Le texte de l’ouvrage est en anglais, tel qu’il a été écrit ; seule l’interface du site est en français.

1. Introduction

Chapitre 1 sur 15 · ≈ 12 min de lecture · 38 paragraphes · 7ᵉ des 40 sections de lecture · rechercher dans l’ouvrage

Globalisation has displaced the effects of decisions far faster than it has displaced the mechanisms of responsibility.

1.1 Territorial decisions, networked effects

Globalisation has displaced the effects of decisions far faster than it has displaced the mechanisms of responsibility. Public policies, financial flows, digital infrastructures, value chains, technologies and climate risks are still governed within determinate legal boundaries; their consequences cross borders, combine, alter the reactions of other actors and may extend over several generations. The central question is therefore no longer only whether a decision is procedurally sound, rational and effective in the jurisdiction in which it is taken. It is also to determine how effects displaced onto third parties become visible, debatable, predictable where this is possible and, where the conditions of proof and of law are met, correctable.¶ 1

A monetary decision taken in a large economy may alter, elsewhere, the cost of refinancing, capital flows, the exchange rate, debt service, monetary policy responses and fiscal space. A digital infrastructure may transform access to information, to work, to credit and to public services, while exerting pressures on the water, electricity, land, competition and data of territories that do not fully participate in its governance. A value chain may distribute production across several continents, concentrate value in certain jurisdictions and displace social or environmental risks towards those whose bargaining capacity is weakest. In such systems, benefits and costs coincide neither in space, nor in time, nor across the same agents.¶ 2

1.2 The problem is not primarily moral: it is institutional

This mismatch is often described as a conflict between those who decide and those who bear the consequences. This formulation is too simple. A central bank expert, a ministry, an agency, a firm or an international organisation acts within a mandate, with an objective function, models, indicators, legal constraints and determinate account-giving mechanisms. Even where external effects are studied, they do not necessarily receive the same weight as the domestic objectives for which the institution must answer. A decision may thus be rigorous, lawful and rational within its own framework, while remaining incomplete with regard to the systemic scope of its effects.¶ 3

At the other end, the affected authorities, firms and populations observe the consequences: depreciation, a rise in the cost of debt, credit contraction, job losses, rising food costs, pressure on water or energy, reduced public investment. Their experience of the harm is real; it is not, however, sufficient to establish the share imputable to an external decision. The structure of the debt, domestic policies, the exchange rate regime, balance-sheet imbalances, the depth of markets, adaptive capacity and other simultaneous shocks may amplify or attenuate transmission. The upstream decision-maker may under-observe the cascade; the downstream actor may overestimate the contribution of a single cause. Both perspectives are situated and partial.¶ 4

The object of an accountability architecture is therefore neither to put the source jurisdiction's intentions on trial nor to accept without examination the account given by the recipient country. It is to construct a common, documented and contestable representation: what was decided; what was transmitted; through which mechanisms; with what uncertainty; who benefited; who bore the cost; what share stems from internal vulnerabilities; and what consequence may be attached to the level of proof attained. An externality presupposes neither malice nor conscious fault. Symmetrically, compliance with a domestic mandate does not render non-existent the effects produced on third parties.¶ 5

1.3 The global accountability deficit

The international system lacks neither organisations, nor expertise, nor norms, nor funding. It produces data, adopts standards, coordinates certain policies, examines disputes and mobilises resources. But these capacities remain sectoral, unevenly distributed and imperfectly connected. An effect may be measured without being qualified; qualified without proportionate consequence; recognised without funding; funded without responsibility being identifiable; or dealt with without the affected persons being able to know the reasons, produce a counter-assessment and obtain a re-examination.¶ 6

The present work gives the name global accountability deficit to the situation in which the transformations required to move from knowledge of an effect to a legitimate response are not all covered, are not entrusted to identifiable duty-holders, or are not connected by procedures capable of preserving provenance, uncertainty, reasons and remedy. The deficit is not to be confused with the absence of a world government, nor with the lack of development finance, nor with the non-implementation of international law. It designates a discontinuity: the functions often exist separately, but no stable architecture guarantees that the evidence survives their linkage.¶ 7

1.4 The theoretical contribution: three completenesses

The central proposition is the Global Accountability Chain. It defines accountability as a property of the complete architecture rather than of an isolated body. Functional completeness requires that the operations necessary to the handling of the case file be covered. Relational completeness requires that each duty-holder of a function answer before an identifiable forum capable of demanding information, explanation, justification, correction and, where the law permits, consequence. Procedural completeness requires that the interfaces carry not a mere number or a final decision, but the source data, the assumptions, the version of the models, the uncertainty, the objections, the dissents and the avenues of re-examination.¶ 8

This theory is situated within established literatures. Accountability has defined the actor-forum relation; global administrative law has brought transparency, participation, reason-giving and review to the fore; rational design has studied the institutional variables suited to cooperation problems; regime complexes and polycentric governance have shown the benefits and the costs of fragmentation; orchestration and experimentalist governance have described forms of action without central hierarchy. The contribution proposed is to connect these achievements within a multi-stage chain, to raise the interface to the rank of an object of governance, and to combine negative mandates, stopping rules, comparison of architectures and procedural preservation of the evidence.¶ 9

1.5 Two loops: responding and preventing

A chain that begins only after the harm has materialised remains insufficient where the mechanisms of propagation are reasonably foreseeable. In interdependent systems, an initial decision alters the conditions to which other actors react; these reactions in turn become shocks. A cascade may thus result from locally justified decisions without having been globally intended.¶ 10

The present work therefore distinguishes two loops. The Global Accountability Chain remains the principal loop: it handles an observed or sufficiently measurable effect, from evidence to redress. The Preventive Accountability Loop is cross-cutting: anticipating trajectories, detecting exposures and thresholds, qualifying an alert, notifying decision-makers, obtaining a reasoned response, preparing anticipatory action and evaluating ex post the quality of the alert and of the response. Prevention is neither a guarantee that no harm will occur nor a transfer of competence to the forecaster. It is a procedural duty of diligence: a material risk, cross-border or intergenerational, reasonably foreseeable and still actionable, must be able to enter into the decision.¶ 11

The two loops share a versioned case file. An alert may remain under monitoring, trigger a capacity or solidarity measure, or open a case file of the chain where effects become observable. Conversely, an effect dealt with ex post may reveal a recurrent mechanism and improve future scenarios. No alert automatically creates pricing, a sanction or a presumption of responsibility.¶ 12

1.6 Functions before institutions

The approach is rights-grounded, function-first. It first distinguishes four registers: legally protected rights; procedural guarantees; decision principles such as proportionality, precaution and subsidiarity; and normative objectives such as equity, sustainability and solidarity. These registers precede the definition of the functions; the choice of bodies comes afterwards.¶ 13

The 7+3 is a first candidate instantiation. The functions are: measuring; qualifying; determining the admissible consequence; applying certain adjustments; financing prevention, transition or solidarity; certifying digital systems; organising remedy. They are framed by the Carthage Charter, the Global Accountability Framework and the Call of Carthage. They should be separated only where their merger compromises the quality of the evidence, independence, contestability or the containment of errors. An identifiable and accountable institutional capacity is necessary for each applicable function; a distinct legal person is not always so.¶ 14

1.7 Negative mandates, stopping rules and routing

Negative mandates constitute the principal safeguard: the measurer neither qualifies nor prices; the qualifier does not reconstruct the science; the consequence authority does not collect on its own account; the financer does not modify the base; the certifier does not govern through the algorithm; the judge does not design the policy he or she reviews. This separation limits role conflicts, but creates interface costs. It must therefore be justified by a demonstrable benefit.¶ 15

Stopping rules give the architecture the right not to conclude. Insufficient data, an unidentifiable counterfactual, an unreconciled overlap, an absent competence or a release liable to produce disproportionate harm must suspend the inference or reduce the status of the result.¶ 16

The S0–S6 classification organises the routing: closure, domestic mitigation, capacity, solidarity, efficiency correction or legal remedy. It prevents measurement from automatically becoming a price and equity from being presented as causality.¶ 17

1.8 The monetary programme as a methodological laboratory

Monetary policy constitutes the originating case and the principal methodological laboratory. The present work decomposes spillover effects into fifteen signed mechanisms, supplemented by a separate stock item. It distinguishes adverse costs and offsets, flows and valuations, visible mechanisms and primitive causal components. This anatomy does not assert that each channel is new; it organises the channels within an auditable system intended for attribution, distribution and remedy.¶ 18

The theory of cumulation studies the superposition of rates, of the curve, of quantitative tightening, of liquidity and of communication; synchronisation among source jurisdictions; the interest-rate, exchange-rate, credit and fiscal reactions of recipient countries; and the exhaustion of buffers. The mechanisms combine and interact without their raw estimates being added before allocation to the primitives.¶ 19

The source-country shock is specified as a multidimensional vector with testable signatures. Bounded partial attribution fixes exposures before the shock, imposes a joint set of coefficients and publishes the bounds. Full-cycle symmetry requires that gains and losses be tracked by date, mechanism and beneficiary; the “100 units injected” protocol separates the registers of reserves, balance sheets, funding, assets, incomes and exposures.¶ 20

1.9 Research questions

The present work is organised around seven questions. RQ1: how to define and measure an accountability deficit without confusing it with a political disagreement or a general lack of institutions? RQ2: which minimal functions and which actor-forum relations are necessary to ensure completeness? RQ3: how to govern the interfaces so as to preserve evidence, uncertainty, reasons and remedy? RQ4: how to distinguish incidence, causality, net loss, equity, solidarity and legal breach, notably in monetary spillovers? RQ5: how to anticipate reasonably foreseeable cascades without governing by forecast? RQ6: how to use artificial intelligence without transferring originary authority to it and without creating an irreversible dependence? RQ7: which architectures, legal pathways, coalitions and pilots make it possible to obtain practical value without coercive world government and with a real possibility of simplification or of rejection?¶ 21

1.10 Method, evidential statuses and materials

The research is interdisciplinary and design-oriented. It combines literature review, institutional mapping, corpus coding, economic formalisation, graduated legal analysis, comparison of architectures, demonstrators, adversarial scenarios and pilot protocols. The document distinguishes verifiable facts, programme results, design proposals, normative commitments, objections and stopping criteria.¶ 22

The empirical materials comprise the register of United Nations instruments, the Sevilla corpus, the taxonomy of externalities, the Tunisian case, the international comparative protocol and the forty-four least developed countries. Their status is differentiated. The counts still rest principally on internal coding; the monetary amounts are not causally validated; the simulations do not prove effectiveness; the legal pathways are mappings to be confirmed act by act.¶ 23

The validation programme requires publication of the corpora and the scripts, double coding, replication, independent legal opinions, bottom-up budgeting, a multi-arm pilot, AI audit, consultation and red-teaming. Transparency does not replace validation; it is its condition.¶ 24

Télécharger ce tableau (CSV)

Télécharger ce tableau (CSV)

1.11 The twenty-five axes of differentiation

The contributions of the present work fall into seven families. The first is the general theory: three completenesses, accountability deficit and the distinction between the theory and the 7+3. The second concerns design: interfaces, function-first method, negative mandates, S0–S6, stopping rules, separation of registers, alternative architectures and epistemic statuses. The third bears on monetary anatomy: fifteen mechanisms, cumulation, synchronisation, buffers and multidimensional shock. The fourth concerns attribution and distribution: bounds, full cycle, non-additivity and tracing of beneficiaries. The fifth is the empirical diagnosis of a system that is dense but discontinuous. The sixth concerns AI and resilience: authority matrix, containment of errors, degraded operation and fast modes under fixed mandates. The seventh bears on adoption and applications: gradualism, Sevilla and preventive peace. The preventive loop runs across these families without automatically becoming a new institution.¶ 25

The twenty-five contributions do not form a single hierarchy. An idea may be highly original, conceptually mature, publishable as a protocol and yet unfit to carry a claim of result. The tables that follow therefore distinguish the function in the narrative from evidential maturity. The M1-M5 scale measures what the present work is today authorised to assert, not the importance of the contribution.¶ 26

The twenty-five contributions fall into seven families.¶ 27

F1 — General theory of the Global Accountability Chain. ¶ 28

F2 — Interface governance and institutional design. ¶ 29

F3 — Anatomy of monetary spillovers. ¶ 30

F4 — Attribution, non-additivity and distributive tracing. ¶ 31

F5 — Empirical diagnosis of global fragmentation. ¶ 32

F6 — AI, data and institutional resilience. ¶ 33

F7 — Applications, adoption and implementation. ¶ 34

1.12 Scope and non-claims

The Humanity Compact is not a world government, a settled universal jurisdiction, a global tax ready to be applied, a theory of strict liability for every cross-border decision, nor a promise of automatic justice. It does not propose to subordinate domestic mandates to a single external calculation. It does not assume that all externalities are monetisable, that every inequity is a global loss or that every alert justifies an intervention.¶ 35

The document proposes a grammar, a candidate architecture and protocols. It seeks to specify what should be measured, by whom, before which forum, with what limits, under what conditions of law and with what capacity for contestation. Its ambition is to be explicit enough to be tested and modular enough to be corrected.¶ 36

1.13 Organisation of the present work

The first part defines the deficit and examines the existing system. The second sets out the theory, the design principles, the functions, the instruments, the preventive loop and the governance of artificial intelligence. The third tests the architecture on a data centre, monetary spillovers, development and peace. The fourth examines the law, the political economy, the pilot, the objections, the costs, the limits and the conditions of rejection. The annexes develop the corpora, the formal model, the empirical protocols, the Tunisian case, the international extension and distributive tracing.¶ 37

The reader should not look in the pages that follow for the confirmation of an already settled regime. The reader should look there for a complete institutional hypothesis, criteria for contradicting it and a programme allowing a collective decision as to what deserves to be retained.¶ 38

@incollection{yaiche_hc_chapitre_1,
  author    = {Yaïche, Nizar},
  title     = {1. Introduction},
  booktitle = {The Humanity Compact},
  year      = {2026},
  version   = {2.3},
  doi       = {10.5281/zenodo.22746533},
  publisher = {Zenodo},
  url       = {https://humanitycompact.org/ouvrage/chapitre-1/}
}

Ouvrage de Nizar Yaïche, ancien ministre des Finances de la République tunisienne. Qui est l’auteur ?