Working manuscript reader
Read-only dated copies of the Manifold Grace working outlines. Each outline has its own link from the relevant roadmap entry.
Papers · Public Gap Register
Synthetic institutional records and false corroboration
Read-only manuscript outline · Snapshot · 15 November 2026
Abstract
This paper investigates synthetic institutional records and false corroboration. Does a decision evidence trail help reviewers detect false corroboration compared with a simple source index? The design below records the proposed contribution and the evidence needed to assess it.
Introduction and question
Does a decision evidence trail help reviewers detect false corroboration compared with a simple source index?
Related work and proposed contribution
Methods and jurisprudence paper on evidentiary lineage. Establish the strongest existing approach and specify what a comparison could add.
Project, evidence and method
Use matched fictional appeals with either independent records or repeated copies from one flawed source. Give all arms identical evidence.
AI evaluation · Study design in development
Score detection, supported facts retained, what the person needs to contest the error, and time. A simpler index performing equally well weakens the proposed contribution.
Results and interpretation
Add findings and supporting evidence here after completing the analysis. Record inconclusive results and cases that weaken the claim.
Limitations and implications
Identify the scope of the conclusion, the strongest competing explanation and what independent review would need to check.
Sources and development record
institutional-evidence-research-roadmap.html#study-1
Add verified references, contributor roles, AI assistance and dated revisions here. First public manuscript date to be recorded if a version is released.
Meaningful review under confidentiality
Read-only manuscript outline · Snapshot · 15 November 2026
Abstract
This paper investigates meaningful review under confidentiality. When can independent scrutiny substitute for a person's opportunity to contest a material allegation? The design below records the proposed contribution and the evidence needed to assess it.
Introduction and question
When can independent scrutiny substitute for a person's opportunity to contest a material allegation?
Related work and proposed contribution
Jurisprudence paper on participation and independent scrutiny. Establish the strongest existing approach and specify what a comparison could add.
Project, evidence and method
Develop hard cases on restricted evidence, summaries and missing records. Analyse AF (No 3), Tariq and Osborn in their respective contexts.
AI evaluation · Study design in development
Check whether AI-assisted summaries preserve the substance needed for rebuttal without revealing restricted material. Defend the jurisprudential conclusion separately.
Results and interpretation
Add findings and supporting evidence here after completing the analysis. Record inconclusive results and cases that weaken the claim.
Limitations and implications
Identify the scope of the conclusion, the strongest competing explanation and what independent review would need to check.
Sources and development record
institutional-evidence-research-roadmap.html#study-2
Add verified references, contributor roles, AI assistance and dated revisions here. First public manuscript date to be recorded if a version is released.
Institutional correction and appeals
Read-only manuscript outline · Snapshot · 15 November 2026
Abstract
This paper investigates institutional correction and appeals. When do review powers lead to a documented correction? The design below records the proposed contribution and the evidence needed to assess it.
Introduction and question
When do review powers lead to a documented correction?
Related work and proposed contribution
Public administration paper on review powers and outcomes. Establish the strongest existing approach and specify what a comparison could add.
Project, evidence and method
Compare appeal files with different evidence access and reviewer powers. Trace verification, direction to correct and follow-through.
AI evaluation · Study design in development
Compare independent human and AI coding of appeal upheld, record corrected and opportunity restored.
Results and interpretation
Add findings and supporting evidence here after completing the analysis. Record inconclusive results and cases that weaken the claim.
Limitations and implications
Identify the scope of the conclusion, the strongest competing explanation and what independent review would need to check.
Sources and development record
institutional-evidence-research-roadmap.html#study-3
Add verified references, contributor roles, AI assistance and dated revisions here. First public manuscript date to be recorded if a version is released.
Administrative burden
Read-only manuscript outline · Snapshot · 15 November 2026
Abstract
This paper investigates administrative burden. What helps students continue an appeal and what leads them to stop? The design below records the proposed contribution and the evidence needed to assess it.
Introduction and question
What helps students continue an appeal and what leads them to stop?
Related work and proposed contribution
Qualitative social research paper on the work of challenging a decision. Establish the strongest existing approach and specify what a comparison could add.
Project, evidence and method
Design interviews and timelines that include people who considered a challenge and never filed. Record learning, compliance and psychological costs.
AI evaluation · Study design in development
Check whether suggested codes preserve context and distinguish participant reports from researcher inference.
Results and interpretation
Add findings and supporting evidence here after completing the analysis. Record inconclusive results and cases that weaken the claim.
Limitations and implications
Identify the scope of the conclusion, the strongest competing explanation and what independent review would need to check.
Sources and development record
institutional-evidence-research-roadmap.html#study-4
Add verified references, contributor roles, AI assistance and dated revisions here. First public manuscript date to be recorded if a version is released.
Meaningful human oversight and defensive administration
Read-only manuscript outline · Snapshot · 15 November 2026
Abstract
This paper investigates meaningful human oversight and defensive administration. Under what conditions does the proposed approach address the problem of meaningful human oversight and defensive administration? The design below records the proposed contribution and the evidence needed to assess it.
Introduction and question
Under what conditions does the proposed approach address the problem of meaningful human oversight and defensive administration?
Related work and proposed contribution
Governance paper on substantive review and procedural delay. Establish the strongest existing approach and specify what a comparison could add.
Project, evidence and method
Test protected time, access to original evidence, override routes and sign-off requirements on fictional decisions.
AI evaluation · Study design in development
Score justified corrections, missed errors, incorrect overrides and time, including whether additional sign-off adds value.
Results and interpretation
Add findings and supporting evidence here after completing the analysis. Record inconclusive results and cases that weaken the claim.
Limitations and implications
Identify the scope of the conclusion, the strongest competing explanation and what independent review would need to check.
Sources and development record
institutional-evidence-research-roadmap.html#study-5
Add verified references, contributor roles, AI assistance and dated revisions here. First public manuscript date to be recorded if a version is released.
Authority and responsibility for automated decisions
Read-only manuscript outline · Snapshot · 15 November 2026
Abstract
This paper investigates authority and responsibility for automated decisions. Under what conditions does the proposed approach address the problem of authority and responsibility for automated decisions? The design below records the proposed contribution and the evidence needed to assess it.
Introduction and question
Under what conditions does the proposed approach address the problem of authority and responsibility for automated decisions?
Related work and proposed contribution
Doctrinal and institutional paper on delegated authority. Establish the strongest existing approach and specify what a comparison could add.
Project, evidence and method
Map instructions, vendor responsibilities, custody, actions and routes to redress in fictional allocations.
AI evaluation · Study design in development
Test whether AI can identify who authorised an action, which evidence supports attribution and where the record is incomplete.
Results and interpretation
Add findings and supporting evidence here after completing the analysis. Record inconclusive results and cases that weaken the claim.
Limitations and implications
Identify the scope of the conclusion, the strongest competing explanation and what independent review would need to check.
Sources and development record
institutional-evidence-research-roadmap.html#study-6
Add verified references, contributor roles, AI assistance and dated revisions here. First public manuscript date to be recorded if a version is released.
Multi-principal alignment and normative uncertainty
Read-only manuscript outline · Snapshot · 15 November 2026
Abstract
This paper investigates multi-principal alignment and normative uncertainty. Under what conditions does the proposed approach address the problem of multi-principal alignment and normative uncertainty? The design below records the proposed contribution and the evidence needed to assess it.
Introduction and question
Under what conditions does the proposed approach address the problem of multi-principal alignment and normative uncertainty?
Related work and proposed contribution
Normative and technical paper on competing duties. Establish the strongest existing approach and specify what a comparison could add.
Project, evidence and method
Build a bounded allocation simulator with stated rights, decision powers and escalation routes.
AI evaluation · Study design in development
Compare weighted preferences, firm constraints and explicit referral. Score violations, overrides and unnecessary referrals.
Results and interpretation
Add findings and supporting evidence here after completing the analysis. Record inconclusive results and cases that weaken the claim.
Limitations and implications
Identify the scope of the conclusion, the strongest competing explanation and what independent review would need to check.
Sources and development record
institutional-evidence-research-roadmap.html#study-7
Add verified references, contributor roles, AI assistance and dated revisions here. First public manuscript date to be recorded if a version is released.
Provenance and changing archives
Read-only manuscript outline · Snapshot · 15 November 2026
Abstract
This paper investigates provenance and changing archives. Under what conditions does the proposed approach address the problem of provenance and changing archives? The design below records the proposed contribution and the evidence needed to assess it.
Introduction and question
Under what conditions does the proposed approach address the problem of provenance and changing archives?
Related work and proposed contribution
Methods paper on source dependence in longitudinal records. Establish the strongest existing approach and specify what a comparison could add.
Project, evidence and method
Build a corpus with known lineage and vary copying, synthetic summaries and missing provenance.
AI evaluation · Study design in development
Measure recovery of known trends and false discounting of independent records.
Results and interpretation
Add findings and supporting evidence here after completing the analysis. Record inconclusive results and cases that weaken the claim.
Limitations and implications
Identify the scope of the conclusion, the strongest competing explanation and what independent review would need to check.
Sources and development record
institutional-evidence-research-roadmap.html#study-8
Add verified references, contributor roles, AI assistance and dated revisions here. First public manuscript date to be recorded if a version is released.
Reproducibility and living research
Read-only manuscript outline · Snapshot · 15 November 2026
Abstract
This paper investigates reproducibility and living research. Under what conditions does the proposed approach address the problem of reproducibility and living research? The design below records the proposed contribution and the evidence needed to assess it.
Introduction and question
Under what conditions does the proposed approach address the problem of reproducibility and living research?
Related work and proposed contribution
Methods paper on changing models and dated evidence. Establish the strongest existing approach and specify what a comparison could add.
Project, evidence and method
Preserve prompts, outputs, model metadata, scoring rules and research snapshots. Record protocol changes and corrections.
AI evaluation · Study design in development
Reproduce scores from preserved outputs, then rerun on available models and report whether the conclusion transfers.
Results and interpretation
Add findings and supporting evidence here after completing the analysis. Record inconclusive results and cases that weaken the claim.
Limitations and implications
Identify the scope of the conclusion, the strongest competing explanation and what independent review would need to check.
Sources and development record
institutional-evidence-research-roadmap.html#study-9
Add verified references, contributor roles, AI assistance and dated revisions here. First public manuscript date to be recorded if a version is released.
Handover and institutional continuity
Read-only manuscript outline · Snapshot · 15 November 2026
Abstract
This paper investigates handover and institutional continuity. Under what conditions does the proposed approach address the problem of handover and institutional continuity? The design below records the proposed contribution and the evidence needed to assess it.
Introduction and question
Under what conditions does the proposed approach address the problem of handover and institutional continuity?
Related work and proposed contribution
Institutional design paper on maintaining a review method. Establish the strongest existing approach and specify what a comparison could add.
Project, evidence and method
Prepare a versioned protocol, scoring guide and training cases. Log reviewer questions and test handover to an independent team.
AI evaluation · Study design in development
Measure scoring agreement, ability to audit earlier findings and preservation of correction history.
Results and interpretation
Add findings and supporting evidence here after completing the analysis. Record inconclusive results and cases that weaken the claim.
Limitations and implications
Identify the scope of the conclusion, the strongest competing explanation and what independent review would need to check.
Sources and development record
institutional-evidence-research-roadmap.html#study-10
Add verified references, contributor roles, AI assistance and dated revisions here. First public manuscript date to be recorded if a version is released.
Independent review with governed queries
Read-only manuscript outline · Snapshot · 15 November 2026
Abstract
This paper investigates independent review with governed queries. Can another researcher check a result without seeing the underlying individual records? The design below records the proposed contribution and the evidence needed to assess it.
Introduction and question
Can another researcher check a result without seeing the underlying individual records?
Related work and proposed contribution
Research governance paper on review without raw access. Establish the strongest existing approach and specify what a comparison could add.
Project, evidence and method
Prototype governed access to a predefined aggregate using fictional records.
AI evaluation · Study design in development
Rerun the aggregate independently and check whether outputs disclose individual details.
Results and interpretation
Add findings and supporting evidence here after completing the analysis. Record inconclusive results and cases that weaken the claim.
Limitations and implications
Identify the scope of the conclusion, the strongest competing explanation and what independent review would need to check.
Sources and development record
institutional-evidence-research-roadmap.html#study-11
Add verified references, contributor roles, AI assistance and dated revisions here. First public manuscript date to be recorded if a version is released.
Proof of a bounded compliance predicate
Read-only manuscript outline · Snapshot · 15 November 2026
Abstract
This paper investigates proof of a bounded compliance predicate. What can a proof over committed records establish about a narrowly specified requirement? The design below records the proposed contribution and the evidence needed to assess it.
Introduction and question
What can a proof over committed records establish about a narrowly specified requirement?
Related work and proposed contribution
Research governance paper on review without raw access. Establish the strongest existing approach and specify what a comparison could add.
Project, evidence and method
Prototype one check over fictional inputs. Document the relationship between the predicate, input authenticity and the wider rule.
AI evaluation · Study design in development
Check whether AI distinguishes verification of the predicate from proof of input authenticity or full compliance.
Results and interpretation
Add findings and supporting evidence here after completing the analysis. Record inconclusive results and cases that weaken the claim.
Limitations and implications
Identify the scope of the conclusion, the strongest competing explanation and what independent review would need to check.
Sources and development record
institutional-evidence-research-roadmap.html#study-11
Add verified references, contributor roles, AI assistance and dated revisions here. First public manuscript date to be recorded if a version is released.
Verifiability of voluntary AI safety disclosures
Read-only manuscript outline · Snapshot · 15 November 2026
Abstract
This paper investigates verifiability of voluntary ai safety disclosures. What can an outsider verify about model-specific safety evaluation claims in voluntary disclosures? The design below records the proposed contribution and the evidence needed to assess it.
Introduction and question
What can an outsider verify about model-specific safety evaluation claims in voluntary disclosures?
Related work and proposed contribution
Global AI governance paper on publicly checkable commitments. Establish the strongest existing approach and specify what a comparison could add.
Project, evidence and method
Trace selected Hiroshima AI Process disclosures to model cards, evaluation reports and independent assessments.
AI evaluation · Study design in development
Compare AI-assisted claim-to-source matching against human coding, including mismatched model versions.
Results and interpretation
Add findings and supporting evidence here after completing the analysis. Record inconclusive results and cases that weaken the claim.
Limitations and implications
Identify the scope of the conclusion, the strongest competing explanation and what independent review would need to check.
Sources and development record
institutional-evidence-research-roadmap.html#study-12
Add verified references, contributor roles, AI assistance and dated revisions here. First public manuscript date to be recorded if a version is released.
Remote verification of a bounded reporting obligation
Read-only manuscript outline · Snapshot · 15 November 2026
Abstract
This paper investigates remote verification of a bounded reporting obligation. What does remotely supplied evidence establish about one reporting obligation? The design below records the proposed contribution and the evidence needed to assess it.
Introduction and question
What does remotely supplied evidence establish about one reporting obligation?
Related work and proposed contribution
Global AI governance paper on publicly checkable commitments. Establish the strongest existing approach and specify what a comparison could add.
Project, evidence and method
Construct fictional logs and evidence packages with known coverage and freshness.
AI evaluation · Study design in development
Score identification of stale, missing and irrelevant evidence, and distinguish technical verification from compliance.
Results and interpretation
Add findings and supporting evidence here after completing the analysis. Record inconclusive results and cases that weaken the claim.
Limitations and implications
Identify the scope of the conclusion, the strongest competing explanation and what independent review would need to check.
Sources and development record
institutional-evidence-research-roadmap.html#study-12
Add verified references, contributor roles, AI assistance and dated revisions here. First public manuscript date to be recorded if a version is released.
Convergence in AI-assisted literature search
Read-only manuscript outline · Snapshot · 15 November 2026
Abstract
This paper investigates convergence in ai-assisted literature search. Does AI-assisted search narrow the evidence and hypotheses researchers consider? The design below records the proposed contribution and the evidence needed to assess it.
Introduction and question
Does AI-assisted search narrow the evidence and hypotheses researchers consider?
Related work and proposed contribution
Meta-research papers on source convergence and ethics review. Establish the strongest existing approach and specify what a comparison could add.
Project, evidence and method
Compare matched search workflows with a shared question and time budget. Include structured database, non-English and archival search.
AI evaluation · Study design in development
Measure source overlap, citation validity and coverage of substantively distinct perspectives.
Results and interpretation
Add findings and supporting evidence here after completing the analysis. Record inconclusive results and cases that weaken the claim.
Limitations and implications
Identify the scope of the conclusion, the strongest competing explanation and what independent review would need to check.
Sources and development record
institutional-evidence-research-roadmap.html#study-13
Add verified references, contributor roles, AI assistance and dated revisions here. First public manuscript date to be recorded if a version is released.
Downstream harms in research ethics review
Read-only manuscript outline · Snapshot · 15 November 2026
Abstract
This paper investigates downstream harms in research ethics review. Which downstream effects are identified by different approaches to ethics review? The design below records the proposed contribution and the evidence needed to assess it.
Introduction and question
Which downstream effects are identified by different approaches to ethics review?
Related work and proposed contribution
Meta-research papers on source convergence and ethics review. Establish the strongest existing approach and specify what a comparison could add.
Project, evidence and method
Compare fictional or retrospective study protocols using participant-focused and wider institutional review.
AI evaluation · Study design in development
Score missed effects and unsupported warnings separately against an independently coded reference.
Results and interpretation
Add findings and supporting evidence here after completing the analysis. Record inconclusive results and cases that weaken the claim.
Limitations and implications
Identify the scope of the conclusion, the strongest competing explanation and what independent review would need to check.
Sources and development record
institutional-evidence-research-roadmap.html#study-13
Add verified references, contributor roles, AI assistance and dated revisions here. First public manuscript date to be recorded if a version is released.
Long-horizon agent reliability
Read-only manuscript outline · Snapshot · 15 November 2026
Abstract
This paper investigates long-horizon agent reliability. Where do failures compound across a multi-step agent workflow? The design below records the proposed contribution and the evidence needed to assess it.
Introduction and question
Where do failures compound across a multi-step agent workflow?
Related work and proposed contribution
Technical paper candidates. Establish the strongest existing approach and specify what a comparison could add.
Project, evidence and method
Use controlled workflows with changing or incomplete records and logged step outcomes.
AI evaluation · Study design in development
Score error propagation, recovery and inaccurate claims of completion.
Results and interpretation
Add findings and supporting evidence here after completing the analysis. Record inconclusive results and cases that weaken the claim.
Limitations and implications
Identify the scope of the conclusion, the strongest competing explanation and what independent review would need to check.
Sources and development record
exploratory-research-roadmap.html#study-1
Add verified references, contributor roles, AI assistance and dated revisions here. First public manuscript date to be recorded if a version is released.
Faithfulness of generated reasoning
Read-only manuscript outline · Snapshot · 15 November 2026
Abstract
This paper investigates faithfulness of generated reasoning. When does displayed reasoning influence a model's answer? The design below records the proposed contribution and the evidence needed to assess it.
Introduction and question
When does displayed reasoning influence a model's answer?
Related work and proposed contribution
Technical paper candidates. Establish the strongest existing approach and specify what a comparison could add.
Project, evidence and method
Run controlled interventions and matched comparisons on a model with suitable access.
AI evaluation · Study design in development
Measure answer changes and transfer to new tasks. Treat fluency as a separate outcome.
Results and interpretation
Add findings and supporting evidence here after completing the analysis. Record inconclusive results and cases that weaken the claim.
Limitations and implications
Identify the scope of the conclusion, the strongest competing explanation and what independent review would need to check.
Sources and development record
exploratory-research-roadmap.html#study-1
Add verified references, contributor roles, AI assistance and dated revisions here. First public manuscript date to be recorded if a version is released.
Recursive synthetic training and loss of coverage
Read-only manuscript outline · Snapshot · 15 November 2026
Abstract
This paper investigates recursive synthetic training and loss of coverage. When does recursive training erode coverage of rare cases? The design below records the proposed contribution and the evidence needed to assess it.
Introduction and question
When does recursive training erode coverage of rare cases?
Related work and proposed contribution
Technical paper candidates. Establish the strongest existing approach and specify what a comparison could add.
Project, evidence and method
Compare synthetic-only, mixed and fixed-reference conditions over training rounds.
AI evaluation · Study design in development
Measure rare-case retention, calibration and performance on independent held-out data.
Results and interpretation
Add findings and supporting evidence here after completing the analysis. Record inconclusive results and cases that weaken the claim.
Limitations and implications
Identify the scope of the conclusion, the strongest competing explanation and what independent review would need to check.
Sources and development record
exploratory-research-roadmap.html#study-1
Add verified references, contributor roles, AI assistance and dated revisions here. First public manuscript date to be recorded if a version is released.
Scarcity framing and shared resources
Read-only manuscript outline · Snapshot · 15 November 2026
Abstract
This paper investigates scarcity framing and shared resources. Does scarcity framing create misplaced expectations of zero-sum trade-offs? The design below records the proposed contribution and the evidence needed to assess it.
Introduction and question
Does scarcity framing create misplaced expectations of zero-sum trade-offs?
Related work and proposed contribution
Behavioural economics and econometrics paper candidates. Establish the strongest existing approach and specify what a comparison could add.
Project, evidence and method
Vary scarcity framing and whether a resource is rivalrous or shareable, holding other scenario details stable.
AI evaluation · Study design in development
Check generated scenarios for ambiguity and compare beliefs and choices across randomised conditions.
Results and interpretation
Add findings and supporting evidence here after completing the analysis. Record inconclusive results and cases that weaken the claim.
Limitations and implications
Identify the scope of the conclusion, the strongest competing explanation and what independent review would need to check.
Sources and development record
exploratory-research-roadmap.html#study-2
Add verified references, contributor roles, AI assistance and dated revisions here. First public manuscript date to be recorded if a version is released.
Policy effects under network interference
Read-only manuscript outline · Snapshot · 15 November 2026
Abstract
This paper investigates policy effects under network interference. How does indirect exposure alter estimated policy effects? The design below records the proposed contribution and the evidence needed to assess it.
Introduction and question
How does indirect exposure alter estimated policy effects?
Related work and proposed contribution
Behavioural economics and econometrics paper candidates. Establish the strongest existing approach and specify what a comparison could add.
Project, evidence and method
Simulate a policy intervention with known exposure and outcomes across a defined network.
AI evaluation · Study design in development
Measure how errors in AI-inferred network links affect policy estimates.
Results and interpretation
Add findings and supporting evidence here after completing the analysis. Record inconclusive results and cases that weaken the claim.
Limitations and implications
Identify the scope of the conclusion, the strongest competing explanation and what independent review would need to check.
Sources and development record
exploratory-research-roadmap.html#study-2
Add verified references, contributor roles, AI assistance and dated revisions here. First public manuscript date to be recorded if a version is released.
Identifying algorithmic pricing patterns
Read-only manuscript outline · Snapshot · 15 November 2026
Abstract
This paper investigates identifying algorithmic pricing patterns. Which pricing patterns distinguish coordinated behaviour from shared demand or cost shocks? The design below records the proposed contribution and the evidence needed to assess it.
Introduction and question
Which pricing patterns distinguish coordinated behaviour from shared demand or cost shocks?
Related work and proposed contribution
Industrial organisation and measurement paper candidates. Establish the strongest existing approach and specify what a comparison could add.
Project, evidence and method
Use simulated retail panels with known generating mechanisms and assess observational feasibility.
AI evaluation · Study design in development
Measure false positives under common shocks and document observationally equivalent explanations.
Results and interpretation
Add findings and supporting evidence here after completing the analysis. Record inconclusive results and cases that weaken the claim.
Limitations and implications
Identify the scope of the conclusion, the strongest competing explanation and what independent review would need to check.
Sources and development record
exploratory-research-roadmap.html#study-3
Add verified references, contributor roles, AI assistance and dated revisions here. First public manuscript date to be recorded if a version is released.
Endogenous measurement and Goodhart effects
Read-only manuscript outline · Snapshot · 15 November 2026
Abstract
This paper investigates endogenous measurement and goodhart effects. When does optimisation of a metric break its relationship to the outcome it measures? The design below records the proposed contribution and the evidence needed to assess it.
Introduction and question
When does optimisation of a metric break its relationship to the outcome it measures?
Related work and proposed contribution
Industrial organisation and measurement paper candidates. Establish the strongest existing approach and specify what a comparison could add.
Project, evidence and method
Build a controlled setting with a published score and independently measured outcomes.
AI evaluation · Study design in development
Test whether an AI optimiser improves the score while the underlying outcome deteriorates.
Results and interpretation
Add findings and supporting evidence here after completing the analysis. Record inconclusive results and cases that weaken the claim.
Limitations and implications
Identify the scope of the conclusion, the strongest competing explanation and what independent review would need to check.
Sources and development record
exploratory-research-roadmap.html#study-3
Add verified references, contributor roles, AI assistance and dated revisions here. First public manuscript date to be recorded if a version is released.
AI adoption and early-career skill formation
Read-only manuscript outline · Snapshot · 15 November 2026
Abstract
This paper investigates ai adoption and early-career skill formation. Under what conditions does the proposed approach address the problem of ai adoption and early-career skill formation? The design below records the proposed contribution and the evidence needed to assess it.
Introduction and question
Under what conditions does the proposed approach address the problem of ai adoption and early-career skill formation?
Related work and proposed contribution
Labour economics paper candidate. Establish the strongest existing approach and specify what a comparison could add.
Project, evidence and method
Map junior tasks to learning opportunities in one profession. Compare unaided and supervised AI-assisted exercises.
AI evaluation · Study design in development
Assess later unassisted performance. Treat historical adoption and long-run employment effects as separate evidence questions.
Results and interpretation
Add findings and supporting evidence here after completing the analysis. Record inconclusive results and cases that weaken the claim.
Limitations and implications
Identify the scope of the conclusion, the strongest competing explanation and what independent review would need to check.
Sources and development record
exploratory-research-roadmap.html#study-4
Add verified references, contributor roles, AI assistance and dated revisions here. First public manuscript date to be recorded if a version is released.
Private compute infrastructure and implementation authority
Read-only manuscript outline · Snapshot · 15 November 2026
Abstract
This paper investigates private compute infrastructure and implementation authority. When do private cloud practices determine how public rules are implemented? The design below records the proposed contribution and the evidence needed to assess it.
Introduction and question
When do private cloud practices determine how public rules are implemented?
Related work and proposed contribution
Global governance and commercial law paper candidates. Establish the strongest existing approach and specify what a comparison could add.
Project, evidence and method
Trace one reporting or evaluation requirement through public rules, provider policies and documented conduct.
AI evaluation · Study design in development
Check extracted obligations and distinguish observed actions from inferred routing motives.
Results and interpretation
Add findings and supporting evidence here after completing the analysis. Record inconclusive results and cases that weaken the claim.
Limitations and implications
Identify the scope of the conclusion, the strongest competing explanation and what independent review would need to check.
Sources and development record
exploratory-research-roadmap.html#study-5
Add verified references, contributor roles, AI assistance and dated revisions here. First public manuscript date to be recorded if a version is released.
Attribution in automated cross-border contracting
Read-only manuscript outline · Snapshot · 15 November 2026
Abstract
This paper investigates attribution in automated cross-border contracting. Which records show who authorised an unexpected automated transaction? The design below records the proposed contribution and the evidence needed to assess it.
Introduction and question
Which records show who authorised an unexpected automated transaction?
Related work and proposed contribution
Global governance and commercial law paper candidates. Establish the strongest existing approach and specify what a comparison could add.
Project, evidence and method
Construct fictional transactions with varying mandates and execution locations. Use the UNCITRAL model law as a starting point.
AI evaluation · Study design in development
Score whether AI distinguishes authorisation, execution and infrastructure location.
Results and interpretation
Add findings and supporting evidence here after completing the analysis. Record inconclusive results and cases that weaken the claim.
Limitations and implications
Identify the scope of the conclusion, the strongest competing explanation and what independent review would need to check.
Sources and development record
exploratory-research-roadmap.html#study-5
Add verified references, contributor roles, AI assistance and dated revisions here. First public manuscript date to be recorded if a version is released.
Attribution after cross-border disruption
Read-only manuscript outline · Snapshot · 15 November 2026
Abstract
This paper investigates attribution after cross-border disruption. What evidence justifies a conclusion about cross-border disruption? The design below records the proposed contribution and the evidence needed to assess it.
Introduction and question
What evidence justifies a conclusion about cross-border disruption?
Related work and proposed contribution
Governance and methods paper candidates. Establish the strongest existing approach and specify what a comparison could add.
Project, evidence and method
Code public post-incident findings and fictional cases with competing explanations.
AI evaluation · Study design in development
Score source accuracy and unsupported attribution separately.
Results and interpretation
Add findings and supporting evidence here after completing the analysis. Record inconclusive results and cases that weaken the claim.
Limitations and implications
Identify the scope of the conclusion, the strongest competing explanation and what independent review would need to check.
Sources and development record
exploratory-research-roadmap.html#study-6
Add verified references, contributor roles, AI assistance and dated revisions here. First public manuscript date to be recorded if a version is released.
Counterfactual inference after a system-wide shock
Read-only manuscript outline · Snapshot · 15 November 2026
Abstract
This paper investigates counterfactual inference after a system-wide shock. Which conclusions survive different plausible counterfactuals for a systemic shock? The design below records the proposed contribution and the evidence needed to assess it.
Introduction and question
Which conclusions survive different plausible counterfactuals for a systemic shock?
Related work and proposed contribution
Governance and methods paper candidates. Establish the strongest existing approach and specify what a comparison could add.
Project, evidence and method
Compare candidate retrospective designs with documented assumptions and spillovers.
AI evaluation · Study design in development
Verify extracted timelines and rival explanations, and test sensitivity to comparison cases.
Results and interpretation
Add findings and supporting evidence here after completing the analysis. Record inconclusive results and cases that weaken the claim.
Limitations and implications
Identify the scope of the conclusion, the strongest competing explanation and what independent review would need to check.
Sources and development record
exploratory-research-roadmap.html#study-6
Add verified references, contributor roles, AI assistance and dated revisions here. First public manuscript date to be recorded if a version is released.
Equity and local heat adaptation
Read-only manuscript outline · Snapshot · 15 November 2026
Abstract
This paper investigates equity and local heat adaptation. Can residents verify who benefited from a local heat adaptation measure? The design below records the proposed contribution and the evidence needed to assess it.
Introduction and question
Can residents verify who benefited from a local heat adaptation measure?
Related work and proposed contribution
Climate governance and procurement paper candidates. Establish the strongest existing approach and specify what a comparison could add.
Project, evidence and method
Trace one comparable commitment in each of two London boroughs through spending, delivery and reach records.
AI evaluation · Study design in development
Check whether AI separates completed delivery, people reached and evidence of reduced exposure.
Results and interpretation
Add findings and supporting evidence here after completing the analysis. Record inconclusive results and cases that weaken the claim.
Limitations and implications
Identify the scope of the conclusion, the strongest competing explanation and what independent review would need to check.
Sources and development record
exploratory-research-roadmap.html#study-7
Add verified references, contributor roles, AI assistance and dated revisions here. First public manuscript date to be recorded if a version is released.
Verifying social value in digital skills procurement
Read-only manuscript outline · Snapshot · 15 November 2026
Abstract
This paper investigates verifying social value in digital skills procurement. Can buyers verify promised digital skills training and its effect? The design below records the proposed contribution and the evidence needed to assess it.
Introduction and question
Can buyers verify promised digital skills training and its effect?
Related work and proposed contribution
Climate governance and procurement paper candidates. Establish the strongest existing approach and specify what a comparison could add.
Project, evidence and method
Trace commitments and delivery records for comparable public contracts.
AI evaluation · Study design in development
Check whether extracted claims distinguish targets, participation and measured benefit.
Results and interpretation
Add findings and supporting evidence here after completing the analysis. Record inconclusive results and cases that weaken the claim.
Limitations and implications
Identify the scope of the conclusion, the strongest competing explanation and what independent review would need to check.
Sources and development record
exploratory-research-roadmap.html#study-7
Add verified references, contributor roles, AI assistance and dated revisions here. First public manuscript date to be recorded if a version is released.
Oxfordshire Wave 6, Lot 3 · First case findings
The published specification establishes the tender baseline. The award and contract details establish signature and a £120,000 value excluding VAT. The June 2026 scrutiny pack reports Wave 6 delivery underway. Its adjacent percentages concern Wave 5.
The council spending and contract register searches yielded no matching Lot 3 payment line. The signed schedules, buyer-validated qualifying starts and outcome counts, and milestone-linked payments remain unverified. Preserve these as separate research questions.
Public working paper: oxfordshire-bootcamps-disclosure-working-paper.html
Companion research project: public-procurement-evidence-gaps.html
Managed retreat and municipal finance
Read-only manuscript outline · Snapshot · 15 November 2026
Abstract
This paper investigates managed retreat and municipal finance. Who bears financial losses when insurance coverage and the local tax base contract? The design below records the proposed contribution and the evidence needed to assess it.
Introduction and question
Who bears financial losses when insurance coverage and the local tax base contract?
Related work and proposed contribution
Climate finance and governance paper candidates. Establish the strongest existing approach and specify what a comparison could add.
Project, evidence and method
Model one jurisdiction using plausible insurance, relocation and fiscal support trajectories.
AI evaluation · Study design in development
Verify extracted obligations and compare how corrected assumptions change loss allocation.
Results and interpretation
Add findings and supporting evidence here after completing the analysis. Record inconclusive results and cases that weaken the claim.
Limitations and implications
Identify the scope of the conclusion, the strongest competing explanation and what independent review would need to check.
Sources and development record
exploratory-research-roadmap.html#study-8
Add verified references, contributor roles, AI assistance and dated revisions here. First public manuscript date to be recorded if a version is released.
Compound climate risks and fiscal stress
Read-only manuscript outline · Snapshot · 15 November 2026
Abstract
This paper investigates compound climate risks and fiscal stress. When does modelling compound hazards change conclusions about fiscal stress? The design below records the proposed contribution and the evidence needed to assess it.
Introduction and question
When does modelling compound hazards change conclusions about fiscal stress?
Related work and proposed contribution
Climate finance and governance paper candidates. Establish the strongest existing approach and specify what a comparison could add.
Project, evidence and method
Compare one-way and coupled scenarios with a bounded physical-to-financial pathway.
AI evaluation · Study design in development
Check source assumptions and separate modelled outcomes from observed impacts.
Results and interpretation
Add findings and supporting evidence here after completing the analysis. Record inconclusive results and cases that weaken the claim.
Limitations and implications
Identify the scope of the conclusion, the strongest competing explanation and what independent review would need to check.
Sources and development record
exploratory-research-roadmap.html#study-8
Add verified references, contributor roles, AI assistance and dated revisions here. First public manuscript date to be recorded if a version is released.
Transboundary climate intervention governance
Read-only manuscript outline · Snapshot · 15 November 2026
Abstract
This paper investigates transboundary climate intervention governance. Under what conditions does the proposed approach address the problem of transboundary climate intervention governance? The design below records the proposed contribution and the evidence needed to assess it.
Introduction and question
Under what conditions does the proposed approach address the problem of transboundary climate intervention governance?
Related work and proposed contribution
Climate governance and legal theory paper candidate. Establish the strongest existing approach and specify what a comparison could add.
Project, evidence and method
Use published assessments to compare evidentiary thresholds for investigation, precaution and compensation.
AI evaluation · Study design in development
Score how accurately AI conveys uncertainty and separates modelled effects, event attribution and legal responsibility.
Results and interpretation
Add findings and supporting evidence here after completing the analysis. Record inconclusive results and cases that weaken the claim.
Limitations and implications
Identify the scope of the conclusion, the strongest competing explanation and what independent review would need to check.
Sources and development record
exploratory-research-roadmap.html#study-9
Add verified references, contributor roles, AI assistance and dated revisions here. First public manuscript date to be recorded if a version is released.
Audit costs and market entry
Read-only manuscript outline · Snapshot · 15 November 2026
Abstract
This paper investigates audit costs and market entry. Under what conditions does the proposed approach address the problem of audit costs and market entry? The design below records the proposed contribution and the evidence needed to assess it.
Introduction and question
Under what conditions does the proposed approach address the problem of audit costs and market entry?
Related work and proposed contribution
Regulatory economics paper candidate. Establish the strongest existing approach and specify what a comparison could add.
Project, evidence and method
Compare fixed costs and detection coverage under uniform and risk-tiered requirements.
AI evaluation · Study design in development
Use matched fictional audit cases and score errors detected, reviewer time and estimated compliance costs.
Results and interpretation
Add findings and supporting evidence here after completing the analysis. Record inconclusive results and cases that weaken the claim.
Limitations and implications
Identify the scope of the conclusion, the strongest competing explanation and what independent review would need to check.
Sources and development record
exploratory-research-roadmap.html#study-10
Add verified references, contributor roles, AI assistance and dated revisions here. First public manuscript date to be recorded if a version is released.
Environmental credit integrity
Read-only manuscript outline · Snapshot · 15 November 2026
Abstract
This paper investigates environmental credit integrity. Under what conditions does the proposed approach address the problem of environmental credit integrity? The design below records the proposed contribution and the evidence needed to assess it.
Introduction and question
Under what conditions does the proposed approach address the problem of environmental credit integrity?
Related work and proposed contribution
Environmental measurement paper candidate. Establish the strongest existing approach and specify what a comparison could add.
Project, evidence and method
Study one credit type with differing baseline, measurement and durability assumptions.
AI evaluation · Study design in development
Check whether AI distinguishes observed physical change from evidence of additionality.
Results and interpretation
Add findings and supporting evidence here after completing the analysis. Record inconclusive results and cases that weaken the claim.
Limitations and implications
Identify the scope of the conclusion, the strongest competing explanation and what independent review would need to check.
Sources and development record
exploratory-research-roadmap.html#study-11
Add verified references, contributor roles, AI assistance and dated revisions here. First public manuscript date to be recorded if a version is released.
Emergency authority and democratic review
Read-only manuscript outline · Snapshot · 15 November 2026
Abstract
This paper investigates emergency authority and democratic review. Under what conditions does the proposed approach address the problem of emergency authority and democratic review? The design below records the proposed contribution and the evidence needed to assess it.
Introduction and question
Under what conditions does the proposed approach address the problem of emergency authority and democratic review?
Related work and proposed contribution
Political science and institutional design paper candidate. Establish the strongest existing approach and specify what a comparison could add.
Project, evidence and method
Compare fixed expiry, renewal and retrospective review in fictional emergency decisions.
AI evaluation · Study design in development
Test authority reconstruction, missing reasons and renewal flags.
Results and interpretation
Add findings and supporting evidence here after completing the analysis. Record inconclusive results and cases that weaken the claim.
Limitations and implications
Identify the scope of the conclusion, the strongest competing explanation and what independent review would need to check.
Sources and development record
exploratory-research-roadmap.html#study-12
Add verified references, contributor roles, AI assistance and dated revisions here. First public manuscript date to be recorded if a version is released.