17 Nov
17Nov


Those who cannot remember the past are condemned to repeat it.

George Santayana, The Life of Reason, 1905


This is the second of two conversations. In the first, I made the strongest case I could for a coalition anchored by ActionSA and the ANC in Johannesburg and Tshwane. Here I take it apart, with the same rigour and the same evidence base. Neither piece is my verdict or advice. Together, I hope, they give readers the full shape of the conflict rather than a slogan from either side. This is conflict analysis, not political punditry. 


Introduction

The talks have moved since the first piece. ActionSA and the ANC held a second round on Saturday, 14 November, on a framework for constituting the councils.[1] On Sunday the ANC also formally opened talks with the DA, despite reported resistance inside the party.[1] ActionSA, for its part, still objects to the DA's mayoral candidates in both cities, Helen Zille and Cilliers Brink.[1] The bloc of ActionSA, the ANC, the EFF and MK remains the most likely outcome.

The case I made for that bloc rested on five pillars: democratic arithmetic, a tested relationship, the contract over the chain, better conflict, and a national dividend. Each pillar is a political reality, whether we choose to see that as significant or not. Each also rests on an assumption that does not survive close inspection and some popular narratives.

My argument here is simple to state. Of all the governments that could be formed from these councils, this one is the thinnest, the most ideologically divided, the most dependent on a record that is weaker than its advocates admit, and the most expensive for the country. In conflict terms, it does not resolve the metros' instability. It postpones it, with job titles.


1. A majority of one is not a government

Start with the arithmetic, because the case for this coalition leans on it most heavily. A four-party bloc of ActionSA, the ANC, the EFF and MK would hold 138 seats against the 136 needed in Johannesburg, and 109 against the 108 needed in Tshwane.[2] That is a margin of two councillors and one councillor respectively.Now set it beside the alternatives. A DA–ANC arrangement would clear the line by 18 seats in Johannesburg and 10 in Tshwane. A DA–ActionSA pact would clear it by 14 in Tshwane and fall just two short in Johannesburg.[2] On the numbers alone, the bloc under discussion is the weakest majority on the table in both cities. Personally I am not in favour of over-emphasizing the numbers game, and case studies certainly show enduring coalitions holding steady on razor-thin margins, but it is also not an insignificant dynamic, as so many of our failed coalitions have shown. Given the past animosities hidden under the surface, such margins also take on a very real risk. 

George Tsebelis gives us the right lens. Every actor whose agreement is needed to change policy is a veto player, and the more veto players a system has, the harder it is to act.[3] With a one-seat margin in Tshwane, the veto does not rest with four parties. It rests with every individual councillor who can stay away from a budget vote. Elections analyst Wayne Sussman made the practical version of this point last week: stitching many parties together suits local government poorly, and handing posts to small partners angers the larger party's own councillors.[4]Ideology compounds the fragility. Paul Warwick's comparative work found that ideologically diverse cabinets collapse sooner than compact ones.[5] Robert Axelrod argued decades ago that durable coalitions join parties that sit next to each other on the main policy dimension.[6] This bloc does the opposite. Herman Mashaba once lodged a parliamentary complaint against Julius Malema over remarks that seemed to encourage illegal immigration,[7] and as recently as this campaign Malema called Mashaba "foolish" over a video on porous borders.[8] Immigration is not a side issue for ActionSA. It is part of its identity. Building a government on that faultline does not remove the conflict. It just moves it into the mayoral committee.


2. Tested, and found wanting

The most persuasive pillar in the first piece was the tested relationship. ActionSA and the ANC have governed together since 2024, the argument ran, and the shadow of the future makes cooperation stable. The difficulty is that a test can be failed as well as passed.Consider Johannesburg, where ActionSA's votes installed the ANC's Dada Morero as mayor in 2024.[9] On his watch, the city's stand-alone audit regressed from unqualified to qualified for 2024/25.[10] The city was at one point reported to have only about two weeks of cash left.[11] Morero admitted that billions had been diverted from Johannesburg Water,[12] while the water utility was losing 44.7% of its water, about R3.8 billion in revenue.[13] The partnership itself broke down in June 2025, when the ANC and its partners removed ActionSA's speaker.[14] ActionSA can fairly say it was never a full governing partner in Johannesburg. But it helped create that administration and then watched it fail. Those ghosts will walk the passages of a new coalition. 

Tshwane is the better story, and it is still not a good one. The Auditor-General issued a qualified opinion for the third year running, and the DA points to a cash deficit of around R4 billion driven by weak revenue collection.[15] Critics warned that the coalition's decision to earmark R2 billion for salary increases and back pay, without testing the matter in the Labour Court, placed an unsustainable burden on the city.[16] Civic groups challenged the city's water tanker spending, which the mayor disputes.[17] 

Most telling for a conflict analyst: the ANC, Moya's own partner, joined the DA in demanding that she release personal financial records relating to renovations at her home.[18] She denies any wrongdoing. But a partner that conducts its disputes through public demands of that kind is not showing the trust the shadow of the future is supposed to produce. Once the DA has been excluded, those wounds will not go away. Axelrod's insight cuts both ways. Repeated interaction teaches cooperation, but it also teaches each side how cheaply the other can be defected against.[6] What these two parties have learned since 2024 is how to use each other, not how to govern together. And voters noticed: the ANC fell from 91 to 63 seats in Johannesburg.[19] What Tshwane voters rewarded was Moya and ActionSA, not the partnership as such.


3. Legitimacy is a conflict variable, not a technicality

The first piece argued, correctly I would think, that a plurality is not a mandate. But that answers a legal question, and the real question is a conflict question. A government can be perfectly lawful and still be experienced as illegitimate by a large share of the people it governs. When that happens, the conflict does not end at the council vote. It moves out of the chamber. This creates a range of conflict outcomes that we keep on adding to the unpaid bill of our coalition failures, from conflict rigidity to distrust, from secondary conflicts to apathy and democracy fatigue. 

Political scientists call the missing ingredient losers' consent. Christopher Anderson and his colleagues showed that democracies stay stable not because winners are happy, but because losers accept the outcome as fair enough to wait for next time.[20] Tom Tyler's work on procedural justice explains when that happens. People accept outcomes they dislike when the process that produced them looks fair, neutral and respectful.[21]Apply that test here. The party that came first in both cities, with about a third of each council, will be excluded. And it will be excluded partly because of personal vetoes on two named candidates, not because of any published disagreement on policy. A Daily Maverick report on the talks observed that a coalition truly reflecting voter intent would run through the ANC, the DA and ActionSA.[1] That is the perception the process creates. It is a perception that will shape how a third of each city treats its government for five years. 

It does not really matter to many to point out that precedent exists in our coalition history – in an important respect it makes it worse. The international precedents I cited in the first piece deserve a second look. New Zealand in 2017 and Portugal in 2015 were indeed governed by coalitions that excluded the largest party. But in both cases the governing partners shared a broad policy direction and campaigned against the same incumbent. Many parliamentary democracies, the Netherlands and Belgium among them, also follow a working convention that the largest party gets the first serious attempt to form a government, usually through an informateur or a royal or presidential mandate.[22] 

I proposed such a plurality convention for Johannesburg in an earlier piece for exactly this reason: it protects legitimacy. The bloc now forming has no shared programme to offer in its place. Its common denominator is who it is against. The conflict environments were also far less acrimonious and identity riven. 


4. Exclusion by label is lazy compromise

The fourth pillar of the first piece borrowed from conflict theory: inclusion as a stabiliser, and a strong DA as a healthy adversary. Turn the same theory around and it indicts the process that is producing this coalition.In my BEYOND COMPROMISE piece (see reference list below) I listed the forms of lazy compromise that postpone conflict instead of resolving it. Two of them are on display this week. The first is exclusion by label. ActionSA has said it is open to the DA, but not with Zille as Johannesburg mayor or Brink in Tshwane.[1] Those are personal vetoes, not policy positions. Scott Atran and Jeremy Ginges showed that once an issue becomes a sacred value, people reject trade-offs that would otherwise be rational, and material offers can even harden their resistance.[23] 

A veto on a named person is a sacred value in miniature. It cannot be traded, so it blocks the negotiation that should have happened. The DA has already indicated how difficult it is to consider this demand. The second is kingmaker extraction. The EFF has form here. In Johannesburg it once refused to vote for a loan the city needed, in retaliation for the ANC's removal of its Gauteng leader as an MMC in Ekurhuleni.[24] That is what partners with a few pivotal seats do when the margin is one or two. They do not merely bargain; they hold the city's budget hostage to settle unrelated scores. The EFF and MK will each be in a position to do this from day one.

Daniel Shapiro's tribes effect describes the deeper risk.[25] When groups define themselves against an outgroup, the identity of the group starts to depend on keeping the conflict alive. A coalition whose most obvious shared identity is "not the DA" will need the DA as an enemy to hold together. Chantal Mouffe's agonism, which I invoked in the first piece, requires that opponents treat each other as legitimate adversaries, not enemies.[26] A government built on exclusion invites the excluded to answer as enemies, and both sides then perform their identities for their bases. That is not better conflict. It is the same conflict, renamed, kicked down the road for a future date. 


5. The national bill

The first piece claimed a national dividend: a DA in opposition in the metros keeps real contestation alive and stops the two largest parties fusing into one establishment. It is an elegant argument. It also underestimates what the metros are being used for.Susan Booysen made the point in 2024, when the same ActionSA–ANC–EFF combination toppled the DA's mayor in Tshwane. She suggested that the Gauteng ANC might be using the metros to prove an alternative to the national GNU formula.[27] That is not how levels of government work in Germany, where the comparison I drew assumed partners who keep their federal and state quarrels apart. 

Here the metro deal is part of the national quarrel. This weekend an NEC member said the ANC had ruled out a national stability pact with the DA, and the Daily Maverick described the two parties as standing at a fork that puts the GNU at risk.[1]

The market reading is blunt. BusinessTech's analysis of the talks set out the worst case: an ANC drifting towards parties that oppose free markets, with the EFF and MK available as national partners if the GNU fails.[2] Business confidence depends on predictable politics at every level,[28] and Johannesburg is not a peripheral town. The Centre for Development and Enterprise warned that its failure would make national success impossible, and that its recovery required a clean break from unstable coalitions.[29] 

A one-seat bloc of four parties is not that break.The cartel argument also deserves a sharper test. Katz and Mair worried about parties colluding to share office everywhere. But the GNU was formed precisely because neither the ANC nor the DA could govern alone, and President Ramaphosa reportedly put the question to his own NEC: how would the ANC work without them?[1] 

The opposite danger is at least as real: polarisation, in which the two largest parties use every metro as a weapon against each other. That is the path this coalition chooses, in the country's economic hub and its capital.


6. The best case for the coalition, answered

A case against that ignores the strongest points for is not worth publishing. Here are the four best arguments from the first piece, and why I do not think they hold.

"Two-thirds of voters chose someone other than the DA" 

True, but a clear majority also chose someone other than ActionSA and the ANC. Together those two parties hold about 39% of Johannesburg's seats and 43% of Tshwane's, and they need the EFF and MK to govern.[2] A non-DA majority is not a programme. Voters for MK and voters for ActionSA did not cast their ballots for the same city.

"The contract matters more than the chain" 

I agree, which is exactly the problem. No published agreement exists yet, and every safeguard I listed in the first piece, from a dispute ladder to a firewall on appointments, has to be accepted by four parties with a one-seat margin. Gamson's Law predicts that partners take portfolios in proportion to their seats.[30] But at a one-seat margin every partner is pivotal, and a pivotal partner can demand far more than its share. The contract that should restrain the coalition will be written by the partners with the most leverage to weaken it. This is coalition conflict basics.

"The alternatives are fragile too" 

Some are. But a DA–ActionSA pact in Tshwane would clear the majority by 14 seats, and a DA–ANC arrangement in Johannesburg by 18.[2] Those are not perfect options, and they carry their own relationship problems. They are, measurably, less fragile than the option being chosen.

"The DA isolated itself by refusing to drop its candidates"

This is the strongest counterpoint, and it is partly fair. My concerns and critiques of the DA’s continued inability to understand and apply identity conflict in its national campaigns and efforts at growing its voter base and national credibility are on record. But the DA can still break the deadlock by finding new candidates acceptable to ActionSA, as Mashaba proposed.[31] But a conflict system that rewards whoever issues the hardest personal veto is not a stable one. Both sides are performing identity here. Rewarding that performance with office does not end it.


A closing word: remember the past

Santayana's warning fits Gauteng's metros uncomfortably well. Johannesburg has had nine mayors and eight coalitions since 2016.[29] Each of those coalitions was formed by parties confident that this time the arithmetic would hold, that this time the partners would behave, and that this time the contract would be honoured. Each time, the thin majority and the personal vetoes won. Coalition-by-numbers is a terrible strategy.The case against an ActionSA–ANC coalition is not that ActionSA cannot govern. Nasiphi Moya's popularity suggests it can. 

It is not that the ANC must be punished, or that the DA is owed anything by coming first. It is that this particular combination brings together every risk factor the evidence identifies: the smallest margin on the table, the widest ideological spread, a weak shared record, a legitimacy deficit, and a national quarrel imported into city hall. This argument has nothing to do with the claimed “DA arrogance”. 

The question is not whether this coalition can be formed. It almost certainly can. The question is whether it can still be standing, and governing well, in 2031. On the evidence, the honest answer is: probably not, and Johannesburg cannot afford to find out the hard way again. In other words, is it the best option on the conflict dynamics that we assessed. 


That ends this two-part conversation. Readers now have both cases at their strongest. The conflict competence is in holding both at once, and in noticing that the real variable in each is not who governs, but how.


RESOURCES AND SUGGESTED READING

Notes

1. Daily Maverick, ANC rules out national stability pact with DA, 15 November 2026.

2. BusinessTech, Coalition 'worst-case' scenario for South Africa is in sight, November 2026.

3. Tsebelis, Veto Players: How Political Institutions Work (Princeton, 2002).

4. EWN, Analyst Wayne Sussman warns GNU-style coalitions won't work in local government, 12 November 2026.

5. Warwick, Government Survival in Parliamentary Democracies (Cambridge, 1994).

6. Axelrod, Conflict of Interest (Markham, 1970); The Evolution of Cooperation (Basic Books, 1984).

7. News24, ActionSA, EFF square off over immigration, 18 January 2021.

8. Ground News, Malema calls Mashaba 'foolish' in sharp retort over border video, 2026.

9. Mail & Guardian, ANC sets its eyes on Tshwane as ActionSA targets speaker positions, 20 August 2024.

10. Sunday Times, Joburg gripped by cash crisis despite mayor's financial optimism, 11 July 2026.

11. EWN, Billions lost, trust declining: can the next Joburg administration turn the city around?, 26 October 2026.

12. Daily Maverick, Resignations demanded as Joburg's water crisis boils over, 26 September 2025.

13. Business Report, City of Joburg admits audit failures as Parliament demands accountability, 9 June 2026.

14. Daily Maverick, ActionSA–ANC Joburg partnership on the rocks after speaker removed, 25 June 2025.

15. Business Report, Tshwane's coalition government faces backlash over Auditor-General's report, 1 February 2026.

16. Rekord, Irregular spending surges as metro struggles to restore financial credibility, 30 January 2026.

17. IOL, Civic group calls for Tshwane to be placed under administration over water tanker spending, 24 October 2025.

18. Briefly News, ANC and DA demand Mayor Nasiphi Moya release financial records over home renovations, 2026.

19. Gauteng.net, 2026 local government election results, November 2026.

20. Anderson, Blais, Bowler, Donovan & Listhaug, Losers' Consent: Elections and Democratic Legitimacy (Oxford, 2005).

21. Tyler, Why People Obey the Law (Yale, 1990).

22. See generally Bergman, Müller & Strøm (eds), Cabinets and Coalition Bargaining (Oxford, 2008).

23. Atran & Ginges, "Religious and sacred imperatives in human conflict", Science 336 (2012).

24. Daily Maverick, Coalitions sans frontières: ANC's tough decisions at municipal level, 19 September 2024.

25. Shapiro, Negotiating the Nonnegotiable (Viking, 2017).

26. Mouffe, On the Political (Routledge, 2005).

27. Booysen, interviewed in The Conversation, South Africa's unity government is being tested, 8 October 2024.

28. Business Day, ANC suffers devastating blow as DA leads in Gauteng, 9 November 2026.

29. Centre for Development and Enterprise, Joburg in Jeopardy, reported in Polity, 18 August 2026.

30. Gamson, "A theory of coalition formation", American Sociological Review 26 (1961).

31. Business Day, Mashaba offers DA Johannesburg mayoralty to break coalition deadlock, 10 November 2026.


Earlier pieces referenced: 

BEYOND COMPROMISE; 

COALITION CHAOS; 

COALITION ORDER FROM CHAOS, part 1.


Full references, further reading material, courses, coaching, study material, workshops / seminars, mediation and representation are available on request.

•     A summary of Andre Vlok’s CV / bio can be accessed at https://www.conflict-conversations.co.za/conversations/2025-online-curriculum-vitae-andre-vlok


Andre Vlok can be contacted at andre@conflict1.co.za for any further information.

(c) Andre Vlok

November 2026

Comments
* The email will not be published on the website.