The Last Mile — Disasters by the Decade

One · The architecture

The machine we built for this

[ DRAFT SCAFFOLD — final essay text to come ]

There is a system. It has a name, a budget, a coordinating agency and a set of acronyms that take a paragraph to unpack: OCHA at the centre, eleven clusters splitting the work by sector, an emergency fund that can disburse in seventy-two hours, and — since 2022 — the Early Warnings for All initiative, which promises that every person on earth will be covered by a multi-hazard warning system by the end of 2027.

Most of it was built in the wake of a specific failure. The cluster system came out of the 2005 Humanitarian Response Review, which followed Darfur. Anticipatory action — money released on a forecast rather than on a body count — came out of the repeated observation that the cheapest moment to act is the one before anything has happened. The architecture is, on paper, a record of lessons learned.

It is also, on paper, extremely good. That is the difficulty this page is about.

Two · The last mile

Where the graveyard is

[ DRAFT SCAFFOLD — final essay text to come ]

In the Ahr valley in July 2021, the European Flood Awareness System issued its warnings days ahead and the forecast was essentially right. One hundred and thirty-five people died in a single German district, in the dark, in a country with the meteorological data on a screen and no siren left standing to broadcast it.

In Valencia in October 2024, the national weather service raised its red warning at half past seven in the morning. The mass alert to mobile phones went out at eleven minutes past eight in the evening. The ravine had been in flood for hours.

The pattern repeats with a consistency that stops looking like coincidence: the forecast is fine, and the last mile — dissemination, comprehension, authority to order the evacuation, somewhere to evacuate to — is where the deaths accumulate.

Figure 1 · The map

Preparedness, by whichever measure you trust

Six international measures of the same thing, none of which agree. Choose one; the map recolours, the legend tells you which way is worse, and the source line tells you who says so and when they last said it.

Layer

Hazard exposure, vulnerability and coping capacity, combined into one risk score.

Hazard

This index is not hazard-disaggregated: one composite figure per country, and no honest way to slice it.

Overlay

Fourteen case studies

The events the essay is arguing from. Full cards below.

Loading…Fetching the layer files.

Structural risk · no unit

—no scale — this layer is not loaded—

no data — the country is not in this source, or not for this hazard

Source not recorded

Keyboard: tab to the map, then ← → to walk countries alphabetically, ↑ ↓ for the nearest neighbour north or south,Enter to open the dossier, Esc to close it. Every value is also in the table at the foot of this figure.

Every country on this layer, as a table (no data loaded)

The base map (world.json) has not been built — run the preparedness pipeline.

Three · Scores against reality

The index that ranked us first

[ DRAFT SCAFFOLD — final essay text to come ]

In October 2019 the Global Health Security Index ranked one hundred and ninety-five countries on their readiness for an epidemic. The United States came first. The United Kingdom came second. Within eighteen months both were among the worst performers in the world on deaths per head.

The index was not incompetent. It measured what could be measured — laboratories, plans, surveillance systems, legal frameworks — and those things were genuinely there. What it could not measure was whether a state would use them: political will, public trust, the willingness of a government to act on its own early data.

Every index on the map above has the same structural problem, and the figure below is an attempt to make it visible rather than to solve it.

Figure 2 · Past against future

What happened, and what was promised

Two composites, built from the same country list. On the left, outcomes already on the record. On the right, capacity as reported by the states that hold it. Set the weights yourself — that is the point of showing them.

A hypothesis, not a verdict. A composite index is an argument with the arithmetic left in. Every component score is printed beside its total; change the weights and the ranking changes, which is the honest thing about it and also the reason not to quote it.

Loading the layers…

Past · observed outcomes

Observed outcomes

Deaths and displacement already recorded. High is good: it means less of both.

Weights drag to re-argue the composite

Recorded mortalitynot loaded

EM-DAT did not load — this component is excluded from the composite below.

Displacementnot loaded

IDMC did not load — this component is excluded from the composite below.

Excluded: Recorded mortality, Displacement. The composite is computed from the 0 components that loaded, and is not comparable with a run where all of them did.

None of this side’s layers have loaded, so there is nothing to rank. The page will not stand in a number for them.

Future · promised capacity

Promised capacity

Coping capacity, adaptation readiness and early-warning coverage, as reported.

Weights drag to re-argue the composite

Coping capacitynot loaded

INFORM did not load — this component is excluded from the composite below.

Adaptation readinessnot loaded

ND-GAIN did not load — this component is excluded from the composite below.

Early warningnot loaded

SENDAI G / MHEWS did not load — this component is excluded from the composite below.

Excluded: Coping capacity, Adaptation readiness, Early warning. The composite is computed from the 0 components that loaded, and is not comparable with a run where all of them did.

None of this side’s layers have loaded, so there is nothing to rank. The page will not stand in a number for them.

Composites computed in the browser from the layer files; components normalised to each source’s own published range and turned so that 100 is always the better end.

Four · The verdicts

There are no natural disasters

[ DRAFT SCAFFOLD — final essay text to come ]

The phrase is older than the discipline, and the official record has been quietly agreeing with it for two decades. Japan’s Diet called Fukushima “a profoundly man-made disaster”. The US Army Corps’ own evaluation task force called New Orleans’ flood defences “a system in name only”. Turkish engineers put the collapse of 2023 on construction amnesties rather than on the fault.

These are not activist framings. They are the findings of commissions convened by the states responsible, and they converge on the same claim: the hazard was natural, and the catastrophe was administrative.

What follows is the record, fourteen times over, with the inquiry named in each case.

Fourteen events, and what the record says

The inquiries, in their own words

Each card names the commission, committee or evaluation that examined the event, because the verdicts are the argument. Where there is no independent inquiry — Myanmar — that is recorded as the finding. Casualty figures are the naming institution’s; several are contested, and the cards say which.

2004

Indian Ocean tsunami

Banda Aceh, Indonesia, and thirteen other countries

≈227,000 dead across 14 countries

What happened

A magnitude 9.1 rupture off northern Sumatra on 26 December 2004 sent waves across an ocean with no warning system in it. Aceh was struck within half an hour; Sri Lanka and India had roughly two hours, Somalia seven. The waves arrived everywhere as a surprise.

What worked

Where knowledge was local, it worked. On Simeulue, the oral tradition of smong — the ground shakes, run to the hills — carried a population of about 78,000 through the wave with a handful of deaths. The Moken sea nomads of the Andaman Sea read the receding water the same way. International giving was, in the end, unprecedented.

What failed

The Pacific Tsunami Warning Center saw the earthquake within minutes and had no mandate, no contact list and no channel to warn the Indian Ocean rim. Countries with two hours of lead time used none of it. The relief that followed was assessed as generous, duplicative, supply-driven and largely deaf to the people it served.

What the inquiry concluded

The Tsunami Evaluation Coalition — a joint evaluation by more than forty agencies, published through ALNAP in 2006 — concluded that the international response was well funded but poorly coordinated and that it too often bypassed and weakened local capacity. Its central recommendation was to strengthen national and local systems rather than the international ones. The Indian Ocean Tsunami Warning and Mitigation System was stood up under UNESCO-IOC in 2006.

Sources (3)

Tsunami Evaluation Coalition, “Synthesis Report: Expanded Summary — Joint evaluation of the international response to the Indian Ocean tsunami” (ALNAP, 2006)
UNESCO Intergovernmental Oceanographic Commission, Indian Ocean Tsunami Warning and Mitigation System (IOTWMS) programme documents
US Geological Survey, M 9.1 Sumatra–Andaman Islands earthquake summary

2005

Hurricane Katrina

New Orleans and the Gulf Coast, United States

≈1,390 dead in Louisiana and Mississippi

What happened

Katrina made landfall on 29 August 2005 as a large Category 3. The storm was survivable; the flood was not. More than fifty breaches in the federal hurricane-protection system put roughly eighty per cent of New Orleans under water, in places for weeks.

What worked

The National Hurricane Center forecast the track and the surge accurately and days ahead. The Coast Guard, operating on local initiative, rescued around 33,500 people. Neighbours in boats did much of the rest.

What failed

The evacuation order assumed a car. It was issued late, and roughly a hundred thousand residents — disproportionately poor, elderly and Black — had no way to comply. Federal coordination collapsed for days. The levees and floodwalls failed below their design storm, at design and construction faults, not at the limits of nature.

What the inquiry concluded

The US House bipartisan select committee titled its 2006 report “A Failure of Initiative”. The Interagency Performance Evaluation Task Force, convened by the Army Corps of Engineers itself, found that the hurricane protection works were “a system in name only” — built piecemeal over decades to inconsistent standards. The independent levee investigation team led from UC Berkeley reached the same conclusion, and put the primary cause of the flooding on engineering and institutional failure rather than on the storm.

Sources (3)

US House of Representatives, “A Failure of Initiative: Final Report of the Select Bipartisan Committee to Investigate the Preparation for and Response to Hurricane Katrina” (2006)
Interagency Performance Evaluation Task Force (US Army Corps of Engineers), “Performance Evaluation of the New Orleans and Southeast Louisiana Hurricane Protection System” (2006–2009)
Independent Levee Investigation Team (UC Berkeley / National Science Foundation), “Investigation of the Performance of the New Orleans Flood Protection Systems in Hurricane Katrina” (2006)

2010

Haiti earthquake

Port-au-Prince and Léogâne, Haiti

Government figures of 220,000–316,000 dead; independent estimates run far lower and the count remains contested

What happened

A magnitude 7.0 earthquake at 16:53 on 12 January 2010, twenty-five kilometres from a capital city of concrete frames built without a seismic code. The presidential palace, the parliament, the cathedral, the tax office and the UN mission headquarters all came down.

What worked

Haitians pulled the overwhelming majority of survivors out of the rubble in the first hours, before any foreign team landed. International urban search and rescue, once it arrived, made about 130 live extractions — the largest such operation on record.

What failed

There was no enforced building code to fail; there was no code. Coordination ran through a cluster system operating in English, in a French- and Creole-speaking country, on a UN base. And the response introduced a second disaster: cholera, absent from Haiti for a century, killing roughly ten thousand people over the following decade.

What the inquiry concluded

The UN Independent Panel of Experts on the Cholera Outbreak in Haiti reported in 2011 that the strain matched South Asian cholera and that sanitation at the Nepalese peacekeepers’ camp on the Meille tributary was inadequate; in 2016 the Secretary-General acknowledged the UN’s involvement and apologised. Separately, the Office of the Special Envoy for Haiti found that of the relief funding disbursed, about one per cent went to the Haitian government and well under one per cent to Haitian firms — the finding that made “build back better” a contested phrase.

Sources (3)

Independent Panel of Experts on the Cholera Outbreak in Haiti, report to the UN Secretary-General (2011)
Office of the Special Envoy for Haiti, “Has Aid Changed? Channelling assistance to Haiti before and after the earthquake” (United Nations, 2011)
Disasters Emergency Committee / Groupe URD, real-time evaluations of the Haiti earthquake response (2010–2011)

2011

Tōhoku earthquake, tsunami and Fukushima Daiichi

Sanriku coast and Fukushima Prefecture, Japan

≈19,750 dead or missing; about 165,000 evacuated from the nuclear exclusion zone

What happened

A magnitude 9.0 earthquake on 11 March 2011 generated a tsunami that overtopped seawalls designed for a smaller ocean. At Fukushima Daiichi the wave flooded the emergency generators, causing a station blackout and, over the following days, three core meltdowns and hydrogen explosions.

What worked

The buildings held. Japan’s seismic code did what it was written to do, and almost all of the deaths were drownings, not collapses. The Japan Meteorological Agency issued a warning within three minutes. In Kamaishi, years of school evacuation drills sent children uphill on their own initiative and pulled adults with them.

What failed

The design tsunami was set by an assumption, not by the geological record, and the operator had internal estimates it did not act on. The nuclear regulator was housed inside the ministry promoting nuclear power. Evacuating hospitals and care homes inside the exclusion zone killed patients that the radiation did not.

What the inquiry concluded

The Fukushima Nuclear Accident Independent Investigation Commission — the first independent commission in the history of Japan’s Diet — concluded in 2012 that the accident “was a profoundly man-made disaster that could and should have been foreseen and prevented”, and attributed it to collusion between the government, the regulators and TEPCO. The government’s own Investigation Committee (the Hatamura committee) reached compatible findings on the operator’s tsunami assumptions.

Sources (3)

National Diet of Japan, Fukushima Nuclear Accident Independent Investigation Commission (NAIIC), official report (2012)
Investigation Committee on the Accident at the Fukushima Nuclear Power Stations of Tokyo Electric Power Company (ICANPS), final report (2012)
International Atomic Energy Agency, “The Fukushima Daiichi Accident” — Report by the Director General (2015)

2013

Typhoon Haiyan (Yolanda)

Tacloban and Eastern Visayas, Philippines

6,300+ dead, 1,000+ missing, 4 million displaced

What happened

One of the strongest tropical cyclones ever measured at landfall crossed the Visayas on 8 November 2013. In Tacloban the storm surge reached five to six metres and behaved, as survivors kept saying, like a tsunami.

What worked

PAGASA forecast the track well and the government pre-emptively evacuated hundreds of thousands of people. That is the reason the toll is in the thousands rather than the tens of thousands.

What failed

The warning used the words “storm surge”, which had no purchase in Waray or Tagalog and which people heard as heavy rain. Some designated evacuation centres — including the Tacloban Astrodome — were inside the surge zone. Local government was itself a casualty: the responders were among the affected, and the response lost its first days to that.

What the inquiry concluded

The Inter-Agency Standing Committee’s Inter-Agency Humanitarian Evaluation of the Typhoon Haiyan response (2014) found a system that mobilised at speed but sidelined national and local structures and communicated poorly with affected people. PAGASA subsequently introduced explicit storm-surge warnings, hazard maps and colour-coded rainfall alerts — a warning failure fixed at the level of vocabulary.

Sources (3)

Inter-Agency Standing Committee, “Inter-Agency Humanitarian Evaluation of the Typhoon Haiyan Response” (2014)
Philippine National Disaster Risk Reduction and Management Council, Yolanda situation and final reports (2013–2014)
PAGASA, storm-surge warning system documentation (post-Yolanda revisions)

2014

West Africa Ebola epidemic

Guinea, Liberia and Sierra Leone

28,600 cases and 11,325 deaths recorded, 2014–2016

What happened

The index case was a child in Meliandou, Guinea, in December 2013. The outbreak was identified in March 2014, crossed three national borders through the Guinea–Liberia–Sierra Leone forest region, and reached capital cities — the first time Ebola had done so.

What worked

Nigeria stopped an imported chain of transmission in Lagos using contact-tracing staff and infrastructure built for polio eradication. Communities themselves changed burial practice, which mattered more than any imported intervention. The ring-vaccination trial in Guinea produced the first efficacious Ebola vaccine.

What failed

WHO declared a Public Health Emergency of International Concern on 8 August 2014, months after its own staff and Médecins Sans Frontières had raised the alarm; MSF had said in June that the outbreak was out of control. Country offices minimised the reports. The three affected states had almost no health workforce to lose, and lost it.

What the inquiry concluded

The Ebola Interim Assessment Panel convened by WHO (the Stocking panel, 2015) concluded that “WHO does not currently possess the capacity or organizational culture to deliver a full emergency public health response.” The Harvard–LSHTM Independent Panel on the Global Response to Ebola, published in The Lancet in 2015, called the delay a failure of global institutions with catastrophic human consequences. Both fed the creation of the WHO Health Emergencies Programme and the Contingency Fund for Emergencies.

Sources (3)

WHO Ebola Interim Assessment Panel, report to the Director-General (2015)
Harvard Global Health Institute / London School of Hygiene & Tropical Medicine Independent Panel, “Will Ebola change the game?”, The Lancet (2015)
UN High-level Panel on the Global Response to Health Crises, “Protecting Humanity from Future Health Crises” (2016)

2020global

COVID-19

Global — pinned at WHO headquarters, Geneva

7 million deaths reported to WHO; ≈14.9 million excess deaths estimated for 2020–2021

What happened

A novel coronavirus reported from Wuhan on 31 December 2019 became a pandemic within ten weeks. Every country had a plan; almost none of the plans described what happened.

What worked

Genomic sequencing published in January 2020 and licensed vaccines within a year — the fastest vaccine development in history. Some states with recent epidemic memory, rather than high index scores, moved early and held.

What failed

The countries at the top of the 2019 Global Health Security Index — the United States first, the United Kingdom second — recorded among the worst per-capita death tolls in the world. The International Health Regulations had no enforcement and rewarded silence. February 2020 was lost. COVAX delivered late, and vaccine supply followed purchasing power.

What the inquiry concluded

The Independent Panel for Pandemic Preparedness and Response, co-chaired by Helen Clark and Ellen Johnson Sirleaf, reported in May 2021 under the title “COVID-19: Make it the Last Pandemic”. It called the pandemic “a preventable disaster”, described February 2020 as a lost month, and found the alert system too slow and too deferential. Its findings drove the negotiation of the WHO Pandemic Agreement, adopted in May 2025 — and the index-versus-outcome gap it exposed is the reason this page treats every score below as a hypothesis.

Sources (4)

Independent Panel for Pandemic Preparedness and Response, “COVID-19: Make it the Last Pandemic” (2021)
WHO, “Global excess deaths associated with COVID-19, 2020–2021” (2022)
Review Committee on the Functioning of the International Health Regulations (2005) during the COVID-19 Response, report to the World Health Assembly (2021)
Nuclear Threat Initiative / Johns Hopkins Center for Health Security, Global Health Security Index (2019 and 2021 editions)

2021

Ahr valley floods

Ahrweiler district, Rhineland-Palatinate, Germany

189 dead in Germany, 135 of them in the Ahrweiler district; 43 dead in Belgium

What happened

A stalled low-pressure system dropped up to 150 mm of rain in a day on saturated ground on 14–15 July 2021. The Ahr rose about eight metres and took bridges, gauges and whole streets with it in the dark.

What worked

The forecast. The European Flood Awareness System flagged an extreme event days in advance and issued explicit warnings to German authorities; the meteorology was, in the assessment that followed, essentially correct.

What failed

The translation from forecast into evacuation. District-level decisions came late or not at all, sirens had been decommissioned in the post-Cold-War decades, and Germany had no cell-broadcast alerting — it was introduced in 2023, after this. Twelve residents of a care home for people with disabilities in Sinzig drowned because nobody moved them.

What the inquiry concluded

The Landtag of Rhineland-Palatinate ran a committee of inquiry (Untersuchungsausschuss) into the flood and the warning chain; the state prosecutor opened proceedings against the Ahrweiler district administrator over delayed warnings, later discontinued. Independent post-event analysis published in Hydrology and Earth System Sciences put the failure squarely in dissemination rather than in prediction — the flood was forecast, and not warned.

Sources (3)

Landtag Rheinland-Pfalz, Untersuchungsausschuss on the July 2021 flood disaster, final report (2024)
Copernicus European Flood Awareness System (EFAS), post-event analysis of the July 2021 floods
Hydrology and Earth System Sciences, post-event review of forecasting and warning in the July 2021 western Europe floods (2022–2023)

2022

Pakistan floods

Sindh, Balochistan and southern Punjab, Pakistan

1,739 dead; 33 million people affected; roughly a third of the country under water

What happened

A monsoon delivering five to eight times normal rainfall in Sindh and Balochistan, on top of accelerated glacier melt, flooded a third of Pakistan between June and September 2022 and left standing water for months.

What worked

The assessment machinery. A joint government–multilateral needs assessment was produced within weeks and became the evidentiary basis for the loss-and-damage fund agreed at COP27 three months later — the first time a specific disaster produced a specific instrument.

What failed

Money. The revised UN flash appeal closed far short of its target while the assessed need ran to tens of billions. Anticipatory-action financing existed at pilot scale against a national-scale event, and floodplain settlement and drainage neglect turned rainfall into inundation that lasted into 2023.

What the inquiry concluded

The Government of Pakistan’s Post-Disaster Needs Assessment, prepared with the Asian Development Bank, the European Union, UNDP and the World Bank in October 2022, put damages at about $14.9 billion, economic losses at about $15.2 billion and reconstruction needs at over $16 billion. World Weather Attribution found climate change had likely intensified the rainfall, while identifying vulnerability — infrastructure, proximity, poverty — as the dominant driver of the impact.

Sources (3)

Government of Pakistan with ADB, EU, UNDP and the World Bank, “Pakistan Floods 2022: Post-Disaster Needs Assessment” (2022)
World Weather Attribution, “Climate change likely increased extreme monsoon rainfall, flooding highly vulnerable communities in Pakistan” (2022)
UN OCHA, Pakistan Floods Response Plan and revision (2022)

2023

Türkiye–Syria earthquakes

Kahramanmaraş, eleven Turkish provinces, and north-west Syria

53,500+ dead in Türkiye and 8,000+ in Syria

What happened

A magnitude 7.8 rupture at 04:17 on 6 February 2023, followed nine hours later by a 7.5 on a neighbouring fault. Tens of thousands of buildings collapsed, many of them pancaking floor onto floor in the manner of unconfined concrete frames.

What worked

Turkish search and rescue and a civilian mobilisation of extraordinary scale; the AHBAP network and ordinary convoys reached towns before the state did. International teams arrived in volume within seventy-two hours.

What failed

Türkiye has had a modern seismic code since 1998 and a stricter one since 2018. It also had repeated construction amnesties — most recently in 2018 — that legalised, for a fee, millions of buildings that did not comply. Military deployment lagged the first day. In north-west Syria, cross-border aid was funnelled through a single authorised crossing and effectively stopped for the first week.

What the inquiry concluded

The Grand National Assembly of Türkiye established a parliamentary investigation committee, which reported in 2023 on building-stock non-compliance and on the coordination of the first seventy-two hours. The Chamber of Civil Engineers of the Union of Chambers of Turkish Engineers and Architects (TMMOB) attributed the scale of the collapse to unenforced codes and to the amnesty regime; prosecutors detained more than two hundred contractors, developers and officials. The finding, in short: the earthquake was natural and the collapse was policy.

Sources (3)

Grand National Assembly of Türkiye, parliamentary investigation committee report on the 6 February 2023 earthquakes (2023)
TMMOB Chamber of Civil Engineers, earthquake damage assessment reports (2023)
UN OCHA, Türkiye and Syria Earthquake Flash Appeals and situation reports (2023)

2023

Derna dam collapse (Storm Daniel)

Derna, eastern Libya

Over 4,300 bodies recovered and thousands more missing; totals from 5,900 to 11,300 have been published and remain contested

What happened

Storm Daniel crossed the Mediterranean and stalled over the Jebel Akhdar on 10–11 September 2023. The Abu Mansour and Al-Bilad dams above Derna failed in sequence in the night, and the resulting wall of water carried entire city blocks into the sea.

What worked

Very little. Libya’s National Meteorological Centre did issue warnings, days ahead, and passed them to the authorities.

What failed

The dams had not been properly maintained since 2002, and a 2022 paper by a hydrologist at Omar Al-Mukhtar University in Derna had warned in print that they would fail catastrophically in a large flood. The warnings that reached residents were contradictory — a curfew told people to stay indoors, in the houses the water took. A decade of rival administrations had left no single authority responsible for the structures.

What the inquiry concluded

Libya’s Attorney-General opened a criminal investigation and ordered the detention of officials from the Derna municipality and the water resources authority; eight were charged over dam maintenance and the failure to evacuate. The World Meteorological Organization’s Secretary-General stated publicly that with a functioning warning and emergency management chain, most of the casualties could have been avoided.

Sources (3)

Office of the Attorney-General of Libya, statements on the Derna dam investigation (2023)
World Meteorological Organization, statements and briefings on Storm Daniel and the Derna floods (September 2023)
UN OCHA, Libya Flood Flash Appeal and situation reports (2023)

2024

Valencia flash floods (DANA)

Paiporta, Catarroja and the Horta Sud, Valencian Community, Spain

≈230 dead, the great majority in the Valencian Community

What happened

An isolated upper-level depression dropped close to a year of rain on parts of the Valencian hinterland on 29 October 2024. The Poyo ravine, dry for most of the year, carried a torrent through towns at the evening rush hour; most of the dead were caught in cars and ground-floor garages.

What worked

The forecast, again. AEMET raised its red warning at 07:31 that morning, and had signalled the risk the day before. The meteorological chain did its job.

What failed

The alert. The regional government sent the ES-Alert mass message to mobile phones at 20:11, by which time the ravine had been in flood for hours and many of the victims were already dead. Gauge telemetry on the Poyo went dark during the critical window, and the regional emergency coordination meeting convened late.

What the inquiry concluded

The investigating court No. 3 of Catarroja opened a criminal case into the delay, naming the regional councillor for justice and interior, Salomé Pradas, and her deputy Emilio Argüeso, on suspicion of homicide and injury by gross negligence over the timing of the alert; the case has proceeded through 2025. The Corts Valencianes and the Spanish Congress both opened inquiry committees. The finding taking shape is not that the flood was unforeseen, but that the ninety minutes between knowledge and message were fatal.

Sources (3)

Juzgado de Instrucción n.º 3 de Catarroja, proceedings on the DANA of 29 October 2024 (2024–2025)
Agencia Estatal de Meteorología (AEMET), warning chronology for the DANA of 29 October 2024
Confederación Hidrográfica del Júcar, hydrological data and reporting on the Barranco del Poyo (2024)

2025

Palisades and Eaton fires

Pacific Palisades and Altadena, Los Angeles County, United States

30+ dead; ~16,000 structures destroyed; insured losses in the tens of billions of dollars

What happened

Two wind-driven fires ignited on 7 January 2025 during a Santa Ana event with gusts near a hundred miles an hour, after an exceptionally dry start to the wet season. The Palisades and Eaton fires became the two most destructive fires in Los Angeles County history within a day.

What worked

The weather service had issued its most severe fire-weather language days ahead and named the danger in unusually blunt terms. Roughly 200,000 people were ordered out, and the great majority got out.

What failed

Not everyone was told in time. In west Altadena, evacuation orders did not go out until the small hours of 8 January, after the fire was already in the neighbourhood — and almost all of the Eaton fire deaths were in that area. There were no sirens. Hydrants lost pressure as demand outran the local system, and utility equipment is under investigation as an ignition source.

What the inquiry concluded

Los Angeles County commissioned an independent after-action review from the McChrystal Group, whose 2025 report found that alerts to west Altadena were issued hours after those to the east and identified failures in the county’s alert and warning process. State and utility-regulator investigations into the ignition of the Eaton fire continue. This is also the event behind the arithmetic further down this page: the insured losses here are of the same order as a full year of global humanitarian appeals.

Sources (3)

County of Los Angeles / McChrystal Group, after-action review of the January 2025 windstorm and wildfires (2025)
CAL FIRE, Palisades and Eaton fire incident records (2025)
California Public Utilities Commission and California Department of Insurance, filings and loss estimates on the January 2025 fires

2025

Myanmar earthquake

Sagaing, Mandalay and Naypyidaw, Myanmar

Official figures of about 3,800 dead; independent estimates are higher and verification is obstructed

What happened

A magnitude 7.7 strike-slip rupture along the Sagaing Fault on 28 March 2025, shallow and running for hundreds of kilometres beside the country’s second city. It struck a state already four years into a civil war, with millions displaced before the ground moved.

What worked

Regional response was fast: Chinese, Indian and ASEAN teams were on the ground within days. Local civil society, community networks and administrations outside government control did most of the immediate rescue, as they had been doing for the war.

What failed

The military authorities continued airstrikes in earthquake-affected areas during the response period, restricted access for international responders and maintained communications blackouts. The disaster also landed in the same quarter as the dismantling of USAID: the American response, historically the largest single bilateral surge capacity in the world, amounted to a three-person assessment team.

What the inquiry concluded

There is no independent national inquiry, and that is itself the finding. The UN Office of the High Commissioner for Human Rights documented continued military attacks in quake-affected areas after the event; the Special Rapporteur on the situation of human rights in Myanmar called for a halt to hostilities and for unimpeded humanitarian access. The record here is composed of UN statements and OCHA reporting rather than of any commission with subpoena power.

Sources (3)

UN Office of the High Commissioner for Human Rights, statements on attacks in earthquake-affected areas of Myanmar (2025)
UN Special Rapporteur on the situation of human rights in Myanmar, statements following the 28 March 2025 earthquake
UN OCHA, Myanmar Earthquake Response situation reports and addendum to the Humanitarian Needs and Response Plan (2025)

Five · The arithmetic

What a life is priced at

[ DRAFT SCAFFOLD — final essay text to come ]

In January 2025, two fires in Los Angeles County destroyed around sixteen thousand structures and generated insured losses in the tens of billions of dollars. In the same period, the UN’s coordinated humanitarian appeals — the bill for every declared emergency on earth — asked for a comparable sum and received a fraction of it.

The comparison is not a moral equivalence and should not be read as one. It is a statement about where the world keeps its money and how quickly it can move it.

The figure below is the second half of that arithmetic: what was asked for, what arrived, and the size of the space between.

Figure 3 · The moral arithmetic

Asked for, and received

The UN-coordinated appeals are the closest thing there is to a global bill for catastrophe. It has never been paid in full.

Loading the funding record…

Loss and damage

Data not loadedPledges to the Fund for responding to Loss and Damage have not been ingested. Until they are, this page will not put a figure here.

Source not recorded

Six · Back to the pictures

What cinema gets wrong

[ DRAFT SCAFFOLD — final essay text to come ]

The disaster film is structurally incapable of telling this story. Its engine is the hazard: the wave has to be visible, the fault has to open, the virus has to do something on camera. Bureaucratic delay does not photograph.

So the genre invents a substitute — one obstructive mayor, one venal executive, one scientist nobody believes — and compresses a systemic failure into a character. It is dramatically necessary and analytically useless. Real inquiries almost never find a villain; they find a committee that met late.

The catalogue of eight hundred-odd disaster films next door is, read this way, an archive of the wrong lesson repeated for a century. The Atlas of Ruin shows where the genre thinks the world ends. This page shows where it actually does.