When Thabo Mbeki and Chris Hani travelled together to Moscow in 1988 to take a short holiday, they were asked, separately, by the Soviet expert on Southern Africa, Vladimir Shubin: "When will you win?" Hani replied: "Ten years more." Mbeki said: "We shall be home in 1990."

Mbeki and Hani were thus in very different places indeed on February 2 1990, when F W de Klerk announced the unbanning of the liberation movements. Mbeki was in Europe, en route to a secret meeting with senior SA intelligence officials at the Palace Hotel in Lucerne, Switzerland, while Hani was in Lusaka, taking part in a meeting of the movement's Politico-Military Council secretariat. When the meeting was interrupted by an official bringing in a report of De Klerk's speech, just received over the wires, Hani was the first to respond. "Nothing has changed," he said. "We need to infiltrate more cadres into the country."

In England, Thabo Mbeki could not have disagreed more. Watching Nelson Mandela's release on TV a week later in the company of Willie Esterhuyse, a Stellenbosch professor and former Broederbonder who was his main intermediary with the SA government and who had set up the secret Swiss meetings, he drew on his pipe and said: "Now there is hope for the country. Now there will be peace."

The difference in the perspectives held by Mbeki and Hani represent the profound disjuncture in consciousness and strategy that characterised the liberation movement in the '80s. The ANC might have talked about "armed struggle" and "negotiations" as two equally balanced pillars of the struggle against apartheid, but neither the Mbekis nor the Hanis believed it, and the dissonance was never resolved.

The schizophrenia is perhaps best characterised by Mbeki's own actions in April 1989, when he led a delegation of ANC officials to an Aspen Institute conference in Bermuda. Also invited were officials of the SA government, and Mbeki held informal discussions with prominent verligte National Party MP Piet Coetser about the possibilities for negotiations.

Once the conference was over, Mbeki boarded a plane and, flying over a veritable Bermuda Triangle of parallel realities, disembarked, ultimately, in Havana. Here he chaired the Seventh Congress of the SA Communist Party, which was to adopt a strategy - of which Mbeki had been on the drafting committee - entitled "The Path to Power", which called for mass insurrection within South Africa. Almost as an afterthought, "The Path to Power" conceded that armed struggle and negotiation could go hand in hand - but insisted that until "the enemy is prepared to talk . . . sights must be clearly set on the perspectives of a seizure of power".

If anyone knew that the enemy was prepared to talk, it was Thabo Mbeki. He had been talking to it just days before in Bermuda, and he had been engaged with it through intermediaries since a Quaker professor from the University of Cape Town, Hendrik "HV" van der Merwe, had visited him in Lusaka in September 1984. He had been speaking to cavalcades of white South Africans - businessmen and Broeders, Afrikaner dissidents and rugby bosses - almost continuously for three years. How on earth could he, in good faith, chair a congress that was to adopt a policy as out of step with reality as "The Path to Power"?

"The problem at the time," Mbeki explains, "was that there was a strong difference of opinion in the ANC about the possibility of [anything] other than an armed seizure of power . . . There were some people who were not only sceptical but hostile to the idea. They saw it as selling out, treachery . . . you couldn't convince them about the fact that in reality the struggle was evolving away from an insurrectionary path."

Realising that he could neither counter the powerful insurrectionist views led by Joe Slovo nor jeopardise the possibility of negotiations by revealing just how far along they were, "the only thing I thought you could do was [to] run these parallel paths, and one of them would lose, inevitably. And I knew the insurrectionary one would lose."

It is precisely this canniness that makes Thabo Mbeki so complicated a figure: on the one hand, he was perceived as being cagey, somewhat deceitful and cynical (by letting the SACP take a militant line so that it could play a certain role and perhaps even discredit itself), but on the other, he chaired the congress so successfully that when it came to voting for the Central Committee at the end of it all, he was elected in first place.

Mbeki now says that the thinking of people like Chris Hani "would revolve around the military struggle" because that was all they were involved in, whereas "some of us would have been exposed to broader [things], to the entire scope, really, of the struggle". Because of his interactions with white South Africans, he understood - as his comrades could not - just how discredited apartheid was becoming in the eyes of the very people who were meant to be buttressing it; thus he knew that its days were limited.

There is something quite premeditated about the way Mbeki set out to win the friendship and loyalty of powerful white South Africans. In New York in June 1986, after a Ford Foundation conference, Mbeki had a drinking session with Broederbond chairman Pieter de Lange that landed up with this paragon of Afrikaner conservatism committing himself to working for national reconciliation on his return to South Africa. Both De Lange and Mbeki are pipe smokers, and Patti Waldmeir, in her seminal book Anatomy of a Miracle , reports that when De Lange's pipe lighter broke down, Mbeki was able to offer his, a gift from North Korea.

The event should go down in the annals of SA reconciliation symbology as even more significant than the fly hook in Roelf Meyer's thumb removed by Cyril Ramaphosa's wife, for Mbeki's pipe became the ANC's most potent negotiating tool: it not only sent up into the atmosphere smoke signals of civility and reason, but it also created a fragrant haze behind which Mbeki could operate with minimal detection.

Whether with Premier chief Tony Bloom, Frederik Van Zyl Slabbert, Willie Esterhuyse or the 50 Afrikaner intellectuals at Dakar in August 1987 (where Mbeki famously declared: "Ek is n Afrikaner "), the modus operandi was one of seduction; the tools charm, intellect and whisky-soaked camaraderie.

It worked. Echoing dozens of others from across the ideological divides of white SA society, Willie Esterhuyse came back to Pretoria after a meeting with Mbeki in December 1987 and told his handlers at National Intelligence that "I'd be prepared to entrust my life to this fellow". He remembers, with a chuckle, "that they were rather upset ".

They were not the only ones. An examination of the minutes of ANC meetings from the late '80s reveals the frustration and annoyance of Mbeki's colleagues with the way he was working. At a National Working Committee meeting on February 22 1988, Chris Hani erupted: "On whose authority has Cde Thabo entered into discussion with these Afrikaner intellectuals? Does the National Executive Committee or the NWC know anything of this? It is very disturbing that a member of the NEC leaves to hold discussion with Afrikaner intellectuals without prior consultations . . ." His comments were received by "general acclamation" and repeated by several people - including Mbeki's own supporters - in NEC meetings over the following two years.

On October 20 1988, after Mbeki had organised a delegation to meet Louis Luyt and Danie Craven in Harare, Joe Slovo registered dissatisfaction that no prior discussions had been held, thereby rendering the NEC "irrelevant ". Pallo Jordan put it more strongly: "A pattern [has] emerged during the year where powers of the NWC are being usurped . . . No small group [has] power to usurp the power of the NWC."

The very cornerstone of Mbeki's "talks" policy, of course, was that it was unmandated: it had the support of Oliver Tambo, but did not seek a mandate from ANC structures because Mbeki and Tambo knew that such a mandate would never have been granted.

And so Mbeki, with Tambo's consent, kept several processes going simultaneously: the large, public meetings with white South Africans - as at Dakar - provided a screen for covert interactions with representatives of the SA government. Meanwhile, while Mbeki engaged in quiet talks with white South Africans along with a small group of trusted comrades, Tambo edged the broader movement toward an acceptance of negotiations - a process which culminated in the Harare Declaration of August 1989, in which the Organi sation of African Unity threw its weight behind the ANC's preconditions for negotiations.

Mbeki's greatest contribution to the history of the SA liberation movement can be characterised thus: that he nudged the ANC away from armed struggle even as he drafted the eloquent resolutions that affirmed it. It was a double act that the orthodoxies of struggle forced him to play. His supporters will say that the times and the work at hand required such deceit; his critics will say that his character was suited to it, and that it could have been done differently. "It's not that we were opposed to the fact that Thabo was talking to the enemy," says one, "it's that we were opposed to the fact that he was not talking to us!"

One way or the other, his approach made him unpopular with many of his own comrades and led to his replacement, in 1991, as head of negotiations by Cyril Ramaphosa.

In the absence of Oliver Tambo, it was Thabo Mbeki who led the ANC from exile in May 1990, but, as always, his position was ambiguous. On the one hand, he was unpopular among the "militants" - led by Joe Slovo and Chris Hani - for the concessions he was making to the government (for it was he who was leading the "talks about talks" over the release of political prisoners and the suspension of armed hostilities), and for his perceived inability - or unwillingness - to carry the mass movement along with him. "The sense we got of Thabo," says one former leader of the United Democratic Front, "was that he was a corridors-ofpower' person who had something of a contempt for the masses and their capacity to move things forward."

And yet these difficult and inchoate times, as the ANC was trying to reconstitute itself, provided the backdrop for Mbeki's moments of greatest - and most appreciated - courage. In December 1990, against the tide of popular opinion, he managed to convince an ANC consultative conference on negotiations that it had to suspend armed struggle as per the Harare Declaration. Six months later, at the ANC's first national congress in Durban in July 1991, late into the night, he persuaded a hostile audience to agree to a phased lifting of sanctions.

This was a triumph of his quiet, articulate persuasion over populism. Interestingly, it was a moment he approached somewhat diffidently: he had to be persuaded to take over from one of his International Affairs officials, Stanley Mabizela, who was not articulating the position particularly well and was failing to convince the delegates. Mbeki unwillingly agreed to recapitulate the argument: he spoke to an utterly silent audience for over 20 minutes, and when he concluded even his fiercest opponents, Slovo included, were on their feet.

But at the very same conference his ambitions for leadership were thwarted.

Although Mbeki had been nominated for the position of deputy president of the ANC by four out of seven regions, Chris Hani made it clear that he would oppose him at the ballot. Wishing to avoid a "divisive" battle between the two, senior ANC leaders attempted to persuade both to stand down, but Hani signalled that he would do so only if Mbeki also did. The message was clear: he would not accept Thabo Mbeki as his deputy president. The deadlock was eventually broken when Walter Sisulu agreed to take the position. Mbeki nonetheless came second to Hani in the vote for the NEC - and Hani then came second to Mbeki in the subsequent vote for the NWC.

In the same elections, Cyril Ramaphosa was elected the party's new secretary-general. The key players in Ramaphosa's recruitment were not internal people, as is often believed, but returned exiles like Joe Slovo and Mac Maharaj. This was partly to bring UDF leadership back into the ANC, but also, very much, a move to unseat Mbeki as leader of the negotiations process, because they believed the former union boss would be both a tougher negotiator and more accountable to the people.

Many from the exile community believe, however, that the real reason was that powerful leaders like Slovo and Maharaj felt that because Mbeki was impervious to their patronage, they needed another young leader to take under their wing.

What followed can only be described as a palace coup. At an NWC meeting in the first week of August 1991, Ramaphosa proposed new deployments in the ANC. Nelson Mandela was out of the country on a visit to Cuba, and both Thabo Mbeki and Jacob Zuma - the new deputy secretarygeneral - were attending a conference at Cambridge, in Britain. In their absence, the NWC approved the proposals, which were then ratified by the NEC, and Ramaphosa replaced Mbeki as the ANC's chief negotiator.

There is a disagreement as to whether Mandela knew this would happen. Senior ANC sources insist that it could never have occurred without his consent, but Mbeki says that "the person who was particularly upset about it was Madiba. He said, how can you take important decisions like that about how to deploy people when I'm out of the country, Thabo is out of the country and Zuma is out of the country, just take decisions, don't consult, [and] proceed to exclude people at the centre of the process?'"

This was perhaps the most difficult moment in Thabo Mbeki's political career - he had just seen his power taken away from him by the very people who had opposed him for wanting to negotiate in the first place. He and Zuma (who had been replaced as head of intelligence by Patrick Lekota and given administrative responsibilities at Shell House) decided not to fight the decision and prepared to exert their influence instead from the ANC's negotiations commission, which Zuma chaired, and from the NWC, which had to rule on all major negotiations policies. From this back room, Mbeki continued to play a significant strategic role, albeit with his lines to the "front" significantly attenuated.

He was, however, a major player at Codesa, which was set up in December 1991. He led the ANC in Working Group III (transitional arrangements), while Ramaphosa led it in Working Group II (constitutional principles). By comparing what happened in these two groups, one can begin to understand the fundamental differences in negotiating style between the two men.

Around Mbeki's table, the NP actually conceded very little, but agreement and compromise were reached with minimal conflict; around Ramaphosa's table, several issues were irreconcilable, finally resulting in the collapse of Codesa.

Certainly, the issues being handled by Mbeki were far less contentious than those being handled by Ramaphosa, but here you could see the two distinct approaches: Mbeki keeps people talking at all costs, while Ramaphosa brings matters to a head through brinkmanship; Mbeki works with what is possible, to establish a beachhead for further advances, while Ramaphosa forces a deadlock to shift the balance of power in his direction.

Those around Mbeki say that if he had been running things, a solution may have been reached a year earlier because of his ability to fashion consensual decisions everyone can buy into. Those around Ramaphosa and Slovo say that if Mbeki had been running things, a solution would never have been reached - precisely because he was not prepared to sharpen the issues via brinkmanship; because his "peace at any price" approach was often less appropriate than showing the other side - as Ramaphosa and Mandela were wont to do - that you meant business.

Mbeki himself says his major disagreement with the negotiating team was over the suspension of negotiations after the Boipatong massacre in June 1992. He argued passionately in the NEC that, as he puts it now, "if you stop these negotiations, we'll have to come back to them [later]. In the meantime, what has happened is that the forces responsible for this violence consolidate. You give them space and it is quite wrong."

But he failed to convince his comrades - he was completely alone, after Boipatong, in disagreeing with Mandela's decision to end all talks with the government.

What brought South Africa back from the brink was, in part, Joe Slovo's public proposal in October 1992 of "sunset clauses" that would give the NP a share of power for five years. But those around Mbeki make the point that it was he, not Slovo, who first introduced the notion of sunset clauses into the discussion - in early 1992, during a bilateral with the NP in Cape Town.

Mbeki recalls that "we were really unprepared" for this meeting, and so, "during a tea break, I said to Cyril, We can't just expect these people to be talking, and we say nothing. But we don't have anything to say!' So I suggested that I would address the matter of what assurances we would give to the NP about transition, [because] their fear of complete loss of power was a major obstacle with regard to them moving forward."

Back in the meeting, he proposed power-sharing, saying, "These are entirely my views and don't represent in any way the views of the ANC" - and Barend du Plessis responded that Mbeki's proposals were known technically as "sunset clauses".

Now that it had been discussed in the bilateral, Mbeki had to take it back to the ANC - which he did with the help of Joel Netshitenzhe, who drafted the proposals. Their most vociferous opponent at the time was Joe Slovo. That Slovo subsequently embraced the idea, popularis ed it and took credit for it still galls many of Mbeki's supporters in the ANC, but one senior negotiator makes the point that "anybody could have come up with the idea of powersharing. In fact, everybody did - including the NP itself! Joe's genius was knowing when to insert it into the debate; when to use it to break a deadlock."

Thabo Mbeki's supporters counter that there need not have been a deadlock in the first place.

In August 1993, accompanied only by Jacob Zuma, Thabo Mbeki slipped out of an NWC meeting at the World Trade Centre to be whisked off, in a hired Fiat Uno driven by his close friend Jurgen Kogl, to a pigeon-racing club in Lynnwood, Pretoria. Kogl, in whose Hillbrow penthouse the Mbekis had been living for the past two years, had set up a secret rendezvous for Mbeki and Zuma to meet three prominent Afrikaners who were rattling the sabres of civil war: General Constand Viljoen and the heads of the Transvaal and Free State agricultural unions, Dries Bruwer and Piet Gouws.

The ANC had recognised that its "sufficient consensus" approach - of talking only to the NP - was going to have the potentially explosive consequences of leaving other critical constituencies out in the cold. And so Thabo Mbeki was tasked, by Mandela, to bring them back in.

Boosted by a new confidence (he had just been elected ANC national chairman in a resounding defeat, at the NEC, over Kader Asmal, who had been nominated by Mandela), Mbeki flourished, once more, in the conditions of secrecy required. He worked in his old, tested way - on the quiet, with his original group of colleagues (Zuma, Aziz Pahad, Joe Nhlanhla, Penuell Maduna) and his original recipe of charm, intellectual persuasion and the belief that adversaries must be kept talking. He had the strategist's understanding of where real power lay - of Viljoen's firepower - and the diplomat's ability to accept where his adversaries were coming from.

His approach to the Afrikaner separatists was particularly inspired. He managed both to undercut them - by demanding, in his trademark style of interrogative scepticism, that they define the Volkstaat, something they were unable to do - and to woo them into the new South Africa, proposing eventually to Viljoen that every vote for his party in the elections would be considered a vote in favour of the Volkstaat, and that on the basis of the results, talks could continue after the elections.

We will never know how the negotiations process - and South Africa's transition to democracy - might have been different had it been driven by Thabo Mbeki rather than Cyril Ramaphosa. One thing is certain, though: Mbeki's primary legacy to South Africa is that, with Zuma and a few others, he has managed to disarm both Afrikaner and Zulu sepa ratists by seducing them into an acceptance of the South Africa he now rules.

He understood the basic conservatism of humanity, its instinct towards inertia; that, once safely inside, the Afrikaner sepa ratists would lose their hunger for the uncertainties of a Volkstaat. He also played a key role in bringing Mangosuthu Buthelezi into the 1994 elections, and it now seems likely that under the Mbeki presidency, the IFP will "come home" to the ANC, thereby eliminating permanently the dangers of Zulu nationalism.

Any anxiety a bout Mbeki's commitment to reconciliation, or his feelings about race, needs to be seen in this context.

But does Mbeki's negotiating approach - keeping people talking and fashioning consensual decisions where everyone compromises and everyone wins - actually create solutions, or does it woo adversaries into thinking they are part of a solution when, in fact, they are not?

This was a question asked repeatedly of Viljoen on the campaign trail by Afrikaners, as it was asked by the comrades of the SACP once it became clear that their "Path to Power" document was irrelevant: were they being deceived?

Thabo Mbeki is not of the Straight Talk school of dealmaking. A master drafter, which Mbeki is, knows how to formulate agreements in such a way that even as you manufacture the illusion of consensus, you inch everyone, perhaps imperceptibly, towards your point of view. Mbeki crafts subjunctive clauses so artfully, balances qualifiers so delicately, that you often come away believing you have a deal even if you are not quite sure what it is.

And you are often not quite sure where he stands (or where you stand), for his great skill, as a negotiator, is getting you to question the very ground beneath the feet of your assumptions. The way Mbeki deploys this skill during his presidency will determine how successfully the negotiator makes the move to statesmanship.

With acknowledgements to Mark Gevisser and the Sunday Times.