The Sovereign Research Agenda: 10 Papers That Will Define the Next Decade

On a spring morning, the paperwork arrived before the power did

In March 2024, the European Parliament approved the EU AI Act: the world’s first broad, risk-based law for artificial intelligence. Long before most citizens had used an autonomous agent to negotiate a bill, book medical tests or move money between jurisdictions, legislators had already begun sketching the perimeter of acceptable machine behaviour. The sequence matters. Governance is arriving before the full social fact of agency has settled.

That inversion is not unusual. Modern societies routinely legislate under technological uncertainty. GDPR was drafted before data portability became an everyday consumer concern. Decentralised finance expanded faster than most securities regulators could classify it. DAOs acquired treasury scale before courts had decided whether they resembled partnerships, unincorporated associations or something genuinely new. In health, advances in biological age measurement and AI-enabled diagnostics are moving more quickly than consent frameworks designed for static records and episodic care.

The consequence is now visible across identity, data, money, health and governance. We do not merely lack finished answers. We lack a coherent research agenda for sovereignty itself: a disciplined account of how persons, communities and institutions retain meaningful authority in an environment increasingly mediated by software, automated inference and cross-border digital infrastructure.

The Sovereign Research Agenda is therefore not a list of fashionable topics. It is a map of the unresolved questions on which practical freedom will depend. It sits within The Sovereign Standard, the broad, open framework for retaining sovereignty in the AI age across human and institutional life. Within that umbrella sits F-ACT — the Framework for Agent Conformance & Trust — a neutral, open, vendor-neutral standard for AI-agent governance built on ASDAR: Authority, Scope, Data, Audit, Revocation. And beneath both sits the implementation layer: the 42 Protocols, Society OS’s deployable mechanism for operationalising sovereignty in practice.

Their animating principle is simple enough to fit on a wall and strict enough to shape infrastructure: govern before execution — not after.

The ten papers below are the work required to make that principle real.

1. Algorithmic constitutions: can rules for software become rules for power?

Every sufficiently important technical system acquires constitutional features. It defines who may act, what they may do, what counts as valid procedure, how disputes are resolved and how authority can be withdrawn. Platform terms of service already function as a kind of private administrative law. Smart-contract protocols encode monetary policy. App-store rules determine access to markets. AI model policies increasingly govern speech, safety and economic participation.

The unanswered question is whether these fragments can be assembled into algorithmic constitutions that are both machine-executable and politically legitimate.

The research challenge has three parts:

• how to express constitutional constraints in forms machines can reliably interpret;
• how to preserve due process, contestability and amendment over time;
• how to prevent technical operators from becoming de facto sovereigns.

There is a real regulatory backdrop. The GDPR established rights around access, correction and erasure. The EU AI Act imposes obligations around risk management, documentation, human oversight and prohibited uses. Digital identity frameworks such as eIDAS 2.0 are moving towards interoperable trust services across the Union. Yet these instruments remain largely external to the software systems they regulate. They tell organisations what must be true, but rarely ensure the software itself is constitutionally structured to make it true by design.

This is precisely where the Sovereign Standard becomes useful. Its claim is not that law should be replaced by code, but that constitutional safeguards should be rendered into governable technical primitives. In the 42 Protocols, the Human-Twin-Agent Protocol addresses the foundational constitutional question: who is acting. WISE Contracts address a second: how rules execute as law-like obligations rather than brittle code paths. The research paper this field needs would define minimum constitutional requirements for software-mediated governance: identity, competence, consent, amendment, evidence and appeal.

A serious algorithmic constitution would have to be complete by construction in one specific sense: every permitted action must have an attributable authority path, and every authority path must have a revocable legal and social basis.

2. Verifiable AI alignment: from promises of safety to evidence of obedience

“Alignment” has become one of the most overused and under-specified words in technology. It often describes a hoped-for correspondence between a model’s outputs and human values. But in governance settings, that is too vague. A tax agent, health assistant or procurement system does not need metaphysical harmony with civilisation. It needs demonstrable obedience to bounded authority.

This is why the key research question is not abstract alignment, but verifiable alignment in context. Can an AI agent prove that it acted under authorised instructions, within approved scope, on permitted data, with a reconstructible audit trail and a live revocation channel?

Current practice falls short. Leading model providers publish safety cards and system cards. NIST’s AI Risk Management Framework provides a useful vocabulary for mapping, measuring and managing risk. The OECD AI Principles articulate broad norms. But these are mostly ex ante documents and ex post assurances. They do not yet create a universal conformance fabric for agents operating across enterprises, states and individual users.

F-ACT exists to fill that gap. Its core, ASDAR — Authority, Scope, Data, Audit, Revocation — is deliberately narrow and technical.

• Authority: who empowered the agent?
• Scope: what may it do, and under what limits?
• Data: what information may it access, derive or disclose?
• Audit: what evidence exists of each consequential action?
• Revocation: how is permission withdrawn instantly and universally?

The corresponding paper should define testable conformance levels for agent systems, using F-ACT’s tiers:

• L0 Unattested
• L1 Declared
• L2 Enforced
• L3 Provable

The next decade’s decisive governance battles will be about who retains authority when software begins to act.

The progression matters. Most current systems sit somewhere between L0 and L1: they declare policies. Very few are truly L2, where enforcement is built into runtime constraints. Fewer still approach L3, where cryptographic proofs, tamper-evident logs or formal verification can demonstrate conformance.

In practical sovereignty terms, this is the difference between asking an assistant not to spend above a threshold and being able to prove it cannot.

3. Decentralised identity at scale: can the self become portable without becoming exposed?

Identity is the load-bearing beam of digital sovereignty. If individuals cannot prove who they are, what they are entitled to do, and which parts of themselves they choose to reveal, then data rights, financial rights and political rights become fragile abstractions.

The technology stack is advancing. W3C standards for Verifiable Credentials and Decentralized Identifiers have matured. eIDAS 2.0 and the European Digital Identity Wallet are pushing state-backed digital credentials towards broad deployment. Mobile hardware now routinely supports secure enclaves and passkeys. Yet the central dilemma remains unresolved: how can identity be portable and interoperable without becoming a universal surveillance surface?

That is not merely a technical issue. It is a governance issue at the junction of privacy law, authentication, anti-money laundering obligations and social trust. A national ID may simplify service access while intensifying centralised visibility. A self-sovereign credential may preserve selective disclosure while struggling to achieve mass usability or institutional recognition.

The paper required here should examine identity as a layered system:

• civil identity issued or recognised by states;
• economic identity for payment, employment and compliance;
• social identity for reputation and community trust;
• agent identity for delegated software action.

The 42 Protocols treat this not as four separate silos but as one orchestrated graph. The Human-Twin-Agent model links the human principal, their digital twin and their governed agents, allowing a person to delegate specific authority without yielding personhood to a platform. In health, that might mean granting an agent permission to gather laboratory results, compare them against recognised clinical guidance and schedule a consultation, while denying it access to unrelated records or payment authority. In finance, it might mean permitting tax preparation but not asset transfer. In civic life, it might mean casting a vote on procedural matters while prohibiting any onward delegation.

At scale, identity sovereignty will depend less on slogans about self-ownership than on the mundane architecture of selective disclosure, liability allocation, revocation and standards interoperability.

4. Cross-jurisdictional smart contracts: when code travels farther than law

Digital commerce routinely crosses borders that legal systems do not. A freelance designer in Lagos, a client in Berlin, stablecoin settlement through a protocol with validators across several continents and an arbitration clause pointing to London: this is no longer exotic. It is daily infrastructure. Yet smart contracts remain notoriously weak at handling the messiest part of law — ambiguity, changed circumstances and conflicting jurisdictions.

The central research question is therefore not whether smart contracts execute. They plainly do. It is whether they can operate across jurisdictions without severing themselves from legitimate legal interpretation.

Courts and lawmakers are beginning to engage. The UK Jurisdiction Taskforce has already argued that cryptoassets can, in principle, be treated as property under English law and that smart contracts can give rise to legal obligations. Common-law courts in several jurisdictions have shown willingness to recognise digital-asset disputes. At the same time, regulators continue to press anti-money laundering, sanctions and consumer-protection duties onto infrastructure that was built to route around central intermediaries.

This is exactly the terrain for WISE Contracts in the 42 Protocols: agreements designed to execute law, not merely code. That distinction is not rhetorical. A code-only contract can move an asset when a condition is met. A law-aware contract must also understand jurisdictional selection, exceptions, evidence standards, emergency suspension, human review and appeal.

A robust paper on this subject would ask:

• how legal prose and machine-readable rules should be linked;
• which events require automatic execution and which require human adjudication;
• how identity, consent and capacity are verified across borders;
• how an agent can be authorised to contract on behalf of a person or organisation.

Without such work, cross-border automation will remain efficient at the margins and brittle at the centre.

5. Regenerative economics: can digital systems measure wealth without consuming the future?

Industrial accounting is very good at counting what passes through markets and very poor at protecting the preconditions that make markets possible. Soil fertility, mental resilience, social trust, unpaid care and long-term public-health capacity still sit awkwardly inside dominant economic models. Yet digital systems are increasingly capable of measuring flows once considered invisible.

The question is whether they can do so without collapsing into technocratic overreach or greenwash by spreadsheet.

This matters for sovereignty because a person or community that cannot govern its data about land, labour, health or local production cannot properly govern its future. The rise of tokenised assets, carbon markets and on-chain treasuries has shown both the promise and the danger. New instruments can widen participation and improve traceability; they can also repackage old extraction in novel wrappers.

A research paper on regenerative economics should be disciplined about evidence. There is credible work from ecological economics, public-health research and development finance showing that long-term prosperity depends on natural, social and human capital, not merely financial throughput. There are practical digital tools for better measurement: remote sensing, auditable supply-chain data, programmable disbursements and transparent treasury management. But there is still no settled governance framework for linking those tools to rights, obligations and local agency.

Govern before execution — not after.

Here HEARTrank becomes important as a trust layer within the 42 Protocols. In a regenerative system, the question is not only what asset exists, but what is trusted about its provenance, stewardship and effects. A community water-restoration programme, for instance, could use governed credentials, WISE Contracts and trust scoring to ensure that funding releases only when independently attested milestones are met, while residents retain visibility into data use and decision rules.

Regeneration worthy of the name requires more than better dashboards. It requires accountable institutions and rights-preserving infrastructure.

6. Trust-network theory: what does legitimacy look like in a world of machine-mediated reputation?

Trust is often discussed as sentiment. In digital systems it is structure. It determines who is believed, who is financed, whose content circulates, whose identity is accepted and whose model outputs are acted upon. Yet today’s trust machinery is fragmented: credit scores, platform ratings, Know Your Customer checks, academic citations, app-store approvals, social graphs and procurement lists all operate as partial, siloed proxies.

The open question is whether trust can be modelled as a governed network rather than a collection of opaque scores.

This has immediate relevance. Financial institutions must assess counterparties. Online marketplaces must suppress fraud. Health systems must know whether a data source is authentic. Agentic systems must decide which tools, APIs, documents and counterparties to rely upon. In all these settings, trust is not binary. It is contextual, revisable and path-dependent.

A serious paper would bring together network science, cryptography, institutional economics and administrative law. It would distinguish identity from reputation, reputation from authority and authority from legitimacy. It would ask how trust should propagate, decay, be challenged and be repaired.

The Sovereign Standard’s contribution is to insist that trust cannot remain a platform black box. Individuals and communities need inspectable grounds for consequential trust decisions. That does not mean every score must be public; it means the basis of trust must be governable. In implementation terms, HEARTrank can be understood as an attempt to make trust legible without making persons transparent. It points towards a system in which attestations, behaviour, institutional standing and revocation signals can be combined under explicit rules rather than hidden heuristics.

In the AI age, trust that cannot be audited becomes power that cannot be contested.

7. Space property rights: the oldest legal question is about to become extraterrestrial again

For centuries, property law has turned on recognisable questions: possession, use, exclusion, transfer, liability. Space reopens all of them under unfamiliar conditions. The 1967 Outer Space Treaty prohibits national appropriation of outer space by claim of sovereignty, use or occupation. The 1979 Moon Agreement proposed a stronger common-heritage approach, though major spacefaring states did not ratify it. More recently, the Artemis Accords have sketched practical principles for interoperability, deconfliction and resource activities.

Yet a deep ambiguity remains. If no state may appropriate celestial territory, what legal architecture governs extraction, habitation, stewardship and dispute resolution beyond Earth?

This belongs in a sovereign-life agenda because the legal logic forged in space will feed back into Earth systems. Questions of commons governance, resource rights, machine autonomy, environmental stewardship and jurisdictional overlap are already present in the digital domain. Space merely strips them to first principles.

The paper needed here would not indulge fantasy. It would examine current treaty law, national space-resource legislation, insurance practice, licensing regimes and the operational realities of satellite infrastructure. It would then ask what a sovereignty-preserving framework looks like when communications are delayed, enforcement is remote and autonomous systems must make local decisions.

The Sovereign Standard is relevant because it distinguishes sovereignty as a retained capacity for legitimate self-governance from sovereignty as territorial statehood. That distinction is essential in space. So too is governed agency. A mining robot, habitat management agent or orbital logistics system cannot simply be "autonomous" in the colloquial sense. It must be tethered to attributable authority, bounded permissions and auditable action through standards akin to F-ACT.

8. Planetary AI ethics: can local values survive global models?

A handful of model architectures, training pipelines and deployment interfaces now mediate language, knowledge retrieval, recommendation and increasingly decision support across the world. This concentration creates a planetary ethical problem. Local communities possess distinct moral traditions, legal norms and risk tolerances. Foundation models, by contrast, are trained on globally scraped corpora and tuned through relatively centralised institutional processes.

The question is not whether universal norms should exist. Some clearly should: prohibitions on fraud, abuse and unlawful discrimination are not merely local preferences. The question is how to preserve moral pluralism under common technical infrastructure.

There are live policy anchors. UNESCO’s Recommendation on the Ethics of Artificial Intelligence sets high-level principles. The Council of Europe’s Framework Convention on AI, Human Rights, Democracy and the Rule of Law seeks a broader legal basis for accountability. National regimes from Brazil to Singapore to the European Union are converging on notions of transparency, safety and human oversight, while differing on enforcement style and institutional design.

The missing paper would explore how ethical constraints can be layered:

• universal red lines;
• jurisdiction-specific legal obligations;
• community-level norms;
• individual delegation preferences.

This layering aligns closely with the Sovereign Standard’s architecture. The broad worldview sets the human and institutional principles. F-ACT defines the agent-governance controls. The 42 Protocols operationalise those controls through identity, trust and lawful execution. Practical sovereignty emerges when a person, family, clinic, enterprise or municipality can configure machine assistance around their legitimate norms without surrendering interoperability with the wider world.

That is a more demanding ambition than content moderation. It is the design of plural civilisation on shared infrastructure.

In the AI age, trust that cannot be audited becomes power that cannot be contested.

9. Quantum governance: what happens when verification itself changes?

Quantum computing remains an uneven field: impressive in scientific promise, limited in present-day commercial impact and often inflated in popular discussion. But one aspect is not speculative. A sufficiently capable quantum system could threaten widely used public-key cryptography, including schemes that underpin secure communication, digital signatures and blockchain-based systems.

That prospect makes quantum governance a current, not future, research problem. NIST has already selected post-quantum cryptographic algorithms for standardisation. Governments and critical infrastructure operators are beginning the long migration towards quantum-resistant security. Because such transitions take years, governance frameworks designed today must anticipate changed verification assumptions.

The paper needed here is not merely about cryptography. It is about institutional continuity under cryptographic transition. How should identity systems rotate keys and preserve trust chains? How should long-lived contracts remain enforceable? What becomes of historical audit if signature assumptions weaken? How can communities avoid being forced into abrupt centralisation simply because the verification layer changes?

The 42 Protocols are useful here precisely because they are an implementation mechanism, not a single cryptographic bet. A Living OS must be able to migrate primitives while preserving constitutional invariants: attributable action, bounded authority, revocation and evidence. The sovereign question is not whether a particular algorithm survives. It is whether persons and institutions retain control during migration.

10. Civilisational continuity planning: how do societies remain governable under stress?

Pandemics, grid failures, cyber disruption, demographic ageing, biosurveillance, climate shocks and institutional mistrust all point to the same neglected domain: continuity. Modern states and firms have business continuity plans. Far fewer societies have credible continuity plans for identity, health governance, social provision and lawful agency under prolonged stress.

Covid-19 revealed the weakness starkly. Governments struggled to balance emergency powers, data sharing, public trust and service delivery. Health systems improvised new digital pathways at speed. Remote work and online schooling exposed dependence on a small number of platforms. The lesson was not that digitisation failed. It was that digitisation without sovereignty becomes brittle when conditions deteriorate.

A continuity paper would ask which functions must survive any crisis:

• identity verification;
• health-record access and consent;
• payment and benefits continuity;
• trustworthy public communication;
• lawful delegated action by software agents;
• local decision capacity when central systems fail.

This is where the Sovereign Stack stops sounding philosophical and starts sounding infrastructural. A resilient society needs governed identity, trust networks, executable legal processes and revocable software delegation. It needs the ability for a municipality, hospital, mutual-aid network or family office to continue operating when upstream services are degraded. In Society OS terms, that is the practical purpose of the Living OS: not novelty, but durable self-governance under changing conditions.

From agenda to institution: how research becomes sovereignty

Research agendas often fail because they confuse conceptual breadth with institutional sequence. These ten papers should not be written as isolated essays. They form a dependency chain.

Identity underpins agency. Agency requires conformance. Conformance depends on audit and revocation. Lawful execution depends on governed identity and trusted evidence. Regenerative economics depends on lawful execution and trusted measurement. Continuity planning depends on all of the above.

Seen in that order, The Sovereign Standard is the umbrella theory of the whole field: a published, stewarded framework for retaining human and institutional sovereignty in the AI age. F-ACT is its technical governance pillar for AI agents. The 42 Protocols are the implementation mechanism: the deployable stack through which the theory becomes systems that people and organisations can actually use.

That nesting matters because too much governance discussion still oscillates between vague values and single-point products. The real task is architectural. A sovereign society needs a worldview, a standard and a stack.

Society OS’s own implementation path reflects this layered logic. The Human-Twin-Agent Protocol addresses attributable identity and delegation. HEARTrank addresses trust legibility. WISE Contracts address law-aware execution. Together they form the Sovereign Trinity at the heart of the 42 Protocols — a practical machinery for deciding who acts, what is trusted and which rules execute.

There is an intellectual motif here as well as an engineering one: 42 years. 42 protocols. 42 papers. Not as branding flourish, but as an admission that civilisational infrastructure is built through sustained sequences of thought, standard-setting and deployment.

The decade ahead

The next decade will not be defined by whether machines can write better emails, generate prettier images or answer more questions. It will be defined by whether societies can preserve legitimate human and institutional authority as machine agency diffuses into everyday life.

That is why these ten papers matter. They are not peripheral future studies. They are the missing constitutional documents of the software age.

If written well, they would help answer the questions already bearing down on citizens and institutions: how a patient delegates to a medical agent without losing privacy; how a worker carries credentials across borders without becoming permanently tracked; how a community funds ecological repair without surrendering control of its data; how a contract can execute globally while remaining answerable to law; how an AI system can prove not simply that it is capable, but that it is obedient.

The best governance technologies do not make people smaller. They make power more attributable, permission more precise and institutions more contestable. That is the promise at the core of the Sovereign Standard, the discipline inside F-ACT and the practical ambition of the 42 Protocols.

The unanswered questions are now clear enough to name. The work of the next decade is to answer them before the infrastructure hardens without us.