It was becoming increasingly clear to me that terrible things happen when our leaders fail to think about data that are outside their typical focus.
Max Bazerman
Introduction
South Africans have put their hopes behind their senior politicians to step up and effectively lead, to lead us out of decades of neglect, poor leadership skills, corruption, and the predictable results of a range of unresolved conflict spirals stretching back decades. The writing has been on the political wall for at least a year now – coalitions will be our new normal, at least for a few years. If that is a given, then coalition politics must be understood to be the highly specialized conflict arena that it is, with its potential for harm, or growth and stability. The first week after the elections has not been inspiring. Partisan bickering, score-settling, open threats and schoolyard insults seem to have been of higher priority for some of these politicians than the urgent work at hand.
This is often, as we have seen in the coalition circuses of recent years, been used a replacement for skilled conflict management. It vilifies and scapegoats the opposition while detracting from an absence of positive solutions and remedies. Our politicians continue to use “experience” and pages from the 80s playbooks as the roadmaps to where we should be. We see it in their public pronouncements, their policies, and in the results that we have to live with. The clock is ticking, South Africa has run out of time, and out of patience. In a hypothetical boardroom discussion with these senior leaders, and if I only had 30 minutes, this is some of what I would have to say.
The wrong problem, solved well, is still the wrong problem South Africa's parties are treating the post-election coalition season as a negotiating problem. It is an identity conflict wearing a negotiating costume, and until the parties recognise the difference, every deal they sign will carry the seeds of its own collapse. The numbers from 4 November are stark. The ANC took 30.76% of the national vote, the DA 24.64%, the MK Party 9.76%, the EFF 6.22% and ActionSA 5.67%. The ANC won no metro outright. The DA governs Cape Town alone but merely leads in Johannesburg, Tshwane, Ekurhuleni and Nelson Mandela Bay. The MK Party leads in eThekwini with 38%. Most of the country's economic engine will be run by coalitions. This is not an academic exercise. Listen to the language that followed. The MK Party's Johannesburg candidate declared the DA a party it will never work with. The Council of the Left offered the ANC partnership on condition it expels the DA and Freedom Front Plus as "reactionary forces". The DA has spent two years warning of a "doomsday coalition". The ANC's secretary-general called for the DA's place in the GNU to be reviewed after the two partners met on opposite sides of a court case. And the ANC has, once again, said it will work with anyone except the DA in Johannesburg while sitting beside the DA in Cabinet. These pronouncements change and contradict, but the cacophony remains.
None of this is the language of interests. Interests sound like water tariffs, procurement rules, housing lists and budget lines. This is the language of identity: of who we are, who they are, and what it would mean about us to be seen with them. Conflict scholarship has a name for disputes of this kind. Edward Azar called them protracted social conflicts, rooted not in divisible goods but in unmet needs for identity, recognition, security and access. You cannot split the difference on a need. You can only meet it, or deny it and pay later. You have to manage it, persuade people, not compromise or strong-arm people.
This is not horse-trading. Resentful compromise now gets you out of this news-cycle, and comes back as more complex, more intractable conflicts with conflict rigidity and increased distrust as bonuses. This piece offers advice from that vantage point, as conflict analysis rather than political advocacy. It is not neutral about everything. My political work makes the dangers of lazy compromise very clear. It takes a hard line against lazy compromise, against using public services as weapons, and against the idea that a coalition is merely a seat-sharing spreadsheet. But it takes no side on who should govern. Its question is narrower and more urgent: how do people who distrust each other this deeply run cities together, or oppose each other without breaking them?
Diagnosis: what is actually being fought overIdentities defined by negation
Herbert Kelman, who spent decades in Israeli-Palestinian problem-solving workshops, described a trap he called the negative interdependence of identities. Each side's sense of itself is built partly on denying the other. To affirm what I am, I must reject what you are. In that condition, any concession feels like self-erasure. South African party politics now runs on this logic. Part of the DA's brand is being the not-ANC and, increasingly, the not-MK. Part of MK's brand is being the not-Ramaphosa ANC and the anti-DA. The EFF and the Council of the Left define their radicalism against what they call the liberal or reactionary right. ActionSA defines its integrity partly by refusing the DA in Johannesburg. Each "never" is a statement to its own base about who it is. That is why the vetoes are so loud and so public: they are not bargaining positions, they are identity performances.
Daniel Shapiro of the Harvard International Negotiation Program calls this the tribes effect: a mind-set in which the other side becomes an adversary to be defeated rather than a party with whom a problem is shared. He names five lures that pull groups into it.
Three are visible in South Africa right now:
Repetition compulsion: the same fights, with the same insults, every election cycle since 2016.
Taboos: it has become unspeakable in some parties even to explore a relationship with certain others.
Assault on the sacred: each side treats something the other says as an attack on what it holds most dear, whether liberation history, constitutional order or the dignity of a leader.
Chosen traumas and the apartheid city
Vamik Volkan showed that large groups bind themselves around a chosen trauma, a historical injury passed down as a marker of identity. South Africa carries several, and they are not equally shared. For many black South Africans, the trauma is apartheid itself and the sense that 1994 delivered political freedom without economic restitution. For many Zulu-speaking voters in KwaZulu-Natal, it now includes the treatment of Jacob Zuma, read not only as a legal matter but as humiliation of a people. For many white, coloured and Indian South Africans, especially in DA strongholds, the formative experience of the last decade is state decay: load-shedding, collapsed municipalities and the fear of being a permanent minority under a government that does not need their votes.
These are not symmetrical histories. Apartheid was a crime against humanity with a clear victim and beneficiary structure, and its effects are measurable today. But a conflict practitioner cannot work only with the trauma that history judges most severe. Every group's fear shapes its behaviour at the table, and an unacknowledged fear does not disappear. It goes underground and steers.
Here is the point the parties keep missing. Local government is where apartheid's legacy is most concrete. Apartheid was, above all, a spatial project: townships placed far from jobs, infrastructure built for some suburbs and not others, bulk services designed for a segregated city. Thirty-two years later, those pipes, roads and distances largely remain. When parties fight over who runs Johannesburg or eThekwini, they are fighting over who controls the machinery that either dismantles the apartheid city or preserves it. That is why municipal coalitions carry such emotional freight, and why technocratic language alone will never, can never, settle them.
The black/white divide under the microscope
The DA has black leaders, black mayors and a growing black vote. MK has a broad base that is not reducible to ethnicity. Yet in the public imagination, and in a good deal of campaign messaging, the coalition fight is being read through race, as much as this is often denied. The DA is cast by its opponents as the party of white interests and capital. MK and the EFF are cast by their opponents as threats to property, order and minority safety. Conflict theory is clear on what happens next.
When political contests map onto ethnic or racial cleavages, Donald Horowitz showed, elections become a census and every result a verdict on group worth. Moderation is punished, because leaders who reach across the line are outflanked by those who promise not to. South Africa is not yet a fully ethnicised party system, but it is drifting toward one. The coalition conduct of the next three months will either slow that drift or accelerate it. In all of this, thanks to the unchecked mechanisms of identity politics, rationality has only a relatively limited role to play.
The ANC's failure as the root conflict
No honest analysis can avoid the ANC's role. For three decades it held a dominant, often unassailable majority. It delivered real gains: grants to millions, housing, electrification and a sound constitutional order. It also presided over state capture, collapsed municipalities, the Eskom crisis, cadre deployment that hollowed out the public service, and an unemployment rate stuck near a third of the workforce. Its local vote has now fallen from roughly 46% in 2021 to under 31%. MK, in large part, is the ANC's own unresolved internal conflict, exported onto the ballot.
The ANC is now in the most dangerous position a party can occupy in a coalition system: the indispensable middle that refuses, or is unable, to choose. It sits with the DA in the GNU and rejects it in Johannesburg. It refuses to attend the Council of the Left yet hears its terms. It postures as the stabiliser while its secretary-general calls for a partner's place to be reviewed. In conflict terms, this is not balance. It is avoidance, and avoidance by the most powerful actor in a system pushes every other actor toward the extremes. It’s messaging is, by design or by systemic ineptitude, all over the show and a part of its own inability to lift its nose from the steep dive it is in.
MK and the DA
The MK–DA relationship is the clearest test of whether South Africa can manage deep conflict. eThekwini makes it unavoidable: MK leads with 38%, the DA is second with 25%, and both are now hunting for partners against each other. The gap between them is real. MK campaigns for radical economic transformation, expropriation of land without compensation, a stronger role for traditional authority, and a reworking of the constitutional order it regards as a settlement that protected white wealth.
The DA defends constitutional supremacy, property rights, market-led growth, fiscal discipline and a merit-based state. These are not misunderstandings to be cleared up over coffee. They are competing theories of what justice requires after apartheid. The acrimony is also real. MK has questioned the legitimacy of electoral outcomes and is challenging the eThekwini count. The DA took legal action in September after an MK leader was reported as threatening to cut off water and electricity to DA voters. DA leaders have said MK has nothing to offer voters and called it a party bent on destabilising Parliament.
Each side tells its supporters that the other is a danger to the city. Recent days have seen positive outreaches and the right noises have been made. So here is the brave conclusion: MK and the DA probably should not govern eThekwini together right now, and nobody should pretend otherwise. A forced marriage between parties that disagree on the basic legitimacy of the constitutional order would produce paralysis, sabotage from within, and a collapse that would deepen the very identity wounds each side claims to protect.
Robert Mnookin's work on bargaining with adversaries makes the point that not every deal is wise, and that the decision to negotiate should rest on a clear comparison of alternatives, not on a moral reflex either way. Timing and sequence are our tools to use here. But "not governing together" is not the same as "not dealing with each other". This is where both parties are failing. The political theorist Chantal Mouffe distinguishes antagonism, where the other is an enemy to be destroyed, from agonism, where the other is an adversary whose right to exist and compete is accepted even as you fight them hard. Democracy, she argues, does not need consensus. It needs conflict to be channelled into agonistic form. Applied to eThekwini, as a particular case study that can guide other decisions also, that means the following. Whoever governs, MK and the DA need a minimum agonistic compact. It would have four elements.
This is a lower bar than coalition. It is also a higher bar than either side currently clears. And it speaks to each side's real interests. MK's voters want to be taken seriously as a political force, not treated as a threat to be contained. The DA's voters want a functioning city and assurance that losing power, if it comes, will not mean losing safety or services. Both of those interests are served by agonism and destroyed by antagonism.
There is one more thing the DA needs to hear. Competence is not a substitute for legitimacy. A party can run an excellent budget process and still be experienced by a large part of the population as an outsider to their history. Whether that is an accurate assessment or experience by the receiver party is largely irrelevant, as identity conflict has very little to do with objective rationality. The DA's habit of framing MK and its voters as the problem to be kept out reads, to many black South Africans, as a familiar story about who is allowed to govern. If the DA wants to lead a majority-black city, it needs to speak to dignity and restitution as fluently as it speaks to clean audits.
And MK needs to hear this. Grievance is a powerful mobiliser but a poor governing programme. Learn from the ANC’s misery. Winning eThekwini's largest share of the vote creates a duty to show that the party can run water treatment works, fix the port city's infrastructure and manage a budget without capture. If MK governs, it will be judged on that, and its supporters will be the first to pay if it fails.
The danger of lazy compromise
Compromise is usually praised as the opposite of conflict. In identity conflicts it is often a way of postponing one. South Africa has seen five kinds of lazy compromise since 2016, and each is on offer again.
Arithmetic coalitions: Parties add up seats to reach 50% plus one and call it a government. There is no shared programme, no agreed priorities and no plan for disagreement. These are the coalitions that gave Johannesburg its revolving door of mayors. They are not partnerships. They are temporary ceasefires with job titles.
Secret deals: Agreements negotiated in hotel rooms and never published let leaders promise one thing to partners and another to supporters. When the gap is exposed, as it always is, trust collapses and the public learns that coalition means betrayal. The Municipal Structures Amendment Bill, which would make coalition agreements compulsory and public, was meant to pass before these elections. It did not.
Unity as silence: The GNU leaders' retreat this month was held with no agenda and a mandate to keep discussions private. Private space for candour is valuable; Kelman's workshops depended on it. But privacy without a disciplined agenda and without agreed outputs tends to produce warm statements and no change. A coalition that avoids its real disagreements does not resolve them. It lets them fester until they explode in the media. Kingmaker extraction: Small parties with a handful of seats demand the mayoralty or key portfolios as the price of support. The result is governments led by parties with the weakest mandates and the strongest incentives to defect. That is not compromise. It is hostage-taking dressed up as democracy.
Exclusion by label
The most corrosive compromise is the one in which parties agree on who to keep out before agreeing on what to do. The "doomsday coalition" and the "reactionary forces" are mirror-image formulas. Both build a coalition around a shared enemy rather than a shared task. Conflict research is unambiguous: groups united only by opposition fragment as soon as the threat recedes, and they need to keep the enemy alive to stay together. The brave alternative is not to refuse compromise. It is to insist on principled compromise, which the Harvard Negotiation Project framed decades ago as separating the people from the problem and focusing on interests rather than positions. But identity conflicts need one more step.
Scott Atran and Jeremy Ginges found in studies of sacred values that material incentives can backfire when offered for something a group holds sacred, while symbolic gestures of recognition can unlock movement. South African parties keep offering each other portfolios when what is often needed first is acknowledgement.
What the parties should do now
Run three tables, not one
Every serious coalition negotiation should be split into three distinct conversations, in this order.
Build the conflict machinery
The guidelines on coalitions published this week (see my earlier article on this) have their pros and cons, and will most probably become a source of further argument. Those considerations can in due course be implemented in conjunction with more formal legislation. As I have argued in earlier pieces, and as other examples (employment law, equity provisions, even the Constitution) show, you cannot legislate effective leadership. The following is a practical, established-through-experience point of departure: Every coalition agreement, national or local, should be public and should contain:
Parliament should pass the coalition bill without further delay. The GNU parties have the votes. Their failure to do so before 4 November is itself a form of avoidance.
Stop negotiating through the media
Much of the current acrimony is megaphone diplomacy: partners briefing against each other, staging walkouts for cameras, and answering private disagreements with public insults. Each statement makes the next concession harder, because leaders become locked into positions they have broadcast to their base. Coalition partners should adopt a simple protocol. Disagreements are raised inside first, with a 72-hour window before anyone goes public. Public criticism targets policies and decisions, never the identity or legitimacy of a partner's voters. Joint briefings are used for joint decisions.
For the ANC: choose, and own your record
The ANC must stop hedging. It cannot be the DA's partner in Pretoria and its enemy in Johannesburg without teaching every voter that coalition commitments are disposable. It should define publicly, and apply consistently, the principles on which it will govern with others. It should also make a genuine, specific public account of what went wrong over 30 years, naming state capture, cadre deployment and municipal collapse. The Truth and Reconciliation Commission rested on the insight that acknowledgement opens doors that argument cannot. A party that admits failure regains the moral standing to lead. A party that blames everyone else loses it.
For the DA: from competence to legitimacy
The DA has earned the right to be taken seriously as a governing party. It now needs to earn trust from voters who do not share its history. That means dropping the language of keeping others out, speaking directly about the apartheid city and how to dismantle it, and showing that it can be a reliable junior partner where voters have put it there. Its strongest card is delivery. Its weakest is the perception that it governs for some South Africans more than others. It is working hard to change these perceptions, it however still does not understand the building blocks of identity conflict, the only tool in the box that will sustainably allow them to grow their current base.
For MK: from grievance to government
MK carries the weight of real anger, especially in KwaZulu-Natal. That anger deserves a hearing in every coalition conversation. But MK must make an unequivocal choice for constitutional means: accept final results, repudiate any threat to withhold services, and present a governing programme that a partner can actually sign. The ANC and the DA should also stop treating its voters as a problem to be quarantined. Excluding a party that leads a metro does not remove its voters from the city.
For the EFF, ActionSA, FF+ and other parties
ActionSA's blanket refusal of the DA in Johannesburg, and the Council of the Left's demand that the ANC expel two partners, are vetoes that serve party identity more than residents. Smaller parties should price their support in services and reforms, not in mayoral chains. A party with 2% of the vote that demands the top job is not a kingmaker. It is a destabiliser.
Adversaries, not enemies
South Africa has done this before. In the early 1990s, people who had tried to kill each other sat down at Kempton Park, and the Mont Fleur scenario exercise brought activists, businesspeople and officials into the same room to imagine shared futures. Collaboration with people you do not agree with, like or trust is not only possible but often the only option. That is the condition of coalition politics now. The settlement of 1994 did not require anyone to give up who they were. It required everyone to accept that the others would remain, and to build rules strong enough to hold their conflicts. The parties of 2026 face the same task on a smaller, more practical scale: water pipes, budgets and committee chairs. The question is not whether they will keep disagreeing. They will, and they should. The question is whether they can disagree as adversaries inside a shared order, or will keep fighting as enemies who each believe the city belongs to them alone.
References and suggested reading
Sources for current events
Some of my own related work
Andre Vlok can be contacted at andre@conflict1.co.za for any further information.
(c) Andre Vlok
November 2026