A New Separation of Powers – How to Legislate Peace

The old separation of powers – executive, legislative, judiciary – was a brilliant invention of the Enlightenment. But it was designed for a world where knowledge was scarce, communication was slow, and the church and the crown were the only institutions with reach. Today, we have universities that generate knowledge at an unprecedented scale, scientific establishments that can advise (or capture) governments, a European Union that sits above national parliaments, and religious institutions that still command the loyalty of billions. The old tripartite separation is no longer sufficient. We need a new architecture: one that acknowledges the distinct roles of philosophy, faith, science, law, and politics. This essay sketches a tentative framework.

Part I – The Flawed Blueprint We Inherited

Montesquieu’s separation of powers was a reaction against absolute monarchy. He argued that political liberty requires that “no one shall be forced to do things that the law does not oblige.” To prevent tyranny, power must be divided among three branches: the legislative (makes the law), the executive (implements the law), and the judiciary (interprets the law). The system has worked, after a fashion, for three centuries.

But the Enlightenment framers did not anticipate the modern research university, the national laboratory, the space agency, or the European Union. They assumed that the church would remain a separate spiritual authority – but they also assumed that it would not govern. They did not foresee a world where scientific expertise would be so central to policy that scientists would become unelected legislators in all but name. They did not foresee a world where the executive would capture the legislature through party discipline, or where the judiciary would become a political battlefield.

We need a new blueprint. Not to replace the old, but to supplement it. The old separation of powers remains essential for the core functions of lawmaking and adjudication. But we must add new pillars to support the weight of modern governance.

Sources for this paragraph:

Montesquieu, The Spirit of the Laws (1748).
Author’s synthesis.

Part II – The Athenian Agora: Philosophy as Open Debate, Not a Council of Elders

In the medieval order, the clergy were the guardians of moral truth. They could not make laws, but they could judge laws against a higher standard. The priest could say to the king: “This law is unjust because it violates divine law.” The king might ignore him, but the moral authority remained.

Today, we have no such institution. The closest equivalent is the university – specifically, the faculty of philosophy. Philosophy departments are the last institutional homes of the question “What is justice?” They train students in ethics, political theory, logic, and the history of ideas. But they have no formal role in governance. They are consulted only if a politician happens to read a book.

But we must not repeat the error of creating a new clerical elite. The model of a council of philosophers – a small group of appointed sages issuing authoritative opinions – is not a European tradition. It is closer to the Chinese system of state‑managed scholarship. That model has its merits, but it is not suitable for the open, pluralistic, and adversarial intellectual culture of Europe.

Europe’s philosophical tradition is Athenian, not Platonic. Athens did not have a council of philosophers. It had the agora – a public square where ideas competed, where citizens argued, where no voice was final and no authority was unassailable. Socrates did not sit on a council; he walked the streets, asked questions, and was killed for his trouble. The power of philosophy is not the power to decree. It is the power to provoke.

Proposal: Instead of establishing an Academy of Philosophy as a parallel judiciary, we should constitutionalise the philosophical agora – a set of institutional guarantees that ensure open, vigorous, and protected debate about first principles. This would include:

Constitutional protection for academic freedom. Universities and philosophy departments must be guaranteed independence from political interference. No government may dictate what questions philosophers may ask or what answers they may propose.
A mandatory platform for philosophical debate in the legislative process. Before any major legislation is enacted, parliament must publish the proposed law and invite written submissions from any philosopher, ethics institute, or citizen. These submissions would be collected in a public philosophical record – not as authoritative opinions, but as evidence of the range of reasonable arguments.
A rotating “Socratic Commission” – not a permanent council, but a randomly selected panel of five citizens, drawn from philosophy departments and learned societies, whose sole function is to question the government’s moral justifications for a law. Their questions would be published, and the government must respond. No votes, no vetoes – only questions.
Public funding for independent ethics councils – modelled on existing bodies in France, Germany, and Finland – that operate as civil society organisations, not state organs. Their opinions would be advisory and non‑binding, but they would be funded to ensure that philosophical critique is not silenced by lack of resources.
A guarantee that no philosophical opinion is treated as authoritative. The state may listen to philosophers, but it may not privilege any school of thought. The only authority is the force of the better argument – and the better argument is determined by public debate, not by the credentials of its proponents.

The Athenian mandate is not “rule by philosophers.” It is rule by reasoned argument. Philosophers have no special claim to power. They have a special claim to ask uncomfortable questions. The constitutional role of philosophy is not to supply answers. It is to ensure that the questions are never forgotten.

The power of philosophy is not the power of the best argument – because no argument can be declared “best” without stopping debate. The power of philosophy is the power of the perpetual question. The agora must remain open. No question is final. No answer is sacred.

Sources for this paragraph:

Author’s proposal, inspired by the Athenian model and the European tradition of academic freedom.
Contrasts with Platonic philosopher‑king ideal.
References to existing ethics councils and public debate mechanisms.

Part III – The Religious Clergy: The Source of First Principles

The European legal tradition is not secular in origin. It is Christian. The canon law of the Catholic Church, the natural law theories of Thomas Aquinas, and the theological jurisprudence of the Reformation shaped the very concepts of justice, rights, and the rule of law that the modern state claims as its own. To relegate religious institutions to mere advisory bodies is to forget this history – and to cripple the foundation of law.

Proposal: Religious institutions – specifically, the historic churches that have preserved the moral and theological tradition of Europe – must be recognised as a source of law. They are not “spiritual pillars of the state”. They are not “advisors”. They are constitutional authorities whose task is to declare the first principles upon which the constitution and all subordinate laws are built.

In concrete terms, the religious institution (e.g., the Catholic Church in its universal teaching office, or an ecumenical council of churches) would have the authority to issue axioms – foundational propositions of natural law, divine law, and moral truth. These axioms are not optional. They are the grammar of justice. Any law that contradicts a declared axiom is not merely bad policy; it is unconstitutional, because it violates the fundamental order of right reason and divine command that the constitution itself presupposes.

This is not theocracy. The religious institution does not administer the state. It does not appoint judges. It does not command armies. Its authority is strictly declaratory: it says what is true about justice, human dignity, and the good. The political branches – parliament, executive, judiciary – then apply those axioms to the concrete circumstances of legislation, adjudication, and governance. The religious institution has no veto, but its declaration is a binding constraint: a law that violates an axiom is no law at all, as Aquinas and Blackstone both taught.

This is the European tradition. It is not a new invention. It is the recovery of a forgotten architecture. The Holy Roman Empire, the medieval kingdoms, and even the early modern states all recognised that the Church had the authority to declare the moral law that bound the secular ruler. The modern separation of church and state was meant to protect the church from state interference, not to silence the church’s voice in public life. By giving religious institutions the power to declare axioms, we restore the balance: the state is free to govern, but it must govern within the limits set by eternal and natural law.

The other religious communities (Jewish, Muslim, Hindu, etc.) would have their own recognised authorities, and their axioms would be respected as binding for their own communities, and as advisory for the general public. Shared inter-faith axioms would be strongly preferred, but in their absence, the Christian axiom would be followed, as this is foundational to the European project and the origin of Enlightenment values. These axioms would be explicitly barred from invading the secular judiciary’s lot.

Sources for this paragraph:

Thomas Aquinas, Summa Theologica, on eternal, natural, and human law.
Blackstone’s Commentaries on the law of nature and the law of God.
Canon law tradition and the relation of church to state.
Author’s proposal, restoring the pre‑Enlightenment understanding of church authority.

Part IV – The Scientific Establishment: From the Laboratory to the Council Chamber

The greatest omission of the old separation of powers is the scientific establishment. Today, governments cannot regulate air pollution without consulting atmospheric chemists. They cannot approve drugs without clinical trial data. They cannot manage a pandemic without epidemiologists. Scientists have become, in practice, an unelected fourth branch of government – but without any formal accountability.

Proposal: The scientific establishment should be organised as a constitutional pillar with three components:

The University System – responsible for fundamental research, education, and the training of future scientists. It would be constitutionally protected from political interference in its research agenda (subject to ethical review).
The Grande École System – modelled on the French grandes écoles, these elite institutions would train the technical leadership of the state (engineers, administrators, military officers). They would be responsible for applying scientific knowledge to practical problems.
The National Laboratory and Space Program System – large‑scale research infrastructures (e.g., CERN, ESA, national labs) are already independent institutions. Their unique position and mandate must be constitutionally recognised and institutionalised. Their directors would be appointed by a scientific council, not by the executive, to insulate them from political pressure and to preserve their long‑term research integrity.

The scientific establishment would have the right to advise the executive and the legislature, and the duty to warn when public policy contradicts established scientific consensus. It would have no veto, but its warnings would be published and debated.

The key innovation is that scientists would be barred from holding political office while serving in the scientific establishment – and politicians would be barred from serving on scientific advisory committees. The separation would be strict, to prevent the capture of expertise by power.

Sources for this paragraph:

Author’s proposal, inspired by the French Conseil d’État, the US National Academies, and the European Space Agency’s intergovernmental structure.

AI and Big Data as Instruments of Governance

Artificial intelligence and big data analytics have transformed the capacity of governments to model, predict, and optimise policy outcomes. They are not a new pillar, but a technological extension of the scientific establishment – a tool that, if properly governed, can dramatically improve the quality of evidence‑based decision‑making while respecting the constitutional separation of powers.

Decision support, not decision‑making. AI systems can analyse vast datasets to identify patterns, forecast trends, and simulate the effects of policy alternatives. They can help the executive allocate resources more efficiently, help the legislature assess the likely impact of proposed laws, and help the judiciary process evidence more systematically. But AI must never be granted final decision‑making authority over human beings in governance. A recommendation from an algorithm is a scientific opinion – admissible, contestable, and never binding.

Information processing and generative AI. The ability of large language models to synthesise, translate, and summarise complex bodies of text has immediate applications for governance. Legislative drafts can be checked for internal consistency. Public comments can be aggregated and categorised at scale. Regulatory guidance can be made accessible to government, citizens and enterprise in plain language. Generative AI can also assist the philosophical agora by producing structured summaries of ethical frameworks, and help the clergy by making theological and pastoral resources more accessible – though the interpretation of values and faith must always remain a human act.

A duty of algorithmic transparency. Any AI system used by the state must be subject to rigorous auditing. Its training data must be documented; its performance must be measured against industry standard benchmarks; its outputs must be explainable in plain language. Algorithmic decisions that affect individual rights – risk assessments in criminal justice, eligibility determinations for benefits – must be reviewed by a human with authority to override.

The data trust as a new institution. Big data governance requires a new institutional form: the data trust. A data trust is a legal entity that holds data for the benefit of specified beneficiaries – in this case, citizens. It would be a shared data lake, purpose-built for the production system in intended purpose, audited for quality and transparency.

No securitisation without oversight. Predictive analytics have legitimate applications in public safety – epidemic forecasting, infrastructure maintenance, disaster response. But when AI is used for surveillance, predictive policing, or social scoring, the burden of justification must be high. The scientific establishment must have the power to audit such systems and to issue public warnings when they are deployed without adequate safeguards. The legislature must explicitly authorise each category of predictive use, and the judiciary must have jurisdiction over challenges.

The European framework as a starting point. The EU’s proposed AI Act establishes a risk‑based classification: unacceptable risk (prohibited), high risk (mandatory conformity assessment), limited risk (transparency obligations), and minimal risk (no obligations). Finland should adopt a similar framework, with the added requirement that any high‑risk AI system used by the state must receive a positive opinion from the scientific establishment.

AI as an aid to deliberation, not a substitute for it. The deepest danger of AI in governance is not algorithmic bias or privacy violation – it is the erosion of deliberation. When legislators rely on AI‑generated summaries, they may lose the habit of reading the source material. When judges rely on algorithmic risk assessments, they may defer to a mathematical output instead of exercising their own judgment. The constitutional principle must be: AI may inform, but only humans may decide. Every AI output used in governance must be accompanied by a human‑readable explanation, and any person affected by a decision based on such an output must have the right to a human rehearing.

Sources for this sub‑section:

EU AI Act (proposed), risk‑based classification system.
Data trust literature (e.g., Open Data Institute, UK).
Algorithmic transparency and accountability research.
Author’s synthesis.

Deep Learning and Emerging AI: Beyond Generative Models

Deep learning continues to evolve at a dizzying pace, and the governance implications are becoming more acute. The year 2026 marks a structural turning point in the evolution of artificial intelligence and governance systems, as organizations across Europe and globally enter the era of the EU AI Act. The limitations of classical AI—its static training paradigm, open‑loop architecture, lack of contextual reasoning, and inability to adapt to wicked, continuously evolving environments—have become impossible to ignore.

Discussions around Agentic AI – autonomous systems capable of decision‑making – are expected to intensify in 2026, raising new regulatory questions. These systems do not simply generate content; they act. They can execute trades, manage supply chains, approve benefits, and, in the near future, make recommendations to welfare service, courts or parole boards. The governance challenge shifts from controlling outputs to controlling agency. A 2025 study on legal priorities for the development of political institutions in the age of AI identified eight legal principles for AI regulation, including state‑level strategic governance, the “human‑in‑the‑loop” principle, anthropological primacy, digital equality, and managed transparency. Within the electoral context, risks such as microtargeting, “dark advertising,” deepfakes, and automated bots undermine electoral integrity and facilitate manipulation of voters’ will. Ensuring the sovereignty and legitimacy of political institutions in the digital age requires the development of national AI models and robust legal regulation, including mandatory algorithmic audits and the prohibition of manipulative technologies.

Three emerging AI capabilities deserve particular constitutional attention:

First, the concentration of compute. Training frontier models requires vast computational resources, which are increasingly controlled by a handful of private firms. This creates a structural dependency – governments that rely on these firms for AI services are effectively outsourcing judgment to corporations whose primary loyalty is to shareholders, not to the public. The constitutional principle must be that sovereign AI capacity – the ability to train and deploy models without foreign dependency – is a matter of national security, not merely technical convenience.

Second, the opacity of deep learning reasoning. Even developers of deep neural networks cannot fully explain why a model produced a particular output. This “black box” problem becomes unacceptable when AI is used for high‑stakes decisions affecting rights and liberties. The duty of algorithmic transparency, discussed above, must be enforced with particular rigor for deep learning systems. Where explainability is impossible, deployment must be prohibited.

Third, the risk of model collapse and drift. Generative models trained on synthetic data – a growing practice as real human data becomes scarce – may degrade over time, a phenomenon known as “model collapse.” The system becomes less capable, more biased, and more confident in its errors. Continuous monitoring, already required under the EU AI Act’s lifecycle governance provisions, must be constitutionally mandated for any high‑risk deployment.

The EU AI Act will impose significant compliance obligations on businesses operating in or selling into European markets, particularly for high‑risk AI systems and general‑purpose AI models, with obligations for high‑risk systems applying from August 2026. The Act introduces a risk‑based classification: unacceptable risk (prohibited or theoretical only), high risk (mandatory conformity assessment), limited risk (transparency obligations), and minimal risk (no obligations). Finland should adopt a compatible framework, with the added requirement that any high‑risk AI system used by the state must receive a positive opinion from the scientific establishment before deployment. This would pilot an EU -wide institutionalization of the scientific establishment as an advisory body.

The Emergence of “Regenerative AI” as a Governance Paradigm (An Example)

Between 2026 and 2030, the entire landscape of artificial intelligence, governance, and organizational decision‑making will be redefined. A new generation of decision systems is emerging – systems that are adaptive, aligned, regenerative, and capable of supporting long‑horizon, multi‑stakeholder reasoning. This class of architectures, which we may call Regenerative AI as a generic term, moves away from the static “model‑centric” approach of the 2020s toward ecosystem‑centric intelligence: multiple agents, continuous loops, human‑AI collaboration, value‑sensitive reasoning, adaptive governance structures, systemic context tracking, and complex feedback cycles.

One illustrative framework (call it a “multi‑pillar regenerative framework”) organizes the essential capacities required for adaptive, responsible, and human‑aligned AI into five categories: (1) Regenerative Sensing – integrating human, environmental, and computational signals into continuously updated situational awareness; (2) Cognitive Alignment – ensuring coherence between human goals, contextual values, and machine reasoning; (3) Adaptive Reasoning & Argumentation – enabling models to navigate conflicts, uncertainty, and multi‑actor decision environments; (4) Temporal Governance – supporting long‑horizon reasoning, scenario continuity, and stability across evolving timeframes; and (5) System Renewal – embedding regenerative feedback loops that allow decisions, models, and processes to learn, self‑correct, and improve.

In the context of the EU AI Act, such a regenerative architecture offers a fundamentally more robust foundation for risk management, lifecycle documentation, conformity assessment, and post‑deployment monitoring, especially for high‑risk AI systems. Architectures of this type are among the first to treat regulation as a structural component, not a constraint; this may become the industry standard. The period between 2026 and 2030 will see: architectures built explicitly for compliance, autonomous monitoring agents, continuous documentation pipelines, real‑time risk dashboards, cognitive safety mechanisms, and embedded oversight modules.

The fundamental principle for governance systems in this new era must be: AI may inform, but only humans may decide. This is no longer a slogan. In a world of Regenerative AI systems that can simulate policy outcomes, forecast long‑term consequences, and adapt to changing conditions in real time, the danger is not that AI will be too stupid to trust – it is that AI will be too plausible to question. The “human‑in‑the‑loop” principle must be upgraded to “human responsibility for the loop” – meaning that no matter how sophisticated the AI, a named human being with legal accountability must sign off on every decision that affects individual rights.

Sources for this sub‑section:

General literature on adaptive and regenerative AI governance.
EU AI Act lifecycle provisions.
Author’s synthesis and generalisation of several proprietary frameworks.

Quantum Computing: The Governance Challenge of the Coming Decade

If AI is the governance challenge of the present, quantum computing is the governance challenge of the next decade – with consequences that may dwarf those of classical AI. Quantum computing is advancing quickly toward real‑world deployment, with the potential to transform sectors ranging from energy and finance to health care and cybersecurity. The policy window is open now. Governments that move decisively – through clear strategy, sustained investment, and international coordination – will shape how quantum technologies are deployed and governed globally.

Quantum computing presents a dual governance challenge: first, the ability of quantum computers to break classical encryption threatens the entire infrastructure of digital security – finance, communications, military systems. Second, the synthesis of quantum computing with artificial intelligence – Quantum‑AI (QAI) – creates agents with both classical and quantum capabilities, including breaking classical encryption and enabling agentic quantum simulation‑driven financial forecasting, creating novel governance challenges. The prospect of an agent exponentially more intelligent than a human, wielding such hybrid power, fundamentally changes the nature of control, demanding sophisticated solutions that go well beyond simple technical alignment.

The 2026 European Quantum Act is a generational opportunity to establish a governance framework that can foster innovation, ensure security, and cement the EU’s role as a key player in the global quantum race. The framework must combine agile, NLF‑style regulation (risk‑based tiers, regulatory sandboxes, principles of proportionality) with an ambitious, Chips Act‑style industrial policy (funding, securing critical mineral supply chains, accelerating lab‑to‑market pipelines). Central to this is a “standards‑first” philosophy, which treats technical standards not merely as tools for safety and interoperability, but as the primary mechanism for embedding democratic values directly into the technology’s architecture.

The concept of constitutionally constrained quantum AI (sometimes called “Quantum‑Resistant Constitutional AI” in the literature) represents the most advanced thinking on this front. The core argument is that if we cannot control a QAI agent tactically, we must constrain it architecturally, embedding a core set of ethical and safety principles (a “constitution”) that are not merely trained into the model but are formally verified and made robust against both classical and quantum‑algorithmic exploitation. This approach argues for a deeply integrated, multi‑layered framework focused on embedding AI ethics, safety, liability, regulatory navigation, stakeholder engagement, and lifecycle auditing for such advanced AI agents.

Sovereign quantum computing has become a strategic priority for countries, some of whom are already investing in the quantum ecosystem and quantum hardware. “Sovereign” in this context means that countries want to retain control of their data and, if possible, keep it within their borders. Placing a quantum computer in the EU, specifically in the Nordic region, means “it complies with EU [data privacy] regulations. Supporting European quantum computing infrastructure is key in acheiving quantum sovereignty.

Constitutional Implications of Quantum Computing

The constitutional principles that emerge from the quantum governance challenge are structural.

The principle of cryptographic sovereignty. As quantum computers threaten to break classical encryption, states must ensure that their critical infrastructure – military, financial, communications – is migrated to post‑quantum cryptography in a coordinated, timely manner. This is not a technical detail; it is a matter of national security and constitutional order.
The principle of post‑quantum transparency. Any QAI system used by the state must be capable of explaining its reasoning in a way that is verifiable by human auditors – even as quantum algorithms become increasingly complex. This may require novel verification techniques; the failure to develop them is a constitutional failure.
The principle of distributed capability. No single actor – whether a private corporation or a government agency – should control the entire quantum‑AI stack. The concentration of quantum capability is a concentration of power that threatens democratic governance. The constitutional order must therefore mandate the distribution of quantum research, development, and deployment across multiple institutions, with robust oversight. Open source is a critical tool in achieving this.
Dual-use technologies ought to be regulated EU -wide.

The constitutional principle for quantum governance must be constrain before capability. Unlike classical AI, where regulation has lagged behind development, quantum computing’s governance must be designed before the technology reaches maturity. The “Quantum Event Horizon” is not a distant abstraction. We are, in essence, rapidly approaching it. As we approach this horizon, the risk of bad actors repurposing these powerful agents – or the agents themselves posing an unmanageable control problem – increases dramatically. Navigating this boundary requires us to act proactively, embedding our values and control frameworks into responsible tech ecosystems now, before we cross a threshold where our ability to steer outcomes is lost.

In summary, the governance of deep learning, emerging AI, and quantum computing requires a multi‑layered approach that operates at three levels: technical (constitutional embedding, formal verification, post‑quantum cryptography), institutional (data trusts, sovereign capacity), and constitutional (binding principles that no algorithm may override). The new separation of powers must extend not only to the classical branches of government but to the algorithms and quantum agents that increasingly shape the conditions of human flourishing. These mandates would be enforced in production systems of public interest.

Sources for this sub‑section:

European Quantum Act (2026).
General literature on quantum‑AI governance.
Sovereign quantum computing strategy papers.
Author’s synthesis.

Part V – The Judiciary Remains Independent

The judiciary is the one pillar of the old separation that needs no fundamental redesign. An independent judiciary – with secure tenure, transparent proceedings, and the power of judicial review – remains essential. The only addition is that judges should be required to consult the philosophical agora (through its public record and Socratic Commission’s questions) and the scientific establishment on cases involving contested moral or technical questions. The opinions of those bodies would be advisory, not binding, but a judge who ignored them would have to explain why.

Sources for this paragraph:

Author’s synthesis of existing judicial systems.

Part VI – The Parliament, the Executive, and the Head of State

The political branches remain the core of democratic governance. Parliament makes laws, the executive implements them, and the head of state (whether monarch or president) serves as the guarantor of the constitution and the symbol of national unity.

But their relationship to the new pillars must be clarified:

Parliament would receive mandatory written submissions from the philosophical agora (the public record of debates and the Academia’s questions), the religious institutions (their declared axioms), and the scientific establishment before voting on any major legislation. Parliament or EU bodies could not legislate any law that breaks these axioms, or fails to answer the Academia’s questions, as these would tantamount to constitutional and treaty violations.
The executive would be required to consult the scientific establishment on any technical regulation. It could not issue a regulation that the scientific establishment had unanimously warned against, period.

Sources for this paragraph:

Author’s proposal.

Part VII – Two Speeds, One Destination: Why a Multi‑Speed Europe Is the Only Path to a European Federation

The European Union has 27 members, soon more. Each has its own history, its own economic structure, its own political culture, and its own appetite for integration. Waiting for all 27 to agree on every major step means moving at the speed of the slowest. In a world of continental powers – the United States, China, India – moving at the speed of the slowest is not moving at all. The solution is not to abandon the weak, but to allow the willing to move ahead. This is the logic of two‑speed Europe. It is not a betrayal of the European ideal. It is the only way to achieve it.

The Prodi Doctrine: A Two‑Speed Europe Is Not a Fallback, but a Necessity

Romano Prodi, former President of the European Commission and former Prime Minister of Italy, has been the most articulate defender of the two‑speed model. In a 2017 interview, he stated bluntly: “My Europe was different from today. Yes to a two‑speed Europe.” He recalled that during the Renaissance, Italian city‑states led in banking, technology, military art and commerce, but no single state could build the new ships needed for the New World, and Italy disappeared from the world map for four centuries. “Now we are in the same situation. Germany, France, Italy – none is large enough for the new globalization.”

Prodi has consistently argued that a two‑speed Europe is not a fallback but a necessity. In a 2025 interview, he stated: “The danger of a split in the European Union is no longer on the table, but I think changes will be slow. Two of the most important decisions have already been taken at ‘different speeds’: Schengen and the euro. I think this will be a general rule for Europe in the future, because in this way we will make progress and, at the same time, respect democracy. There is no alternative.”

He has called for “a core group of 10 countries” to organize common defense as a pilot project. His core insight is that waiting for all 27 to agree means not moving at all.

The French Proposal: A Coalition of the Willing

The French position, articulated most forcefully by President Emmanuel Macron, builds directly on Prodi’s logic. In February 2026, European leaders formally embraced the idea of a two‑speed union as the fastest way to break the political impasse over economic reforms needed to reboot the European economy. Macron set a summer deadline for a broad deal.

The European Commission, under Ursula von der Leyen, has taken up the idea, considering a two‑track approach that would allow a smaller group of at least nine EU members to press ahead using the treaty mechanism of enhanced cooperation. As Ursula von der Leyen explained, “Often we move forward with the speed of the slowest and the enhanced cooperation avoids that.”

The first movers would be the six leading economies: France, Germany, Italy, Spain, Poland and the Netherlands. The French Ambassador to Bulgaria, Éric Lebédel, clarified the two principles: “no country should be excluded from an area in which it wants to move forward, but no country should be able to block those who want to go faster and further.”

The Federalist Horizon: From Two Speeds to One Destination

The two‑speed Europe is not an end in itself. It is a pathway toward the long‑standing federalist vision of a “United States of Europe.” The manifesto of the Spinelli Group, signed by Mario Draghi, Emmanuel Macron, and many others, argues that “a federal Europe capable of swift, democratic decision‑making, rooted in our unique socio‑economic model” is the only way to preserve European sovereignty.

Draghi has presented federalisation not as a political choice but as a “historical necessity”, warning that without deeper integration Europe faces deindustrialisation and strategic irrelevance. Czech President Petr Pavel called in 2026 for a “United States of Europe”, arguing that a federal superstate is “probably the only solution for Europe if it wants to remain relevant.”

The German proposal of a “concentric circles Europe” built around a federal core of the Inner Six has been a recurring theme. Once a critical mass of countries has integrated, the federal union becomes self‑reinforcing. The periphery either converges or accepts that the centre will move without them. The ultimate goal remains what Prodi called “the Europe I dreamed of”: a genuinely federal union, capable of acting on the world stage as a single power.

Constitutional Implications for the New Separation of Powers

The new separation of powers must be multi‑level and multi‑speed:

Multi‑level: The same institutional pillars (philosophical agora, religious institutions as sources of axioms, scientific establishment, independent judiciary, parliament, executive, head of state) would exist at both national and EU levels, with clear rules for which level has primacy in which domain, guided by the principle of subsidiarity.
Multi‑speed: Member states that wish to integrate more deeply in a given policy area may do so through enhanced cooperation or through intergovernmental treaties. A core group of federalising countries could draft a Federal Compact – a treaty open to all, but initially binding only on those who ratify it. Over time, as more countries ratify, the federal core becomes the Union.

This is not a call for a European super‑state imposed from above. It is a recognition that climate change, migration, trade, defense and research are inherently cross‑border. The architecture of subsidiarity – that decisions be made as closely to the citizen as possible – is precisely what makes multi‑speed integration legitimate.

The European Union is not a state. It is a laboratory. The new separation of powers must treat it as such: an evolving, multi‑level, multi‑speed political order that is neither a confederation of sovereign states nor a federal super‑state, but something in between – and something that may, in time, become the first genuine post‑national federation in human history. The architecture of wisdom must be humble enough to accommodate this uncertainty and bold enough to provide a framework for its resolution.

Sources for this section:

Romano Prodi, Euronews interview (2 March 2017); Euractiv interview (24 October 2025).
Euronews (12 February 2026); Zawya (16 February 2026).
French Ambassador Lebédel (BNT, 24 March 2026).
Spinelli Group Manifesto.
Mario Draghi, cited by Jacek Saryusz‑Wolski (2026).
Petr Pavel (2026).
Enhanced cooperation provisions (Article 20 TEU).
Author’s synthesis and proposal.

Part VIII – What Remains Laïc

The French concept of laïcité – the strict separation of religion from the state – is one of the great achievements of modern political thought. It freed the state from clerical control and freed believers from state coercion. It created the space for religious pluralism by refusing to privilege any one confession. But laïcité was designed for a world where the church was the only organised spiritual power. Today, we face new forms of orthodoxy: ideological conformity in universities, partisan loyalty that mimics religious fervour, and the unacknowledged theology of “progress” that permeates technocratic governance.

What remains laïc in our new framework? Four things.

1. The state must not establish any religion. The philosophical agora is not a church. The religious institutions declare axioms, but they do not administer the state. The state does not pray.

2. The state must not interfere in the internal affairs of religious institutions. The clergy are independent in matters of doctrine, liturgy, and spiritual discipline. The state may regulate the temporal affairs of religious organisations (property, taxation, employment law) but may not dictate what they believe.

Defenses Against Extremism

The prohibition on state interference is not a licence for religious institutions to become lawless enclaves. Every constitutional order has the right – and the duty – to defend itself against those who would use religious authority to incite violence, subvert democratic institutions, or violate the fundamental rights of citizens. Non‑interference is a presumption, not an absolute.

The European Court of Human Rights has recognised that the state’s duty of neutrality does not require passivity. States have a positive obligation to ensure that religious groups can coexist freely, and to maintain neutrality when inner disputes arise between rival denominations. The same duty obligates the state to protect the rights of individuals – including vulnerable members of religious communities – against abuses committed under the cover of religious authority.

Defences against extremism take a bounded form.

Legal red lines. The state may criminalise incitement to violence, hate speech, and the advocacy of terrorism – regardless of whether the speech originates in a pulpit. These are not theological judgments; they are ordinary criminal law applied equally to all. The state may also prohibit practices that violate the fundamental rights of others (e.g., forced marriage, female genital mutilation), even when those practices are claimed as religious obligations. The prohibition on interference does not extend to the enforcement of general, neutral criminal laws.

Financial transparency and accountability. The state may require religious organisations to disclose their funding sources, particularly those originating from foreign states or entities with a documented history of supporting extremism. This is not interference in internal affairs; it is the enforcement of anti‑money‑laundering and counter‑terrorist financing regulations that apply to all non‑profit entities. Foreign funding that comes with strings attached – the promotion of a particular ideological line, the exclusion of certain voices from religious education – is a legitimate subject of state scrutiny.

A duty of care for institutional integrity. Religious institutions that claim the privilege of autonomy must also accept the responsibility of self‑governance. The state may require that such institutions have transparent governance structures, accessible complaints mechanisms for victims of abuse, and clear procedures for the removal of clergy who incite violence. The state may not choose the next bishop, but it may insist that the process for choosing the next bishop is not rigged by a foreign power or a corrupt faction. EU policy briefs on counter‑radicalisation have warned that “restrictive interference in religious institutions can be ineffective and even counterproductive when it comes to addressing radicalisation”, while “benefits can accrue from states nurturing supportive and positive relations with religion”.

Religious literacy and dialogue as preventive tools. Finland’s national action plans for the prevention of violent extremism have recognised the role of religious literacy – the capacity to understand religious diversity without reducing it to security threats. The state’s best defence against extremist recruitment is not surveillance but positive engagement: supporting moderate religious leaders who counter extremist narratives, funding interfaith dialogue, and ensuring that the state’s own counter‑radicalisation programmes do not securitise entire communities. The French government has stated that “dialogue between the state and Muslim religious representatives should be based on trust and responsibility”, noting that “Muslims reject the distortion of their faith by extremist ideologies”.

No theological policing. The red line is sacred. The state may not vet sermons for orthodoxy. It may not remove a religious leader for preaching a doctrine that is merely unpopular, provocative, or politically inconvenient. It may not appoint or remove clergy. The defence against extremism operates at the boundary of action, not at the boundary of belief. A preacher who calls for peaceful withdrawal from politics is protected; a preacher who calls for the bombing of a parliament is not.

The ultimate defense: the loyalty of citizens, not institutions. The most reliable defence against religious extremism is not the surveillance of mosques or the vetting of imams. It is the cultivation of a citizenry that is attached to democratic institutions, that can distinguish between faith and fanaticism, and that will not trade its freedom for a false promise of heavenly reward. The state cannot create such a citizenry through regulation alone. It requires the cooperation of families, schools, civil society, and – yes – religious institutions that choose to be part of the solution rather than part of the problem.

Sources for this sub‑section:

European Court of Human Rights, Metropolitan Church of Bessarabia and Others v Moldova, positive duty of state neutrality.
EUI Policy Brief: “state‑religion partnerships can reduce socio‑political alienation and disenfranchisement”.
French Interior Minister Retailleau: state may support but not interfere; Muslims reject extremist distortion of their faith.
Finland’s National Action Plans for the prevention of violent extremism: religious literacy approach.
Author’s synthesis of legal and policy principles.

3. The public square must remain open to all voices, religious and secular. Laïcité is not a gag order. It is a guarantee that no single voice can silence the others. The agora speaks reason. The clergy speaks faith (and declares axioms). The scientist speaks evidence. The citizen speaks opinion. All are heard. None commands.

4. The separation must be mutual. Just as the state does not dictate to the church, the church does not dictate to the state – but the church’s declared axioms are binding as a matter of constitutional principle. The clergy may not veto specific laws, but a law that violates an axiom is unconstitutional. The agora may counsel; it may not command. The scientific establishment may warn; it may not legislate. However, the scientist under the proposed model can constrain the Parliament from cutting the rainforest. The Parliament cannot legislate to allow cruel treatment of the poor. The Executive cannot unilaterally go to war. The axioms declared by the clergy as Montesquieu would have put it, are pre-political constraints that invalidate secular laws or actions violating them. The scientist cannot lobby their pet project. The philosopher cannot convert everyone.

What remains laïc is the neutrality of the political sphere in matters of worship and doctrine. Parliament is not a church. The executive is not a seminary. The courts are not a council of bishops. They are human institutions, bound by human laws, accountable to human citizens. But they are also bound by the moral axioms declared by the religious institution – because those axioms are the foundation of the constitution itself. These axioms are never specific enough to invade space from other institutions. The clergy cannot order everyone to wear a certain garment, or dictate family law on non-members, for example. Crucially, this would mean ex-communication of foreign state-controlled cleargies from the axiomatic pre-political process. This would be ensured by impartiality and anti-commingling clauses.

Laïcité is not hostility to religion. It is separation of the church and state.

Sources for this paragraph:

French law of 1905 on the separation of churches and state.
Constitutional Council of France, laïcité jurisprudence.
Author’s synthesis.

Part IX – Conclusion: The Architecture of Wisdom

The old separation of powers was designed to prevent tyranny. It succeeded, more or less. But preventing tyranny is not enough. We also need to ensure that governance is wise. Wisdom requires knowledge (science), values (philosophy), meaning (religion), and justice (law). No single institution can supply all four.

The framework proposed here is tentative. It is not a blueprint for immediate implementation, but a provocation to think more systematically about the roles of different forms of knowledge and authority in a complex society. The clergy cannot be reduced to lobbyists. The philosophers cannot be reduced to commentators. The scientists cannot be reduced to contractors. Each has a distinct function, and each must be given a distinct institutional home – but the philosophers’ home is not a council. It is an agora. It is a perpetual debate. And the religious institution’s home is not a department of state. It is a source of law – the source of first principles.

The architecture of wisdom is not a machine. It is a garden. It requires careful tending, constant pruning, and a willingness to let some parts grow wild. But without architecture, the garden becomes a swamp. We have been gardening without a plan for too long. It is time to draw the map.

Sources for this paragraph:

Author’s conclusion.

Sources for this blog post (by section):

Part I: Montesquieu (1748); author’s synthesis.
Part II: Author’s proposal, inspired by Athenian agora and the European tradition of academic freedom; contrasts with Platonic model.
Part III: Author’s proposal – religious institutions as sources of law, declaring axioms; based on Aquinas, Blackstone, canon law tradition.
Part IV (Scientific Establishment): Author’s proposal; EU AI Act; data trust literature; algorithmic transparency research; General literature on adaptive and regenerative AI governance; European Quantum Act; general quantum‑AI governance literature; sovereign quantum computing strategy papers; author’s synthesis.
Part V: Standard judicial independence literature.
Part VI: Author’s synthesis of parliamentary and executive governance.
Part VII: Romano Prodi (2017, 2025); Euronews (2026); French Ambassador Lebédel; Spinelli Group; Mario Draghi; Petr Pavel; enhanced cooperation provisions; author’s synthesis.
Part VIII: French law of 1905; Constitutional Council jurisprudence; ECtHR Metropolitan Church of Bessarabia; EUI Policy Brief; French Interior Minister Retailleau; Finland’s National Action Plans; author’s synthesis.
Part IX: Author’s conclusion.

End of post.