prisoners, torture, maintenance of 'no-go' zones, laying of mines, etc. will arise at an early stage in most negotiated settlements to armed conflict. The mediator's responsibility is to draw the parties' attention to the appropriate international law provisions, while conceding that the terms of the agreement will ultimately be the responsibility of the parties themselves. With issues of amnesty in particular, the mediator may simply point out that such provisions will have limited applicability outside their territories and may serve to undermine their status and international support.

Anticipating implementation issues

A mediator should anticipate putting implementation modalities on the agenda for negotiations from the start so that the parties are thinking about this issue throughout the process instead of scrambling to address it after agreement has been reached. Armed groups do not want to be held hostage to their adversary's cumbersome legal procedures and seldom respect the imperatives of legal continuity. Nonetheless effective practice requires that new rights and structures should be protected from challenges to their legality and this was the option taken in South Africa, Burundi and Sudan. In Sri Lanka and Nepal this option poses serious impediments to a settlement.

Training and resources

Parties who are experienced or trained in negotiation are more likely to engage constructively in a negotiation process. They are more likely to have the confidence to be risk-takers – and negotiations is a risk-taking business – and be bound by the discipline of negotiations. Government parties are not necessarily skillful in negotiating or understanding tactical compromises and how they can be used to effect concessions. Accordingly they are in equal need of exposure to negotiating skills. In South Africa the evident and unexpected initial disparity of negotiating skills between the ANC and the National Party was initially a source of some satisfaction to the ANC. This changed however when the ANC recognised that adversaries who are poor negotiators make for a poor negotiating process.

What is true for negotiating skills is equally true for questions of substance. It is in the mediator's own interest to arrange for or to provide training for the parties. In some instances joint training can be an effective first exercise in a collaborative engagement between the parties, as was successfully employed in selected technical areas in Sudan and Burundi. This however is unusual, and there are also good reasons for conducting training separately, not least the need for candid exchanges.

Getting good negotiating mandates

The mediator should, as far as possible, ensure that the representatives of the parties, particularly the armed groups, have a mandate to negotiate. Dealing with messengers from a leadership not present at the talks – and thus not subject to the discipline of the process and the logic of its compromises – is to be avoided.

Moreover, the negotiating teams should not necessarily include the highest level of leadership of the respective parties. The leaders of the parties can be left out of the tough and undignified bargaining that may take place in acrimonious exchanges at the negotiating table. Apart from questions of dignity and personal security there is the important objective of having recourse to a higher level of decision-makers once deadlock is reached. This process was effectively used in South Africa to create layers of deadlock-breaking mechanisms, while in Sudan it was the eventual presence of the highest level of leadership at the talks that secured real movement in the negotiations once the negotiating teams (with their very limited mandates) had reached a seemingly intractable impasse.

Managing internal divisions

Understanding the relationship between party leadership and their rank and file is critical. The South African negotiations took their rhythm from the need to allow party leaders to consult with their regular members. This was critical in order to engage popular support for a new social contract. This in turn guarantees that the leaders are able to deliver their members and supporters in any resulting compact. This is particularly true of armed groups whose negotiators need to keep close to their commanders, rank and file, and the various hard-line factions in the movement.

The mediator should always be alive to the need for the parties to have ownership and even management of the process. As the peace process unfolds it may be possible to achieve ever greater levels of collaboration and joint decision-making in regard to the management of the process. The failure to achieve this sense of ownership may result in a situation where one party is able to walk away from the table claiming that the process is a "UN" or "Norwegian" or "foreign" project. Establishing collegial management committees to arrange daily agendas or investigate problem areas is one way to increase ownership of both the process and the text. In this regard the mediator is increasingly required to adopt a background role. This is what happened in the Naivasha talks in Sudan. 'Ownership' is not a soft 'feel good' consideration. It is about the hard realities of implementation and sustainability. Throughout the engagement process mediators must remember that the process must belong to the parties and that they are assisting the parties with their process, not the other way around.