Aid Obstruction — When Institutions Stop Relief and Silence the Witnesses

The Compassion Benchmark measures how institutions recognize, respond to, and reduce suffering. Most of the country index's lower band reflects institutions that fail to help — weak capacity, poor service, neglect. This briefing isolates a sharper and escalating harm class: institutions that have crossed from failing to help to actively blocking relief and suppressing the monitors who would document it.

The distinction matters because the benchmark scores conduct, not just outcomes. A state that cannot feed its people because it is poor is a capacity failure. A state that can permit relief and instead bombs the convoy, bans the agency, expels the food-aid officials, attacks the burial team, and denies the fact-finders entry is committing an active harm — and it is doing so against the two things the benchmark is built to protect: the people in need, and the evidentiary record that would hold the institution to account.

The cohort in frame is seven states. Six are at the absolute 0.0 floor — Sudan, Israel, Russia, Myanmar, South Sudan, Yemen — and one, the Democratic Republic of the Congo, sits just above at 2.3. This is a cross-referenced read of the published record against current public evidence (June 2026). It interprets the existing scores; it does not change them. The central thesis: aid obstruction is a defining feature of the floor cluster, and its most corrosive variant — the suppression of monitors — is an attack on the accountability record itself, which is precisely why these institutions sit where the scale has no room left to fall.

Composites and dimension vectors recomputed directly from rankings[] in site/src/data/indexes/countries.json. The country index holds 193 entities; the scored catalog across all seven indexes is 1,156.

The structural fact: the six floor states carry an identical vector — every dimension at the minimum anchor, which the canonical composite renders as exactly 0.0. The floor is not "a very low score"; it is the simultaneous collapse of the entire profile to the anchor, reserved for conduct that leaves no remediation surface to credit. Active obstruction of relief is one of the clearest routes to that state: when the response itself is being attacked, there is no partial structure left to score upward. The DRC at 2.3 — a near-anchor profile that has not fully collapsed — is the instructive contrast, and §5 examines why.

The benchmark's eight dimensions let the obstruction pattern be read as a distinct signature, separable from the generic repression profile (which concentrates in EQU/ACC/INT). Obstruction concentrates in four:

• ACT (Action) — blocking the response. The institution does not merely fail to act; it acts to stop others. Expulsion of WFP officials in Sudan, the banning and deregistration of relief agencies in Gaza, the targeting of aid convoys and food-aid warehouses in Ukraine.
• BND (Boundaries) — violating the protected status of aid, medical and press actors. Humanitarian workers, medical teams, burial teams and journalists hold a protected status under international humanitarian law. Striking a UN-marked convoy, a maternity hospital, or a WFP warehouse, and attacking burial teams, is a direct boundary violation — not collateral failure but the crossing of the line the dimension is built to measure.
• ACC (Accountability) — suppressing the monitors. The most corrosive variant. Declaring fact-finders persona non grata, banning UNRWA, deregistering NGOs, and detaining and harassing humanitarian workers attacks the evidentiary record itself. This is the dimension collapsing by intent: the institution acts to prevent measurement.
• AWR (Awareness) — denying the crisis. Refusing to acknowledge famine conditions, disputing humanitarian agencies' figures, and asserting compliance while access is blocked. The denial is not ignorance; it is the front edge of the obstruction.

This fingerprint is what separates the obstruction cohort from states that are merely poor or merely repressive. It is also what makes the harm legible as a class rather than a set of unrelated tragedies — and the evidence below shows it escalating across four theatres at once.

Sudan (0.0) — the most acute live case. On 18 June 2026, the UN's top human-rights official warned that an imminent Rapid Support Forces offensive against El Obeid (al-Obeid), capital of North Kordofan, risks large-scale atrocities, with roughly 500,000 civilians at risk, including more than 100,000 internally displaced. At least 29 countries raised the alarm at the UN Human Rights Council, citing dozens of drone strikes over two weeks targeting fuel stations and trucks and cutting off basic services, and reiterating the need for unhindered humanitarian access. This follows the October 2025 fall of El Fasher after an 18-month starvation siege — which the UN Fact Finding Mission determined on 19 February 2026 bore "the hallmarks of genocide" — and the RSF's declaration of two senior WFP officials as personae non grata with 72 hours to leave. Famine is declared in at least five locations. The obstruction signature is complete: relief blocked (ACT), protected actors and infrastructure targeted (BND), food-aid leadership expelled and access denied (ACC), crisis ongoing amid contested figures (AWR).

Israel / Gaza (0.0) — obstruction during a ceasefire. Israel banned UNRWA from operating on its territory in January 2025 and, since March 2025, has blocked the agency from taking humanitarian assistance directly into Gaza. New registration requirements, administered by a ministry that has taken a hardline stance, have moved to deregister and suspend international NGOs — the entities that provide over half of Gaza's food assistance, support most field hospitals, and deliver the majority of shelter and non-food aid. Human Rights Watch documented in May 2026 that Israel curbed aid and killed civilians during the ceasefire period that took effect 10 October. The UN human-rights chief called the suspension of aid groups "outrageous." That obstruction persists during a ceasefire — when capacity is not the constraint — is what marks it as an active harm rather than a wartime byproduct.

Russia / Ukraine (0.0) — striking the relief infrastructure. On 14 May 2026, Russian FPV drones twice struck a clearly UN-marked humanitarian convoy moving between Kherson and Ostriv. In May 2026 a precision Iskander missile hit the World Food Programme warehouse in Dnipro — the second strike on that facility in six months — destroying food assistance for an estimated 130,000 frontline civilians. A maternity hospital was struck on 1 February 2026. Between January and April 2026, at least 62 incidents affecting humanitarian personnel, vehicles, supplies and facilities were recorded. The repeated, marked, deliberate targeting of protected humanitarian assets is the BND/ACT obstruction signature in its most kinetic form.

DR Congo (2.3) — obstruction inside a contested-territory health emergency. During the 2026 Central Africa Ebola epidemic, a burial team was attacked in South Kivu on 6 June 2026 and forced to abandon a coffin; another safe-and-dignified-burial team was attacked in Mongbwalu (Ituri), and five workers were taken hostage at entry/control points after false accusations of spreading Ebola. Contact tracing reaches only roughly 20–45% of contacts against a WHO requirement of 90%, and ongoing conflict in Ituri restricts surveillance-team movement and the secure transport of lab samples. The obstruction is real and lethal to the response — but a substantial share is driven by non-state armed groups and community-level violence rather than by a state apparatus deliberately closing the door. §5 explains why that distinction holds the DRC at 2.3 rather than 0.0.

The cross-cutting toll. 2025 is on course to be the deadliest year on record for aid workers; provisional data recorded 265 killed by mid-August, up nearly 50% year-on-year, after a record 383 in 2024 (181 in Gaza, 60 in Sudan). Preliminary 2025 data records some 950 human-rights defenders, journalists and trade unionists killed or forcibly disappeared — more than double a decade earlier. These are the witnesses. Their removal is the ACC harm at population scale.

The floor mechanics. The canonical composite renders an all-anchor profile [1×8] as exactly 0.0. Reaching it requires every dimension to collapse to the minimum simultaneously — reserved by ruling for conduct that leaves no remediation surface to credit. Active obstruction is one of the cleanest ways an institution removes that surface. A state with a contested, reversible policy can be scored above the floor because there is something partial to credit. A state expelling food-aid officials, banning relief agencies, and bombing the convoys has removed the partial structure entirely — the response it would be credited for is the thing it is destroying.

The floor is also a ceiling on visible deterioration. This is the methodological crux. Because Sudan, Israel, Russia, Myanmar, South Sudan and Yemen are already at the all-anchor minimum, none of the June 2026 obstruction events can move their scores. The El Obeid build-up, the Dnipro warehouse strike, the Kherson convoy attack, the Gaza NGO suspensions — each re-confirms the floor rather than registering as movement. The number's silence is itself a finding: the scale has run out of room before the conduct has run out of escalation. Severity past the floor is not captured by the composite; it is captured only in the record. That is one reason a thematic read like this one exists — to make visible the deterioration the number cannot.

Why suppressing monitors is uniquely corrosive. Blocking aid harms the people in need now. Suppressing the monitors harms something structural: the accountability record itself. The benchmark — and every accountability institution like it — depends on independent observation: fact-finding missions, humanitarian agencies, journalists, NGOs. When the RSF declares WFP officials persona non grata, when UNRWA is banned and NGOs deregistered, when fact-finders are denied entry, when 950 defenders and journalists are killed or disappeared, the institution is not merely failing to be measured — it is acting to prevent measurement. This is the ACC dimension collapsing by intent. And it is self-sealing: an institution that destroys the witnesses corrodes the very mechanism by which it could ever be credited with recovery. The floor's exit path runs through restored, verified, independent monitoring — exactly the thing this variant is built to destroy.

The 2.3-versus-0.0 line. The DRC's near-anchor 2.3 is the cohort's load-bearing contrast. Its profile has not fully collapsed because a meaningful share of the obstruction is attributable to non-state armed groups and active-conflict dynamics rather than to a state apparatus deliberately closing the response — there remains a thin, contested remediation surface (a functioning, if besieged, national response partnered with WHO and MSF). The 0.0 states have crossed where the DRC has not: the obstruction is policy, executed by the institution being scored. That line — state obstruction versus contested-territory collapse — is precisely what the 2.3-versus-0.0 gap is holding.

• Sudan / El Obeid is the highest-value live trigger. If the RSF assault proceeds against the 500,000 civilians at risk, it will re-confirm Sudan's floor and deepen the documented record — but the score cannot move below 0.0. Watch instead for the accountability response: a Security Council designation, an expanded fact-finding mandate, or — conversely — a further expulsion of monitors, which would be the clearest signal of the corrosive variant intensifying.
• Gaza's NGO deregistration is the access bellwether. Whether international NGOs are restored or permanently removed during the ceasefire determines whether the obstruction is a temporary wartime posture or a settled policy. The latter would harden the floor placement; a verified, sustained reopening of independent relief and monitoring is the only documented route toward an exit.
• Russia's targeting pattern is the BND test. Continued deliberate strikes on marked humanitarian assets (a third Dnipro warehouse hit, further UN-convoy attacks) extend the record. A cessation of such targeting would be the first observable BND-recovery signal — none is currently visible.
• The DRC is the cohort's cusp case. If state authorities move from contested-territory collapse to deliberate closure of the Ebola response — or if the response stabilizes and monitoring is restored — the DRC is the cohort member most likely to move off, or further toward, the floor. It is the entity where the state-obstruction-versus-conflict-collapse line will next be tested.
• The monitor toll is the leading indicator. The trajectory of aid-worker and journalist killings — already at record highs in 2024–25 — is the single broadest signal of whether the obstruction class is expanding beyond this cohort. A continued rise would suggest the pattern is generalizing across the lower band, not confined to these seven states.

• Canonical scores (ground truth): site/src/data/indexes/countries.json — all composites, bands and dimension vectors for Sudan, Israel, Russia, Myanmar, South Sudan, Yemen (0.0 floor) and the Democratic Republic of the Congo (2.3) were recomputed directly from rankings[] and reconcile with the canonical formula. Country index n=193; scored catalog n=1,156 across 7 indexes.
• Formula / methodology: canonical composite baseComposite + integrationPremium per site/scripts/lib/scoring.mjs::computeCompositeFromDimensions; site/src/data/dimensions.ts (8 dimensions). Allegation-versus-adjudication distinction observed throughout (UN Fact Finding Mission determinations cited as adjudicated findings; ongoing events cited as documented incidents).
• Sudan / El Obeid: UN News — "'Stop this madness': Rights chief warns of impending atrocities as militia closes in on El Obeid" (18 Jun 2026) · Al Jazeera — "At least 29 countries raise alarm about atrocities in Sudan's el-Obeid" (18 Jun 2026) · UK Government — UN Fact Finding Mission report on El Fasher / Sudan Core Group statement (Feb 2026) · HRW World Report 2026: Sudan · AOL/Reuters — "Sudan's military expels top UN food aid officials"
• Israel / Gaza: Human Rights Watch — "Gaza: Israel Curbs Aid, Kills Civilians During Ceasefire" (19 May 2026) · Al Jazeera — "Which aid groups is Israel banning from Gaza now" (31 Dec 2025) · Times of Israel — "UN rights chief slams 'outrageous' Israeli suspension of aid groups in Gaza" · UN News — "Gaza aid lifelines under strain" (Jan 2026)
• Russia / Ukraine: France 24 — "'Aid workers are targeted': Russian drones strike UN convoy in Ukraine" (19 May 2026) · UN News — "Ukraine: Russian attack destroys humanitarian food aid in Dnipro" (May 2026) · French MFA — "Russian strikes on a bus and a maternity hospital" (2 Feb 2026)
• DR Congo / Ebola: MSF — "Ebola disease outbreak in DR Congo: MSF response, key facts, and timeline" · WHO — "Ebola outbreak — DRC 2026" · Wikipedia — "2026 Central Africa Ebola epidemic"
• Cross-cutting toll (aid workers / monitors): OCHA — "World Humanitarian Day: Attacks on aid workers hit another record" · UN News — "Rights defender killings hit record high" (Jun 2026) · Save the Children — "World Humanitarian Day: 2025 set to be deadliest year on record"