THE 100 ARCHIVE PROJECT

Purpose and Scope

1.1  What the Archive Is

The 100 Archive Project is a public record of civic consequence in Nigeria's Fourth Republic. It documents how power is exercised, identifies responsibility across the accountability chain, preserves evidence trails, and enables diagnostic analysis of institutions and democratic systems.

The Archive is not a scandal vault. It is not a news site. It is not a political tool. It is a civic record — the kind of permanent, sourced, structured documentation that Nigeria's democratic history has never had and that every functional democracy requires.

1.2  Scope

Geography

Nigeria first, built to scale to African countries using the same data schema. See Chapter 28 for the scaling framework.

Timeframe

Nigeria: 1999 to present. The Archive covers the Fourth Republic from its beginning. 1999 is the correct starting point because it frames the Archive as a democratic record — an assessment of how Nigeria has governed itself under its current constitutional order.

Actors Covered

The Archive covers public institutions and non-state actors where they materially shape public outcomes. This includes:

• All federal and state government institutions, ministries, departments, and agencies
• The judiciary at all levels
• The National Assembly and state legislatures
• INEC and state electoral bodies
• Security agencies — police, military, DSS, EFCC, ICPC, and others
• Contractors and private companies receiving public funds or performing public functions
• Political parties in their institutional conduct
• Corporations where their conduct materially affects public rights or public funds
• International organisations and development partners where their programmes involve Nigerian governance

Case Types

Failures, contested events, policy decisions, court judgments, reforms, positive precedents, and institutional conduct — including symbolic conduct where it has accountability implications.

1.3  What This Document Governs

This methodology is binding on all full-time and part-time editorial staff of the Office of the Citizen Foundation; all members of the Editorial Board; all journalist contributors; all state archive partners; all public contributors; all researchers, fact-checkers, and interns. Every person with editorial responsibility must confirm in writing before beginning work that they have read and understood this document.

1.4  Relationship to the Archive Constitution

This methodology is the operational companion to the Archive Constitution. Where this document and the Constitution conflict, the Constitution prevails. Where this document and the Inter-Entity Agreement conflict, the Inter-Entity Agreement prevails. This document supplies the operational detail the Constitution does not.

The Three-Tier Accountability Framework

2.1  The Core Question

Every case must attempt to answer three questions. Where a question cannot be answered, the gap becomes part of the case record — explicitly and permanently.

2.2  Evidence Indicators by Tier

Every tier attribution must be grounded in a Class A or Class B source as defined in Chapter 7. Class C and D sources may corroborate but do not establish tier attribution. The table below specifies which document types establish each tier. A researcher who cannot point to a document of the appropriate type for each attributed tier has not met the evidentiary standard.

2.3  The Non-State Actor Extension

Where a contractor, private company, political party, or other non-state actor materially shapes the outcome of a governance failure, they are named in the entry as a contributing actor — not as a responsible party in the three-tier sense, but as an identified participant whose conduct the Archive documents. The public official who authorised, contracted, or supervised the non-state actor remains the Tier 3 Operational responsible party.

2.4  Cross-Administration Responsibility

Where a failure spans multiple administrations, accountability is mapped to each administration for the period during which it held authority. The entry does not adjudicate between competing political narratives. It documents specific, evidenced failures at each point in the timeline. See Section 25.3 for the full cross-administration protocol.

Case Taxonomy — 12 Sectors

3.1  The Twelve Primary Sectors

Every case is assigned one primary sector and may carry secondary sector tags. The sectors are the spine of the Archive. They prevent thematic drift and enable sector-level analysis. The taxonomy is fixed in Version 1.0 and may only be amended by the Editorial Board.

3.2  Secondary Tagging

A case may carry up to three secondary sector tags where its subject matter crosses multiple domains. A procurement case involving health equipment carries primary tag: Procurement and Contracting; secondary tags: Health and Public Safety, Public Finance and Budgeting. Secondary tags enable cross-sector analysis without distorting sector statistics.

3.3  Governance Outcome Classification — Mandatory

Every case must be assigned one of five Governance Outcome Classifications. This field is mandatory and must be completed before a case proceeds to Stage 2 of the editorial review. Its purpose is structural: it forces the research team to actively look for institutional functioning, resilience, and success — not only failures. Without this field, selection pressure will naturally skew the Archive toward documented collapses, and the archive will become a blacklist regardless of editorial intent.

Annual publication targets by classification are set in the Annual Work Plan. The Annual Bias Audit reviews the distribution of published classifications. An archive that publishes only Institutional Failure and Institutional Misconduct is not an accurate record of Nigerian governance — it is a partial record of its worst moments. The classification field is the structural mechanism that prevents that distortion.

3.4  Pattern Detection Tags

In addition to sector tags and the Governance Outcome Classification, every case is assigned a minimum of one and a maximum of five Pattern Detection Tags. Pattern tags capture the structural governance failure pattern — the mechanism by which the failure occurred — independent of the sector in which it occurred. They enable cross-sector analysis: the same failure pattern appearing in procurement, health, and security simultaneously is a systemic finding that sector tags alone cannot surface.

The standard Pattern Detection Tag vocabulary is as follows. Tags outside this vocabulary require Editorial Board approval before use.

Case Qualification — The Threshold Rule

4.1  What Qualifies as a Case

A case qualifies for the Archive if it is an event, decision, judgment, failure, reform, or institutional act with a definable time and place involving identifiable institutions, and it meets two of the following four triggers:

This threshold is deliberately designed to allow a range of case types — child deaths, judicial misconduct, procurement failures, and policy decisions — to sit in the same archive without the archive becoming noise. A case does not need to be dramatic. It needs to be documentable and civic in consequence.

4.2  What Does Not Qualify

Cases that meet only one trigger do not qualify, regardless of their emotional weight or media prominence. The Archive is not a memorial and is not a news aggregator. Where a case is rejected at threshold, the rejection is documented with the threshold assessment recorded.

The Case Importance Score (CIS)

5.1  Purpose

The Case Importance Score exists to prevent the Archive from becoming emotionally driven and statistically unrepresentative. Without a scoring system, the Archive will fill with the most dramatic, the most recent, and the most media-prominent cases — not the most important. The CIS creates a documented, auditable basis for editorial prioritisation.

The CIS is used for three purposes: selecting the first 100 entries; deciding what gets fast-tracked within the publication queue; and weighting analysis in the Annual State of Accountability report.

5.2  The CIS Formula

The CIS is the sum of four equally weighted dimensions. No dimension carries greater inherent weight than any other. Equal weighting reflects the principle that democratic harm, human harm, financial harm, and systemic harm are each legitimate grounds for civic significance — neither superseding the other.

Every dimension must be scored independently. A researcher who scores D1 before fully assessing D2–D4 risks anchoring all scores to the first assessment. The correct sequence is: gather all evidence relevant to all four dimensions; score each independently; sum. The CIS Scoring Form in the Research File records each dimension score separately with its primary source anchor.

5.3  CIS Scoring Anchors — Dimension by Dimension

Every integer on every dimension has a defined anchor. Partial scores are not used. A case scores 3, not 3.5. Where evidence is borderline between two integers, the lower score is assigned and the reason for the boundary decision is recorded in the CIS Scoring Form.

D1 — Democratic Impact

D2 — Human Impact

D3 — Financial Impact

D4 — Precedent Value

5.4  CIS Band Classification and Rules

5.5  CIS Reliability and Calibration

Because the CIS is used to allocate publication slots and weight aggregate analysis, scoring consistency across researchers and cases is a methodological requirement, not a quality aspiration. Two reliability controls operate on the CIS.

First: during the pilot phase and at each new researcher induction, a calibration exercise is run. All researchers score the same three benchmark cases independently. Acceptable calibration requires that each researcher's CIS total for each benchmark case falls within two points of the established benchmark score. A researcher whose scores fall outside this band on more than one benchmark case is retrained before working on live cases.

Second: the Research Director reviews the CIS distribution across published cases quarterly. A dataset where more than 60% of published cases score in the High band, or more than 30% score in the Low band, is a signal of scoring drift. The Research Director conducts a recalibration session with all active researchers when drift is detected. The recalibration findings are documented in the quarterly audit record.

Publication Slot Allocation

6.1  The 4-4-2 Rule

For every ten cases published, the Archive reserves publication slots by CIS band. This rule is mandatory and may not be overridden by editorial urgency, news cycle pressure, or any other consideration.

The 4-4-2 rule keeps the Archive truthful without letting tragedy dominate it. If only high-impact cases were published, the Archive would over-represent dramatic events at the expense of systemic patterns that appear in medium and low CIS cases.

At the end of each calendar quarter, the Research Director reviews the published set against the 4-4-2 allocation. Where the ratio has drifted — for example, because of a cluster of high-CIS events in one quarter — the next quarter's publication plan is adjusted to rebalance.

6.2  Within-Sector Balance

The 4-4-2 rule applies within each of the 12 sectors as well as across the archive as a whole. No sector may have all its entries in the High band, regardless of availability. Sector balance targets are set in the Annual Work Plan and reviewed quarterly by the Research Director.

Evidence Standards and Source Classification

7.1  The Naming Standard

Researchers who fail to name a responsible party where primary evidence clearly supports it are as much in breach of this methodology as researchers who name a party without sufficient evidence.

7.2  Evidence Hierarchy

Primary Sources — Class A (Highest)

• Court judgments, tribunal rulings — federal and state
• Audit reports: Auditor-General (federal and state), NEITI, NNPC audits
• Budgets, appropriation acts, official Federal Gazettes
• Procurement records, contract award notices, tender documents
• National Assembly and State Assembly Hansard
• Official correspondence, circulars, and directives bearing official seal
• FOI Act responses bearing official letterhead and date stamp
• EFCC and ICPC formally published investigation findings
• INEC official publications and gazette notices
• CBN policy circulars and official regulatory notices

Primary Sources — Class B

• Government press releases and official media statements
• Annual reports of government institutions
• Parliamentary committee reports formally tabled
• Regulatory agency published notices

Secondary Sources — Class C

• Published investigative journalism from named outlets with editorial standards, where the underlying documents are cited
• Peer-reviewed academic research with sourcing
• Reports from credible international organisations (UN agencies, World Bank, Transparency International, Amnesty International) with documented methodology

Secondary Sources — Class D

• Credible Nigerian news reporting without named primary sources
• Civil society organisation reports with partial documentation
• Legal filings and court documents in proceedings that are still ongoing

Citizen Sources — Class E

• Testimony from named individuals with documented personal knowledge
• Community reports with corroborating detail
• Photographic or video evidence with verified provenance and chain of custody

Not Admitted

• Anonymous source testimony
• Social media posts (see Chapter 15 for the specific conditions under which social media may be cited as Class C/D)
• Leaked documents of unverifiable provenance
• Hearsay and rumour regardless of how widely held
• Reports from organisations with an undisclosed interest in the outcome

7.3  Minimum Publication Threshold

To publish a case as documented, the Archive requires a minimum of one independent Class A source, plus at least one supporting source of Class B or higher. This is the floor. Sensitivity Tiers (Chapter 9) impose higher thresholds for recent and current-administration entries.

Where the best document cannot be obtained, the researcher must: state what the best document would be; state all steps taken to obtain it (FOI filed, portal searched, agency contacted, date of attempt); and mark the case as incomplete on that evidence dimension. The documented attempt is itself a form of accountability.

7.4  The Best Missing Document Field

Every Research File must include a Best Missing Document section that records: what primary source should theoretically exist for this case; what legal or procedural requirement mandates its existence; and what steps were taken to obtain it. Where the document's absence is itself evidence of a systemic failure — a contract file that should legally exist but does not, a budget variance report required by statute that was never filed — the absence is documented as primary evidence of structural failure.

7.5  Corroboration Requirement

No single source, however authoritative, is sufficient alone. Every factual claim in an Archive entry must be independently corroborated. Independence means the sources have no common origin: two newspapers reporting the same government press release are one source with two citations, not two independent sources.

Confidence Ratings

8.1  The Five Confidence Levels

Every published case carries a confidence rating. The rating acknowledges that evidence exists on a spectrum. An entry published at Moderate confidence with its limitations stated is more honest than refusing to publish until Verified evidence is available. The confidence rating is visible to all readers on the case page.

8.2  How Confidence Ratings Are Assigned and Revised

Confidence ratings are assigned by the Research Director at Stage 1 of the editorial review and confirmed by the Editorial Lead at Stage 3. Ratings may be upgraded as new primary evidence is obtained. Upgrades require a change log entry. A rating may not be downgraded without a formal editorial review.

The Sensitivity Tier System

9.1  Why Differentiated Standards Exist

Not all entries carry the same legal and reputational risk. An entry documenting an abandoned road contract from 2005 carries different risk from an entry naming a sitting minister for a failure in the current fiscal year. The core evidence standard applies to all entries. The editorial process, right-of-reply window, legal review requirements, and publication authorisation level differ based on sensitivity.

A case's Tier is assigned at proposal stage and reviewed at Stage 3 of the editorial review. If new information during research elevates sensitivity — for example, a named official is reappointed to a current government position — the case is re-classified upward and the corresponding process applied immediately.

Entry Selection and Triage — Phase 1

10.1  Intake Channels

Cases enter the Archive through four channels:

• Internal research: generated by the Foundation's research team through their own sector monitoring
• Journalist submission: cases submitted by journalist partners under the Contributor Protocol
• CSO submission: cases submitted by civil society organisations with a documented evidence pack
• Public tip (Phase 2 onwards): cases submitted through the public portal after the first 100 entries are published

10.2  Triage Decisions

Every intake is assessed within 10 working days against the threshold rule (Chapter 4) and assigned an initial CIS score (Chapter 5). Triage decisions are:

• Accept for research: threshold met, CIS assigned, case type confirmed, researcher assigned
• Hold for evidence: threshold potentially met but insufficient initial evidence; submitter notified with specific evidence requirements
• Decline: threshold not met; submitter notified with written reasons; declined cases are logged and retained

10.3  The Entry Proposal Form

Every accepted case requires a completed Entry Proposal Form before research begins. The form records: proposed entry title; sector; approximate date range; geographic scope; proposed responsible parties with titles; initial evidence inventory; initial CIS score; initial Sensitivity Tier; and the assigned researcher.

10.4  The Selection Principle

Cases are selected based on CIS score and evidence quality, applied identically regardless of the identity of responsible parties. A case is not selected because it is politically convenient. A case is not rejected because it will attract controversy. The criteria are documented civic consequence and available evidence. Nothing else.

10.5  The 12-Pilot Launch Protocol

The Archive does not launch publicly with 100 cases. It launches publicly after completing and validating 12 pilot cases — one from each of the 12 sectors. This sequencing is not optional and is not subject to acceleration on grounds of urgency, external pressure, or funder timelines.

The reason is straightforward: launching with 100 cases is how projects collapse. Research quality degrades under volume pressure. Methodology gaps that are inexpensive to fix in 12 cases become structurally embedded across 100. A single poorly evidenced entry in the first 100 gives institutions a credible basis for challenging the entire Archive's methodology rather than the specific entry. The credibility the Archive needs for the next decade is built in the first 12, not the first 100.

The 12 pilot cases are published simultaneously at launch. They are not published one at a time as they are completed. Publishing them together demonstrates depth and credibility from the first day. Publishing them individually creates a period of public operation during which the methodology is still being tested — which is the opposite of the launch conditions the Archive needs.

The Research Process

11.1  Research Standards

Research for Archive entries is not journalism. It is closer to legal preparation. The researcher's task is to assemble evidence that meets a documented standard, not to construct a narrative. The entry does not exist until the evidence exists. The entry is written from the evidence — never toward a conclusion.

11.2  The Research File

Every case must have a Research File retained permanently. The Research File contains:

• The approved Entry Proposal Form
• All source documents in original and archived form
• The Evidence Log listing every source reviewed, admitted or rejected, with the reason for each decision
• The Best Missing Document section (Section 7.4)
• The FOI request log (Chapter 16)
• The right-of-reply correspondence and any response received
• All editorial review records including blind review certification
• The legal review record (for Tier 2 and Tier 3 cases)
• The publication authorisation record
• The Editorial Traceability Log (Section 11.5)
• All subsequent update and correction records, each with version number

Research Files are never deleted. Minimum retention: ten years from publication. Where a case is subject to legal proceedings, the relevant Research File is retained until all proceedings are finally resolved.

11.3  The Evidence Log

The Evidence Log records every source consulted, whether or not ultimately cited. For each source: type, title, date, class assigned, whether admitted, the specific claim it supports, and any reason for rejection. The Evidence Log is the documented record of research good faith. In any legal challenge, it demonstrates that the researcher considered all available evidence — not only the evidence that supported a predetermined conclusion.

11.4  The Documentation Availability Note

Each published case includes a Documentation Availability note visible to readers. It states, honestly, what types of primary evidence were available for this case and what was not obtainable. This is the Archive's published acknowledgment of its own evidential limits on a case-by-case basis. It is not a disclaimer — it is a transparency mechanism.

11.5  The Editorial Traceability Log

Every published case must have a completed Editorial Traceability Log in its Research File. The Log is a named, dated, personnel-specific record of every editorial action taken on the case. Its purpose is legal: 'our editorial process' is a defence in principle; 'this specific case was researched by [name] on [date], verified by [name] on [date], and published by authorisation of [name] on [date]' is a defence in fact.

The Standard Case Template

12.1  Mandatory Fields

Every published case contains the following fields. No field may be omitted. Where a field cannot be completed, a specific notation explains why.

12.2  The How to Read This Entry Guide

Every case page includes a brief plain-language guide for ordinary readers explaining: what the three accountability tiers mean in this case; what the Governance Outcome Classification means; what the confidence rating means and what evidence exists; what the entry status means; and how to use the Citizen Action. This guide is not editorial content — it is a comprehension tool. It is templated and standardised across all entries, adapted only to reflect the specific classification, tiers, and status of the case.

12.3  The Three-Layer Entry Structure

Archive entries are structured in three layers. Each layer serves a different audience and a different purpose. The three-layer structure is mandatory for all entries. It ensures that the entry is simultaneously accessible to an ordinary citizen reading the first layer, useful to a policy researcher reading all three, and legally robust as a documented record.

Layer 1 must be written first and reviewed separately from Layers 2 and 3. A researcher who cannot write a clear, sourced 300-word factual account of the event in plain language has not yet understood the case sufficiently to write the analytical layers. Layer 1 is the test of comprehension, not just of writing.

12.4  Entry Version Numbering Convention

Every entry carries a version number. The convention is:

Every version is permanently accessible in the Change Log on the case page. Readers can navigate to any previous version. The version number appears on every entry page alongside the date of the current version and the date of original publication.

12.5  Institutional Network Map

Where a case involves three or more institutions, an Institutional Network Map is mandatory. Where a case involves one or two institutions, it is recommended but optional. The Map documents the chain of institutional relationships through which the failure or outcome occurred: which institution had authority over which, which contracted or delegated to which, and at which point in the chain the documented failure occurred.

The Map is displayed visually on the case page as a simple chain diagram and documented in text in the Research File. It serves two purposes: it prevents the entry from obscuring complex accountability chains behind a simplified top-level attribution, and it feeds directly into the Accountability Index's system-level network analysis. Over time, the aggregate network maps across all cases reveal which institutions are consistently central to governance failures — not because they are always the primary responsible party, but because they appear at the intersection of multiple institutional chains.

Document Preservation and Source Archiving

13.1  Why Document Preservation Is Critical

Government documents in Nigeria disappear — from official websites, from physical archives, from institutional records. This is not always accidental. The Archive assumes that any primary source it relies upon may cease to be publicly available after publication, and preserves its own copy of every source document before an entry is published.

13.2  The Three-Copy Rule

Every primary source document used in an Archive entry must exist in three forms before the entry is published:

• The original URL or physical location, documented with the date accessed
• A downloaded copy stored on the Archive's secure server in its original format
• An archived copy submitted to the Internet Archive Wayback Machine, with the archive URL documented in the Research File

Entries citing sources without all three copies in the Research File may not be published.

13.3  DocumentCloud Annotation

Where a primary source is a long document, the specific pages and paragraphs relied upon must be highlighted in the Archive's DocumentCloud account. The entry cites both the full document and the annotated passage. This prevents the Archive from being accused of misrepresenting a document's contents and allows readers to verify specific claims without reading entire reports.

13.4  Post-Publication Monitoring

All primary source URLs cited in published entries are checked quarterly. Where a URL has gone dead, the archived copy is promoted to primary citation status and the dead URL is noted with its last-accessible date. Where a document disappears from a government website after publication — particularly where the timing coincides with a dispute or media report — that disappearance is documented as a new change log entry.

13.5  Record Retention

All source documents are retained for a minimum of ten years from the date of publication. Where a case is the subject of legal proceedings, all related documents are retained until all proceedings are finally resolved plus five years.

Protecting Sources and Whistleblowers

14.1  The Commitment

No source who provides information on the understanding that their identity will not be disclosed will ever be identified — not to the Editorial Board, not to funders, not to any government authority, not to any court except where all legal means to resist disclosure have been exhausted and a final unappealable court order compels it. This commitment is not conditional on the accuracy of the information provided.

14.2  Source Categories

Category A — Named Sources

Sources who have explicitly consented to identification. Must sign a Source Consent Form. Researcher must confirm the source understood the potential consequences.

Category B — Attributed Sources

Identity known to researcher and Research Director, described in entry by function only: 'a former official of the Federal Ministry of Finance.' Identity stored in Research File in encrypted form accessible only to the Research Director.

Category C — Confidential Sources

Identity known only to the researcher. Not recorded in any form that links source to entry. Category C source information may only generate leads to primary sources — it may never alone support attribution of responsibility.

14.3  Digital Security

All communication with sources who have not consented to open identification must use end-to-end encrypted channels. The Archive maintains a Signal number and a SecureDrop instance. Researchers must not contact sources via unencrypted email, standard phone calls, or social media direct messages for Archive-related communications.

14.4  Legal Demands for Source Identification

Any demand from police, SSS, EFCC, a court, or any authority to identify a source triggers the following immediate sequence:

• Report immediately to Research Director and Institute legal advisers
• Provide no information to any authority without legal advice
• File for legal relief from any disclosure order before complying with it
• If all legal challenges fail and a final unappealable order is made, consult Institute Board and Editorial Board before complying
• Any compliance is the minimum required by the order and is immediately publicly disclosed

Social Media and Viral Claims as Evidence

15.1  The Default Position

Social media posts — including posts from official government accounts and verified official accounts — do not by default qualify as primary evidence. They may serve as leads to primary evidence. They may appear as Class C or D contextual sources. They do not alone support attribution of institutional responsibility.

15.2  Conditions for Social Media Citation

Social media content may be cited only where all of the following conditions are met:

• The account is verified and the post's authenticity can be confirmed independently
• The content constitutes an admission, a statement of intent, or a documented action by a named official in their official capacity
• The post has been independently archived using Wayback Machine or archive.today before possible deletion
• The post is cited as Class C or D and is corroborated by at least one Class A or B source

15.3  The Viral Claim Test

Before committing research resources to a claim that originated as viral social media content, the researcher must apply the Viral Claim Test:

• What is the earliest traceable origin of this claim?
• What primary evidence would exist if this claim were true? Has that evidence been sought?
• Who benefits from this claim being believed? Does it serve a partisan interest?
• Has it been fact-checked by a credible organisation? What was the finding?

A claim that fails the Viral Claim Test is not pursued. The test result is logged in the research record.

Freedom of Information — Requests and Non-Responses

16.1  The FOI Act as a Research Instrument

The Freedom of Information Act 2011 supersedes the Official Secrets Act of 1911 and gives any person the right to request information from any public institution. The institution must respond within 7 days. Non-response is a criminal offence under Section 7 of the Act, punishable by a minimum of one year imprisonment. The Archive files FOI requests as a standard research step — not as a last resort.

16.2  FOI Request Protocol

Every FOI request filed by the Archive must:

• Be addressed to the specific institution by legal name and the name and title of its head
• Specify the exact documents or information requested in precise language
• Cite Section 1 of the Freedom of Information Act 2011 as legal basis
• Request response within 7 days as required by law
• Be sent by registered post or email with delivery confirmation, retaining proof of delivery
• Be logged in the Research File with date sent and delivery confirmation reference

16.3  The FOI Escalation Ladder

Nigeria's FOI compliance rate is poor. Without a defined escalation sequence, requests stall indefinitely and the research timeline collapses. The FOI Escalation Ladder converts institutional silence into a documented, escalating accountability record. Every rung of the ladder is actioned automatically — not at the researcher's discretion.

16.4  Non-Response as Evidence

Within the original case entry, the documented non-response is cited as evidence of the institution's approach to transparency. It does not, alone, establish that the requested information exists or that it would have been damaging. It establishes that the institution refused to comply with its legal obligation to disclose. That fact is relevant to the case's confidence rating and to the Accountability Index's Transparency dimension score.

16.5  The FOI Tracker

The Archive maintains a public FOI Tracker listing every FOI request filed, the date filed, the institution, the information requested, the date of response (or the running count of days without response), and the response received. The FOI Tracker is updated monthly and is a standalone public output independent of the case database. See Chapter 27 for the full Outputs framework.

Handling Living and Former Officials

17.1  The Foundational Principle

The office is the subject. The person is the attribution. This applies equally to living and deceased officials, current and former office-holders. The evidence standard does not change based on the current status of the named person. What changes is the pre-publication process.

17.2  Deceased Officials

Named without a right-of-reply requirement. The entry includes a notation that the named party is deceased as at the date of publication. Where family members submit a response on behalf of a deceased official, that response is published in the Response field at the Archive's discretion.

17.3  Former Officials Who Deny Their Role

A blanket denial without primary source evidence does not affect the attribution. The denial is published in full in the Response field. The primary source evidence establishing authority is restated alongside it. Where the denial raises a specific, evidenced counter-claim — not a blanket denial but a claim that identifies a different responsible party with supporting evidence — the entry is placed Under Review while the counter-claim is investigated. See Chapter 22 for the full disputes process.

17.4  Sitting Officials — Elevated Process

Officials who currently hold any public position are subject to the Tier 3 process regardless of whether the entry relates to their current role. Three Class A sources, 21-day right-of-reply, full Editorial Board vote, legal sign-off. This is not greater protection — it is greater rigour, because entries about sitting officials carry greater public significance and require the highest evidential standard.

The Right-of-Reply Process

18.1  Mandatory, Not Courteous

The right-of-reply process is not a courtesy extended to powerful people. It is a legal safeguard, a journalistic standard, and a demonstration of the Archive's commitment to accuracy. An entry published without a completed right-of-reply process is not a valid Archive publication and will be removed pending completion of the process.

18.2  The Right-of-Reply Notice

Sent to every person named as a responsible party before publication. The Notice includes:

• A description of the entry — the failure documented, approximate date range, and the role attributed
• A summary of the primary source basis for the attribution
• An invitation to provide corrections to factual inaccuracies, primary source evidence that materially changes the attribution, or a statement for inclusion in the published entry
• The deadline: 7 days (Tier 1), 14 days (Tier 2), 21 days (Tier 3)
• The Archive's contact details and named point of contact
• The statement that non-response does not prevent publication but will be documented

The Notice is sent by email to the most recent official contact, by registered post to the last known official address, and through their official spokesperson where available. Proof of delivery is retained in the Research File.

18.3  Responding to Responses

• Acknowledge receipt within 48 hours
• Assess whether the response raises new primary source evidence that materially affects the entry
• If evidence is material: pause publication, investigate, amend if warranted
• If the response is a denial or political objection without new evidence: incorporate verbatim in the Response field
• Confirm to Editorial Lead within 5 working days of deadline whether entry is cleared to proceed

18.4  Non-Response Protocol

No response received within the window: the entry proceeds to publication. The published entry records: 'Right-of-reply notice sent to [name] on [date]. No response received by [deadline date].' A non-response is never treated as an admission. It is a documented fact.

The Four-Stage Editorial Review — Phase 2

19.1  Overview

Every entry passes through four mandatory stages before publication. No stage may be skipped. No single editor may act as reviewer at more than two consecutive stages for the same entry. The four stages catch four different categories of error that a single review cannot reliably identify: evidential, structural, linguistic, and legal.

Stage 1 — Research Submission and Checklist

The researcher submits the complete draft entry and Research File to the Research Director, who reviews the Entry Submission Checklist confirming: source documents archived in three-copy format; responsible parties identified to the required standard for the assigned Sensitivity Tier; Evidence Log complete; right-of-reply process completed; entry text free of ethnic, religious, or political party framing; CIS assessed and appropriate for sector queue. Entries failing the checklist are returned with written reasons.

Stage 2 — Blind Responsibility Review

The Stage 2 reviewer receives the entry with all personal names of responsible parties replaced by their official titles only. The published entry always carries full names — this is an internal working document only. The reviewer certifies in writing: 'I have reviewed this entry with responsible party names redacted. I confirm that the evidence supports the institutional attribution described. I confirm I would apply the same standard to any person in these roles regardless of their identity. I confirm I have not sought to identify the named parties during this review.'

An entry the reviewer cannot certify is returned to research with written reasons.

Stage 3 — Full Entry Review

Names restored. The Editorial Lead reviews for: accuracy of all factual claims against cited sources; accuracy of official titles and tenure dates; compliance with bias prohibitions; clarity and precision of language; entry status and confidence rating accuracy; completeness of the Response field; and Cross-Administration Bias Check for cross-administration entries (is accountability distributed in proportion to the evidence?).

The Stage 3 reviewer asks of every sentence: does this sentence — explicitly or implicitly — code the failure in terms of ethnicity, religion, or political party? Any sentence that fails this question is revised before Stage 4.

Stage 4 — Legal and Conflict Review

Reviewed for legal risk under Nigerian and UK defamation law and for undisclosed conflicts of interest. The Stage 4 reviewer confirms: all named parties were given the right-of-reply; all factual claims are sourced to archived sources; no claims could be characterised as opinion rather than evidenced fact; no conflict of interest was undisclosed; public interest defence requirements are met. Tier 3 entries require Institute legal adviser written clearance before the Editorial Board vote.

19.2  Inter-Coder Reliability Protocol

Any methodology that produces numerical scores — the CIS, the Confidence Rating, the Accountability Index dimensions — must demonstrate that those scores are consistent across researchers. A score assigned by Researcher A must mean the same thing as the same score assigned by Researcher B for a comparable case. Without that consistency, the scores cannot be aggregated, compared, or cited in the Annual Report as meaningful findings. This is the inter-coder reliability requirement.

The Archive applies inter-coder reliability procedures to three scored elements: the CIS; the Governance Outcome Classification; and the Confidence Rating. Each operates as follows.

The Reliability Standard

The minimum acceptable reliability threshold for each scored element is a percentage agreement of 80% at band level (High/Medium/Low for CIS; the five categories for Governance Outcome; the five ratings for Confidence) across a set of test cases. For the CIS total score, agreement within two points on the 0–20 scale counts as agreement. For Governance Outcome Classification and Confidence Rating, exact category match is required.

Where the Archive has capacity to calculate Cohen's Kappa — the standard inter-rater reliability coefficient used by Varieties of Democracy and comparable index methodologies — the minimum acceptable Kappa is κ ≥ 0.70 for all three scored elements. A Kappa below 0.60 indicates insufficient reliability and triggers mandatory recalibration before any new cases are published.

The Reliability Testing Procedure

Reliability is tested in two contexts: at researcher induction (before any researcher publishes a live case); and quarterly across the active researcher cohort (to detect scoring drift).

• A set of three benchmark cases is maintained by the Research Director. These are completed, published cases whose scores have been agreed by the full editorial team and the Advisory Board. The benchmark scores are the reference standard.
• Each researcher independently scores all three benchmark cases without access to the reference scores or to other researchers' scores.
• The Research Director calculates the percentage agreement and, where possible, Cohen's Kappa between each researcher's scores and the benchmark scores.
• Any researcher whose agreement rate falls below 80% on any scored element is not cleared to work on live cases until a documented recalibration session has been completed and a re-test conducted.
• Quarterly reliability tests use the same benchmark cases. A researcher who was cleared at induction but whose quarterly scores fall below threshold is suspended from scoring until recalibrated. The suspension and recalibration are recorded in the Traceability Log.

Handling Score Disagreements in Live Cases

Where two researchers independently score the same live case element and their scores disagree by more than two points on the CIS, or by one category on Governance Outcome Classification, or by one rating on Confidence Rating, the disagreement is resolved as follows:

• Both scores are recorded with their reasoning in the Research File.
• The Research Director reviews both scorings against the primary sources and makes a final determination, which is also recorded with reasoning.
• Where the Research Director cannot determine the correct score, the case is escalated to the Editorial Lead. The Editorial Lead's determination is final.
• Recurring disagreements on the same dimension across multiple cases are a signal that the scoring anchor for that dimension requires clarification. The finding is documented in the Annual Methodology Review.

Entry States and Lifecycle

20.1  Entries Are Permanent

Archive entries are never deleted. They evolve through documented states. Every version is archived permanently and publicly accessible. The change log records every update with date, change, reason, and authorising editor. Version transparency is what distinguishes the Archive from journalism.

20.2  Version Numbering and History

Every published version of every entry carries a version number following the convention defined in Section 12.4: v1.0 for original publication; v1.x for minor factual updates; v2.0 for material changes to accountability mapping, responsible party attribution, or Governance Outcome Classification; v3.0 for resolution. All previous versions are permanently accessible on the case page. Readers can navigate to any version in the change log. No version of any entry may be deleted.

The version history of the Archive's entries is itself an accountability record. An entry that moves from v1.0 (Institutional Failure) to v3.0 (Resolved) documents the full arc of a governance failure and its accountability outcome. That arc — from documented failure to documented resolution — is one of the most important things the Archive can show.

Public Contribution and Moderation — Phase 3

21.1  Timeline for Opening Public Contribution

The public contribution portal opens after the first 100 editorial entries are published and the moderation process has been tested in practice. This sequencing is deliberate: the Archive needs a proven methodology, a published track record, and a tested moderation team before accepting submissions from the public.

21.2  The Two-Stage Contribution Pipeline

Stage 1 — Public Submission

Any person may submit through the public portal. The form collects: description of the governance failure; proposed responsible parties with official titles; at least one source link or document; the submitter's name and contact details (held for verification, not published); and a declaration of good faith. Submissions appear in a public Proposed Entries queue labelled as unverified. The queue is not searchable by name — preventing it from being used to spread unverified allegations.

Stage 2 — Verification and Moderation

Every proposed entry is reviewed by an editorial team member within 21 days. Outcomes:

• Accepted for research: full methodology applied, researcher assigned
• Returned for development: specific evidence requirements communicated to submitter in writing
• Declined: written reasons published in the proposed queue alongside the entry

21.3  State Archive Partner Submissions

Partners managing state sub-archives under the formal Partner Protocol submit entries at Stage 2 — they have already applied the research methodology before submission. Reviewed within 14 days. Published with dual attribution.

Disputes, Corrections, and Withdrawal — Phase 4

22.1  Who May Submit a Dispute

Any person named as a responsible party, or their legal representative, or family members where the named party is deceased. Third parties with no direct connection may submit public comments but not formal disputes.

22.2  Requirements for a Valid Dispute

• Submitted in writing to the Archive's designated disputes address
• Identifies the specific entry by title and publication date
• Identifies the specific factual claim or claims disputed — not a general objection
• Provides primary source evidence in support — a bare denial is not a valid dispute
• Submitted within 12 months of publication or 12 months of the named party becoming aware

22.3  The Dispute Process

• Acknowledged within 5 working days
• Entry marked Under Review within 24 hours of a valid dispute being accepted
• Research Director investigates with reference to the original Research File
• Archive responds within 21 working days: entry corrected with notice, or dispute rejected with full explanation
• Unresolved disputes escalated to full Editorial Board within 30 days
• All outcomes published
• Disputes involving current legal proceedings referred to Institute legal advisers before any response is issued

22.4  Corrections

Every factual error — however identified — is corrected promptly. The correction is published in the change log. The original incorrect text is preserved in version history with a correction notation. The Archive does not silently amend entries. Corrections that change the attribution of responsibility require Editorial Board sign-off and are published as a Superseded version.

22.5  Withdrawal — Two Grounds Only

• A final, unappealable court judgment establishes that the entry contains a defamatory falsehood that cannot be defended on public interest grounds
• The Editorial Board by majority vote determines that the entry contains a fundamental error that cannot be corrected and that continued publication causes unjust harm to a named party

Withdrawal is not deletion. The entry is replaced with a withdrawal notice explaining the grounds. The original entry remains in version history with a prominent withdrawal notice. All withdrawals are disclosed in the Annual Bias Audit.

Bias Controls and Skewness Management

23.1  The Honesty Principle

The Archive's data will skew toward what is documented, what courts touch, what leaks, what gets media attention, and what FOI can reach. This is not a neutral selection. The Archive does not pretend otherwise. It publishes its bias controls and its acknowledgment of selection bias. That transparency is what separates a credible civic record from propaganda.

23.2  The Five Bias Controls

1. Coverage Targets

Each year, the Annual Work Plan sets sector and geographic targets — not quotas, but documented targets — to avoid accidental overfocus. Targets are reviewed quarterly. Significant deviation from targets triggers a Research Directive.

2. Disclosure Index

Every case includes a Documentation Availability note stating what types of evidence were accessible and what was not. Over time, the aggregate Disclosure Index reveals which sectors, states, and institutions are most and least documentable — itself a finding about transparency.

3. Non-Response as Data

FOI non-responses are documented as entries in the FOI Tracker and as case data. Institutions that consistently fail to respond to FOI requests are identified in the Annual Report. The non-response rate is a transparency metric published alongside case data.

4. Quarterly Sampling Audits

The Research Director conducts a quarterly audit of the published set, reviewing: sector balance; region and state balance; party and timeframe spread; evidence class distribution; CIS band distribution; and Sensitivity Tier distribution. Audit findings are published in the Annual State of Accountability report.

5. Published Limitations Section

The Archive publishes a Limitations section on its website, updated annually, explaining what it currently overrepresents and why. This section is not a disclaimer — it is an honest accounting of what the Archive can and cannot show, and what that means for interpreting its data.

23.3  The Annual Bias Audit

The Archive Constitution requires an Annual Bias Audit conducted by an Advisory Board member who has not served as an editor during the year under review. The Audit measures and publicly reports: distribution of entries by geopolitical zone; distribution by state; distribution by administration and governing party; number of Research Directives issued; number of entries returned at Stage 2 blind review; number of entries withdrawn; and any departures from this methodology with explanation.

Writing Style and Language Standards

24.1  The Historical Record Standard

Archive entries are written in historical record style with legal-brief discipline. The purpose of this style is not aesthetic — it is strategic. This tone travels internationally, protects the Archive legally, and distinguishes it from journalism and advocacy. It is the style of a civic institution, not a newsroom.

24.2  The Four Style Rules

1. Neutral Voice

Third person throughout. No 'we.' No editorial position. The entry describes what happened, who was responsible, and what evidence shows. It does not tell readers what to conclude.

2. Clean Chronology

Events are stated in chronological order with specific dates or date ranges. Where a date is approximate, this is stated explicitly ('approximately April 2018, based on audit records dated May 2018'). The timeline is the structure.

3. Explicit Sourcing

Every factual claim is attributed to a named source. The pattern: '[Claim] (Source: [Class A source, title, date]).' Claims that cannot be sourced are not made.

4. Strict Fact/Inference Separation

Facts and inferences are explicitly labelled. Facts carry source citations. Inferences are introduced with the formulation: 'The available evidence suggests...' or 'Based on [evidence], a reasonable inference is...' This separation protects the Archive legally and maintains its credibility as an evidence-based institution.

24.3  What Is Prohibited in Archive Language

• Adjectives implying editorial judgment: 'massive,' 'shocking,' 'corrupt,' 'brazen'
• Characterisations of motive or intent unless documented in a primary source
• The word 'allegedly' — if evidence supports a claim, make it; if it does not, do not
• Ethnic, religious, or political identity of responsible parties as framing or context
• Moralising language of any kind
• Imprecise official titles — titles must be stated exactly as they were at the time

24.4  Attribution Language — Preferred and Avoided Phrasings

The single most common source of defamation risk in accountability writing is not the underlying factual claim — it is the language used to connect the named person to the documented event. The following table specifies preferred Archive phrasings and the corresponding language to avoid. The preferred phrasings are not softer or more equivocal. They are more precise. They express what the evidence actually shows — no more, no less — which is both more defensible and more credible.

Nigerian-Specific Challenges

25.1  Documents Disappearing from Government Websites

The three-copy archiving rule in Chapter 13 is the primary response. Additionally: all primary source URLs in published entries are checked quarterly. Where a URL goes dead after publication — particularly where timing coincides with a dispute or media report — the disappearance is documented as a change log entry. The timing of document removal is itself a data point.

25.2  Officials Denying Roles After Leaving Office

Addressed by the Naming Standard and the Research File. Attribution is based on documentary evidence of authority, not the named party's own characterisation of their role. Blanket denials are published in the Response field. Specific, evidenced counter-claims are investigated. The Research File's documentation of the basis for attribution is the Archive's defence against revisionist accounts.

25.3  Multiple Administrations Sharing Responsibility

Cross-administration entries map accountability to each administration for the period during which it held authority. The timeline structure names each responsible party with their specific tenure dates and documents the specific failure attributable to each period. The Cross-Administration Bias Check at Stage 3 confirms that accountability is distributed in proportion to the evidence — not concentrated on one party for political reasons.

25.4  Widely Reported But Unverified Claims

Ten newspapers reporting the same claim does not make it a primary source. Widely reported claims without accessible primary evidence are held at Proposed status with Confidence Rating: Under Review. They may only be published at Moderate confidence where: the harm category is severe enough that public interest demands publication despite evidential limitations; the Editorial Board votes to publish; the Institute's legal advisers confirm defensibility; and the entry carries prominent disclosure of its evidential limitations.

25.5  Security Environment

Entries involving governance failures in areas of conflict or criminal activity require: prior Research Director and Country Director approval; assignment only to researchers with appropriate training; a security briefing before fieldwork; regular check-in protocol during fieldwork; no disclosure of entry subject to persons outside the editorial team; and immediate reporting of any threat or intimidation to the Institute. Any security threat connected to Archive work is documented and, where appropriate, publicly disclosed.

Governance

26.1  Publication Authority

The Editorial Lead has final publication authority for Tier 1 and Tier 2 entries, subject to the four-stage review. The full Editorial Board has publication authority for Tier 3 entries by majority vote. No single individual — including the Founder — holds a veto over publication decisions.

26.2  Advisory Board Role

The Advisory Board reviews: methodology (not individual cases); balance audits; risk controls; and Annual Report drafts. Advisors do not approve individual cases. The Archive takes full institutional responsibility for all published entries. Advisors are not liable for case content.

26.3  Separation of Governance and Editorial Authority

The governance entities — the Institute's Board, the Foundation's Board, the Advisory Board, and the Founder — govern the institution. The Editorial Board governs editorial decisions. These are constitutionally separate authorities. No governance body may direct, instruct, or attempt to influence editorial decisions. See the Archive Constitution, Article 5, for the full editorial independence provisions.

26.4  Conflict of Interest

Every editor, researcher, and Advisory Board member maintains a current Conflict of Interest Declaration. Any conflict relevant to a specific case requires recusal from all decisions about that case. Recusals are recorded in the Research File and disclosed in the published entry. Undisclosed conflicts of interest constitute a material breach of this methodology.

Outputs

27.1  The Archive

The primary output. The structured, searchable database of all published cases, publicly accessible, with full version histories, Research File summaries, and confidence ratings.

27.2  The FOI Tracker

A standalone public database of every FOI request filed by the Archive. Updated monthly. Displays: institution, date filed, information requested, response deadline, response received or running non-response count, and eventual outcome. The FOI Tracker is an output independent of the case database — it documents the Archive's engagement with the Nigerian state's transparency obligations regardless of whether the FOI request generates a publishable case.

27.3  Annual State of Accountability Report

Published each calendar year. The Archive's cases are the raw material. The Report provides: sector-by-sector analysis of documented governance performance; institution-level accountability patterns; geographic analysis by state and zone; cross-administration comparisons where the evidence permits; the full results of the Annual Bias Audit; the Limitations section; and the FOI compliance analysis. The Report is published in full under open licence.

27.4  Institution-Level Accountability Index

A later-phase output. Based on accumulated case data, the Index will rate institutions across dimensions including: responsiveness to FOI requests; frequency of documented accountability failures; history of corrections and improvements; and compliance with legal transparency obligations. The Index will be released when the Archive has sufficient case depth to make institution-level comparisons statistically meaningful — estimated at Year 3.

27.5  Sector Reports

Periodic thematic reports on specific sectors — elections, judiciary, procurement, education. Published as case data accumulates to a threshold sufficient for sector-level analysis. These reports are co-published with media and civil society partners where appropriate.

Scaling to Africa

28.1  The Schema Architecture

The Archive's data schema is designed for continental scaling from its first version. Every case field that is Nigeria-specific — state, LGA, institution name — is a variable within a broader geographic and institutional framework. The accountability chain (Policy / Institutional / Operational) is universal. The case taxonomy applies to any democracy. The evidence hierarchy reflects international standards.

28.2  What Changes by Country

• Countries become a mandatory field in the case schema
• The region taxonomy expands to cover the relevant geographic subdivisions of each country
• Institutions are mapped to country-specific types within the same institutional categories
• The FOI legal framework and minimum response window are updated to reflect each country's legislation
• The defamation legal risk assessment is updated by jurisdiction
• Local partner organisations are identified and brought into the state partner model

28.3  What Stays the Same

The three-tier accountability framework. The case taxonomy. The threshold rule and the CIS. The evidence hierarchy. The confidence rating system. The editorial review process. The bias controls. The entry template. The writing style standard. The source protection commitment. The editorial independence architecture. These are the portable civic infrastructure that makes the Archive replicable.

28.4  Expansion Sequencing

Year 3: Begin pilot with one additional African country, selected on the basis of: civil society partner capacity; FOI legal framework; English or French language output capacity; and political environment that permits independent accountability documentation. The pilot runs for 12 months under the existing Nigeria methodology before any Africa-wide launch.

Annual Methodology Review

29.1  The Review Process

This methodology is reviewed annually by the Editorial Board in consultation with the Foundation's research team, state archive partners, and journalist contributors. The review assesses whether the methodology is producing the intended quality of entries; whether any rules have proved unworkable; whether new Nigerian-specific challenges have emerged; and whether the CIS, evidence scoring, and confidence rating systems are appropriately calibrated.

29.2  What Cannot Be Changed by Annual Review

The following elements require a two-thirds majority of the full Editorial Board to amend, as specified in the Archive Constitution. The Annual Review may not recommend changes to them unilaterally:

• The Naming Standard
• The prohibition on ethnic, religious, and political party framing
• The minimum evidence requirements for each Sensitivity Tier
• The Blind Review requirement
• The Cross-Administration Rule
• The source protection commitment in Chapter 14
• The editorial independence provisions
• The right-of-reply requirement

29.3  Publication of Review Outcomes

The outcome of every Annual Methodology Review — including any amendments adopted and the reasons for them — is published in full on the Archive website within 60 days of the review. All previous versions of this methodology are permanently archived. The version history of this document is itself an accountability record.

Worked Example — A Complete Coded Case

30.1  Purpose of This Chapter

This chapter presents a single complete Archive case coded step by step through every stage of the methodology. The case is illustrative: it is a composite drawn from documented patterns in Nigerian public infrastructure governance, not a publication-ready entry about any specific real event. All names are generic role titles. All figures are illustrative within documented ranges.

Researchers should use this chapter as their primary operational reference when working on a first case. Every field, every scoring decision, and every attribution in a real case should be traceable to a rule in this methodology. Where a researcher cannot find the rule that governs a specific decision, they raise it in the weekly editorial meeting. The decision is then made and recorded in the Methodology Decisions Log.

30.2  The Case Facts

In 2018, the Federal Government of Nigeria appropriated ₦6.4 billion for the construction of a 47-kilometre rural road connecting three LGAs in a northern state to the state capital. The contract was awarded in March 2018 to a private construction company. By December 2021, the project was 12% complete according to an inspection report commissioned by a National Assembly Committee. No further work had been recorded on site by July 2023. The construction company had received ₦4.1 billion — 64% of the contract sum — in documented payment tranches. The Federal Roads Maintenance Agency (FERMA) was responsible for project supervision. The supervising zone director changed twice during the project period.

30.3  Step 1 — Case Qualification

Apply the two-of-four threshold rule from Chapter 4.

Result: Two of four triggers met (Institutional Relevance + Public Impact). The case qualifies. Entry Proposal Form opened.

30.4  Step 2 — Sector and Classification

30.5  Step 3 — CIS Scoring

Score each dimension independently against primary source evidence before summing.

30.6  Step 4 — Evidence Classification and Research File

Best Missing Document: The FERMA zone director's project supervision reports for 2018–2023 would establish who specifically authorised payments against incomplete milestones. FOI request filed [date]. See Step 6.

Three-Copy Rule: All five sources above exist in original URL form, downloaded server copy, and Wayback Machine archived form. Archive URLs documented in Research File.

30.7  Step 5 — Accountability Mapping

30.8  Step 6 — FOI Request Log

30.9  Step 7 — Confidence Rating

Assessed against the five confidence levels in Chapter 8.

The case has: one Class A source establishing the appropriation (Tier 1 Policy); one Class A source establishing FERMA's supervisory mandate (Tier 2 Institutional); Class A payment records establishing the financial dimension; and a Class A parliamentary report establishing the 12% completion finding. The contractor's non-performance is established by the parliamentary inspection report. The only gap is the missing FERMA supervision reports (Best Missing Document). The gap affects the Tier 3 operational detail within FERMA, not the core attribution.

Confidence Rating assigned: Strong (S). The core accountability picture is supported by multiple Class A sources. The missing supervision reports would add operational detail but would not materially change the fundamental attribution of failure to FERMA and the contractor.

30.10  Step 8 — Right-of-Reply

30.11  Step 9 — The Entry in Three-Layer Format

Layer 1 — THE EVENT

Between 2018 and 2023, a ₦6.4 billion federal road construction project connecting three local government areas to the state capital remained 12% complete while ₦4.1 billion of the contract sum — 64% — was paid to the contracting company. The Federal Roads Maintenance Agency was responsible for supervising the project. No documented supervisory intervention preventing the payments against incomplete milestones has been found.

Layer 2 — THE INSTITUTIONAL CHAIN

THE EVIDENCE: [See Section 30.6 above — five sources, all Class A or C, all triple-archived]

THE OUTCOME: As at the date of publication, no formal investigation has been announced. No payment recovery has been documented. The project remains inactive. FERMA has not responded to a formal FOI request filed on [date] seeking project supervision records. A standalone FOI Non-Response entry has been opened under Sector 11.

Layer 3 — THE GOVERNANCE QUESTION

THE ACCOUNTABILITY QUESTION: What project milestone verification procedure did FERMA apply before certifying payment tranches to the contractor between 2018 and 2021? Who within FERMA signed the payment certificates that the Federal Ministry of Finance used to authorise disbursements? What supervisory findings, if any, were produced by the two FERMA Zone Directors during their respective tenures?

DIAGNOSTIC NOTE: This case reflects a documented pattern in Nigerian infrastructure procurement: payment systems that release funds against contractor claims without verified project completion. FERMA's supervisory role is statutory but the supervision records that would show whether that role was exercised are unavailable, either because they were not produced or because they have not been disclosed. The accountability gap sits at the intersection of payment certification and project supervision — a junction where neither agency has full visibility of what the other does.

30.12  Step 10 — Sensitivity Tier and Publication Authorisation

Tier assessment: The most recent named official is the Minister of Works (2018) — a former official. The FERMA Zone Directors are former officials. No sitting official is named. No Tier 3 process is required. The case is classified Tier 2 — involves financial impact above NGN 1 billion and named former officials. Publication authority: Editorial Lead + one Board member sign-off, following four-stage review.

Traceability Log: All fields completed — researcher, evidence collector, Stage 1 through Stage 4 reviewers named and dated, right-of-reply handler, publication authoriser. Research File complete. Entry cleared for publication.

Labelling and Metadata Doctrine

This chapter is the classification standard for every entry in the Archive. It governs what researchers fill in, how they fill it in, and the rules that govern each decision. It is binding on all researchers, editors, reviewers, and technical systems.

Where this chapter and any earlier chapter of this Methodology appear to conflict, the earlier chapter governs — this chapter is the operational classification companion to the research and editorial standards set out above. All conflicts should be reported to the Research Director rather than resolved independently.

The controlled vocabulary fields in this chapter draw their permitted values from the Registry Bank. Do not invent labels. Do not paraphrase labels. Use the exact string as it appears in the Registry.

A standalone Labelling and Metadata Doctrine document (v2.1) exists as a separate researcher reference. That document and this chapter contain identical substantive content. The standalone document has a different update cadence — it is updated when classification rules change. This chapter is updated when the Methodology is revised. Where they diverge, this chapter governs.

31.1  The Structural Model

This chapter is built around cases. The Archive does not catalogue general government activity — it records documented governance issues that require accountability analysis.

A case is a distinct governance accountability issue with a definable time period, identifiable responsible parties, and documented public consequence.

A project provides optional context for cases that occur within a defined public programme. Projects do not contain accountability analysis. They are stored in the Registry Bank and linked to cases by Registry ID.

Events are individually sourced and dated occurrences within a case timeline. They are the factual record from which the three-layer narrative is built.

Registry records are reusable data objects for entities appearing across multiple cases — institutions, officials, projects, and political administrations. Reference them by Registry ID rather than re-entering their details.

31.2  Metadata Families

Fields are organised into eight families. Every field belongs to one family. The families govern the database schema.

Field status throughout this chapter:

31.3  Identity Metadata

Sector codes used in the Case ID:

31.4  Case Title Rules

Three rules govern every case title. Failure on any one is grounds for revision at editorial review.

Rule 1 — Neutral descriptive language

Describe the governance issue. Do not imply guilt, assign blame, or characterise intent.

Rule 2 — Specificity

A reader must be able to distinguish this case from all others in the Archive from the title alone. Name the institution or programme, the governance domain, and the approximate period or key event.

Rule 3 — Stability

Do not change a published title unless: new evidence fundamentally changes the nature of the issue (requiring v2.0); the named institution is officially renamed; or a factual error is found. Cosmetic edits after publication are prohibited. Titles are permanent reference points for academic citation and cross-archive linkage.

31.5  Case Summary Standard

Every summary must be 120–180 words and contain five elements in this sequence.

31.6  Classification Metadata

All values in this section are drawn from the Registry Bank. Use the exact label string as it appears in the Registry. Do not paraphrase.

31.7  Harm Classification

31.8  Project Context

Complete this section only where the case arises from a defined public programme or infrastructure initiative with its own budget and institutional mandate. Not all cases involve projects.

31.9  Responsibility Chain

Record attribution at three tiers. Each tier is populated independently. Where attribution cannot be established at a tier, record the reason explicitly — do not leave blank. See Chapter 2 for the full tier framework and evidence indicators by tier.

31.9.1  Fields — Applied Per Tier

31.9.2  Attribution Language Rules

The Basis of Attribution field must be written in evidence-anchored language. These rules apply to every attribution statement regardless of tier.

31.9.3  Attribution Confidence Levels

Drawn from the Registry Bank. Applied per tier. Governs what language may be used in the published entry for that tier.

31.10  Evidence Metadata

31.11  Event Timeline

Every case must include a structured event timeline. The timeline is not a narrative — it is a dated, sourced factual record from which the three-layer narrative is built.

31.12  Accountability Response Metadata

Where a formal accountability response to the case event has been documented, record it here. Accountability responses include investigations opened, prosecutions commenced, audit findings issued, court orders made, sanctions applied, or formal institutional corrections.

31.13  FOI and Transparency Metadata

Every FOI request filed in the course of research must be documented here. The FOI record serves two purposes: it establishes the Archive's evidence collection procedure; and it is itself accountability data — institutional disclosure behaviour is part of the governance record.

31.14  Publication Metadata

Publication metadata governs the workflow status of the case and records the editorial decisions that led to publication. These fields are completed and updated by the Research Director and Editorial team, not by the researcher.

31.15  Case Linkage

Where a case has a documented relationship to another Archive case or Registry record, the linkage must be recorded. Linkages enable the analytical systems in Chapters 2, 3, and the Accountability Index to surface cross-case patterns.

31.16  Administration Metadata

Every case must be linked to the political administration during which the primary failure occurred. Where a case spans multiple administrations, link to each with its specific period.

31.17  Uncertainty Values

Where a field value cannot be established from available evidence, use one of the following uncertainty values. Do not leave fields blank and do not estimate values. The appropriate uncertainty value is itself a finding.

31.18  Prohibited Labels

The following words and phrases must not appear in any published field — including titles, summaries, narratives, event descriptions, or attribution statements — unless they appear verbatim in a court judgment or official government finding being directly quoted with full attribution.

31.19  Case Creation Protocol

Before a case is created, determine whether the event requires a new independent case or should be recorded as an event addition to an existing case, a child case, or a related case linkage. Creating unnecessary new cases fragments the accountability record and degrades cross-case analysis.

31.20  Case Split and Merge

Split

A split applies when an existing published case is revealed by new information to contain two or more distinct accountability questions.

• The original Case ID is retired to 'Archived — Superseded' and remains permanently accessible.
• Two or more new cases are created with new Case IDs, each linked to the retired original.
• New cases begin at v1.0.
• The reason for the split is documented in the change log of the retired case and in the Research Files of the new cases.
• Research Director approval required.

Merge

A merge applies when two entries are determined to represent the same governance question — either through an original case creation error or because subsequent evidence shows the issues were not independently distinct.

• Both original Case IDs are retired to 'Archived — Superseded' and remain permanently accessible.
• A merged case is created with a new Case ID, linking to both retired originals.
• The merged case begins at v1.0.
• Research Director approval required.

31.21  Version Control

Every version update must record: the version number, date of update, researcher responsible, specific fields changed, reason for change, and the primary source that triggered it. All previous versions are permanently accessible.

31.22  Amendment Rules

Requires Full Editorial Board Vote

• The five-class evidence system — any modification requires a consistency review across all existing case records.
• The Governance Outcome Classification five-category system.
• The twelve-sector taxonomy.
• The Case ID format.
• The Attribution Confidence level definitions.

Research Director May Amend (with Editorial Lead sign-off)

• Adding new entries to the Registry Bank where a new category is genuinely needed.
• Adding new Event Types, Response Types, or Harm Types.
• Updating Registry Bank records — institutional names, contact details.
• Workflow stage requirements and field status designations.

END OF METHODOLOGY

The 100 Archive Project  ·  Editorial Methodology and Operating Standard  ·  Version 1.3  ·  2026

Office of the Citizen   ·  Freely available for citation and reproduction with attribution