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

Extended abstract

Page liminaire 2 sur 8 · ≈ 9 min de lecture · 19 paragraphes · 2ᵉ des 40 sections de lecture · rechercher dans l’ouvrage

This abstract is intended for academic readers, for evaluators and for research institutions. It sets out the question, the method, the provisional results, the contributions and the limits without conflating a specified programme with a validated result.¶ 1

Problem statement. Globalisation has extended the reach of public, financial, commercial, technological and infrastructural decisions well beyond the jurisdictions and the generations that take part in their formation, while accountability mechanisms remain principally territorial, sectoral and fragmented. An effect may be observed without being qualified, qualified without proportionate consequence, recognised without financing, or handled without the affected persons having effective access to information, to counter-assessment, to reasons and to redress. The present work names this discontinuity the global accountability deficit. It does not interpret it as a simple moral opposition between source jurisdictions and recipient countries: a decision may be lawful, technically rigorous and rational with respect to a domestic mandate while producing effects that do not fully enter its objective function; conversely, the existence of downstream harm does not suffice to establish exclusive causality, fault, a net global loss or an automatic claim.¶ 2

Question and theoretical proposal. The document formulates the Global Accountability Chain as a theory of the completeness of cross-border and intergenerational accountability. An architecture is said to be complete when it satisfies three cumulative requirements: functional completeness, covering the necessary transformations of the case file; relational completeness, linking each duty-holder to an identifiable forum capable of demanding explanation, justification and correction; and procedural completeness, preserving at the interfaces the provenance of the data, the assumptions, the model versions, the uncertainty, the objections, the dissents and the avenues for review. The contribution is therefore not the invention of each link taken in isolation, but their organisation into an end-to-end chain, governed by rights, by negative mandates, by stopping rules and by comparison with lighter architectures.¶ 3

Positioning. The framework draws on the literature on actor-forum accountability, global administrative law, the rational design of international institutions, regime complexes, polycentric governance, orchestration and experimentalist governance (Bovens, 2007; Grant and Keohane, 2005; Kingsbury, Krisch and Stewart, 2005; Koremenos, Lipson and Snidal, 2001; Ostrom, 2010; Abbott et al., 2015; de Búrca, Keohane and Sabel, 2014). It proposes four specific extensions: treating the interface as an autonomous institutional object; defining functions by their prohibitions as much as by their competences; recognising the institutional right not to conclude or not to price; and subjecting the proposed architecture to a test of parsimony, of cost, of contestability and of resistance to capture.¶ 4

Method and materials. The research combines institutional analysis, political economy, international law, welfare economics, organisational design, data science and the governance of artificial intelligence. The approach is rights-grounded, function-first: legally protected rights, procedural guarantees, decision principles and normative objectives are distinguished before the identification of the functions and of their housing. The research record comprises: a register derived from 136 analytical records relating to the instruments of the United Nations, de-duplicated into 99 primary instruments; a functional mapping of 130 initiatives of the Sevilla Platform for Action; a taxonomy of twenty families of externalities; a decomposition of monetary spillovers into fifteen signed mechanisms; a qualitative comparison of competing architectures; an integrated demonstrator relating to an artificial intelligence data centre; a model of bounded partial attribution; and protocols for international extension, for distributive tracing and for piloting.¶ 5

Provisional descriptive results. The internal coding currently available is consistent with the hypothesis of an international system dense in norms, programmes, expertise and financing, but asymmetric in the continuity of its functions. The register lists 99 distinct primary instruments, of which 6 are binding, 8 consultative and 85 political or non-binding. In the Sevilla corpus coded as at 12 July 2026, 71 initiatives have financing as their principal function, as against 3 for rating; none is principally devoted to adjudication and none to direct legal constraint. These results indicate a possible break between the production of evidence, qualification, consequence, financing and redress; they become robust only after publication of the corpus, double coding, measurement of intercoder agreement, examination of negative cases and comparison with competing nomenclatures.¶ 6

Monetary contribution. The monetary sub-programme transforms the general notion of spillover into an auditable anatomy. It distinguishes fifteen mechanisms distributed across monetary and financial channels, imported price shocks, real-economy effects and offsets, while separating stock variations from attributable flows. The theory of cumulation holds that tightening instruments overlap and interact, that the defensive reactions of receiving jurisdictions may amplify the shock, that synchronisation among the major central banks strengthens the global factor, and that the exhaustion of reserves, of fiscal space and of bank margins may shift the adjustment onto public investment and prolong the harm. The source-country shock is specified as a multidimensional vector with testable signatures; attribution is partial and bounded, mechanism by mechanism, under joint restrictions; the primitive components prevent double counting; gains and losses are tracked over the full cycle and distributed among sovereigns, central banks, banks, firms and households before aggregation.¶ 7

The discipline of attribution. The present work separates intention, incidence, causality, domestic vulnerabilities, net loss, equity and legal breach. The S0–S6 classification operates as a routing mechanism and not as an automatic scale of severity: certain situations call for closure, for monitoring or for domestic mitigation; others may justify a consent-based solidarity; only certain classes open the possibility of an efficiency correction; breaches of an applicable obligation fall within the forums and remedies of law. Accounting reconciliation, causal identification, economic attribution and normative or legal qualification constitute four distinct validations. A zero accounting residual establishes neither the counterfactual, nor causality, nor responsibility.¶ 8

Institutional architecture. The Humanity Compact proposes a first instantiation termed the 7+3: seven candidate functions — to measure, to qualify, to determine the admissible consequence, to apply certain adjustments, to finance prevention or solidarity, to certify digital systems and to organise redress — framed by the Carthage Charter, the Global Accountability Framework and the Call of Carthage. The number seven is neither sacred nor presumed optimal. A function may be carried by an existing institution, an accredited network, a protocol or a distinct body. Separation is justified only where it protects evidence, independence, contestability or the non-propagation of errors better than a merged architecture. Where guarantees are equivalent, the lightest configuration must prevail.¶ 9

Two loops. The main chain is completed by a Preventive Accountability Loop, a cross-cutting loop of prevention and alert. The Global Accountability Chain answers the question of how to handle an observed or sufficiently measurable effect, from evidence to redress. The preventive loop acts where mechanisms, exposures and vulnerabilities make a cascade reasonably foreseeable: anticipating, detecting, qualifying the alert, notifying, obtaining a reasoned response, preparing a possible anticipatory action and assessing ex post the false positives, the false negatives and the effects of the alert itself. It does not create a scientific veto power, does not impose a substantive policy and does not constitute an eighth institution taken as given. It proposes a procedural duty of diligence, of examination, of notification, of response and of learning.¶ 10

Artificial intelligence and resilience. The institutions are designed as tooled by artificial intelligence but governed by identifiable human authorities. An authority matrix distinguishes assistance, recommendation, validation and decision; data governance, the plurality of models, error containment, algorithmic assurance and degraded-mode operation aim to prevent a technical failure or a single supplier from becoming an incontestable authority. In a crisis, speed must come from pre-established rules, public thresholds, prepared financing, automatic expiry and maintained redress, not from a general suspension of the guarantees.¶ 11

Applications and adoption. The framework is put to the test on four fields of differing status: the artificial intelligence data centre tests the internal coherence of the interfaces; the monetary case constitutes the principal methodological and empirical programme; Sevilla provides an operational application centred on the actionability of development finance; preventive peace remains an exploratory extension founded on a theory of change. Adoption is gradual and grounded in consent: data protocols, comparable measurements, simulations, advisory chambers, contracts, national platforms and pilots precede any binding effect. The pilot must compare the existing process, a light federated protocol and the 7+3, on the same case files, data and pre-registered indicators.¶ 12

Scientific status and limits. The document establishes a problem of discontinuity precise enough to be studied, proposes a theory, specifies several models and protocols, and verifies by simulation the internal coherence of certain interfaces. It does not yet establish the exhaustiveness of the seven functions, the empirical superiority of the 7+3, the general causality of the monetary estimates, the legal compatibility of each act in every jurisdiction, the full cost, the assent of the dominant actors, resistance to capture strategies, the predictive value of the preventive loop or field effectiveness. Its credibility depends on the opening of the corpora, the publication of the code, independent replication, act-by-act legal opinions, the costing of the architectures, a comparative pilot, a red-team and the effective participation of the populations concerned.¶ 13

Conclusion. The central contribution is a transportable grammar of accountability: covering the necessary functions, linking each holder of power to a forum, preserving evidence and uncertainty at the interfaces, preventing without governing by forecast, and preferring the lightest solution that protects the guarantees. The 7+3 is an open institutional hypothesis; the monetary sub-programme is a research infrastructure to be estimated and replicated; the preventive loop is a proposal to be tested. The Humanity Compact should be judged less by the ambition of its institutions than by the quality of the conditions it creates for it to be contradicted, simplified or rejected.¶ 14

Keywords. Cross-border externalities; global accountability; institutional design; interfaces; prevention; monetary policy; partial attribution; governance of artificial intelligence; development finance; preventive peace.¶ 15

JEL codes. F02, F33, F38, F53, H23, H87, K33, O19.¶ 16

Abstract.¶ 17

Abstract. The globalisation of public policy, financial flows, value chains and digital infrastructure has extended the reach of decisions beyond the borders and generations that bear their consequences, while the institutions that could hold such decisions to account remain fragmented. This monograph formulates the Global Accountability Chain as a theory of functional, relational and procedural completeness: an architecture is accountable only when every necessary function is covered, when duty-holders answer to identifiable forums, and when the interfaces between functions preserve evidence, uncertainty, reasons and avenues for redress. Combining institutional analysis, political economy, international law, welfare economics and data governance, it proposes a first instantiation — the “7+3” — of seven functions and three instruments, tooled by artificial intelligence but governed by identifiable human authorities. A provisional diagnosis maps a dense but discontinuous international system; a formally specified model of bounded partial attribution decomposes monetary spillovers mechanism by mechanism without over-attribution. The contribution is a transportable grammar for evaluating any accountability architecture, together with negative mandates, stopping rules, and a discipline that separates measurement, causation, equity and law. Empirical results remain provisional; the framework is offered as a refutable proposal, not an established result.¶ 18

Keywords: cross-border externalities; global governance; accountability; institutional design; international financial architecture. JEL classification: F02, F33, F38, F53, H23, H87, K33, O19; political economy; procedural rights; development finance; AI governance; resilience.¶ 19

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

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