Famine as a Scored Event — One Hunger Evidence, Three Different Scores

The Compassion Benchmark measures how institutions recognize, respond to, and reduce suffering. Almost every other famine ranking measures the disaster: how many people are hungry, how many tonnes of food are needed, how large the funding gap is. This briefing isolates a different question — the only one the benchmark is built to answer: when famine is formally declared, what does the benchmark do, and which institution does it hold responsible?

The thesis: famine in 2026 is a governance event, not a weather event. In Sudan and Nigeria, the documented driver is conflict, siege, and blocked humanitarian access — not crop failure. And because the benchmark scores institutional conduct rather than humanitarian tonnage, the same category of evidence (an IPC Phase 5 finding) produces three different outcomes across the cohort, depending entirely on how the responsible institution behaved. Sudan's famine reinforces a floor that cannot fall further. Nigeria's famine already moved a score and crossing a band. Somalia's projected famine is a future trigger the benchmark watches but has not yet scored. The contrast is the teaching content.

IPC — the Integrated Food Security Phase Classification — is the United Nations–backed system that rates hunger on a five-phase scale. Phase 5 is the worst: "Catastrophe" at the household level, and "Famine" when it reaches an area-wide threshold of starvation, acute malnutrition, and death. A formal Phase 5 / famine declaration is a structured, evidence-reviewed institutional finding — which is precisely why the benchmark treats it as adjudicated-equivalent evidence (see §4).

Composites and dimension vectors recomputed directly from rankings[] in site/src/data/indexes/countries.json and reconciled with the canonical formula (v1.2). All three primary entities reconstruct to 0.0pt drift.

The structural fact: three states, three identical-category triggers (IPC Phase 5 declared/projected), three different benchmark outcomes. The differences are not about the famine's size — Somalia's projected famine threatens 6 million in crisis, larger than Nigeria's confirmed Phase 5 caseload — but about institutional conduct and where each entity already sits on the scale. The cohort is the cleanest available demonstration that the benchmark scores the institution, not the tonnage.

Per-entity history from site/public/data/history/<slug>.json (generated 2026-06-24).

Nigeria's trajectory is the teaching exemplar. The history shows the famine signal building for weeks, held below the scoring bar, then crossing it on a single decisive ruling:

• 2026-06-01 → 06-16: Repeated boundary-watch and "documented" entries at composite 21.9 (Developing). The lean-season escalation — "52.8M food insecure; Borno 15,000 at Catastrophe level for first time in a decade" (2026-06-01) — was classified as a seasonal shock with no documented state perpetration, and a January 2026 WFP source was repeatedly excluded under the freshness ruling (outside the 14-day window). The boundary-watch text names the exact conversion trigger: "IPC Phase 5 famine-area confirmation converts the … boundary watch."
• 2026-06-19: The FAO-WFP Hunger Hotspots report (Jun 17) supplies a fresh, in-window, threshold-based confirmation: ~15,000 in Borno at Phase 5 Catastrophe, 34.8M at Phase 3+. Tier-A scored event, delta −3.9, status band-crossing-proposed. Nigeria crosses Developing → Critical to 18.0.
• 2026-06-20 → 06-24: Five consecutive floor-confirmed / hold entries at 18.0. Each cycle the same Borno Phase 5 evidence re-surfaces; each cycle it is correctly screened as a double-count (§3e-bis(4)). The 2026-06-25 assessment makes the double-count screen decisive and explicit.

The teaching point: the benchmark waited. It did not move on the weather-framed seasonal shock, the out-of-window data, or the regional appeal. It moved on the specific, fresh, threshold-based IPC finding — and then it refused to move again on the same fact.

Sudan's record is the inverse. The 2026-06-25 assessment confirms 0.0 and logs the new severity as floor reinforcement: "IPC Phase 5 famine declared in El-Fasher and Kadugli; 20 additional areas at risk," alongside the RSF drone campaign on El Obeid (day 14+, 50+ civilians killed, 500,000+ displaced). The assessment is explicit: "New adverse evidence is evidence-tier reinforcement, not a composite move (no lower score exists)." The famine is recorded as reinforcing the AWR/ACT/EQU floor drivers — it strengthens the case for the floor designation without changing the 0.0.

Somalia's history shows the forward-trigger discipline operating. The 2026-05-31 entry carries a NATURAL-DISASTER-OVERLAY ruling: famine risk "driven by drought, Al-Shabaab aid blockage, and $852M funding gap — federal government cooperating with UN/WFP." The 2026-06-23 boundary-watch entry names the trigger precisely: "famine declaration (IPC Phase 5) constitutes adjudicated-equivalent trigger for a scored change." As of 2026-06-25, the Buur Hakaba famine is projected, not declared — so the score holds at 4.7 and the trigger stays armed.

This is the analytical heart of the briefing. Each rule is illustrated by exactly one cohort member.

Why a famine declaration clears the bar while an accusation does not. The benchmark's default discipline is pre-adjudication restraint: an allegation, a complaint, or a forecast is evidence-tier — it is logged and watched, but it does not move a score until a structured institutional process reaches a conclusion. A criminal charge is not a conviction; a lawsuit is not a ruling.

An IPC Phase 5 / famine declaration is the food-security equivalent of that conclusion. It is not a single agency's press release — it is a multi-agency, threshold-based, evidence-reviewed classification (the IPC Famine Review process), the closest thing in the humanitarian system to a verdict. That is why it functions as adjudicated-equivalent: it is strong enough to move a score where a hunger warning or a regional appeal is not.

Nigeria is the proof. For weeks the record carried famine warnings and an out-of-window dataset at 21.9 without moving. The score moved only when the fresh, in-window FAO-WFP/IPC Phase 5 confirmation landed (2026-06-19). The trigger is the finding, not the suffering — the suffering was present the whole time.

The benchmark scores who CAUSED the harm, not who it hit, and not merely the territory it happened in. This is the single most important distinction in the cohort, and Sudan and Nigeria sit on opposite sides of it.

• Sudan — active perpetration. The famine in El Fasher and Kadugli is downstream of an RSF siege. UN agencies are explicit on causation: "the drivers of hunger are clear: conflict, displacement and blocked humanitarian access," and the towns are "largely cut off by conflict." The WFP/FAO framing — "conflict still decides who eats and who does not" (Nov 2025, persisting into 2026) — names a deliberate act, not a natural shock. The famine is scored against the conduct of the besieging party, which is part of why Sudan sits at the absolute floor.
• Nigeria — prevention failure, not perpetration. Boko Haram and ISWAP — non-state armed groups — are the conflict drivers. The Nigerian state did not create the insurgency and is not documented as the primary perpetrator of the civilian harm. Nigeria is scored for failing to prevent, protect, and prioritize response to foreseeable catastrophic need. That is a real failure, but a categorically lighter one than perpetration — which is exactly why its downgrade was a measured −3.9, concentrated in Action, Equity, and Systemic Thinking, rather than a collapse to the floor.

The IRC supplies the cohort's load-bearing cited frame: the looming Nigerian famine is "entirely man-made" and "entirely preventable." That phrase is a humanitarian-agency characterization of causation — it is cited, not editorialized — and it is precisely what the perpetrator-vs-prevention distinction operationalizes.

At the bottom of the scale, new famine evidence reinforces the designation; it does not move the number. Sudan's profile is the all-anchor minimum — every one of the eight dimensions at 1.0 — which the canonical formula renders as exactly 0.0. There is no lower score. A famine declaration in El Fasher and Kadugli is qualitatively new severity, and the record captures it — but as evidence-tier reinforcement of the floor designation, not as a composite delta. The scale has run out of room before the conduct has run out of escalation.

Somalia (4.7) sits a notch above the absolute floor and behaves the same way for the same structural reason: its 2026-06-25 confirmation logs the Buur Hakaba projection and the 1-in-10 WFP coverage collapse as near-floor reinforcement, not a move. The number holds; the record deepens.

The benchmark has no "humanitarian emergency" switch. A famine is decomposed into the specific care behaviors it implicates, scored across the eight dimensions. Across the cohort the same fingerprint recurs:

• Action (ACT) — responsiveness and proportionality to catastrophe-level need; collapsing pipelines and looming assistance cuts. Nigeria AC2 (proportionality) scored 1/5: "15,000 at Phase 5 / 34.8M at Phase 3+ outstrips response."
• Equity (EQU) — priority for the most vulnerable, who are the ones slipping into Phase 5. Nigeria EQ2 scored 1/5; this is Nigeria's joint-weakest dimension at 12.5.
• Systemic Thinking (SYS) — root-cause orientation for a decade-plus recurring catastrophe. Nigeria SYS 12.5; the recurrence is the signal that this is structural, not a one-off shock.
• Boundaries (BND) — the protected status of relief and the question of leaving besieged populations without alternatives. This is the axis the Aid Obstruction briefing (2026-06-19) examined in depth; in Sudan and Somalia, Al-Shabaab taxation of food deliveries and RSF siege are the BND failures.

The point: "famine" is not a label the benchmark applies. It is a pattern of failures it reads into the dimensions — which is why two famines can produce different scores, and why the same famine can implicate different dimensions in different states.

Set the three side by side and the benchmark's value proposition is visible in a single frame.

The tension this surfaces: an outside reader, looking only at the humanitarian tonnage, would expect Sudan (famine confirmed, atrocities ongoing) and Somalia (6 million in crisis) to show the most movement and Nigeria the least. The benchmark does the opposite — Nigeria is the only one that moved. That inversion is not a flaw; it is the entire point. Movement on the scale tracks the changing state of institutional conduct and where the entity already sits, not the magnitude of the disaster. Sudan and Somalia cannot move because they are already at or near the floor; Nigeria moved because it crossed a band boundary from a position that still had room to fall.

Somalia is the closest entity in the entire portfolio to a famine-triggered score change. The 2026-06-25 digest names the window precisely: a formal IPC Phase 5 famine declaration in Buur Hakaba is projected for end-June through September 2026, with the first plausible declaration date around 2026-06-30. The current evidence is a famine risk analysis under a worst-case scenario (failed Gu rains, rising prices, aid shortfall) — the first such analysis since the 2022 crisis, when famine was averted by a massive scaled-up response.

What to watch:

• If a formal IPC Phase 5 famine is declared in Buur Hakaba with documented state-response failure, that is an adjudicated-equivalent trigger that would convert Somalia's near-floor 4.7 to a scored change. The NATURAL-DISASTER-OVERLAY ruling (drought-driven, state cooperating with UN/WFP) is the offsetting factor that currently restrains the magnitude; the attribution question — how much is drought, how much is Al-Shabaab blockage, how much is state-response failure — will determine the size of any move.
• Sudan El Obeid remains the sharpest mass-casualty risk in the portfolio (drone campaign day 14+, El Fasher precedent), but Sudan is at 0.0 — any escalation deepens the floor record without moving the number.
• Nigeria's August 2026 lean-season peak is the next observable point; if access barriers persist or conditions worsen with fresh in-window data, a further dimensional adjustment is possible. The current 18.0 will hold against any re-presentation of the already-scored Borno event.
• The Sahel calibration question (Mali 12.5 vs Burkina Faso 6.3) is a separate open review and is deliberately kept secondary here.

The exit path from a famine-driven score, for any cohort member, runs through the same gate as entry: a structured, in-window, threshold-based finding of sustained improvement — not the passage of time, and not aid delivered under duress.

Canonical record (read-only):

• site/src/data/indexes/countries.json — composites recomputed: Sudan 0.0, Nigeria 18.0, Somalia 4.7, Mali 12.5, Burkina Faso 6.3
• site/scripts/lib/scoring.mjs::computeCompositeFromDimensions (v1.2) — all three reconstruct to 0.0pt drift
• site/public/data/history/{nigeria,sudan,somalia}.json — longitudinal trajectories
• research/assessments/{sudan,nigeria,somalia}-2026-06-25.md; research/assessments/nigeria-2026-06-19.md
• research/digests/2026-06-19.md, research/digests/2026-06-25.md
• Prior briefings (differentiation): research/special-briefings/aid-obstruction-2026-06-19.md, research/special-briefings/floor-and-critical-2026-06-11.md

Fresh web evidence (June 2026 grounding, Tier-3+):

• IPC — Famine confirmed in El Fasher and Kadugli, 20 areas at risk: https://www.ipcinfo.org/ipcinfo-website/countries-in-focus-archive/issue-137/en/
• WFP — "Famine conditions confirmed in Sudan's El Fasher and Kadugli"; "conflict still decides who eats and who does not": https://www.wfp.org/news/famine-conditions-confirmed-sudans-el-fasher-and-kadugli-hunger-and-malnutrition-ease-where
• FAO — same confirmation: https://www.fao.org/newsroom/detail/famine-conditions-confirmed-in-sudan-fasher-and-kadugli-as-hunger-and-malnutrition-ease-where/en
• WFP/USA (OCHA/FAO/WFP/UNICEF) — Somalia famine risk emerges, Buur Hakaba IPC AMN Phase 5, 6M in crisis: https://wfpusa.org/news/joint-news-release-ocha-fao-wfp-unicef-un-agencies-warn-of-worsening-hunger-and-malnutrition-crisis-in-somalia-as-famine-risk-emerges/
• EC Knowledge4Policy — Somalia Famine Review, Burhakaba/Baidoa: https://knowledge4policy.ec.europa.eu/publication/somalia-famine-review-ipc-analysis-conclusions-recommendations-burhakaba-baidoa_en
• IRC — "Nigeria: Risk of Man-made Famine Threatens Millions" ("entirely man-made"): https://www.rescue.org/press-release/nigeria-risk-man-made-famine-threatens-millions
• IPC — Borno, Nigeria famine-risk alert: https://www.ipcinfo.org/ipcinfo-website/resources/resources-details/en/c/1141574/

This briefing interprets the existing Compassion Benchmark record against current public evidence. It does not change any published score. Where it surfaces a scoring question, that question is filedand not applied. Entities never pay for inclusion, score changes, or suppression of findings.