The Constitution and the Constitutional Court make short work of extreme arguments on both sides of the debate over affirmative action.

The misalignment between the FW de Klerk Foundation and Hermann Pretorius of the Institute of Race Relations (IRR) is not, at its core, a disagreement about whether apartheid caused profound and continuing harm. Both sides accept that South Africa still carries the economic and social wreckage of racially engineered exclusion.

The real disagreement concerns constitutional method, political honesty, and policy design: how should the Constitution be read, how should democratic redress be understood, and at what point does remedial law risk becoming a stale and damaging substitute for genuine upliftment?

The Foundation’s position draws an essential constitutional distinction. We reject reducing all race-conscious law into a single category of “race laws”, while also acknowledging that South Africa cannot indefinitely sustain a governance model built around permanent racial administration.

This middle ground is not evasive. It is the most legally serious and morally responsible position in the present debate.

The “poison of apartheid and the antidote of post-1994 redress” are not the same thing, even if both operate within a legal environment still shaped by race. Pretorius’s interpretation, by contrast, treats section 1’s commitment to non-racialism as though it nullifies section 9(2)’s explicit permission for restitutionary measures. That misunderstands the Constitution.

Negotiated Tensions

South Africa’s constitutional order is an architecture of negotiated tensions. It must be read holistically, not as a set of winner-takes-all clauses.

The IRR and Pretorius are correct to warn that the endless bureaucratisation of race is dangerous. We agree with that concern. But his argument overreaches by collapsing apartheid-style legal racialisation, anti-discrimination law, data-gathering provisions, and remedial transformation measures into one flattened category. He then claims that the Constitution itself forbids a legal architecture repeatedly upheld by the Constitutional Court. The result is not constitutional clarity. It is ideological compression.

The Foundation’s position is more difficult, but more persuasive, because it holds three constitutional truths together. First, redress remains necessary because history still weighs heavily on the present.

Secondly, non-racialism is not decorative rhetoric but a binding constitutional destination. Thirdly, the current policy mix, especially rigid racial targets and continuing race classification across law and administration, has become increasingly blunt, patronage-prone, and socially corrosive. That is why the Foundation has argued that South Africa should move away from permanent racial classification and why we have warned that these instruments increasingly do more harm than good.

Then, the IRR and Pretorius’s constitutional claim rests on a half-truth. It is true that section 1 is foundational and more heavily entrenched than most of the Constitution. Section 1 defines South Africa as a democratic state founded on “human dignity, the achievement of equality and the advancement of human rights and freedoms”, as well as “non-racialism and non-sexism”. But none of that establishes section 1 as a free-standing veto over section 9(2). It proves only that section 1 matters profoundly.

Section 9 says something equally important. Section 9(1) guarantees equality before the law and equal protection and benefit of the law. Section 9(2) then provides that equality includes the full and equal enjoyment of all rights and freedoms and that legislative and other measures “designed to protect or advance persons, or categories of persons, disadvantaged by unfair discrimination may be taken”. The text therefore does not oppose equality and remedial measures. It places remedial measures inside the constitutional commitment to equality.

The Constitutional Court made this unmistakable in Minister of Finance v Van Heerden. Its formulation has become canonical: section 9(1) and section 9(2) are “both necessary and mutually reinforcing” and the Constitution “embraces for good reason a substantive conception of equality inclusive of measures to redress existing inequality”. That is fatal to Pretorius’s central thesis.

One may legitimately debate the limits, duration, design, or fairness of particular remedial measures. One may not truthfully argue that the Constitution treats substantive equality and non-racialism as mutually exclusive poles.

Harmonisation

That is why harmonisation matters. The Constitution must be read together. The Constitutional Court has not interpreted section 1(b) as annulling the machinery of section 9(2). It has insisted on reading the provisions holistically and coherently.

Also, section 195 strengthens this position. It requires public administration to be governed by efficiency, accountability and good human-resource management, while also requiring it to be “broadly representative of the South African people”, with employment practices based on “ability, objectivity, fairness and the need to redress the imbalances of the past to achieve broad representation”. This provision is devastating to both extremes. It rejects Pretorius’s absolutist anti-classification reading because the Constitution itself combines representation and redress. But it also rejects crude quota politics because representation must be pursued together with ability, fairness, objectivity, and effective administration. The Foundation’s position sits inside that constitutional middle.

This is why the Foundation’s broader constitutional ethos matters. Even worthy ends must be pursued through lawful and constitutionally disciplined means. A constitutional democracy depends not only on good outcomes, but on lawful processes. Redress without legality degenerates into arbitrary engineering. Non-racialism without redress degenerates into historical amnesia. The Constitution requires neither. It requires disciplined transformation.

Not Without Value

The IRR’s Race Law Project is not entirely without value. As a database of statutes that make race legally relevant, it has descriptive utility. Its public materials refer to 324 racial Acts in total and 145 operative laws, of which nine have been “deracialised”. Its methodology says the index does not judge whether a law is good or bad; it asks only whether race or proxies for race have legal effect. But that neutrality is overstated.

The index counts “operative” laws even where some originally racial pre-1994 laws are no longer enforceable because of the 1993 and 1996 Constitutions. The headline number of 145 includes laws the project itself marks as “deracialised”.

This is where the Foundation’s criticism matters. The phrase “race law” collapses morally distinct legal phenomena into one emotionally explosive category. A democratic measure aimed at remedial inclusion is not the same as apartheid legislation designed to degrade, exclude, and dominate. Once intent, context, and function are ignored, anti-discrimination provisions, redress mechanisms, and historical remnants can all be flattened into the same ideological object. Even if the index is descriptively broad, its public deployment is not politically innocent.

The problem is aggravated by the IRR and Pretorius’s rhetoric. His May 2026 response accused the Foundation of contemplating the return of the “potloodtoets” (pencil test), described our position as an “absurde vorm van regstellende rassisme” (absurd form of affirmative racism) and derided it as “liegbekkerige kwasi-nie-rassigheid” (sloppy quasi-non-racialism). Whatever one thinks of the legal dispute, this is not the language of careful constitutional engagement. It weakens the IRR’s claim to principled constitutional seriousness, because a project supposedly committed to analytical sobriety chose polemic over persuasion.

Yet the Foundation should not deny the part of the IRR critique that is correct. The continued legal salience of race is troubling. The statute book still contains many measures in which race remains legally relevant, showing that South Africa has too often treated racial classification as an administrative shortcut. The maturity of the Foundation’s position lies in refusing to accept the IRR’s false equivalence while also refusing to defend the status quo.

This distinction between constitutional principle and policy design is clearest in employment equity. The government’s defence of stronger employment equity rules is not invented from thin air. The 2023/24 Commission for Employment Equity reported that whites held 62.1% of top management posts overall and 65.1% in the private sector, while Africans held only 17.2% despite making up 80.7% of the economically active population.

Limits

But the law also limits remedial programmes. In South African Police Service v Solidarity obo Barnard , the Constitutional Court stressed that remedial measures must advance those disadvantaged by past discrimination, but must not “unduly invade the human dignity” of those affected by them. In Solidarity v Department of Correctional Services, the Court emphasised that representivity must account for regional as well as national demographics. In Minister of Justice v South African Restructuring and Insolvency Practitioners Association, rigid ratios in an appointments policy were accepted as quotas not permitted under section 9(2).

That jurisprudence does not justify abolitionism. But it does justify criticism of overreach. The point is not that all race-conscious redress is unconstitutional. The point is that rigid demographic engineering, detached from actual disadvantage, merit, operational need, and administrative consequences, can become unconstitutional, irrational, or destructive.

The broader social setting matters. South Africa’s official unemployment rate stood at 32.7% in the first quarter of 2026. Multidimensional child poverty for children aged 0-17 was still 57.3% in 2023. South Africa remains defined by mass poverty and exclusion. In such an environment, redress that degenerates into compliance arithmetic without building human capability is morally troubling, because the cost is paid through weaker growth, poorer service delivery, and delayed upliftment for the poor majority whom redress is meant to serve.

That is why the Foundation’s criticism of “blunt instruments” and “social engineering” matters. We do not deny the need to remedy history. We argue that some state instruments increasingly frustrate the constitutional destination they were meant to serve.

Needs-Based Upliftment

The strongest constitutional path now lies in needs-based upliftment. The Foundation has argued for the smart use of data to identify actual socio-economic vulnerability rather than relying indefinitely on racial classification. That approach is constitutionally honest because it accepts that section 9(2) permits redress. It is empirically precise because it targets disadvantage where it is lived: in weak schools, jobless households, poor municipalities, collapsing infrastructure, disability, orphanhood, household income, and territorial deprivation. It is also morally stronger, because South Africa cannot become genuinely non-racial by endlessly bureaucratising race.

The Foundation’s position should therefore be stated with confidence: apartheid’s legacy must be remedied; the Constitution permits and indeed demands substantive equality; but the future of that project lies in socio-economic upliftment grounded in evidence, legality, competence, and human dignity, not in the indefinite reproduction of racial categories by the democratic state. That is not a retreat from justice. It is the most credible way to complete it.