It has taken three peace accords – and, in February 2002, the killing of Jonas Savimbi – for the arms to be silenced for good in Angola. There are several ways to look at this trajectory, extended over 12 years. The simplest observation is that the two first agreements – those of Bicesse in May 1991 and Lusaka in November 1994, both reached under the auspices of the international community – resulted in a resumption of the war, with ever more deadly consequences for the civilian population; while the third one – the 2002 Luena Memorandum – in which the international community played a minimal and largely symbolic role, not only succeeded in formalising a cessation of the fighting but effectively brought an end to the cycle of wars that has devastated Angola since its independence in 1975. This is how the Angolan government has chosen to portray the course of events, while the international community has opted to see Luena as also resulting from its patient efforts to bring peace.

How should these failures and this success in reaching peace be interpreted? Looking back, the answers seem to lie in the combination of two factors: a deadly pursuit of military victory and hegemonic power by the two warring parties, and a situation of 'too many interests' among the 'real' international community. In Angola the interests of this 'real' international community of great powers and transnational corporations have always provided the context for and strongly influenced the attitude of the 'official' international community (the United Nations); this was the case firstly with respect to the Bicesse and Lusaka accords and their implementation, and later in relation to the 'real' international community's support not just for the Angolan government but also, silently, for the military option, and the abandoning of any attempts at negotiations.

These interests provide the basic explanation for the shortcomings and failures of international interventions. Although they certainly changed over time, they remained constant in reinforcing the sidelining of the needs for peace and democracy of those forces that one may call 'unarmed' – the Angolans who did not recognise themselves in the hegemonic intentions of the warring parties (including a number of members of those parties). These interests made it possible for both 'armed parties' to treat society as they wished in their attempts to achieve total power by any means.

This article will briefly indicate which issues were at stake for the national and international players with regard to the Bicesse and Lusaka accords, and how they developed and changed following the end of the Cold War. It will also assess their impact on the 'peace process', up to the point where it was replaced by a 'war process'. Ultimately, the precious result of peace was achieved, but the way it came about has obvious implications for the very nature of this peace.

Foreign interests have played a crucial role from the war for independence onwards. Yet they themselves did not create the divisions within Angolan nationalism – divisions that stem basically from a power struggle between different elites – although they certainly exacerbated them. During and after the great civil and international war of 1975-76 that brought the Marxist Popular Movement for the Liberation of Angola (MPLA) to power, Angola became embroiled in foreign interests, both regional and international. For essentially strategic reasons, the importance of this 'regional conflict' for the two patrons of the Cold War explains the war's ferocity.

It also explains the situation and the state of mind of the main protagonists at the start of the negotiations in the late 1980s: both had been able to wage a 'rich man's war', thanks to the resources at their disposal – oil for the government; and the military, political and financial support of their allies for both. Although very different, both possessed the means to dominate society and did not need to care much for the population. They were reluctant to democratise and had no inclination towards mutual accommodation. In particular, the Union for the Total Independence of Angola (UNITA), formed by the war and a totalitarian culture, was in a victorious mood, because it had forced its enemy to deal with it, and 'its side' had won the Cold War. It believed in force as the ultimate means of gaining power. As for the MPLA, under the mantle of a militarised party-state, it had for some years been involved in a kind of 'wild socialism' based on arbitrariness, privilege and massive corruption by the ruling clique, and the abandonment of the people to growing misery. It had been hugely discredited even in the eyes of its own supporters, making a failure of every attempted economic reform and rejecting any political change. Its power rested on widespread opposition to UNITA or Savimbi and the control of the country's oil riches and the resources of the state, and thus on the maintenance of the one-party system.

The first accord, signed in New York in December 1988, was an international one. It did not tackle the internal dimension of the conflict nor even entail a cessation of some of the ongoing foreign interventions: indeed, surprisingly for an agreement aimed at solving a 'regional conflict', it did not include a single clause on ending the support from the USSR and the US to either side. It took place in the context of the final decline of the Soviet Union and while the US (the mediators of the agreement despite being themselves a crucial party to it) had the upper hand internationally. Under mounting military pressure from UNITA and with the international tide against it, the MPLA reluctantly entered into negotiations.

The peace accord signed in Bicesse in May 1991 closely resembles other accords of that time in many of its features, such as the objective of democratising the political institutions and deciding between the two warring sides through an electoral process preceded by a 'transition'. Obviously, the Angolan players, the MPLA in government and the UNITA rebels, were bellicose and dictatorial, but no more so than their counterparts in places like Cambodia or Mozambique. What makes the Angolan case different is the interplay of national and international issues at stake. This also helps to explain why there were no corrections or adjustments to the intrinsic shortcomings of this general model, common to the 'first wave' of peace processes, which could have prevented the bloody failure of Bicesse.

Official negotiations opened in Portugal in 1990, under the auspices of a Troika composed of Portugal, the USSR and the US, with the latter pulling most weight. The negotiations forced the MPLA to abandon Marxism-Leninism early on and adopt a multi-party system, and eventually, in May 1991, and despite remaining 'the government', to sign a peace agreement with a UNITA that was recognised as a 'party' on the same footing as itself. The accord forced the MPLA to hold elections after a transition period during which the demilitarisation of the two forces and the formation of a single army should take place, to ensure that the vote would be respected.

The unarmed actors in Angola ('moral', political, civic, etc.) had no role in the negotiations or any say in the implementation of the accords. At the time, the principles of conflict management did not generally acknowledge them, as it was more a matter of using electoral means to settle the scores left by the Cold War. However, in Angola, this scheme was more complicated, and resulted in an even more peculiar situation: although the UN had been involved in the successful implementation of the New York Accords, it was not called upon until the very end of the negotiations. Bicesse came at a time when there was a lot of talk about a new international order and when the UN, after the Cold War, could regain its prominent role. Yet the international organisation remained marginalised in the negotiations and in the process established by the peace agreement: in the text itself, it merely had the status of 'invitee'! The Troika, on the other hand, installed itself in the driving seat of the process, the CCPM (Joint Political and Military Commission), and thus the interests of the three member countries and even more the balance of power between them, greatly in the US's favour, prevailed over the UN. The influence of the Troika could thus continue to dominate up to the end.

Neither of the belligerents with whom the Troika and the UN were to share the responsibility for the peace process had yet abandoned its search for hegemony.

Neither the MPLA nor UNITA were interested in reconciliation or democratisation. Above all, the MPLA wanted to avoid defeat, and it had only accepted certain conditions reluctantly and under pressure. UNITA only wanted peace because it was certain – as was almost everyone else at the time, including the MPLA – that it would win the elections, and achieve its aim of gaining state power. The three countries 'managing' the accords could not have had any illusions. In such circumstances, one can see the importance of the international community's role during the transition period if it were to succeed in establishing lasting foundations for peace.

What, then, can be said of this accord, described as exemplary by some of its international promoters, and greatly welcomed by the Angolan population; and why did it fail?

The failure can be analysed in different ways. The external 'crafters' retrospectively pleaded that they lacked a good understanding of the two Angolan sides, particularly of the 'loser' in the process: UNITA. Yet this is only credible with regard to the UN and its belatedly appointed Special Representative, Margaret Anstee. She was also the first to accept the international community's responsibility, but emphasised the lack of resources (people, money, mandate). The conclusion she reached was serious: the UN should never have conducted a peace process under such conditions. But why, then, did the external actors behave so irresponsibly? An analysis of the reasons for their intervention in the Angolan peace process in the first place may shed light on their establishment of a process that was 'not quite' identical to others in that period and on its failure. In particular, it will also explain why corrections and adaptations that could have avoided its eventual collapse were not made during implementation.

The transition was placed entirely in the hands of the two armed parties. Transitory political rules were not established, nor was a coalition government (which would have avoided the dangers of a 'winner takes all' outcome) decided on for the period following the elections. The parties themselves, and particularly an influential UNITA, rejected the option of a coalition government before the elections that could have secured minimum standards of impartiality in preparing the elections.

Bicesse incorporates features of other peace agreements of that time, features that have elsewhere caused some difficulties during their implementation as well as in the post-election period. But in Angola the problems lay beyond the weaknesses and ill-conceived ideas of this first wave of peace processes. In fact, for the US (with the consent of the two others), peace was not the first and only aim. The peace process was perceived more as a route for UNITA to come to power. This outcome, almost taken for granted, would be achieved with minor political and financial costs, thanks to the central role attributed to the Troika. That is mainly why the UN's mission was so marginal, and the means at its disposal so derisory, especially in comparison with its contemporaries in Namibia and Cambodia. That is why an early date for the elections was set, despite being a totally unrealistic time frame for the completion of all the necessary tasks. That is why there was indifference about the type of pre-election government, enabling the MPLA to paradoxically maintain its grip on government (total in relation to any other Angolan actor, and only mitigated by the clauses and organs of the peace process) until the elections, as UNITA did not want to risk losing the credit it had by taking responsibility in government. The only thing that mattered to the US was to 'accompany' the victory predicted for UNITA, while for the other international actors the involvement of the US was a sufficient guarantee that events would run smoothly.

The structure of the agreement enabled the different parties to 'preserve their current status'; out of government, UNITA had no negotiating power other than its military force, while all the reins of transitory power and resources of the party-state remained in the hands of the MPLA. The international community did not push either side to change during the implementation of the agreement. UNITA maintained control over some of its areas and 'its' people and had no intention of losing its only asset by disarming, and the international community turned a blind eye. Meanwhile, noting the international community's lack of interest in its democratic obligations, the MPLA quickly mobilised its forces to avoid losing everything by losing the vote: it went on to victory thanks to its access to funds, total control of the administrative apparatus and the state-owned media, the mobilisation of legal and illegal resources, and the establishment of a paramilitary force. Violations multiplied and went unpunished; amid the laxity of the international community, a logic of radicalisation set in. This provoked growing scepticism within the population, transforming UNITA's assets (its arms and the 'culture' associated with them) into liabilities: its arrogance, its blatant retention of its weapons, and its thirst for revenge. Coupled with the enormous efforts (in expenditure and means) of the 'party in power', these factors rallied support to the MPLA that it did not have at the time of Bicesse. In these circumstances, the postponement of the elections would have been desirable (which is what the UN subsequently did in Mozambique), as the minimal political and military conditions for them to take place and be respected had not been met. Yet the opposite decision was made. Regardless of the fact that the UNITA army had not been dissolved and that the government had set up a new special police force, the date initially agreed for the elections was considered untouchable. Thus, the UN solemnly declared the two armies dissolved, put a so-called single army in place and went to great lengths to accelerate the process of voter registration.

The elections took place at the scheduled time and gave the MPLA a clear victory (54 per cent of the votes) over UNITA (34 per cent). José Eduardo dos Santos (49.7 per cent) recorded a smaller and insufficient victory over Jonas Savimbi (40.7 per cent). The UN – whose credibility was now increasingly questioned – finally labelled the results "generally free and fair". A significant number of voters believed the accusations of fraud made by UNITA and other parties, even if on the part of UNITA's leadership the claim represented a denial of any possibility of defeat. Of course, those who voted MPLA demanded that the ballot be recognised, and – as others also still wanted the vote to be respected, and the war to be avoided – the government easily managed to mobilise and galvanise its supporters with this UN backing. Powerless, the international community tried to negotiate at least a modification of the conditions for the second round of the Presidential vote, to make the playing field less unequal. But UNITA had already put itself on marching order throughout the whole country and was using this as blackmail and preparation for war. For the MPLA, it was unthinkable to be held back on its journey to victory, or, now that it had national and international legitimacy, to consider any kind of power-sharing.

The events afterwards were and still are very contested but documentary and oral evidence allows them to be described as follows: while the discussion over these conditions was taking place, and UNITA was building its war machine throughout the country, the government denounced an attempt by UNITA to take power in the capital city and organised a 'pre-emptive coup' in Luanda and several provincial capitals. As UNITA had legal and illegal weapons in Luanda, there was heavy fighting, and in three days thousands were reportedly killed. An enormous majority of the dead were connected to UNITA and included a number of the organisation's Luanda-based politico-military leaders as well as soldiers and civilians (militants and even ordinary voters).

Having failed to ensure respect for the outcome of the elections, the international community simply ignored these killings in an attempt to restart negotiations. But an armed power struggle had already broken out – for the 'ratification' or the 'rectification' of elections that had been viewed by the two contestants simply as the final battle for ultimate power. And the international community was at that point in no condition to stop those confrontations from ushering in a resumption of the war. It chose on one hand to confirm its past actions, i.e. the elections and their outcome, and thus to hold UNITA responsible for the war, and on the other to try in vain again and again to present the two parties with new proposals for the resumption of negotiations (these early proposals being the basis of those which would finally be accepted at Lusaka).

The war unleashed tremendous violence. As UNITA had remained armed while the government army had disintegrated, and with paramilitary forces clearly insufficient to face the rebellion, it initially gained considerable advantage. As long as UNITA retained military superiority, it rejected any new terms of agreement proposed by the international community. But times changed, and the government used its increased legitimacy to urge the international community for support. The end of the Cold War and the discovery of considerable offshore oil reserves that had hitherto been exploited in partnership with the MPLA government were strong additional factors in its favour, while it seemed clear that UNITA would be unable to maintain control over the capital even if it won on the battlefield. It was thus time to take this new situation into account in all respects (legal status, business prospects). The US, fully supportive of UNITA until the elections, now gave the signal for change. When by mid-1993 UNITA again rejected an accord (the Abidjan Protocol), the US finally recognised the Angolan government, opening the way for UN sanctions on UNITA. This general re-alignment in favour of the 'legitimate government' led to a gradual reversal of political, diplomatic as well as military forces in its favour.

Under pressure from the international community and from a reversal of military fortunes, UNITA issued a communiqué in October 1993 reaffirming the validity of the Bicesse Accords, opening the way for talks between the two sides in Lusaka in November. Over the next twelve months, and amidst continuing heavy fighting (and much death), the two sides conducted talks facilitated by a team led by new UN Special Representative Alioune Blondin Beye and representatives of the Troika.

In November 1994, after a series of considerable military setbacks, UNITA (but not Savimbi personally) was forced to sign the Lusaka Protocol. This agreement did not however constitute the ratification of a defeat: the international community ensured that military defeat was avoided, and had learned some lessons from the failure of Bicesse. First, the vanquished party should have a place in power for the outcome to be politically acceptable. Second, the armed factions involved should not have the military means to change the course of events. And additionally, disarmament should be taken seriously and adequate resources allocated to it. The outcome of the elections was not reconsidered, and only the second round for the Presidency was on the agenda.

In the context of the election results, UNITA was now considered an illegitimate rebellion, thereby ending the symmetry between the two parties that characterised Bicesse. Only UNITA was urged to disarm and demilitarise in order to be integrated in a government army left intact by the accord. The quasi-exchange that structured the accord – disarmament for participation in the government – conformed to this change in legal status and to the lessons learned. Two other elements further increased the government's legitimacy. Firstly, despite the agreement, the sanctions imposed in 1993 to compel UNITA to negotiate were not lifted. Secondly, the 'triple zero' clause contained in the Bicesse Accords (repeated in an annex to the Lusaka Protocol) – a clause forbidding both parties to re-arm themselves and other countries to supply them with arms – was not part of the Protocol itself. This meant that, in terms of the text of the accord and of the 1993 UN resolution imposing sanctions, re-armament constituted a real breach for UNITA, while it was no longer a problem for the government, at least not in legal terms. Furthermore, there was nothing in the texts of Lusaka prohibiting foreign countries from re-arming the 'legitimate government'. In other words, there was not just a structural asymmetry, but also an actual disequilibrium of rights and obligations with respect to the crucial issue of demilitarisation.

While this agreement, with its somewhat improved mandate and resources for the UN than at Bicesse, could have helped avoid the failure of 1991, it still exhibited important shortcomings. It gave no role to the unarmed forces – those not part of the struggle for power, who were the ones primarily concerned with peace. And although the UN was attributed a larger role, the Troika was still at the heart of the operation. Most importantly, the agreement was signed in a context of absolute mistrust between the parties, and it overlooked the fact that both parties were utterly determined not to abide by the rules of the game if necessary. The UNITA leadership still believed in the force of arms. The elimination of its leaders and supporters in 1992 only served to further radicalise it politically and consolidate this conviction. It would disarm neither in advance nor unilaterally, as long as its survival and political position were not secured as it wished. For Savimbi and a part of the leadership, this meant being in power. As for the government, for two years of war it had succeeded in functioning just as before even under a multi-party system (and had soon after elections created a so- called but hollow 'government of national unity' with small allied parties). It therefore completely rejected the notion of being held accountable and only agreed to share power formally with UNITA in the Government of National Unity and Reconciliation (GURN) set up by Lusaka so as not to alienate the international community. Under the new and very favourable circumstances, it embarked on highly profitable but predatory and opportunistic business practices. Despite the war, it managed to attract partners interested in oil, war imports and any other viable trade. These practices of wild enrichment and corruption went unpunished and were accompanied by utter and increasing misery for the general population.

Now forced to rely solely on the diamonds under its control to preserve its military apparatus, and determined not to disarm, UNITA systematically procrastinated and violated an agreement that it deemed unfavourable and hoped to be able to renegotiate on the basis of its continued military strength. Abusing its position, the government subverted and bypassed the Protocol and did not fulfil some important obligations with regard to the police and the military. In the words of a UN official, "UNITA violated the agreement by day, the government by night". These unpunished infringements fuelled mutual distrust and led to re-armament on both sides. UNITA did this secretly with the help of arms and diamond dealers, as well as some remaining friendly governments. Initially, the government bought arms through illegal or covert channels too, but soon it did business and cooperated with governments, albeit in a discreet manner. The main preliminary condition of the peace process remained the disarmament of UNITA. Of course, the international community 'understood' – given the nature of the Angolan political economy – that UNITA needed both economic and political guarantees before it would disarm. Partly for this reason, it twice accepted the notoriously false 'declaration' of disarmament by UNITA, as it also chose, despite all evidence to the contrary, to accept that the GURN really was a government of 'national unity and reconciliation'.

At its formation in April 1997, this government was 'united' and 'reconciled' only in name. It contained 'UNITA members' who had been co-opted, through the terms of the agreement, into a government in fact led exclusively by the MPLA. The power-sharing was therefore as fictitious as the demilitarisation of the rebel movement. While numerous heads of state once again praised the step reached with its inauguration, the Angolan population did not celebrate this time. It knew its masters or its enemy, and no longer had faith in the international community. Given the structure of the accords, with their fundamental imbalances, and the implicit but notorious agendas of the 'Angolan parties', one would have needed a highly political as well as a resolute and balanced international community to avoid the hardening of the impasse. The situation turned even sourer when the government decided to intervene militarily in the neighbouring Congos (moves prohibited by the Protocol) to help bring friendly governments to power, without provoking an international reaction.

Both parties undeniably prepared for a new confrontation and tensions on the ground increased. In 1998, the government judged that it would be politically, diplomatically and militarily capable of waging a real war. Recognising the failure and futility of its policies, the international community admitted that both sides were violating the terms of the agreement, and reiterated that the primary responsibility lay with UNITA for not disarming in the first place. Without the power to stop the downward spiral of militarisation or the violations by either side, this reiteration had absolutely no implication for preserving peace. It did have implications, though, for increasing the political legitimacy of the government. In June 1998, the Security Council reinforced sanctions once again, this time to include diamonds, the nerve of UNITA's war machine. Finally, at the end of 1998 the government launched what it called its 'war for peace', a war against an enemy "who had never respected the accords" and whose leader, Savimbi, the government did everything to have internationally criminalised.

It is impossible to understand the government's objectives in this 'last' war without considering its 'internal' policies. Four years of 'neither peace nor war' had enabled the MPLA, which dominated all the 'democratic institutions', to intensify the pillaging of public resources with impunity. In partnership with foreign interests and under the auspices of a Presidency with reinforced powers, it was able to partly privatise them for its own profit. There was in fact intense international competition for the benefits firstly of oil and also of all other viable business. But this exercise in clientelismo, which has as its corollary the poverty and the necessary absence of rights (even those recognised by law) of the overwhelming majority, was only possible if real democratic processes were 'neutralised', to prevent opposing political forces from capitalising on the growing dissatisfaction. To the extent to which the MPLA controlled all the wheels of the state, the public and private economy and the media, the 'unarmed opposition' was neutralised, caught between co-option, repression and impotence. This left the armed opposition of UNITA. The ultimate aim of the government's military strategy was to neutralise UNITA politically, since a militarily annihilated UNITA would be unable to negotiate, and inevitably therefore, incapable of weakening the control of 'the party in power' under the pretext of a transition process. A military solution would thus be most effective in creating the conditions in which the MPLA alone could determine, firstly the very terms of peace, as well as the time frame for elections and the political, economic, social future of Angola, without encountering the 'normal' threats to its system of power that peace would bring.

To achieve this, the government could rely on the fact that it was the legitimate power and that the rebellion was refusing to disarm. To the extent that the international community was both guarantor of this legality and of the accords, and indifferent to the realities of the MPLA's governing practice, the government could count on its close ties with a series of powerful foreign partners and on the acquiescence of an important section of the 'real' international community in its war – involving substantial political support, but also direct though discreet military support from certain friendly countries. But it wanted more: in pursuit of its ambitious – and ultimately political – objectives, it also needed the official international community to confer legitimacy on the war and abandon all attempts at dialogue. The UN did not completely follow this line, refusing to recognise UNITA Renovada – formed by dissidents claiming to 'replace' Savimbi's UNITA, but with no autonomy from the regime – or to declare Savimbi a war criminal. But it did break off all contact with UNITA, whose foreign missions were ordered to close, and the Security Council's Sanctions Committee even attempted to cut off its means of political expression.

Anxious to play a role in a future process to end the conflict and not to leave Angola as a failure, and consequently concerned not to antagonise the government, the UN abandoned its mandate – to search peace through negotiations – and even ceased its efforts to deliver humanitarian aid to areas under UNITA control (its obligation under international humanitarian law). Sanctions were enforced with remarkable tenacity, in particular from 2000 onwards, after the first military victories by the government, and justified by continued reference to the 'principal responsibility' of UNITA for the impasse, and by the very sanctions agreed. These soon began to constitute a real contribution to the government's war effort, making it difficult for UNITA to get supplies and forcing it to 'survive by its own means'.

The result was first a humanitarian tragedy, with heavy use of a scorched-earth policy either as a means of survival by UNITA or to achieve military victory by the government. After the greatly desired elimination of Savimbi and the military victory it reinforced, the result was also a 'make-believe' negotiation. This did not give way to any kind of political transition, and thus, as desired, the government retained sole control over the future of Angola, with the most favourable terms it could hope to secure for itself in the context of peace.

In the early 1990s, the international community chose to ignore the hegemonic interests of the two parties and the militarisation of a UNITA that wanted power at all costs. At the end of the decade, it was the same indifference to the needs and aspirations of Angolans, and then towards the military and predatory nature of governance in Angola, that gave free rein to the game of economic and strategic interests. These interests had not weakened, but rather reoriented themselves and, once in conformity with the election results, inclined more heavily to the government's advantage. But what remained at the end of the process in the eyes of many Angolans was the powerlessness and disrepute of the UN: once the 'political UN' was no longer of any use to it, the government's camp strongly reproached it for not having done enough during or after the war, and for wanting to meddle with its 'sovereignty'; UNITA condemned its partiality during the second period (but avoided mentioning the first). The civic forces that had mobilised to bring an end to the war by peaceful means, but with whom the UN did not engage, resented having been abandoned in their efforts to reach a goal that should have been shared by the international organisation. The result was not only harmful to the UN. More seriously for Angolans and Angola, it also meant that the minimum conditions for a militarily achieved peace to be converted into democratisation and a just and durable civil peace were not met. The foreign investors and partners, and their governments (the 'real' international community) were not too concerned: they were not bothered at the height of the war and there was now sufficient stability for their business requirements, not to mention the fact that the reconstruction of the country provided a huge opportunity to make even larger profits.