Manifold Grace

RESEARCH METHODOLOGY · DEVELOPMENT NOTE · 4 NOVEMBER 2026

Mapping Systems, Investigating Problems and Developing Proposals

Explore the Interdisciplinary Research Atlas → — a programme-to-research crosswalk with filters for academic themes, disciplines and selected Manifold Grace pages, while keeping academic recognition separate from thematic mapping.

Manifold Grace uses an iterative, interdisciplinary approach to independent research and public-interest proposals. A broad initial map can make the boundaries, actors, dependencies, terminology and contrasting perspectives of a topic visible before selectively investigating specific questions. Findings then inform revised maps, project designs, policy options, feature requests, evaluation plans and teaching resources. This is a declared working approach, not a claim that every topic has been comprehensively surveyed or every proposal independently validated.

Methodological transparency: Research questions determine the methods used, not the other way round. Broad mapping is a starting hypothesis about the structure of a field, not its final taxonomy. Different projects may share methods while retaining distinct purposes, evidence standards, regulatory requirements and outcomes.

Project-level research and engineering: The 39 learning projects now have a shared case-study structure covering historical evidence, annotated code, scientific methods, comparative contributions, architecture, ethics, assurance and accessible release. Each retains its own editable dossier and evidence records. Open the research workbench →

A revisable research cycle

1. Landscape mapping

Identify the scope, systems, actors, historical context and relevant subject areas; mark unknowns and omitted perspectives.

2. Perimeters and taxonomy

Separate boundaries, categories, mechanisms, roles and competing definitions. State why particular classifications are useful and where they fail.

3. Selective investigation

Investigate problems at the required depth through primary records, empirical research, technical inspection, legal analysis or other domain-appropriate methods.

4. Multi-perspective synthesis

Compare relevant institutional, practitioner, technical, policy and affected-person perspectives without treating each claim as equally supported.

5. Design and proposals

Develop policy options, regulatory questions, feature proposals, prototypes or independent projects, keeping desired outcomes separate from demonstrated effects.

6. Test, revise and record

Evaluate evidence, assumptions, feasibility, safety, costs, accessibility and outcomes. Preserve original versions and date subsequent revisions.

7. Teach contextually

Create accessible explanations from mapped projects: introductory landscape, intermediate applied case, specialist method and independently verified results where available.

The cycle is deliberately non-linear. A detailed case can expose a missing actor or flawed category; a failed test may require new evidence or a changed project boundary. Mapping broadly does not oblige the author to develop every mapped branch.

Four complementary structures

StructureRoleExample
Knowledge mapShows relationships, overlaps and dependencies.AI governance ↔ evidence systems ↔ education ↔ public accountability.
TaxonomyClassifies concepts according to an explicit purpose.Different AI architectures or forms of regulatory duty.
Systems architectureExplains components, mechanisms, controls and information flows.Evidence retrieval → verification → authorised action.
Teaching pathwayOffers learner-oriented routes through mapped material.Overview → contextual example → specialist investigation → evaluation.

Policy and regulatory governance research

For a policy or regulatory question, establish the applicable jurisdiction, responsible bodies, historical versions of rules, institutional powers, affected populations, competing policy objectives, equality and accessibility considerations, relevant evidence and routes for challenge or oversight. Record alternatives and unintended effects rather than assuming that a preferred proposal will work. Stakeholder consultation, legal interpretation and empirical policy evaluation require different methods and should not be combined into a single truth score.

A policy recommendation remains a recommendation until it has been considered, authorised or implemented by an appropriate decision-maker. Where evidence is incomplete, distinguish observation, hypothesis, inference, legal question and proposed reform.

Technical feature requests and independent projects

Feature requests may originate in a practical barrier, comparative research, a gap in existing services or a broader system map. Describe the problem, affected users, existing alternatives, proposed capability, constraints, risks and possible test criteria. A submitted request is not automatically adopted, technically novel, independently implemented or attributable as a cause of subsequent industry change. Link to dated original submissions and responses without retrospectively rewriting them.

Independent projects may advance further through prototype, experiment, review and published findings; others may remain conceptual. Both can be recorded, provided their status is plain.

Connection to CESI — where relevant, not universal

The Chronological Ethical Self-Inventory (CESI) is an existing reflective framework that connects life-course chronology with evidence, ethical interpretation, capability, opportunity and institutional recognition. It can inform research into how individual experience is represented and assessed, particularly where time, institutional context or unequal access matter. It is not presented as a universal method for mathematical AI evaluation, engineering control systems, every policy field or all independent projects.

The substantive relationship is methodological overlap—attention to chronology, evidential interpretation, assumptions and recognition—rather than an assertion that every feature request or research topic is derived from CESI. Other work may use causal modelling, system architecture, doctrinal legal research, participatory methods, design science or experimental testing as more suitable primary approaches.

Multi-perspective analysis without overfitting

Projects may converge on shared concerns such as reliable evidence, fair process, accountable decisions and human agency. That overlap should be documented when it adds explanatory value. It should not erase disagreements, exaggerate causal relationships, imply uniform adoption of one ideology or force unrelated topics into a single theoretical model. Name similarities and differences separately, specify what is established by records and keep unverified interpretations provisional.

Project-level transparency checklist

  1. Research question, domain, scope, omissions and current maturity.
  2. Why each method, taxonomy and comparison was selected.
  3. Source types, dates, provenance, contradictory evidence and limitations.
  4. Relevant stakeholder perspectives, including those that disagree.
  5. Working assumptions, causal claims and alternative explanations.
  6. Separate labels for mapping, proposal, implementation, testing and verified outcomes.
  7. Changes over time, original contribution records and attribution boundaries.
  8. What may be taught responsibly and what remains unresolved.

Research · CESI · Feature Requests and Contribution Register · Cross-sector Evidence Verification · Verifiable Advanced Intelligence · Information Science · AI Product Feedback

Statement drafted 4 November 2026 as a transparent methodological development. This description does not change the historical wording, submission dates or status of previously recorded research and feature requests.