Conclusion — From moral aspiration to structural responsibility

Front matter 7 of 8 · ≈ 10 min read · 27 paragraphs · section 22 of 40 · search the book

Institutional responsibility begins when explanation by intention no longer suffices.

1 From the displaced effect to institutional responsibility

The starting point of this book is a mismatch that has become structural: decisions cross borders faster than accountability. A policy may be lawful and technically justified within the scope that produced it, while altering the choices, the prices, the balance sheets, the resources and the development possibilities of actors who took no part in its formation. This mismatch does not automatically turn the decision-maker into a wrongdoer, nor the affected actor into a creditor. It creates a more fundamental requirement: to build a common procedure making it possible to know what occurred, what could reasonably have been anticipated, through which mechanisms, with what uncertainty, who benefited, who bore the cost and what response is admissible.¶ 1

Institutional responsibility begins when explanation by intention no longer suffices. It does not deny domestic mandates; it places them back within an environment of interdependence. It does not moralise about the recipient; it examines its vulnerabilities, its choices and its capacities. It does not presume correction; it organises the evidence, the adversarial process and the possibility of not concluding.¶ 2

2 The theoretical contribution that must outlive the organisation chart

The most durable contribution of the Humanity Compact is neither the number seven nor the names of the institutions. It is the theory of the three completenesses. An accountability architecture must cover the necessary functions; connect each duty-holder of a function to a forum capable of demanding explanation and correction; and preserve at the interfaces the provenance, the uncertainty, the reasons, the objections and the avenue of redress. This grammar can be applied to a single organisation, to a network, to a federated protocol or to several institutions.¶ 3

The governance of interfaces constitutes a decisive contribution. An architecture sometimes fails less through the incompetence of its components than through what disappears between them: an assumption becomes a figure, a figure becomes a rating, a rating becomes a price, and uncertainty ceases to be visible. The case file must therefore be portable, versioned and contestable; each function signs only the transformation for which it answers.¶ 4

3 The safeguards: negative mandates, routing and stopping

Negative mandates limit power by construction: the measurer does not price; the qualifier does not reconstruct the science; the financer does not control the base; the certifier does not decide; the judge does not design the policy. This separation protects against certain conflicts of interest, but it is not costless. It creates interfaces, delays and costs. The 7+3 can therefore be preferred only on condition that it demonstrates a value superior to that of a common protocol, an accredited network or a merged architecture.¶ 5

The S0–S6 classification and the stopping rules give the system the capacity to recognise that not all effects justify the same response. A pecuniary transfer, a domestic friction, a global loss, an inequity and a legal violation are not interchangeable categories. To measure does not oblige one to price; to prevent does not oblige one to prohibit; to finance does not establish fault; and the absence of sufficient evidence must be able to lead to a reasoned closure.¶ 6

4 The monetary programme: a research infrastructure, not yet a verdict

The monetary programme provides an anatomy in fifteen mechanisms, a theory of the superposition of instruments and of reactions, a multidimensional shock model, a bounded partial attribution, a discipline of non-additivity and a distributive tracking of the complete cycle. It makes it possible to pose with greater rigour the question of who benefits from an easing, who bears a tightening, at what moment and through which channel.¶ 7

This methodological architecture is advanced; it is not yet a general causal estimation. The reconciled reference case establishes accounting consistency. The coefficients, counterfactuals, interactions, thresholds and distributions must be estimated, subjected to sensitivity analysis, opened and replicated. To empirical partners, the scientific offer is clear: here is a protocol explicit enough to be executed, contested and improved together. The value of the programme will depend on its capacity to publish also the null results, the unidentified channels and the cases where external attribution becomes weak.¶ 8

5 Prevention: shifting accountability upstream without governing by forecast

The Preventive Accountability Loop extends the chain to the moment when the harm can still be reduced. It organises anticipation, detection, qualification, notification, reasoned response, anticipatory action and learning. It responds to the problem of cascades: several rational decisions may alter one another’s constraints and produce a systemic consequence that no one has explicitly chosen.¶ 9

This loop does not authorise a scientific authority to govern. It requires only that material and reasonably foreseeable risks can be examined in time. The decision-maker retains their mandate; the alert retains its uncertainties; false positives, false negatives and self-fulfilling effects are assessed. If a light information-sharing protocol produces the same value, if the alerts are not actionable or if they cause more harm than they avert, the loop must be reduced or abandoned.¶ 10

6 Artificial intelligence and resilience: using and constraining

Artificial intelligence can broaden the capacity to measure, spot inconsistencies, explore scenarios and bring small administrations closer to large ones. It can also concentrate data, standardise categories, render errors opaque and create a dependence on a few suppliers. The framework therefore chooses neither technological optimism nor refusal of the tool. It imposes an authority matrix, an identifiable human validation, the plurality of models, auditability, the containment of errors, the right to an independent counter-assessment and the capacity to operate without AI.¶ 11

The same doctrine holds for crisis. Speed must not come from an indefinite suspension of the rule, but from prepared procedures, explicit thresholds, pre-arranged financing, an expiry and oversight. A resilient architecture preserves, under pressure, the trace, personal responsibility and the avenue of redress.¶ 12

7 Development and peace: distinct applications, distinct statuses

The application to Sevilla shows how the chain can become an infrastructure of actionability: making visible the stage of an initiative, the quality of its financing, the gates that remain to be passed, the responsibilities and the corrective loop. It is probably the most immediately pilotable application.¶ 13

Preventive peace extends the framework to the way in which undocumented grievances, with no forum and no response, can feed mistrust. This proposal remains exploratory. Conflicts also arise from incompatible interests, from strategies, from coercion and from power relations that measurement does not resolve. The chain becomes an infrastructure of peace only if it improves the handling of grievances without producing surveillance, stigmatisation or escalation.¶ 14

8 Law, consent and the limits of power

This book identifies graduated vehicles: methodological cooperation, internal policy, contract, legislation, treaty. This mapping does not amount to a validation of competence. Any binding measure must identify the legal basis, the competent authority, the procedural guarantees, the mechanism of incorporation, recognition, enforcement, immunities and the avenues of redress. Analogies with the prevention of transboundary harm do not create a general legal duty applicable to monetary policies.¶ 15

Legitimacy does not result solely from the consent of States. Affected populations must be able to obtain access to information, to an independently funded counter-assessment, to reasons and to re-examination. Conversely, participation does not create a general veto absent from the applicable law. Dual representation — State and direct — must be organised without confusing their legitimacies.¶ 16

9 Adoption: beginning with standalone value, not with the complete regime

A global architecture cannot wait for universal agreement. It must begin with what creates a value of its own: data protocols, provenance registers, comparable measurements, alerts in shadow mode, experimental ratings, shadow prices, advisory chambers, data governance contracts, re-examination mechanisms and pilot financing. Early entrants must receive capacity, information, legal certainty or financing that justifies their participation without turning the incentive into capture.¶ 17

The pilot must compare three routes on the same case file: the existing process, a light federated architecture and the 7+3. It measures cost, delay, errors, revised decisions, quality of the interfaces, comprehension, redress, independence, performance without AI and the capacity to conclude correctly that there is no activation. The pilot must not render an organisation irreversible; it must make it possible to decide whether it deserves to exist.¶ 18

10 What is established, specified, proposed and open

The document establishes that transboundary and intergenerational effects create an accountability problem precise enough to be studied. It presents internal descriptive results consistent with a functional asymmetry. It specifies a theory, principles, a candidate architecture, an attribution model and test protocols. It verifies by simulation the internal consistency of certain case-file passages.¶ 19

It does not yet establish the exhaustiveness of the functions, the superiority of the 7+3, the general causality of the estimates, the effectiveness of prevention, universal legal compatibility, the full cost, resistance to capture, acceptance by the powerful or the effect on peace. These limits are not peripheral reservations. They define the work on which the change in the status of the project depends.¶ 20

The reasonable political decision is not yet to approve a definitive global architecture. It is to approve a common method for testing it: to open the evidence, to compare the existing process, a light federated solution and the 7+3, then to incorporate only those functions that demonstrate a superior value at a proportionate cost. A useful international resolution would therefore first of all be a resolution of procedure, of pilot and of return before States; any eventual legal constraint would come afterwards, function by function and by consent (§12.7).¶ 21

11 The collective validation programme

The next stage must be conducted by teams that have no interest in confirming the model. The corpora and codebooks must be opened; the coding replicated; the monetary programme executed across several countries and cycles; the models, data and scripts published; the legal avenues examined by separate specialists; the architectures costed; the preventive loop tested in shadow mode; the AI prototype audited; the affected populations consulted; the negative results published.¶ 22

The Humanity Compact will become more credible not when it has eliminated all objections, but when it has organised the conditions under which they can modify it. A mature institutional theory must be able to survive the disappearance of some of its institutions, the merging of its functions and the abandonment of its least-performing hypotheses.¶ 23

12 From moral aspiration to a verifiable institutional obligation

The ambition of the project is not to moralise global decision-making, nor to ask decision-makers to answer for what no institution enables them to know. It is to stop entrusting the consideration of others to the personal quality of the decision-maker alone. A trustworthy architecture does not demand omniscience. It demands knowing what must be measured, declaring what remains uncertain, responding to reasonable alerts, giving reasons for consequences and accepting that affected persons may contest the path that led to the decision.¶ 24

The Humanity Compact is neither a verdict against interdependence nor a promise of automatic justice. It is a proposal for an infrastructure of trust: for every displaced effect, a piece of evidence; for every reasonably visible risk, a contestable alert; for every power, a mandate and a limit; for every uncertainty, a trace; for every decision, a statement of reasons; for every challenge, a forum.¶ 25

The core of the project must be able to survive the revision of its organisation chart. If a federated protocol, a three- or five-body architecture or existing institutions better protect the evidence, prevention, redress and independence, that solution must prevail. The success of the Humanity Compact is not measured by the retention of the number seven. It is measured by the capacity of the framework to identify the lightest solution that makes effects visible, risks anticipable, decisions explicable and errors correctable.¶ 26

“A structure becomes human when it forgets neither those it does not see, nor the risks it could have seen, nor the errors it has still to correct.”¶ 27

@incollection{yaiche_hc_conclusion_from_moral_aspiration_to_structural_respo,
  author    = {Yaïche, Nizar},
  title     = {Conclusion — From moral aspiration to structural responsibility},
  booktitle = {The Humanity Compact},
  year      = {2026},
  version   = {2.3},
  doi       = {10.5281/zenodo.22746533},
  publisher = {Zenodo},
  url       = {https://humanitycompact.org/ouvrage/conclusion-from-moral-aspiration-to-structural-respo/}
}

A book by Nizar Yaïche, former Minister of Finance of the Republic of Tunisia. Who is the author?