Category: Uncategorized

Lessons for global catastrophic risk from the Arctic Circle Assembly 2026, beyond AI

TLDR/Summary

• The Arctic Circle Assembly (Reykjavík, Iceland, 8–10 October 2026) showed a region no longer insulated from the world. Nuclear risk, energy shocks after Hormuz, great-power posturing and Greenland were all on stage; New Zealand is not insulated either.
• Read my companion blog reporting on AI risk as discussed at the Assembly. With my own inferences about domination, utopia vs dystopia, and the possibility of societal collapse.
• The Atlantic Meridional Overturning Circulation (AMOC) was the Assembly’s clearest case of a catastrophic risk under deep uncertainty. The science cannot say when or whether it will tip, but every new line of evidence moves the risk the same way.
• AMOC is not only a current. Sea ice, ice algae, copepods and fish stocks are changing now, with consequences for carbon storage and for human food.
• Science is doing its job; decision-making under deep uncertainty is policy’s job. Policymakers waiting for certainty are applying the wrong test. No-regrets actions in food, fuel and preparedness pay off whatever the ocean does.
• Iceland shows how a risk reaches the agenda. A scientists’ letter, a ministerial report, the National Security Council, and well-crafted op-eds that offered a minister the chance to lead.
• Every system New Zealand depends on is a current that can tip. Geopolitics, supply chains, resources and social cohesion need the same continuous monitoring, long-term funding and Plan B that AMOC is getting.
• Iceland is counting its stocks; New Zealand should too. Iceland’s reports on fuel, feed and food reserves are the model, and its own history shows what happens to an island that cannot secure its shipping.

Introduction

The Arctic Circle Assembly fills Reykjavík’s Harpa concert hall each October with hundreds of sessions. I sampled perhaps 5%: a Wednesday pre-event “Connected Ocean Observing: ARIA x Reykjavík Institute”, run by the UK’s Advanced Research and Invention Agency (ARIA) with the Reykjavík Institute, and the first two of three days, chosen through a global catastrophic risk lens.

This post is therefore partial. My companion post covers the AI sessions, including Tristan Harris’s plenary, the best session I attended. My thanks to Arctic Circle and its founder, former President Ólafur Ragnar Grímsson, for an event where anyone may ask a question. Here I survey the geopolitics and infrastructure sessions, then use AMOC as the main case. The test throughout is the one I applied at SRA Nordic last week: what lessons transfer to global catastrophic risk, and what should New Zealand learn?

The geopolitical weather

The day one plenary set the tone. Canada’s Foreign Minister Anita Anand spoke of sovereignty, respect for Arctic peoples, security, economy, and environment, and cited Prime Minister Mark Carney’s Davos case for “value-based realism”: with the threat environment changed, collaboration among like-minded states is now a necessity. That fits our view that cooperation on catastrophic risk evolves most readily at the regional level, since individual jurisdictions can’t manage cross-border risks alone. Greenland’s representatives set out their view of sovereignty and way of life in the face of US pressure. “Nothing about Greenland without Greenland.”

The US session, with the Governor of Alaska and the Chair of the US Arctic Research Commission, was about hard assets: ports, shipping lanes, icebreakers and missile defence, and partners to help pay for them. Some of the language was painful and misguided. More telling was the omission. Neither speaker mentioned climate change until asked by a member of the audience, and the answers deflected.

“Can the Arctic stay calm while the world burns?” took the cascading view and applied it to any far-flung region of the world. Lassi Heininen, a long-time scholar of Arctic politics, recalled that Reagan and Gorbachev met in Reykjavík 40 years ago this month and nearly agreed to abolish nuclear weapons.

Today no arms control treaty or serious negotiation is current, and with the stresses of climate change added, he is “bloody worried” about the risk of nuclear war. Francesco Sassi, a Norway-based energy analyst argued that security must now include energy security: after Hormuz, and a possible US export ban, Arctic oil and gas production could rise and new alliances form. Others traced Russia–China cooperation on pipelines, shipping and coast guard patrols.

The verdict was that the Arctic cannot be insulated from these shocks. Neither can New Zealand. Prices are not coming down, and politicians need a realist view, and to communicate expectations with society. Unfortunately the audience questions stayed with energy, not nuclear spillover, which is where the catastrophic tail sits.

Infrastructure that serves two purposes

The session I would most like New Zealand’s regional councils to have seen was on dual-use infrastructure in Iceland’s West Fjords, a remote, sparsely populated region much like the West Coast where I live. Moderator Gylfi Ólafsson described how the community, as the security picture darkened, developed a shared vocabulary and a report to Parliament that prioritised local investments by cost, feasibility and dual use; search and rescue, roadside mobile coverage and fibre came out on top. But other investments in maritime upgrades, radar stations, submarine cable security, and a relocated airport are all topical. Brynja Huld Óskarsdóttir argued that dual-use infrastructure is the fastest route to defence resilience: it keeps people living there, and presence is both deterrence and capability. Patti Bruns (Arctic Mayors’ Forum) and Peter Clarkson (Mayor of Inuvik) agreed: there is no Arctic security without healthy Arctic communities.

What was missing was food and liquid fuel, the two inputs that decide whether a remote community functions in a crisis. But the community process deserves as much attention as the list. New Zealand could take this approach independently, adopting NATO-style baseline resilience requirements without NATO, and building priorities through local citizen deliberation of the kind our charity Islands for the Future of Humanity are proposing to pilot in Reefton, NZ.

AMOC: what science can and cannot say

The Atlantic Meridional Overturning Circulation (AMOC) carries warm water north and is a large part of why north-west Europe and Iceland are so mild for their latitude. A sharp weakening would cool the North Atlantic even as the world warms, affecting weather, sea level, fisheries and agriculture across the hemisphere.

ARIA’s pre-event was the most candid account of uncertainty I heard all week. ARIA’s Forecasting Tipping Points programme (£114 million, led by Sarah Bohndiek and Gemma Bale) aims to build an early-warning system for the North Atlantic, combining established observing arrays with cheaper autonomous instruments and new models, and is now moving to its second phase.

Dr Graeme MacGilchrist (University of St Andrews) organised the science into four questions. Can AMOC collapse? Yes, in models, but rates and mechanisms are unclear. Has it? Evidence for past collapses is ambiguous. Is it changing now? The instrumental record is about 20 years, too short to say. Will it? Persistent model biases affect how sensitive the system is to forcing. Early warning, he said, is elegant theory that is not yet validated, and today’s observing system is a collection of projects with precarious funding.

The next morning’s panel, organised by Iceland, the UK and the Netherlands, gave the same message more briefly. Femke de Jong (NIOZ) explained how few lines of instruments cross the Atlantic. Laura Jackson (UK Met Office) said the AMOC is likely to weaken through this century, with abrupt collapse unlikely before 2100, but a tipping point is plausible and a much weaker state could follow. Steingrímur Jónsson (Marine and Freshwater Research Institute, Iceland) argued for sustained monitoring. Other tipping points, such as the West Antarctic Ice Sheet, may occur first and interact.

Not just a current

A session on the Arctic marine food web showed that the changes are already under way. Geraint Tarling (British Antarctic Survey) explained how sea-ice algae and copepods drive a polar biological carbon pump, and how a seasonal lipid pump locks carbon away below 1,000 metres for centuries.

Loss of sea ice, invasion by Atlantic species and ocean acidification are weakening it, a triple whammy that feeds back into warming. Antje Boetius (Monterey Bay Aquarium Research Institute) described returning to sites where long filaments of ice algae once hung beneath the ice and finding none, replaced by swarms of gelatinous life. Sea ice now forms in different places and drifts faster and thinner, and multi-year ice, with its years-long ecosystems, is gone.

The lesson is that “AMOC risk” is not one variable. It is a coupled system of currents, ice, plankton and fish, ending in food prices, livelihoods and cod stocks. When I asked whether AMOC monitoring and biological monitoring are linked, the answer was that new devices make more possible. Ocean flora and plankton may be the canary in the AMOC mine, and North Atlantic monitoring should include them if it does not already.

The risk keeps moving one way

On day two Johan Rockström (Potsdam Institute for Climate Impact Research) called AMOC the “poster child” tipping point and noted that six of the climate system’s tipping elements are in the Arctic, with interactions we do not understand. Recent studies project limited weakening of 18–43% by 2100 on one approach, and 51 ± 8% under intermediate emissions when models are constrained by observations; others find collapse after 2100 in many models. Rockström has put the probability of a drastic slowdown or collapse at 10% or more. That is a non-trivial risk to habitability and human life on both sides of the Atlantic.

A later session asked whether a Chief Planetary Science Officer could help turn such science into policy. Luc Bas holds that role for Belgium. Rockström noted that the entire range of AMOC uncertainty has moved upwards. Even conservative estimates now carry real risk, and even the lower impacts are severe. This is a pattern across global risk: as evidence accumulates, the news from a human point of view mostly gets worse. Uncertainty is not a reason to expect the benign end.

Science is doing its job; now policy must do its own

At the morning panel, one diplomat opined that science takes a long time, which makes it hard for policymakers. That is the wrong framing. Scientists are providing what they can: the mechanisms, the evidence and the uncertainty around it. Deciding under that uncertainty is not a scientific problem. It is a policy skill, decision-making under deep uncertainty (DMDU), with established methods: robust strategies that perform acceptably across many futures, adaptive pathways with signposts and triggers, and stress-testing against scenarios.

Policymakers should be expected to have these technical skills. Iceland’s Minister Jóhann Páll Jóhannsson has said it plainly: “We cannot afford to wait for definitive, long-term research before acting.”

The answer under deep uncertainty is no-regrets action: measures worth taking whatever the ocean does. The UK speaker at ARIA’s session put it well: there will never be a government department for AMOC, but many bad things could happen, so build resilience across them. Food strategy, fuel stocks, fisheries adaptation and migration planning all qualify.

I note that this is the logic of our own NZCat project, which contemplated the impacts of a Northern Hemisphere nuclear war on Aotearoa/New Zealand. We concluded that many of the same consequences for societal systems exist across many global risks. Our mantra was “solve for nuclear winter, solve for all”: prepare vital systems for the severe case and you are prepared for most of the rest.

In peripheral discussions I learned how AMOC reached the top of Iceland’s agenda. In 2024, 44 scientists wrote to the Nordic Council of Ministers. The Icelandic Minister for the Environment, Energy, and Climate asked his department for a report, tabled it in Parliament and referred the issue to the National Security Council, which in 2025 added AMOC to its list of threats, the first climate phenomenon to be treated that way. Alongside this, a few people wrote careful op-eds that framed the issue as one of security, for food and livelihoods, and left space for someone to step up. The minister took it by the horns and became the hero of the story: a win for everyone. The lesson echoes Harris’s session and last week’s point about whoever frames a risk first. A few well-crafted pieces, repeated in every talk and article, do more than a stream of quick items.

The security framing is spreading. The minister wrote on AMOC in the second issue (Haust 2026) of Tímarit Vörðu, Iceland’s first journal of national security, launched at an event adjacent to the Assembly on 9 October by its editor Kristinn Árni L. Hróbjartsson.

The issue argued for the importance of Icelandic-flagged ships and low taxes on air and sea transport for transport security, so that commercial operators keep Iceland in their networks. It also asked where Iceland’s increased NATO commitment is actually going. The minister also argued that a security lens lifts climate out of left and right, rich and poor, and can build cross-class coalitions. That matches what we found in our survey of New Zealanders, where a majority favoured government plans for global catastrophic risks, and a dedicated agency.

Every stressed global system is a current

AMOC is a chaotic system that can tip, collapse, or decay without collapsing, and it is monitored only because people built the instruments, funding and institutions to watch it. The systems New Zealand depends on are currents too. Geopolitics, trade and shipping, fuel and fertiliser supply, critical minerals, the information environment and social cohesion can each tip, collapse or slowly degrade, and they interact.

We monitor almost none of them continuously with validated metrics and a view to tipping points. AMOC monitoring teaches four things. Watch constantly, with long-term dedicated funding, not a collection of short projects. Collaborate internationally, because no one country can see the whole system. Keep the information clean and shared. And prepare Plan A and Plan B, with triggers agreed in advance for moving from one to the other. That is the core of our proposals for New Zealand’s national risk assessment: an accountable entity, monitoring of complex and stressed global systems, a register that maps vulnerabilities to costed options, and realistic assumptions about resources, energy and geopolitics.

What resilience looks like: Iceland counts its stocks

Iceland is doing the unglamorous part. A 2022 task force in the Prime Minister’s Office found that no rules set minimum stocks of food or farm inputs and that nobody tracked how much food importers and retailers held.

Two follow-up reports for the government fill in the detail. The Agricultural University’s 2024 report on emergency stocks for food production ranks fuel first, then electricity, then imported feed: without fuel, stockpiles of everything else are of little use. It calls Iceland’s oil position “unacceptable” and recommends at least three months of fuel, feed and veterinary medicines, and spreading the timing of fertiliser imports, 85% of which now arrive between March and May.

The University of Iceland’s 2025 proposals set state stocks for three weeks to six months of isolation, and seven days for every household; Iceland grows about 1% of the grain its people eat, and its only grain mill closed in 2025. Both were presented to a government seminar in November 2025, alongside a public food security dashboard.

Iceland’s electricity is fully renewable and islanded, but its transport, fishing fleet and aviation depend on imported oil, with no mandatory stockholding yet in force.

These problems of island isolation are old. In 1262 Icelandic chieftains swore allegiance to the King of Norway, ending the Commonwealth. Exhausted by civil war, they asked in return for peace, law and reliable shipping, a promise of ships each summer.

An island that cannot secure its transport links eventually trades away its independence to someone who can.

This is a warning for New Zealand.

New Zealand imports all its refined liquid fuel, most of its fertiliser and much of its grain, and holds minimum fuel stocks of a few weeks. We have no mapped picture of national food stocks and no minister who owns food security. We should be doing what Iceland is doing, and at least as seriously. New Zealand needs a serious plan for rationing onshore diesel stocks across a year of disruption, and a national food security assessment and plan.

Lessons for New Zealand

1. Judge policymakers on decisions under uncertainty, not on waiting for certainty. Science on AMOC, and most catastrophic risks, is doing what it can. Officials need DMDU skills and should use them.
2. Lead with no-regrets actions. Fuel stocks and local supply (eg minimum local biodiesel production capacity), food and seed reserves, fertiliser timing and community resilience pay off across AMOC, nuclear winter, volcanic winter, pandemics and trade shocks.
3. Monitor the currents we depend on. Fund continuous, long-term watching of geopolitics, supply chains, resources and social cohesion, with Plan B and triggers agreed in advance. With a zoomed out ‘big-picture’ view many of these are deteriorating with implications for what is going to be possible for governments to do.
4. Count our stocks. Follow Iceland: map liquid fuel, feed, fertiliser, medicine and food holdings, legislate minimum levels, and give households clear guidance.
5. Tell the story well, then repeat it. Iceland’s AMOC designation came from a well-framed narrative that let a minister lead. Risk communicators should aim for that, not unfocused volume.

Closing

I left Reykjavík struck by how far a small nation has come in a year on a risk it cannot control alone: watching it, framing it as a national security risk, and counting what it would need. New Zealand faces its own currents. The first step is to start observing them.

By Matt Boyd and Nick Wilson

TLDR / Summary

• This year’s SRA-Europe Nordic Chapter conference (Reykjavík, 1–2 October 2026) was small, well run, and focused on risk in a time of transition across the Nordic–Baltic region.
• Two key lessons transfer to global catastrophic risk (GCR) mitigation: unfamiliar risks are framed by whoever speaks first, and national risk assessment should start from the critical functions society needs, not from a list of hazards.
• The Grindavík volcanic eruption is a live experiment in managed relocation. Iceland bought out a town at 2.6% of GDP. Every high-risk community needs a relocation plan in the drawer, and New Zealand will face the same questions.
• Resilience needs to be built into designs, from islanded grids to protecting Iceland’s saga-era norms, beats resilience bolted on after the shock. Developing volunteer capacity and norms around volunteer service can supplement resilience measures.
• Small nations can act. Our post-conference roundtable with Icelandic colleagues was about what Iceland and New Zealand can do, alone and together and it revealed a repertoire of options.

Introduction

Two weeks after Cambridge, Nick Wilson and I were in Reykjavík for the Society of Risk Analysis (SRA)-Europe Nordic Chapter conference, held at Edda, the Institute for Icelandic Studies, on 1–2 October. It is a different kind of meeting from the Cambridge Conference on Catastrophic Risk: a few dozen people, mostly in one room, and a theme of risk “in a time of transition” rather than global catastrophe as such. The organisers pulled a coherent programme out of a very diverse membership, and the intimacy meant conversations continued over coffee and dinner rather than ending at the microphone. Many thanks to Þórhallur Ingi Halldórsson for organising an excellent event.

Because this was not a global catastrophic risk (GCR) conference, this post is written through two lenses: what transfers to global catastrophic risk, and what Iceland is learning that New Zealand will need to know. Everything else is passed over quickly.

Geothermal energy and resilience framing

Katherine McComas (Cornell) opened with a keynote talk on risk perception and “next generation” geothermal. Conventional geothermal supplies less than 1% of global energy and is largely confined to places like Iceland or New Zealand with shallow geothermal sources, but enhanced geothermal systems (EGS) drill into hot rock at 4–7 km, where heat is almost universal. The technology is arriving faster than the acceptance research. For example Fervo’s Utah project has drilled to nearly 6 km and began delivering power to the grid this year. The heat can also be used in “district heating” systems and could help overcome dependence on fossil fuels.

The catch is induced seismicity, McComas’s specialism. Her surveys find most people know little about EGS; associations are mostly positive, but respondents also associate risks, with earthquakes top of the list. Because the technology is unfamiliar, impressions are easily destabilised, and whoever frames it first sets the template. UK respondents reached for links to “fracking”, and negative connotations, even when the survey never mentioned it. Trust, consistently, predicts acceptance.

Two things were left hanging. First, probability. An audience member asked what the actual risk of a damaging earthquake is, and the question deserves a direct answer. The 2017 Pohang earthquake in South Korea (magnitude 5.4, 82 injured, around US$290 million in damage) was found by a government commission to have been triggered by fluid injection at an EGS site 600 metres from the epicentre. Second, McComas’s own work is supported by DEEP, a European-US programme whose stated purpose is “de-risking” EGS. That is public money and legitimate research, but worth knowing when weighing a talk that leans towards acceptance. She pinned positive messages on anecdotes of an Icelandic colleagues association of earthquakes with “energy”, and mass crowd (dancing) induced seismic measurements, but these allusions lacked the robustness of her negative connotation survey results.

The key relevance for global risk, is her framing one. Nuclear winter, catastrophic grid loss and engineered pandemics are, for most citizens, as unfamiliar as deep geothermal. Suggested solutions are also unfamiliar, and people are easily misinformed about risk and resilience. The first credible framing wins, and if risk researchers do not supply it, someone else will.

Whose acceptance, whose transition?

The first plenary took the same question to wind power. Frederic Bouder (University of Stavanger) compared the Netherlands, where turbines proceeded despite debate over noise and flicker, with Norway, where opposition became a national zeitgeist once Sami communities asked what benefit they get from national grid infrastructure in their backyard. Social consensus, he argued, can no longer be assumed. Gisela Böhm (University of Bergen) found citizens’ mental models of the energy transition poorly aligned with policy (carbon capture ranks third in Dutch policy and near the bottom for the public), and Aistė Balžekienė (Kaunas University of Technology) found governance and participation deficits in the green and digital transitions across every Baltic country studied.

The roundtable focused on participation, noting that informing is not empowering, and telling people a concern is “made up” gets nowhere; engaging with the belief does. Finland’s choice of a nuclear waste site that was acceptable to its host community rather than technically optimal came up, as did Vilnius’s (Lithuania) first citizens’ assembly. This is the deliberative strand of our own governance plan, and it was good to hear it in a region where high social trust has arguably given governments an easy ride in the past.

In-built resilience

From the afternoon sessions on Day 1, we pick out four talks with wider application. Kari Mäki (Tampere University) turned resilience logic upside down: instead of hardening the whole electrical grid from the top, identify critical assets, match them with storage or backup, and build “resilience nodes” that operate normally but can island themselves and black start independently. This is the modular, decentralised design that our review of GCR resilience found to matter, and it applies directly to New Zealand’s long, thin grid that is exposed to diverse natural hazards as well as large scale risks such as solar storms and cyber-attacks, which would likely have heterogeneous impact across the nation.

Maija Nikkanen (University of Helsinki) is stress-testing Finland’s protein system against a “Disease X” pandemic scenario: collaboration across stakeholders in preparedness works only with shared understanding and trust, and views of the risk landscape diverge sharply between sectors. A talk on real estate risk and non-linear processes offered a line for all risk models: future risk is not only about shocks, it is also amplified by using poor or incomplete models of the risk.

Sigurður Emil Pálsson (University of Iceland) gave one of the most enjoyable talks. The historical Icelandic commonwealth had law and courts but no king and no enforcement, so order rested on reputation, kin and reciprocity, and resilience had to be social before it could be institutional. Traditional norms of being watchful in unfamiliar situations, using measured words that do not inflame conflict, and treating hospitality as survival infrastructure are critical lessons for any age. An eighteenth-century Icelandic household manual carries the same advice. This is evolved resilience, cultural adaptations selected and transmitted over centuries, exactly the inheritance our metacrisis paper argues modern societies are discarding without realising what they are throwing away. His practical point was that resilience protects function: what can be quickly restored is a less attractive target for disruption.

Michael Stieber (Harz University) described expectation gaps in Antarctic overwintering teams: harm comes when expectations collide with reality. We kept wondering how far this scales. If expectation gaps are a risk, governments that try “not to scare people” about catastrophe may be manufacturing exactly this kind of harmful gap between expectation of the future and future reality.

Total defence, democracy, and strategic islands

The security plenary asked whether the Nordic–Baltic turn to “total defence” could actually undermine the democratic values it exists to protect. Sara Bondesson (Lund University) listed the mechanisms: secrecy, limits on protest, blurred civil–military lines, centralised governance. Her project will work backwards from the values a society wants to hold in thirty years. Estonia, the discussion noted, is building military and civil defence together, which aligns them but may not serve resilience to non-military hazards.

Anna Moraczewska (Maria Curie-Skłodowska University) proposed Greenland and Gotland as “risk observatories” where systemic risks become visible first: climate acceleration, critical minerals extraction, military mobility, infrastructure stress. This idea has merit; it is the horizon-scanning function of anticipatory governance, which we support in the domain of global risk mitigation. But when asked why islands specifically feature as her bellwethers – her answer was that they are small and easy to study and vulnerable to early degradations. But a couple of sentinel sites is not sufficient. We need a wider array of observatories, island and otherwise, feeding a dashboard of global stress monitoring, resources, trade, energy use and availability, misinformation, political forms, and social indicators that say what they imply for society over one, ten and fifty years. A global array of information tracking where and how human systems are being tipped into systemic risk. We are monitoring atmospheric climate, we need to monitor all climates of global risk.

Grindavík and the importance of scenario planning

Matthias Kokorsch (University Centre of the Westfjords) gave the keynote to which New Zealanders should pay attention. In November 2023 Grindavík in Iceland, population about 3,800, was evacuated as magma moved beneath it and lava entered the town that January. The state bought out more than 90% of homes at a cost Kokorsch put at €820 million: about €2,000 per Icelander, €215,000 per Grindavík resident, and 2.6% of GDP.

The critical role of volunteers in a crisis was a recurring theme through the conference. Volunteers, released by their employers, contributed thousands of hours of support during the Grindavík crisis, and form the backbone of Iceland’s highly trained search and rescue services. Norms around volunteer obligations, training and resourcing, and release-from-work, can greatly enhance national resilience capabilities.

Two years on, a few dozen residents have returned to Grindavík, the school is being restored inside the hazard zone, but there is still no clear end state to this situation.

Evacuation is an event, Kokorsch said, but relocation is a process, and the hard questions are political, not scientific. Can a village be closed? Who decides, at what threshold, with how much public money? Icelanders disagree about whether the response was right, and disagreement is not along party lines. A house is easy to price; a community is not, and whether people see €820 million as a bailout or as reciprocity for potential future aid elsewhere shapes whether they think it was worth it. Then there is precedent. Having paid once, will the state be expected to pay again, and why should nature-driven loss of habitability trigger collective compensation when other losses, such as economic closures, stay private?

His conclusion was right. The Grindavík eruption forced a decision in a way that slow-onset change like sea level rise and recurrent flooding due to climate change never will. Every high-risk community should have a relocation plan in the drawer, built through dialogue before the crisis, so that planning preserves choices rather than predetermining abandonment. A good plan gives people more than five minutes and more than one option for leaving. Adda Þóreyjardóttir Smáradóttir (ÖRUGG) added what recovery assessments usually miss: for children the dominant harm is psychosocial, slow and cumulative, and much of it is produced by the recovery process itself.

New Zealand has its own version, in flood-prone towns after Cyclone Gabrielle and coastal communities facing managed retreat, and we have resolved none of Kokorsch’s questions. Who decides, at what threshold, and whether relocation is planned with communities or imposed on them are questions for citizen deliberation, before the next event rather than after.

Beyond immediate impacts of volcanoes and potentially severe disruptions to local communities, volcanoes also pose a global risk, with large magnitude eruptions showing statistical association with famines around the world and dynastic collapses. One of us (Nick) detailed our findings on these points in a presentation, combined with a case study from the Tambora eruption of 1815.

From hazards and events to systems and functions

The talk with the most direct lesson for national risk assessment came from Böðvar Tómasson (University of Iceland). Most national risk assessments start with events and end in a matrix, which hides the strength of knowledge behind each score, the tails, and the couplings that let one scenario cascade into another, and aims protection at infrastructures rather than the functions they serve. His alternative starts from basic needs, core values and vital societal functions, and asks what continuity of each requires. This is close to what we argued New Zealand’s National Risk Register should do, and to the vulnerability-first approach of the AGILE project presented in Cambridge and earlier this year at SRA-Alicante.

Islands as a hedge, and two roundtables

My own talk (Matt), argued that remote island nations may escape the direct impact of many global catastrophes, as the Covid-19 mortality record and volcanic- and nuclear war-cooling evidence suggest, yet remain particularly vulnerable through trade dependence. Food self-sufficiency is necessary but not sufficient when fuel, manufacturing and sectoral interdependence can drive an “ungraceful degradation” even where calories persist. Our proposal has been to prepare islands as nodes of persisting complexity, which protects their own populations and insures humanity’s ability to rebuild.

Following this talk I (Matt) took part in two panel discussions on the lessons from Grindavík as well as what small island nations like Iceland and New Zealand can do to mitigate global catastrophic risk. Key themes included the similarities between Iceland and New Zealand in terms of geology, social cohesion, remoteness and dependence on trade and international agreements. There is potential for much fruitful collaboration between Icelandic and New Zealand researchers, policymakers and practitioners on national and global risk.

Lessons for GCR, and for New Zealand

1. Frame first. Unfamiliar risks are defined by whoever speaks first, trust decides acceptance, and good ideas can be pre-empted by messaging from those with vested interests. Catastrophic risk needs credible, early public framing. National risk assessments and vulnerability assessments, should not be secretive documents or processes.
2. Assess functions, not events. Start national risk assessment from the societal functions that must continue, keep severe tail risk in the mix and lay out cascading impacts and couplings explicitly, mapping vulnerabilities to resilience building options.
3. Put a relocation plan in the drawer. Grindavík shows that relocation is a political process with precedent effects. Plan it with communities before the hazard forces a five-minute evacuation decision.
4. Build resilience into the design. Islanded grid nodes for energy and communications, and inherited social norms show that resilience already present outperforms resilience bolted on afterwards.
5. Small nations can act effectively. Iceland and New Zealand are both remote, trade-dependent and hazard-exposed, but with relatively high social capital and the ability to act faster than large states. We left Reykjavík with the beginnings of a shared research agenda.

(In-depth reporting 15-20min read)

TLDR/Summary

• The Cambridge Conference on Catastrophic Risk (17–18 September 2026) took the governance challenges shaping tomorrow’s world as its theme, across AI, nuclear weapons, the information ecosystem, pandemics and climate.
• Congratulations to the Cambridge Centre for the Study of Existential Risk (CSER) on another excellent event.
• Nuclear risk got the attention it deserves. No arms control treaty is in force for the first time since 1972, and no targeting plan accounts for nuclear winter, yet states could cut risk unilaterally without suffering strategically.
• AI’s impact on information ecosystem risk now has case studies for what was predictable a decade ago. Although almost nothing structural has been done since, and we should ask why.
• Pandemics are societal problems with biological triggers. Leaders must be able to admit when evidence has changed, and change course.
• The climate panel offered the most concrete governance proposals. But largely missing were discussions of systems like food, energy, resource limits, and accountability for national and cross-border governance of catastrophic risk.
• Human extinction is unlikely, but global catastrophe could already be underway. We should plan for it, and I argue we need an evolutionary lens to explain why known solutions are so rarely adopted and refocus our efforts where they can succeed.

Introduction

Last week I attended the Cambridge Conference on Catastrophic Risk (CCCR 2026), hosted by the Centre for the Study of Existential Risk (CSER) at Magdalene College. Congratulations and thanks to Jess Bland, Sonja Amadae and the whole CSER team for another excellent event. Global risk research is fragmented across disciplines and institutions, and CSER’s steady work in convening it is a public good the rest of us rely on. Few other meetings put nuclear strategists, pandemic epidemiologists and AI researchers in the same room.

I wrote up the 2024 conference largely as a survey of the discussions. This time I’m a bit more critical of omissions and lost opportunities. The theme was governance, so I apply a simple test: what problems were identified, what solutions were proposed, and what was missing. Noting that I am a co-author on four papers behind two of the posters presented at the conference (including the excellent poster by Florian Ulrich Jehn (see below) that was runner up for ‘Best Poster’). In the discussion that follows I’ll also be drawing on arguments from a new paper I’ve written with Nick Wilson, which provides a formal definition of the “metacrisis”, and explains why known solutions to global risk so rarely succeed.

Pre-conference Keynote: Max Tegmark on “A Better Path for AI”

On the evening before the conference proper, Max Tegmark, MIT professor and president of the Future of Life Institute (the group behind the 2023 AI “pause” letter and films such as Artificial Escalation), gave a keynote by Zoom. He was unexpectedly optimistic.

AI risk has gone mainstream, he argued. Literally the previous day, Bernie Sanders and Steve Bannon had shared a stage in Washington to call for curbs on AI, and the Pro-Human AI Declaration is gathering signatures. Tegmark’s case was for trustworthy, verifiable “tool AI” in place of AGI and he introduced a theory of verification proof. Systems can be highly autonomous, highly general or highly capable, and combining any two of these gives us useful tools. What we should not build is the combination of all three.

I was less sanguine. Calls for regulation are rising, but US President Donald Trump insists that the only guardrail needed is a “smart president”. The unanswered audience questions in the Zoom chat clustered on enforcement: not how do you build aligned AI, but how do you stop the other kind of AI being built when it is more profitable?

Former US President Barack Obama gave a coincident nuanced take on the day of the CSER conference. This is clearly an issue for our times, though I think that a topic often neglected in discussion of AI risk or AI solutions is the issue of physical resources. Minerals prices are rising, finds are becoming smaller and harder to extract, and geopolitical turmoil creates conditions for piracy and destruction of production capacity. I suspect all this will constrain what is possible, and its price, and catalyse a lot more conflict, a point I return to below.

Nuclear war and game theory: Sonja Amadae on geopolitics in 2026

Thursday opened with Professor Sonja Amadae, Director of CSER. Not enough people are worried about nuclear war, she observed, despite a hot war in Europe and other overt conflicts. Arms control has lapsed, nuclear spending favours preparing for war over reducing its risk by roughly 40 to 1, and doctrines that reject the nuclear taboo are resurgent.

A central tension turns on how states frame the security dilemma. In a prisoner’s dilemma defection dominates. In a stag hunt mutual cooperation is best for everyone, and the only problem is trust. The audience was encouraged to participate in such a game. In pairs, each of us privately ranked the four outcomes of a two-player nuclear game, talked to try to infer our counterpart’s ranking, and made a single move.

Two things became clear. If you are confident of your opponent’s worldview, coordination is easy; the danger lies in misreading it. But a great deal also turns on the probability of catastrophe attached to mutual arms race (defection, defection) outcomes. How this is rated determines the expected value of defection versus a subjugation outcome if a cooperative move is denied (cooperate, defect).

Amadae’s closing question is the right one: how do we turn geopolitical prisoner’s dilemmas into stag hunts? You do not persuade players to be nicer; you change the payoffs so that defection stops winning, as verified arms control did in the late 1980s. That idea, that governance must reshape what the relevant environment rewards, runs through the rest of this post and our new preprint providing an evolutionary analysis of the “metacrisis”.

Nuclear risk has not gone away (Panel discussion)

Many catastrophic risk programmes now give most prominence to AI risk. CCCR 2026 gave nuclear war a plenary, a panel and a cluster of lightning talks. This is justified given the risk seems to have been rising, and it is one of the most plausible routes by which AI could actually produce a global catastrophe. Demetrius Floudas noted in his lightning talk that 2026 is the first year since 1972 without a single nuclear arms control treaty in force. Furthermore, the lesson of the Iran war is that a state can be punished for being close to a bomb, which incentivises secretly making one, and announcing later, the only way to guarantee safety.

The panel made three key points:

International governance is collapsing. Mette Eilstrup-Sangiovanni (University of Cambridge) noted that institutions turn one-shot games into repeated ones, yet the “non-proliferation regime complex” is fragmenting with renewed proliferation, institutional paralysis and a funding crisis at the International Atomic Energy Agency. No single new treaty will fix this and the challenge is to manage a complex system of nuclear states, with far more investment in verification.

Nuclear winter is deeply uncertain, and war plans ignore it. Madeline Berzak (University of Chicago Existential Risk Laboratory) reported her new work. Nuclear winter depends on firestorms that loft soot into the stratosphere, which depend on what is targeted and what burns. Yet canonical models still lean on Hiroshima: a small bomb over a low-rise wooden city. Modern cities sprawl, rise higher and are full of plastics that yield two to three times the soot of wood. She observed that no targeting plan has ever accounted for the risk of nuclear winter.

Unilateral choices can cut risk without sacrificing security. James Acton (Carnegie Endowment) described a risk problem of entanglement, for example the same satellites and command-and-control systems serve conventional and nuclear operations, so an attack on them in a conventional war could look like the opening of a nuclear one. States need not wait for cooperation. Separating nuclear from conventional command and control, restraint in targeting, and care about where AI is used all reduce total risk, and each can be done independently, without suffering strategic setback.

An omission in the nuclear war panel discussion surprised me: nobody mentioned the UN’s Independent Scientific Panel on the Effects of Nuclear War, the first UN study of its kind since 1988, which reports in 2027. Nick Wilson and I will present our work on nuclear war vulnerability and resilience options to the panel’s regional meeting in Bangkok on 21 October.

AI and the information ecosystem (Panel discussion)

The AI panel, chaired by Alexandru Marcoci (CSER), skipped speculation on superintelligence and asked how AI is impacting the information environment now. Sam Stockwell (Alan Turing Institute) brought evidence through his team’s incident repository which has logged some 60 crisis events in which AI degraded the information environment, mostly through fabricated “evidence” from the scene of real world events.

Moira Nicolson (UK Cabinet Office) showed how this reaches government. Social media is a distracting megaphone for a small minority, so MPs systematically underestimate public support for issues like climate policy.

Giulio Corsi (Leverhulme Centre for the Future of Intelligence) explained why it will get worse. At present most AI “slop” never wins attention, but agent swarms can: they coordinate, persist, game the early-engagement signals that feeds use for ranking, and hold conversations, which move people where static content does not.

The panel indicated that remedies will need to focus on source labelling (through provenance standards such as C2PA), structures that make claims meet pushback rather than echo, and undercutting the monetisation of inflammatory content.

I agree with all of this. Though I have agreed with it for nine years. In December 2017 I gave a conference talk arguing that AI-driven content targeting threatened democracy, freedom, and free will. In 2018 I blogged at length that cultural evolutionary processes explain how information spreads according to three things individual biases assess (content, source and frequency) and that “all three of these key features can easily be manipulated, at scale, and with personalization”. I proposed “healthy content” labelling akin to food labels, cryptographic verification of media (which is what C2PA now attempts), and outlawing the impersonation of humans. The panel’s three talks map onto that triad.

Even back then I was not breaking ground, as my research had drawn on the evidence and recommendations of others.

So the question I most wanted answered was not addressed. Given these problems were staring us in the face a decade ago, why has almost nothing structural happened, and why expect anything different now, moving forward? The problem has been diagnosed repeatedly. Where were the panel’s concrete policy proposals and pathways to implementation, with a plan to overcome the barriers and pushback, including the defection of agents in the game playing an outlaw strategy for personal gain at cost to cooperative human systems?

Digital media platforms are selected for engagement, and engagement rewards emotive, tribal and familiar content over accurate content. The actors who profit are the ones who would have to change, the harm is diffuse and deferred, and we reach for fixes that change what people know (media literacy, fact-checking) rather than what the system rewards, this is a ‘selection environment’ problem, and a problem that requires an evolutionary governance lens.

My 2018 argument in one sentence was that cheap, personalized, AI-enabled deception would break the connection between what society believes and what is actually true; and because accurate collective beliefs are necessary for solving every other major problem, securing the information environment should be treated as a high-priority form of global catastrophic-risk mitigation. At the time I ranked action to mitigate degradation of the information climate as more of a priority than action on the actual climate, since success in the latter depends on success in the former.

But there is a deep loss in the dynamics underway. Faithful transmission of adaptive knowledge, with error correction, is a cumulative cultural achievement. Ritual, religion, apprenticeship, editorial gatekeeping and peer review all functioned, whatever else they did, as fidelity checks on the information environment. We have swept many aside in two decades without asking why they evolved or understanding the importance of their function. Move fast and break things is an information free-for-all and is not conducive to cumulative adaptation, the process which is humanity’s superpower.

I call the result cognitive niche destruction: erosion of the shared map of reality on which every other risk response depends. The panel’s example of catastrophe was a faked general announcing escalation. The slow version worries me more: leaders misreading the public, year after year, while climate action, or an understanding of the risk, and the systemic stresses humanity faces, stalls.

Pandemics: societal problems with biological triggers (Panel discussion)

Friday’s panel on the future of pandemic preparedness, chaired by Charlotte Hammer (CSER), shirked a discussion of pathogens and medical countermeasures, and that was exactly the point.

Philip AbdelMalik (WHO) put it best: pandemics are fundamentally societal problems with biological triggers. The hard part is making response decisions under uncertainty. Mika Salminen (Finnish Institute for Health and Welfare) spoke from the decision-maker’s chair noting that infectious disease (and other) experts are expert in one domain, but a decision-maker must weigh all the domains and decide. Katharina Lauer (University of Duisburg-Essen) observed that most people assume AI’s pandemic threat is creation of a worse pathogen. But preparedness is mostly about gathering information, coordinating decisions and effecting a response, and those are what AI can disrupt (see the information ecosystem discussion above).

In discussion, the panel noted that when prevention succeeds nobody notices, and institutions can be degraded precisely because they worked silently, so success needs to be advertised. On AI-enabled bioweapons the panel was cautious: much of the alarm comes from organisations with a commercial interest in AI’s power, and designing an agent is a long way from building and delivering one. Pandemics that kill billions are unlikely, they argued, because extreme-mortality pathogens tend to be self-limiting.

The most important remark concerned political leaders, who must be able to admit that the evidence has changed, or that they were wrong, and change course. Much harm has been and is done by dogmatic or ideological adherence to a strategy. This is a structural problem, not a personal one. Today’s political discourse treats updating as weakness (the “U-turn”, the “flip-flop”), so leaders are selected for doubling down and flawed information keeps propagating. Another ‘selection environment’ problem. A practical response could be to decide in advance, and publicly, which signals would trigger a change of strategy, so that changing course means executing the plan rather than confessing failure.

Climate: featuring governance proposals (Panel discussion)

The theme of the conference was “governance challenges shaping tomorrow’s world” and to be honest by Friday afternoon I had begun to wonder where the governance proposals were. This panel supplied some, and it was the clearest session of the conference.

Elena Kavanagh (University College Cork) used the Atlantic Meridional Overturning Circulation (AMOC) risk to show how we fail to govern tipping points. Depending on which study is reported, collapse is anywhere from 0 to 100% likely: an “uncertainty trap” in which unresolved signals justify inaction. Institutions are incremental by design, and none operates at the scale of the tipping point. Crises and wars have historically driven institutional innovation. She asked if scientific foresight could do that work before the shock? Her answer repurposes existing institutions across three phases (see Figure): prevention, which works only before a threshold is crossed; impact governance, which must begin before crossing tipping points; and stabilisation afterwards.

Kennedy Mbeva (CSER) argued that climate risk is a governance failure, and that “common but differentiated responsibility” is quietly becoming common but shifted responsibility. The Paris Climate Agreement won participation by weakening differentiation, leaving new burdens without the means of implementation, while decarbonisation itself exports harm through mineral extraction. He invoked Thomas Hale’s “catalytic cooperation”: cooperation built by first-movers who change the costs and benefits for everyone else.

Dante McGrath (University of Cambridge) treated solar radiation modification (SRM), as “a stress test for global governance in the age of overshoot”. Exceeding 1.5°C is now inevitable, and carbon removal is far short of what is needed. SRM is “a bad idea whose time has come”. But it fights the fever without curing the infection, redistributes risk regionally creating winners and losers, and threatens termination shock. He set out four futures: no SRM; multilateral SRM (legitimate but slow); minilateral SRM (faster but less legitimate); and rogue deployment. Governance must be built step by step, from research to deployment. His summary of the carbon problem: “Reduce, Remove, (Reflect?)”.

In discussion, the distinction between treaty negotiation and treaty implementation stood out: we agree, then fail to deliver. An unimplemented treaty changes nothing that the world actually rewards or punishes. And African leaders lack access to African research on SRM, because nobody funds it.

Posters and lightning talks

With thirty-odd posters I can only be selective, and I have favoured those offering something a government could use.

Governing vulnerability. UCL’s branch of the Horizon Europe AGILE programme, another branch of which I encountered in Alicante this year, distilled 41 interviews with senior resilience experts into recurrent failure modes. Mhari Gordon told us of short political cycles, accountability gaps, “disaster amnesia”, just-in-time brittleness, and centralised command that slows local action. The team’s conclusion was that high-impact, low-probability crises result from a failure to understand and address systemic vulnerabilities, not simply failures of prediction. Their levers include no-regret capabilities, safe worst-case assessments, unfamiliar exercises and cross-border preparedness. This was a refreshing systems-focused rather than hazard-focused vulnerability take.

Middle powers and citizens. Mariana Beselga asked how middle powers can turn presence into leverage over frontier AI. Across six countries and four governance processes, participation was near universal but established coordination rare. The only process with a documented effect on developer behaviour was the Hiroshima AI Process, through its OECD reporting framework. Leverage comes from converting commitments into instruments located where behaviour can change, and measuring the result. Giuseppe Dal Prá presented the first Odyssean Process on AI, which combines expert elicitation, decision-making under deep uncertainty and citizen deliberation.

Resilience and food. Florian Jehn presented No Place to Hide?, on which I am a co-author. Four modifiable factors impact regional resilience to global catastrophe: democracy, decentralisation, preparedness and food system resilience. Trends are heading the wrong way on all four. Alongside it, Madeline Berzak’s XLab team revisited nuclear civil defence in “With More Than Shovels”. Some argue that civil defence might perversely raise the probability of war if it looks like first-strike preparation, but measures that are invisible to adversary planning and protect recovery capacity escape that paradox. Agricultural readiness looks positive with the authors’ calculations finding that each $1 to $20,000 in preparation investment could save a life as well as benefit pandemic, volcanic winter and climate disaster scenarios.

Also notable. Catherine Diyakonov’s poster on coercive diplomacy in the Russo-Ukrainian war, voted best poster by attendees, showed how Russia used gas and Black Sea grain as leverage short of force. And Nick Wilson presented our own Covid-19 work, drawing on three papers we’ve written analysing excess mortality in the pandemic (click for video of Nick’s talk).

Reflections: existential catastrophe is unlikely, “mere” catastrophe is not

The host of CCCR 2026 was the Centre for the Study of Existential Risk, yet across three days almost nobody argued that humanity faces extinction. The field has quietly migrated from existential to catastrophic risk, and I think rightly. Even the most pessimistic nuclear winter models leave billions dead, not everyone. The most lethal pandemics would likely be self-limiting (following potentially catastrophic destruction). Humans have survived radical climate shifts before, because we are dispersed across diverse niches, though many individual human groups and civilisations have certainly perished. And nobody articulated a pathway from AI to extinction that does not run through the pathway of one of the other catastrophic risks.

Catastrophe is another matter. Work such as the Cascade Institute’s Polycrisis CORE model describes a world system whose plausible trajectories converge on basins of decline. The conference’s own evidence points the same way. No nuclear arms control treaty is in force. Emissions have risen since the Paris Agreement. There is still no binding regulation of frontier AI. The information ecosystem has degraded for a decade in ways that were foreseen.

Sober risk analysis has to accept that if we could not act on these across ten or thirty years, the reasonable expectation is that we will not act in time now. A serious global catastrophe this century should be the planning assumption, not the tail.

Did the conference give us governance? To some degree. Our own research across the last decade suggests that governing global catastrophic risk needs seven things:

1. An accountable national coordinating entity;
2. A national risk assessment that maps vulnerabilities to costed options;
3. Cross-sector and novel scenario exercises;
4. Local and citizen deliberation;
5. Grounding in integrated global models and evolutionary dynamics;
6. Structures supporting regional international cooperation;
7. Realistic assumptions about resources.

The conference covered vulnerability, exercises, deliberation and coalitions well. Almost nobody addressed who, in any given country, is actually accountable for these problems. But I want to draw attention to two further gaps.

Resources. Every future discussed, from abundant tool AI to decarbonisation or SRM at scale, assumes material and energy availability that may not be realistic. A recent paper in Resources Policy finds that the AI buildout is an infrastructure materials story rather than a semiconductor materials story. Data centre copper demand, mostly for grid reinforcement, rises from about 2% of global refined output in 2025 to 6.5% by 2030, roughly the annual consumption of the United States. Copper hit a record near US$14,800 a tonne in May. AI deploys in months, while mine openings take decades and are getting deeper and smaller.

As Nate Hagens argued in his “superorganism” paper, the global economy behaves like an energy-hungry organism that must keep growing, and depletion steadily raises the energy cost of extracting what remains.

This bears on the politics of AI, renewables and sustainable transitions, and various high compute intensity solutions to information pollution too. Global growth is already constrained by frictions on liquid fuel, energy and material supply, growing worse through 2026 (Ukraine, Russia, Hormuz, Red Sea) in ways that have yet to fully hit economies.

AI is seen as the way to sustain growth, which helps explain Washington’s resistance to regulation. My view is that the same frictions will constrain AI. The more likely catastrophe is not runaway superintelligence but a flattening and then decline in growth, with conflict over what is left. Resource limits are a handbrake on the worst AI futures, but also on the solutions. None of these global system stresses and potential associated cascading catastrophes featured much at the conference.

Food system. The conference was organised by hazard, not by the systems those hazards would break, which is how groups such as the Cascade Institute and the Accelerator for Systemic Risk Assessment (ASRA) approach global risk. So the system through which most catastrophes would actually kill people, namely food, barely appeared (there were a few exceptions, but it deserves a panel). The omission was striking given the venue: two weeks earlier the UK’s Environment Secretary had advised households to keep food stores at home, ahead of what the Met Office expects to be the largest El Niño since the 19th century (cynics might argue this call was a way to prepare against Russian hacking of logistics systems without scaring the public).

My country, New Zealand’s, version of the problem is complacency: we export food, but it does not follow that we are food secure, because production depends on imported diesel, fertiliser, seeds and parts and produces a monoyield of milk products. Our research shows New Zealand could feed itself through even a severe global catastrophe, but only with prior planning for fuel, crop mix and distribution.

Why has so little worked to mitigate the risk of global catastrophe? The conference said much about what is going wrong, and less about why known solutions are not adopted. Our new paper defining a “meta-crisis” argues that institutions, technologies and norms evolve under selection pressures that reward what succeeds locally and immediately.

We see platforms selected for engagement, states for relative advantage, firms for extraction, politicians for never changing course. Meanwhile the conditions that let risk-reducing adaptations accumulate (variation, modularity, a stable selective environment, faithful transmission of what works, and suppression of those who gain by defecting) are degrading together.

Appeals to the good sense of individual actors, through rational plans, and moral arguments cannot fix this. Changing what the environment rewards, and protecting the mechanisms that contain defection, can. The Montreal Protocol and verified arms control did exactly that. In our paper we diagnose the ultimate not merely proximate problem, and recommend the actual governance approaches that are needed to address global catastrophe risk.

Lessons for moving forward

1. Plan for the catastrophe as well as prevention. Kavanagh’s framework puts impact governance before the threshold, and XLab’s first tier shows that cheap, no-regret resilience exists. Preparing for failure is part of governing the risk, not giving up on prevention, and anticipatory governance is the key.
2. Act where cooperation can evolve. Global agreement is the hardest case. Unilateral risk reduction (Acton), coalitions of first-movers (Hale, via Mbeva), middle-power coordination (Beselga) and minilateral arrangements (McGrath) are more tractable, provided legitimacy and protection against rogue actors are built in. In our meta-crisis paper we explain why this is predictably so.
3. Change what systems reward. The information ecosystem is the test case: end the monetisation of inflammatory content and the gaming of ranking signals, rather than asking billions of users to be more discerning. Humanity’s ability to coordinate based on effective and adaptive information to overcome global risk depends on it.
4. Fix accountability. Every country needs a named entity responsible for catastrophic risk across policy silos and electoral cycles. For New Zealand we have drafted a Parliamentary Commissioner for Catastrophic Risk Bill.
5. Organise around systems as well as hazards. Food, energy, materials and information are where catastrophes converge. I would love to see the next CCCR give them the main stage.

I left Cambridge with little hope that prevention alone will succeed in the face of its abject failure to date.

I have more conviction that the field’s next task is to explain why this is so, and with that wisdom to build resilience to the coming outcomes where building is still possible.

Thanks again to CSER for starting that conversation.

Introducing our new preprint on evolutionary governance for systemic and global catastrophic risk

By Matt Boyd

TLDR/Summary

• Nothing about human society and its generation of global risk makes sense except in the light of evolution. I’ve been building that claim through my recent posts and conference talks.
• A new preprint I’ve developed with Nick Wilson, and now under peer-review, brings all this together in a technical paper: The Metacrisis as a Degradation of Evolvability.
• Five research communities working on global risk are converging, often unawares, on the extended evolutionary framework. We name it, systematise it from first principles, and draw out what it implies for governance of global risk.
• Risk-generating architectures recur wherever certain evolutionary dynamics obtain; the metacrisis is the simultaneous degradation of five conditions of evolvability (variation, modularity, selective-environment stability, transmission fidelity, and outlaw/defector suppression)
• Global risk governance must therefore act on those conditions, above all on suppressing defectors.
• The evolutionary lens is not the answer to everything, but our paper deliberately pushes it as far as we can, because evolution keeps seeping through the gaps in other explanations and joining them together.
• The upshot is a change in what governance is for: less for specifying solutions, and more about engineering, nudging and continually readjusting the conditions under which good solutions can arise, spread and hold.

Two intellectual strands twisted together

This post has been a long time coming. In 2007 I wrote a Masters thesis in philosophy, on cultural evolution and the information environment we have created. That was back when “meme” still had a technical definition, Dawkins’s unit of cultural inheritance, and did not yet mean a picture of a cat that goes viral. My thesis argued meme theory was useful in understanding how culture evolves.

My 2011 PhD took the next step: technologies play a privileged causal role in psychological development, and as we invent new ones, we change the developmental environment for present and subsequent generations, thereby driving evolution of the mind, cumulative niche construction applied to human cognition.

Both theses were under Kim Sterelny, whose work on cumulative culture runs through our new preprint too.

For over a decade since, I have worked on global catastrophic risk, studying pandemics, nuclear winter, island refuges, and national risk assessment. It was seemingly inevitable that the two strands; evolution and global risk would meet.

Dobzhansky said nothing in biology makes sense except in the light of evolution.

I’ll appropriate that: nothing about human society, and about why it keeps generating risks we know how to avoid, makes sense except in the light of evolution.

Regular readers of this blog will have seen the argument assembling. In February I asked whether there is really a “meta”-crisis. In May I presented an early version of this argument at a European risk-analysis conference. In July I wrote a definitional field guide following a chain of four questions:

• What could go badly wrong? That’s global catastrophic risk.
• How do failures spread? That’s systemic risk, where a tightly-coupled world is a generator of hazards, not a bystander.
• How do crises compound? That’s polycrisis, the causal entanglement of crises such that the whole is worse than the sum of its parts.
• And why do we keep generating a fragile world?

I now turn to this fourth question.

Rising global risk despite known fixes

The major global hazards were identified decades ago, and so were fixes: disarmament, emissions reduction, health-security investment, limits to extraction. The Club of Rome laid much of it out in 1972. Yet emissions have risen since the Paris Agreement, arms-control treaties have lapsed, and by most measures the world is more exposed. Humanity is not ignorant of what threatens it. What’s missing is second-order understanding: why we consistently fail, and at what level to intervene.

Five fields, one grey elephant

Our new preprint begins by outlining the communities working on these problems of global risk. National risk assessment and disaster risk reduction handle hazards, mostly natural, mostly after the fact. Global catastrophic risk studies confront rare but devastating events. Systemic-risk analysis maps feedback-rich networks and names risk-creating actors. Polycrisis scholarship traces entanglement and attractor states, including the Cascade Institute’s “Illiberal Decline” and “Mad Max” basins. Behavioural and evolutionary science speaks of traps, ratchets and a “human behavioural crisis”.

Each field is like a blind monk with a hand on an elephant, describing in rich detail what they feel, but unable to articulate the whole. Each field reaches for the same vocabulary of adaptation, selection, co-evolution, traps, ratchets. Some is generic systems talk. A surprising amount is specifically evolutionary, used by fields with no evolutionary self-conception. That is latent Darwinism and it signals intellectual convergence.

Evolution is not a metaphor here

In risk science, grey elephants are a metaphor for big obvious risks, charging right towards us, that everyone can see plainly, yet no one acts.

In our pre-print evolution is not a metaphor. We deploy a precise version of evolution, not a simplified high school one, and certainly not the discredited “Social Darwinism” that confused selective success with moral merit.

Natural selection is an abstract process that arises wherever there is heritable variation, differences in success that depend on it, and differential persistence as a result. Dennett called this substrate neutrality: the logic doesn’t care what the variants are made of. Genes are one substrate. Firms, institutions, legal codes, professional norms, technologies and ideas are others, to varying degrees. Bicycle design lineages show the adaptive-radiation signature of biological clades, and no designer intended it.

This instantiation comes in degrees, and where any human system sits on Godfrey-Smith’s continuum of “Darwinian populations” is an empirical question. But human systems are also designed, meaning that intention and the invisible hand of evolution operate simultaneously. Because no actor can intend outcomes everywhere and all the time, much of what results in human systems evolved rather than being designed.

This is where the social sciences fit, and our framing does not replace economics, political science or the study of human decision-making. Those describe real mechanisms, and human choice is a genuine causal element in the system. What evolution adds is the ultimate level of explanation, in Mayr’s sense, not how each mechanism or complex system feature works but why it exists and persists, shaping power and possibility.

Deliberate decisions exist in a selective environment that determines which decisions get retained and copied. The framework is “extended” because inheritance runs through several channels at different speeds (genes, memes, laws, education syllabi, technologies), because we actively construct the niches that then impose selection back on human activities, and because not all cultural change is adaptive. For example, prestige, conformity and familiarity spread variants regardless of merit.

Why we keep building a fragile world

Our new paper groups global risk by examples of its driving dynamics rather than by hazard.

• Multilevel selection. What’s adaptive for one actor (individual, firm, nation) is often destructive for the commons that sustains it. Muir bred hens individually for egg yield and got flocks that feather-pecked and cannibalised themselves. Selecting whole cages for productivity produced cooperative, higher-yield flocks. Scale that up and you have excessive carbon emissions, overfishing, aquifer depletion, and antimicrobial resistance. Firms are rewarded for maximising extraction even when run by people who grasped the consequences decades ago.
• Arms races. Red Queen dynamics where everyone must invest more merely to hold position. In frontier AI, safety spending is individually costly and competitively detrimental if an organisation’s rivals don’t match it, so the field races to deploy. Non-reckless actors are simply outcompeted. Global risks from nuclear weapon stockpiling and gain-of-function research share this structure.
• Lock-in and fitness traps. Configurations stable under current pressures but globally suboptimal, can be hard to leave because improving requires first getting worse. Fossil fuel infrastructure is the obvious case, where build out has constrained changes that are now possible.
• Transmission bias. Cultural evolution selects for what is readily learned and passed on, not for what is true, and engagement-optimised platforms erode the shared map of reality on which collective action depends and the information scaffolding which centuries of cultural evolution had built.

These are not accidents or failures of will. They are predictable outputs of evolutionary selection at several scales at once. Where the selective environment has been deliberately re-engineered, on the other hand, the Montreal Protocol, verified arms control from the late 1980s, Ostrom’s management of the commons, the dynamics have sometimes been beaten. In each case a suppression or coordination mechanism was built and protected, converting locally advantageous defection into a losing move.

Evolvability, and Metacrisis defined

Evolvability is a system’s capacity to generate, transmit and retain complex adaptations. Biological theory and computer science treat it as something that must be built into a system, not assumed. It rests on five conditions:

• Variation: a supply of diverse candidate solutions
• Modularity: semi-independent parts that can change without dragging the whole down
• Stability of the selective environment: pressure consistent enough for adaptation to accumulate
• Fidelity of transmission: reliable inheritance of what works, with error-correction
• Outlaw/defector suppression: containment of entities that gain locally by breaking cooperative rules or exploiting commons

A definition of the metacrisis follows: the metacrisis is the simultaneous degradation of these conditions in the systems meant to govern or mitigate global risk.

This is not any particular crisis, or even a polycrisis, but a crisis of the capacity to adapt to crisis at all. That is why governance fails consistently across unrelated domains.

Each evolvability condition is observable. Food-system homogenisation onto a few crops and chokepoints is a variation failure. The Tōhoku earthquake, Suez canal blockage and Hormuz fuel crisis rippling through stripped-out supply chains are modularity failures. Policy reversal each electoral cycle is a stability failure that selects against long-horizon investment. The pandemic plans and stockpiles accumulated after SARS in 2003 and the 2009 pandemic, then largely lost before 2020, are an inheritance fidelity failure.

Importantly, evolvability is value-neutral, given that a cancer can be highly evolvable inside the body. So, capacity for evolvability must be paired with a fitness function worth sustaining, and what fast cultural evolution actually selects for is rarely wellbeing or planetary sustainability, but profit, votes, engagement or military advantage. Also, evolvability sits at an interior optimum. If there is too much variation then inheritance is destroyed, if too much stability then maladaptation can become entrenched. As an example, German scientific forestry optimised what it could count and discovered, rotations later, the ecological relationships it had discarded.

Outlaws, and the lesson from cancer

Defector/outlaw suppression matters most, not only is suppressing outlaws constitutive of evolvability (multiplication of rogue genes that degrade organism fitness, across generations, is a biological example of ‘outlaws’), but outlaw activity can undermine the other necessary criteria. At every major evolutionary transition, from cells to organisms to societies, the higher-level unit exists only because it continually constrains, regulates, or suppresses, lower-level strategies that benefit themselves at the expense of the whole.

That is a decisive argument against purely non-interventionist, market-only solutions to large scale risk.

An outlaw need not be conscious or act illegally, since firms externalising costs, states leaving treaties, platforms amplifying misinformation, or farmers applying excess nitrogen fertilizer, may be responding to legitimate incentives.

Increasingly, though, outlaws attack the suppressors. At the height of the US opioid epidemic, wholesalers did not evade the DEA, instead they disabled it through lobbying, a revolving-door attorney and a 2016 statute. All this is the governance analogue of a tumour co-opting the immune cells sent to kill it.

Cancer’s evolutionary biology supplies four transferable principles against this kind of defection, namely:

• Redundant, multi-layer suppression (single point solutions are insufficient)
• Calibrating rather than maximising enforcement, since maximal enforcement selects for resistant evasion, just as chemotherapy can select for resistant tumours
• Protecting the suppressors from capture
• Suppressing early, before risk generating actors become too big or organised to constrain

Global overshoot, and walking some of it back

Unconstrained market dynamics constitute a selection environment with a fitness function of short-run return measured in profits. Under it humanity has transgressed several planetary boundaries, oversupplied what pays (extraction, coupling, engagement, throughput) and undersupplied what doesn’t (slack, buffers, prevention, institutional memory). No institution was mandated to preserve resilience, so resilience was competed away, then we struck Covid-19, the Russian Invasion of Ukraine, the US assaults on Iran, and potentially worse global shocks to come.

Reducing catastrophic risk means walking some of this back. We need to incentivise retreat from planetary boundaries, rebuild buffers, accepting that some efficiency was borrowed against the future. The evolutionary framing doesn’t make that easy. It explains why it hasn’t happened on its own and won’t at the global level.

Our pre-print lays all this out, describing the current dynamics, explaining why they persist, and offering a pathway forward.

New Zealand in the present

My book club is currently reading the Helen Clark Foundation’s 2026 release “Facing Up to Our Future”, which consists of twenty short policy chapters on the long-run problems no single government seems able to solve. What strikes me is how much of our pre-print’s machinery is in there, independently arrived at across multiple fields of policy and economy, but never framed in evolutionary terms such that the big picture dynamics are clear.

The book’s framing is often a diagnosis of the metacrisis, speaking of political regimes structurally incapable of addressing decades-long problems, which is a claim about degraded adaptive capacity rather than about any one policy. The opening economic chapters then supply the conditions one by one.

• Variation. Competition policy and industrial policy are, in these terms, about the supply of candidate solutions. New Zealand’s supermarket duopoly is a monoculture, with two dominant variants there is little for selection to work on, evolvability is throttled at the source, lock-in takes hold to the detriment of the higher-level system, namely New Zealand society.
• Modularity. Niche construction (in our paper’s sense) has created entrenchment in the energy market, incumbents whose infrastructure and integration and coalitions favour their own continuation and cannot be teased apart one by one to change, limiting what the system can become.
• Stability of the selective environment. The book’s recurring complaint about inconsistent, reversible policy is a stability failure. Investment flows to long-horizon adaptative solutions only where this action is predictably rewarded over time.
• Defector suppression. New Zealand’s regulatory environment is full of tools that lack bite, such as a passive Commerce Commission, captured regulation and high barriers to market entry. This is suppression architecture that has stopped suppressing what disrupts the cooperative whole, permits externalities, allows fragilities to expand, and fails to preserve modularity, or variation.

Industrial policy to mitigate the contribution to global risk made across various industries, as well as investment in appropriate infrastructure, skills and training can shape evolvability. But only if paired with a meaningful fitness function, ie incentives and punishments tied to risk reduction and wellbeing, otherwise we simply accelerate adaptation in whatever direction the market was already heading.

Unfortunately, market solutions alone will never overcome the outlaw/defector problem, because the market is where defection is rewarded. Good industrial policy grounded in evolutionary theory and tied to global risk reduction is the alternative, and is necessary otherwise, in the end, we all lose.

The point is not that policy accounts need our pre-print. It is that the relevant dynamics are already and everywhere operating. There is some understanding of them across standard systems and economic thinking. But the evolutionary dynamics need to be understood better, deeper and more broadly, because they are foundational rather than incidental to reducing global systemic and catastrophic risk.

Credit to others, and our contribution

None of this work comes out of thin air. Arnscheidt and colleagues have named risk-creating actors as a primary driver of global catastrophic risk. Liu and Renn call for biomimetic designs “tested by natural selection”. Søgaard Jørgensen’s group catalogues Anthropocene traps and now speaks of a “co-evolutionary struggle” between resilience capacity and the polycrisis. Hämäläinen’s cybernetic “complexity gap” is a functional analogue of evolvability. Waring and colleagues treat the Anthropocene as an evolutionary ratchet. There are many other examples.

These are real advances, mostly arrived at inductively and in fragments. What no one has done is articulate the complete framing from first principles, ie evolution as the substrate-neutral generative process, evolvability as an organising concept, and tied these to a formal definition of the metacrisis in these terms, with suppression as constitutive rather than optional. Several of these frameworks omit outlaw suppression entirely. Synthesising all this is what our pre-print attempts.

Our deliberate overreach

Our paper is intentionally provocative and probably pushes the explanation a little further than is warranted. Not all cultural change follows Darwinian dynamics tightly, and institutions, unlike tumours, can foresee consequences and redesign themselves. Every concept we import from biology is because of the shared selective structure, not the disanalogous features.

But this extended framework supplies machinery for the loose cases too, namely niche construction, transmission biases, developmental plasticity, multiple inheritance channels, and so on. Our aim was to cast the net wide, to show there is a broader way of thinking about global systemic and catastrophic risk that supplies coherence and ultimate explanation, providing the why to supplement the what.

Towards action

Our approach would shift the object of global risk governance. We don’t specify precise solutions, but instead recommend engineering, nudging and continually readjusting the evolutionary factors so that adaptive solutions to specified fitness problems, evolve naturally:

• Shape the fitness function: make biosphere persistence and human wellbeing what actually wins and stop feeding risk-creating actors with subsidies and support.
• Stabilise the selective environment: with wise durable policy contracts insulated from electoral cycles.
• Curate variation and modularity: mandated buffers, supply-chain and planetary-boundary examiners embedded in firms on the bank-examiner model, modular refuges, rural revitalization, or transition towns that preserve capability through catastrophe.
• Protect transmission of adaptations: ring-fenced prevention budgets (eg pandemic PPE), institutional memory, error-correction of the information commons.
• Above all, build and protect multi-layer suppression: early, calibrated, capture-resistant control of defectors that win locally at the expense of humanity.
• Act regionally and at bloc level, where cooperation is more evolvable than at global scale.

Because each evolvability condition is observable, the five can form a dashboard tracking the direction of travel of global risk-governance capacity, and the framework makes falsifiable predictions, for example interventions that robustly alter the selection environment should outperform those that just alter actors’ knowledge or goals.

How this fits alongside existing work

Our work is integrative and complementary, not competitive. The Cascade Institute’s Polycrisis CORE model maps the global system attractor basins. The evolutionary account explains the forces pulling toward them and which levers change the gradient. The Accelerator for Systemic Risk Assessment has built assessment capacity, and evolvability supplies what to assess and why. Oliver and colleagues’ systemic-risk “watchpoints” are a step toward a dashboard. Richardson et al.’s normative principles in the context of global risk are where the fitness function gets its ethics. Complexity science explains how systems become fragile, and the evolutionary account explains why fragile architectures keep being selected despite full awareness.

The risk community has spent enormous effort cataloguing what is going wrong. An evolutionary reading names, monitors and builds the conditions that determine what can be redesigned to go right. The invisible hand has been running on our institutions all along. The task is to make it visible, and to direct it.

Read our new preprint here (currently under peer-review)

TLDR/Summary

• Our new review paper, No place to hide? Regional resilience and vulnerability to global catastrophic risk (Jehn et al. 2026), is published in the peer-reviewed journal Global Challenges.
• The paper reviews the literature on what makes places resilient to pandemics, nuclear war, large volcanic eruptions, near-Earth object impacts, massive cyberattack, geomagnetic storms, and high-altitude electromagnetic pulse (HEMP).
• We organise these into three pathways to harm: abrupt sunlight reduction scenarios (ASRS) including nuclear, volcanic or impact winter; global catastrophic infrastructure loss (GCIL), which is the prolonged loss of electricity, industry and the systems they carry; and global catastrophic biological risk (GCBR), namely pandemics and engineered pathogens.
• Almost all global catastrophic risk (GCR) research has been about hazards. This review addresses the other half of the risk equation, vulnerability and exposure, and is the first to map geographic, institutional and infrastructural resilience factors across the full hazard set together.
• Our blunt conclusion is that no place on Earth is resilient against all kinds of GCR. Australia, New Zealand and Japan are the only countries with evidence in the literature sourced for resilience across all three pathways (and only Australia is directly named in all three), yet all remain dependent on continued trade for commodities like fuel, medicine and fertiliser.
• Across the different risks, common resilience factors that show up most are geographic isolation (eg islands), self-sufficiency (especially in food production), high governance quality (more democratic and lower inequality) and decentralisation to mitigate single point catastrophic failures (eg impacting trade or food supply).
• Governance quality mitigates every hazard we examined: stable democracy, high state capacity, low inequality, low corruption and social cohesion appeared as resilience factors for every GCR category, and they reinforce rather than undermine each other.
• One important contribution is not the map but the trade-offs. Many resilience factors that protect against one pathway expose a country to another. Isolation is a pandemic asset and a GCIL liability. A large industrial base builds resilient-food capacity and digital fragility simultaneously.
• Governance quality and decentralisation stand out as both modifiable by policy and largely free of cross-GCR trade-offs
• For Aotearoa New Zealand (NZ) this finding needs to discipline the development of resilience. Being less exposed is not the same as being a refuge. Geography got NZ to the start line; only intention and policy will get us to a truly resilient state in the face of global catastrophe.

Global Catastrophe as Risk Pathways

The global catastrophic risk field has grown fast over the last decade. This research has been good at analysing hazards, how much soot a nuclear war injects into the stratosphere, what a Carrington-class solar storm does to electrical transformers, how a large magnitude volcanic eruption perturbs the monsoon. Research is far weaker on the question a policymaker in Wellington, Reykjavík or Montevideo actually asks:

Given all this hazard, how exposed are we, and what can we do about it?

That is the gap our review set out to fill. We ran a systematic OpenAlex scan that returned 3,200 candidate papers, screened down to 551, then to 147 read in full. We closed the remaining gaps by bringing in subject-matter expertise on societal collapse, volcanism, cyberattack and food systems, which added a further 77 expert-identified papers, for a final included corpus of 152.

Our synthesis is organised around the three harm pathways rather than a list of hazards, because many different hazards converge on the same failure. Nuclear winter and a Tambora-scale eruption (volcanic winter) both end in a crop-yield problem. Cyberattack, high-altitude electromagnetic pulse (HEMP) and a severe geomagnetic storm all end in a dark grid.

Planning for the pathways, rather than specific hazards, means that most of the risk portfolio is covered.

Regular readers will recognise the move. It is the same logic behind our call for a national vulnerability register rather than a hazard register organised around the potential system failure, not the diverse range of triggers and mapped to a set of desired resilience options.

We intentionally did not build an index

We deliberately resisted collapsing all this into a composite GCR resilience score. Trade-offs (see below) are difficult to weigh, we do not yet have the evidence to know whether low inequality matters more than domestic fertiliser production, or island status more than state capacity.

That said, a lot of robust evidence informed this review. Covid-19 supplied a genuine natural experiment: islands recorded 64.8 excess deaths per 100,000 against 194.3 for non-islands and jurisdictions running explicit exclusion and elimination strategies recorded negative excess mortality of -2.1 per 100,000 against 166.5 elsewhere. Democracy and lower inequality both predicted better health (and sometimes macroeconomic) outcomes. That is real evidence about real resilience factors, and it should inform decisions.

Important trade-offs exist

The frequent resilience factors (isolation, self-sufficiency, governance quality and decentralisation) will not surprise anyone in this field. What is new is that these factors are not necessarily additive. They can pull against each other, and we tabulate the tensions explicitly in the paper. For example:

• Geographic isolation lets a country seal a border against a pandemic, but it also puts it a long way from imported spare parts, expertise and medicine after infrastructure loss.
• A large industrial base is what allows breweries and paper mills to be repurposed into resilient food production during an ASRS, but it is also what creates dependence on the complex, digital, globally-sourced supply chains that a GCIL destroys.
• Renewables reduce fossil fuel dependence, but wind and solar output falls during a sunlight reducing scenario.
• High-yield conventional agriculture provides a bigger food buffer and a harder fall when fertiliser stops arriving.
• Islands offering isolation frequently come with active volcanoes/earthquakes, long coastlines and tsunami exposure.
• And the very inefficiency that makes meat production environmentally costly also makes livestock a substantial latent food reserve in a catastrophe.

No volume of outcome data will deduce the right resilience investments for a country, because the question is not only what works. It is how much of one good thing a society is willing to give up for another, under uncertainty, over what time horizon, and for whose benefit. Those are value judgements.

This is precisely why deliberative democratic process, such as citizens’ assemblies, are not a soft add-on to resilience policy but the mechanism by which it becomes legitimate. There are many available avenues to greater resilience. Societies need to see the evidence, deliberate over the trade-offs, and choose.

A theoretical divergence

We also make an important theoretical point, in that mainstream resilience science, following Folke and Walker, prizes connectivity. Our review finds that in the GCR context the ability to disconnect can be protective. Connectivity is genuinely ambivalent, granting access to diverse resources while also transmitting failure.

This aligns with what the Cascade Institute has been arguing about global systemic risk. Their stress–trigger–crisis model separates the slow-moving stresses that quietly erode systemic resilience from the fast triggers that tip a system over, and singles out homogenisation and hyper-connectivity as mutually reinforcing drivers that leave tightly linked global systems prone to synchronous failure.

Dense interconnection is not the opposite of fragility, because past a threshold it is fragility.

Developing refuges against GCR so humanity can reboot civilisation

Previously I have proposed that a refuge is not simply somewhere that happens, by chance, to be less affected by a catastrophe, and suggested three necessary features:

• intention – a deliberate purpose both to protect a population and to preserve or restore the capacity for technological and institutional flourishing.
• compartmentalisation – separation from the harm, achieved by whatever means, islandness, remoteness and surface-independence are merely routes to it, and the real requirement is that the compartment’s outcome be decorrelated from the world’s.
• sustainability – the compartment holds, and complex societal function persists, for as long as necessary.

Baum and Adams, in their study of Western Australia and China as Covid-19 pandemic refuges (Risk Analysis, 2023), work from a considerably broader definition, ie merely a place providing safety or avoiding significant harm, and identify refuges by observed outcome. On that reading a jurisdiction becomes a refuge because case numbers stayed low, and can drift in and out of the status as curves move.

The distinction is not academic hair-splitting, because it changes what a government should do. If refuge is an outcome, you find out whether you were one afterwards. If refuge requires intention, compartmentalisation and sustainability, it is a set of preparations made in advance, and our new review’s central finding reads very differently. “No place to hide” is a statement about incidental safety, and incidental safety was never refuge in the first place.

What the new review adds is a precise inventory of where each candidate’s compartment leaks. Covid was an “underkill” event, and generalising from it to genuinely catastrophic threats is hazardous. Much resilience work remains to be done.

Where this leaves New Zealand

Pre-Covid-19 work by Nick Wilson and I prioritised island nations as pandemic refuges, scored candidates against 14 macroindices in 2021 (Australia 0.71, New Zealand 0.64, Iceland 0.58), and in 2023 ran a food self-sufficiency threshold analysis under severe nuclear winter.

Our new, and international, multi-author review vindicates that programme. Australia and New Zealand (with Japan) are the only countries appearing resilient across all three GCR pathways. It also disciplines the work, as New Zealand carries active volcanoes/earthquakes, a small industrial base, near-total trade dependence for liquid fuel, limited pharmaceutical manufacturing capacity, and a position, alongside places like Tasmania, Ireland, Scandinavia and Canada, inside the high-likelihood zone for severe space weather damage.

That last point shows that a fall-back refuge chosen for one hazard family can be exposed in another, which strengthens the case for multiple prepared refuges rather than one.

New Zealand features prominently in the GCR resilience literature for a reason. It is remote, temperate, food-producing, democratic, low-inequality, rich in hydro and geothermal energy, and capable of enforcing a border. This combination holds almost nowhere else (Australia is a related case).

These resilience features are a genuine national asset, but are currently unmanaged as resources providing safety in the face of global catastrophic risk. The gap between “less exposed” and “prepared” is entirely composed of things we could choose to do: onshore liquid fuel resilience and a domestic biofuel capability; a trans-Tasman pharmaceutical arrangement and sovereign shipping assets; diverse and distributed food production and strategic reserves; a hardened, islandable grid with black-start capability; backup communications for payments and logistics coordination; seed banking (including landrace banking for old-fashioned/“wild-type” varieties); urban and near-urban agriculture; pre-agreed criteria for when and how to enter a state of deliberate isolation, and whether that state is open or closed.

GCR should also be written into the national risk assessment rather than omitted from it and information publicly disseminated about hazards, vulnerabilities, and resilience investment options.

None of this requires relocating the country. All of it is achievable within a decade, and most of it pays dividends in ordinary times.

This is precisely the premise on which we founded Islands for the Future of Humanity, the non-partisan charity that grew out of the 2023 NZCat project on NZ’s vulnerability and resilience to nuclear war.

The evidence recognises islands as potential refuges, but island nations face real self-reliance constraints, and the work is to turn those vulnerabilities into strengths.

Until now that was largely an inference from our own studies. This new review is the international literature converging on it independently, from a different methodological direction, with a broader hazard set and eight authors across five countries.

Part of Australasia’s prominence no doubt reflects where researchers have chosen to look, and other countries with similar profiles remain understudied, so wider research on regional GCR resilience would be very welcome.

Decentralisation as a key intervention

There is a further way to read the finding that governance quality and decentralisation are the two factors that help against everything catastrophic. In recent work with Nick Wilson on the metacrisis, I argue that humanity’s deeper problem is a degradation of evolvability, ie the conditions under which risk-reducing solutions can be generated, retained and selected for. Five conditions matter: variation, modularity, stability of the selective environment, fidelity of transmission, and suppression of defectors who gain locally while degrading the whole.

Read against that framework, the results of our new Review stop looking like a coincidence. Decentralisation is modularity. It provides semi-independent components that can fail without bringing down the wider system. Governance quality, trust and low inequality are what sustain faithful information transmission and defector suppression. Crop and climate diversity are variation. And a refuge, properly understood, is best seen not as insurance against a named hazard but as a modular societal node that preserves capability and variation through catastrophe and can reseed adaptation afterwards. This protects evolvability itself, and the capacity to rebound.

The above also explains why the trends we flag as concerning (democratic backsliding, rising inequality, supply-chain concentration) degrade resilience against every hazard simultaneously. They are not separate problems but the erosion of the same underlying conditions.

This framing also guides action, as cooperation is far more readily evolvable at regional and bloc scale than globally. This is because the competitive pressure that produces internal cohesion does not exist between planet-scale entities.

All this means the highest-leverage work available to a small country is not to wait for global agreement. It is building deep, redundant, pre-negotiated resilience arrangements with Australia and our Pacific neighbours.

Conclusion

Nowhere is safe from global catastrophe. That finding should not be softened. But exposure is not fate, and the factors that matter most, and matter across hazards in a threat-agnostic way, are the ones squarely within the power of governments willing to look past the electoral cycle to nurture true resilience.

In light of all the above, and in an election year, New Zealand’s task in the face of uncertain times and potential global catastrophe reduces to a single need: stop relying on the accidental benefit of remote geography and start building intentional resilience.

A practical follow-up to the “Reality of Everything” Symposium and other recent discussions of humanity’s predicament

TLDR/Summary

• Events like Wellington’s recent “Reality of Everything” Symposium diagnose the world’s predicament vividly, but seldom pause to define the terms doing the heavy lifting: global catastrophic risk, systemic risk, polycrisis, metacrisis.
• This guide follows a through-line, from the hazards we can see down to the drivers we usually can’t: what could go badly wrong (global catastrophic risk), how failures spread (systemic risk), how they entangle and compound (polycrisis), and why we keep generating them (the metacrisis).
• For a climate-focused audience the key move is this: carbon emissions and climate change are one thread, not the whole cloth, one of several global catastrophic risks, spreading and compounding through a tightly-coupled world.
• Beneath all of it sit behavioural and evolutionary drivers, the ultimate causes of the proximate symptoms policy usually addresses. Defined precisely, the metacrisis is not “many crises at once” but the degradation of our collective capacity to adapt, introduced briefly here and developed elsewhere.

A superb diagnosis, with the terms left undefined

Gatherings like the Reality of Everything Symposium are strong on diagnosis: speaker after speaker showed, compellingly, that human society is in overshoot, that planetary boundaries are being breached, and that our crises are interlocking rather than separate.

What such events seldom do is define the words carrying the argument. Global catastrophic risk, systemic risk, polycrisis and metacrisis sometimes get used almost interchangeably, partly blurring into an anxious hum.

That blurring matters, because it is difficult to prioritise within, or act upstream of, a problem space you cannot name.

This blog is a short field guide, organised as a chain of four questions, each reaching a little deeper than the last.

First question: what could go badly wrong? (Global catastrophic risk)

Start with the hazards themselves. Global catastrophic risk (GCR) is the study of events and processes large enough to overwhelm civilisation’s capacity to cope: nuclear war, engineered pandemics, extreme outcomes from artificial intelligence, famine-creating volcanic eruptions, and runaway climate change.

These risks are sometimes defined as those that would put the lives of 10% of humanity at risk, up to and including human extinction. The field’s contribution is a long horizon and a willingness to take seriously events that are unlikely in any given year but so consequential their expected harm can dwarf that of familiar disasters, a blind spot the UN’s own 2025 Global Assessment Report and Pact for the Future now concede.

For readers who arrived through climate, this is the first and most useful reframing: climate change is not a category of its own, apart from “other” risks, but one member of the GCR family. Treating it as one thread among several, rather than the whole cloth, is what lets the rest of the picture come into focus (we develop this widening of the lens here in the context of Aotearoa New Zealand).

Second question: how do failures spread? (Systemic risk)

Knowing which hazards could be catastrophic is not the same as understanding how a shock propagates. That is the work of global systemic risk, and it involves a genuine shift in perspective: instead of seeing the global system as an innocent bystander that receives hazards from outside (eg pandemics, storms, or volcanic eruptions), systemic-risk thinking treats the system itself as a generator of them.

Liu and Renn distil the hallmarks: densely connected feedback networks, in which effects loop back to change their own causes; strong nonlinearity, so a small nudge can trigger a disproportionate response once a tipping point is crossed; causation that jumps across sectoral and national borders; and deep uncertainty, because the system’s behaviour cannot be read off from any single part.

A useful mental image, repurposed from Siegenfeld and Bar-Yam, is a hundred ladders leaning against a wall. Tie them together for “efficiency” and each ladder becomes steadier on its own, less likely to fall. But if one does fall, it now drags all the others down with it. That is tight coupling: the very connections that make a system efficient in calm times make it fragile in a crisis. The 2011 Tōhoku earthquake rippling through just-in-time supply chains, and the 2021 Suez Canal blockage halting roughly a tenth of world trade from a single stuck ship, are textbook cases of triggers tipping stressed systems into crisis.

Two refinements are worth carrying forward. The Cascade Institute’s stress–trigger-crisis model separates the slow-moving stresses that quietly erode a system’s resilience (over-connection, homogenisation, concentration) from the fast-moving triggers that finally tip it over. And Arnscheidt and colleagues name risk-creating actors, those who extract a local advantage while loading cost onto the whole system, as a primary driver of global vulnerability. We return to this idea below.

Third question: how do crises compound? (Polycrisis)

Systemic risk explains how one system tips over. Polycrisis describes what happens when several do so at once and turn out to be wired together. Lawrence and colleagues give the field its working definition, the causal entanglement of crises across global systems in ways that significantly degrade humanity’s prospects, and specify how that entanglement happens, through three pathways.

The first is common stresses: a single underlying pressure, climate change being the obvious one, simultaneously weakens food, water, migration and security systems, so separate crises flare up together not by coincidence but from a shared root. The second is domino effects: a crisis in one system directly triggers one in another, as when a regional conflict becomes an energy shock, which becomes a food-price shock, which becomes political instability somewhere else entirely. The third, and most insidious, is inter-systemic feedback: two crises reach back and worsen each other in a loop, each making the other harder to resolve.

Run those pathways together and you get the defining feature of a polycrisis, the whole is genuinely worse than the sum of its parts, precisely because the parts are no longer independent. The years 2020–2023, when a pandemic, a European war, and energy, food and inflation shocks all fed into one another, are the case study everyone now reaches for.

Liu and Renn clarify the complementarity: systemic risk research specifies the conditions under which cascading failure becomes possible, polycrisis research tracks the cascades once entangled, and neither yet explains why governance consistently fails to adapt.

Somewhat soberingly, the most comprehensive systematic review of polycrisis to date catalogues the component crises and their drivers, then characterises many of those drivers as asserted rather than analysed. The field has mapped the entanglement in impressive detail but cannot yet say why it keeps happening, or why most of these crises, flagged as far back as the 1970s (eg Club of Rome report), remain unresolved. Which brings us to the deepest question.

Fourth question: why do we keep generating a fragile world? (From what to why)

There is a pattern. Each of the first three questions excels at describing what is happening and how it unfolds; none quite explains why humanity keeps generating the conditions behind catastrophic, systemic and polycrisis risk in the first place.

Biology offers a precise way to name the gap. Ernst Mayr distinguished proximate causes, the immediate mechanism, as in “the bird migrates because shortening days trigger a hormonal response”, from ultimate causes, ie the evolutionary reason the mechanism exists at all. Crucially, these are not two links in one chain but two answers to different questions about the same behaviour: how it works, and why it was selected for.

Almost everything governments fund, and almost everything discussed at events like Reality of Everything, acts on our behaviours and their fallout. Cut emissions, reduce consumption, diversify trade, stockpile fuel, plan for the next storm. This work is necessary. But it operates on the expression of humanity’s drives and their downstream consequences while leaving the ultimate question untouched: why do humans reliably develop the appetites for growth, status and accumulation whose collective expression produces these risks?

Suppress one proximate expression and the underlying disposition, its evolutionary rationale and sustaining forces still intact, tends to resurface through another. That is why symptom-by-symptom effort keeps falling behind: we are managing the outputs of drives whose reason for being we have not addressed.

The fields already working at the ultimate level are behavioural and evolutionary sciences. The same two questions extend beyond genes. Institutions and norms are culturally transmitted traits with their own proximate mechanisms, meaning how they operate today, and their own ultimate explanations, meaning why competition between groups selected and spread them. That dual structure is what earns “institutional” its place alongside the evolutionary drivers. These fields trace some of today’s crises to a “human behavioural crisis“, a suite of drives toward growth, status and consumption that were once adaptive and are now exploited at planetary scale.

Reach further back and the pattern is evolutionary. Competition between human groups, cultural niche construction, and technological evolution have ratcheted us toward ever-greater scale and environmental control, while rewarding strategies that pay off locally even when ruinous globally, the “risk-creating actors” of the systemic-risk literature, now seen from below. This is the multipolar trap, in which each player, acting rationally in its own interest, helps produce an outcome nobody wants.

So what, precisely, is the Metacrisis?

Here the vocabulary finally separates cleanly. A crisis is a single system in trouble. A polycrisis is many crises causally entangled, worse than the sum of their parts. The metacrisis is the layer beyond both: the ultimate, upstream drivers, behavioural, institutional and evolutionary, that keep generating and re-generating the whole mess.

I think this can be tightened one step further. Rather than a vague gesture at “everything, but deeper,” the metacrisis can be defined as the degradation of the conditions that allow risk-reducing solutions to accumulate and persist, a crisis not of any particular risk but of humanity’s capacity to adapt to large-scale risk at all.

In evolutionary language that capacity is called evolvability, and when its supporting conditions erode together, good solutions stop compounding and governance fails consistently across otherwise unrelated domains. Where the more familiar framing locates the metacrisis in a crisis of collective sense-making, of how we perceive and make meaning, this one locates it a step wider, in the adaptive machinery that would let good solutions take hold and accumulate at all; sense-making is then one input to that capacity rather than the whole of it.

I flag this lens for completeness; it is deliberately provocative and, for now, a theoretical framework rather than the settled account, developed in current work and presented in nascent form at recent risk-science conferences. Watch this space for my imminent pre-print research paper.

Why this matters, especially if you came via climate concern

If you arrived at the ‘Reality of Everything’ and global risk through climate change, the upshot is liberating rather than deflating. Climate change (CO2/radiative forcing) is one of nine planetary boundaries, and the climate crisis is one expression of a larger predicament driven by common upstream forces.

Seeing the through-line, hazards, stresses, how they spread, how they compound, and why we keep making them, lets you do two things at once: keep pushing on the proximate work that is urgent and real, while recognising where climate action connects to other risk like pandemic preparedness, trade and supply-chain resilience, and the health of the institutions on which all risk mitigation depends.

Naming the space is the difference between fighting symptoms one at a time and acting on the mechanisms that produce them. That is the conversation we hope events like Reality of Everything can grow into next, and it is one this small field guide is meant to kick start.

Thanks again to the Symposium organisers and participants.

By Matt Boyd (with Mike Cassidy reporting on the workshop)

TLDR/Summary

• The 2026 Volcanic Impacts on Climate and Society (VICS) Conference brought together the VICS community, a working group associated with PAGES: Past Global Changes, to consider past, present and future eruptions.
• For a global catastrophic risk audience, the central message was clear: volcanic eruptions are not merely local geological disasters. Large eruptions can, and have, become global climate, food, trade, financial and political shocks.
• More importantly, large volcanic eruptions are rarely “single-cause” disasters. They typically act as amplifiers: triggering or worsening existing vulnerability across food systems, markets, disease dynamics, migration, state fragility and geopolitical instability.
• That makes such eruptions particularly important for global systemic risk, catastrophic risk, and polycrisis thinking, as a large volcanic eruption is a plausible trigger that could tip already stressed critical global systems into crisis, and there is historical precedent.
• The VICS Conference also sharpened a second point we have made in our work, which includes analysis of the Tambora eruption and volcanic winter and nuclear winter and trade disruption, namely that abrupt sunlight reduction is not only a climate problem, it is a civilisation-support problem.
• The interdisciplinary workshops hosted at VICS provided a template for how academics, policymakers and emergency management personnel can use scenarios to better understand and prepare for the global and cascading effects of large volcanic eruptions.
• With next year’s 2027 VICS Conference being held in Taupō New Zealand – this country has an extra opportunity to bring particular focus to all these issues.

Past: volcanic eruptions as amplifiers, not sole causes

The first day of the VICS Conference focused on the past, treated as the best available laboratory for understanding how societies respond to sudden climate shocks.

A repeated structure appeared. We see eruption, atmospheric perturbation, regional climate anomaly, crop harvest stress, grain price spikes, disease, migration, and unrest. But the causal chain was never simple. One eruption does not produce one global outcome. The same event may bring cold and wet conditions to one region, warm and dry conditions to another, and little obvious disruption elsewhere. Latitude, season, eruption chemistry, injection height, background climate state, crop calendars, trade networks, political capacity and social vulnerability all typically matter.

This is exactly why volcano risk is a global systemic-risk problem. Many past eruptions have caused reductions in global mean temperature. A global-mean temperature number can be useful, but people do not eat global means. They eat crops grown in particular regions, during particular seasons, embedded in particular market and political systems.

Some of the strongest evidence for volcanic effects and response came from combining “archives of nature” and “archives of society”: ice cores, tephra records, sulphur (and its isotopes), tree rings, cave records and marine records alongside diaries, parish burial records, grain prices, market records and historical texts. That fusion allows researchers to ask not only “did an eruption occur?” but “when, where, how high did the emissions go, what climate response followed, and how did societies mediate the harm?”

Examples ranged from Icelandic eruptions and their climate fingerprints, to 17th-century northern Europe, to China, Japan and the early 19th-century crisis years around the Tambora eruption. The 1690s in northern Europe were particularly instructive: Scotland, Finland and Norway all experienced cold, harvest failure and very high mortality, but outcomes varied. Norway’s grain imports and social-political context appear to have buffered some impacts. The lesson is not “eruption equals famine.” It is “eruption plus vulnerability plus failed response equals catastrophe.” The figure below is a systems diagram showing the cascading impacts of the 1831/1835 eruptions (as described in Richard Warren’s talk).

Past–present: the wrong volcano may be in our risk model

Day two pushed the evidence closer to present-day risk. A key theme was that high-latitude and mid-latitude eruptions may matter more than the standard catastrophic tropical-volcano mental model suggests.

Ally Peccia’s work on the 43 BCE Okmok II eruption in Alaska was striking in this regard. A remote Aleutian eruption appears to line up with cold, famine and major documented crises across the late Roman Republic, Egypt, China, Sri Lanka and India. The key scientific issue (especially for thinking about future eruptions) is not simply how large the eruption was, but how much sulphur reached the stratosphere, how it spread, and how sensitive the Northern Hemisphere climate was to that loading.

This matters because the world’s preparedness imagination still tends to picture a Tambora-like tropical eruption. Tambora remains the benchmark, and rightly so. But VICS reinforced that remote extratropical volcanoes can be especially globally consequential. They may “punch above their weight” hemispherically.

If risk assessments focus too narrowly on the volcanic explosivity index (VEI), tropical latitude, or direct local exposure, they may miss the events most relevant to global food and trade disruption, the cascading systemic effects.

Anja Schmidt’s keynote made this even sharper. She argued that the VEI is not well suited to climate-risk thinking because it does not reliably track sulphur release or climate response. A proposed Volcano-Climate Index would instead rank eruptions by climate effect. This is the kind of shift risk governance often needs: from measuring the hazard in geological terms to measuring what actually matters for human systems, thereby allowing risk assessment in terms of likelihood of catastrophic climate impacts.

Schmidt also flagged a striking probability: a substantial chance (70%) of Pinatubo-scale eruption induced cooling before 2100. The risk is not just the global cooling. It is estimating cascading hazards and socio-economic consequences in a hyperconnected world.

That is where volcano science needs to meet food-system modelling, shipping, finance, public health, conflict studies and national resilience planning. This is important because the Pinatubo eruption in 1991 (in the Philippines) had a greater impact on global food production (maize down 9%) than any severe El Nino, though this is complicated by the fact that the eruption may have triggered an El Nino!

In today’s even more interconnected, food trade-dependent, and conflict-ridden world, the implications for food security would be even more serious.

There is also an uncomfortable analogue between large volcanic eruptions and nuclear war (and its stratospheric soot injection from burning cities), as well as geoengineering as a response to climate heating. Nuclear winter, volcanic winter and solar geoengineering all involve sunlight reduction, altered precipitation and non-uniform regional effects. They differ enormously in cause and governance, but they stress some of the same civilisational-support systems: food, trade, energy, trust and political coordination. We contend that the likelihood (and therefore the food/trade/societal resilience policy response) to climate altering shocks, needs to account for the aggregate probability of severe volcanism, nuclear war, geoengineering policy, and other drivers.

Future: the next eruption happens in a warmer, more connected world

The future-focused day on day three of the Conference, made the global catastrophic risk relevance explicit.

Markus Stoffel returned to Tambora as the benchmark: the 1815 eruption, roughly 1°C Northern Hemisphere cooling, failed harvests, grain price spikes, the “year without a summer,” and severe global consequences.

But the more important point is that a Tambora-scale eruption today would not occur in the world of 1815. It would occur in a world of more than eight billion people, tightly coupled trade, just-in-time supply chains, financial contagion, geopolitical tension, fragile trust, and climate-stressed food systems.

That is the polycrisis point. The eruption is not the whole disaster. It is the initiating shock in an already stressed system.

May Chim’s modelling, as summarised at the Conference, suggested that global warming may amplify future eruption impacts: a Pinatubo-sized eruption later this century could cool the surface more than the same eruption in a pre-industrial climate. This is because of altered plume height, aerosol behaviour and ocean heat dynamics. In other words, climate change may not simply be one background risk among many. It may change the behaviour of other risks.

James Dalziel’s insurance and financial-risk talk showed how poorly prepared institutions remain. Recent volcanic losses look small compared with storms and floods, but the tail is large. Lloyd’s/Cambridge-style scenarios suggest a plausible extreme volcanic event could produce multi-trillion-dollar economic impacts over five years, with losses travelling through insurance, aviation, shipping, supply chains, food systems and healthcare.

Mike Cassidy’s presentation brought the global catastrophic and volcanic risk cases together. Policymakers under-prioritise volcanic risk because eruptions seem rare, local and geophysical. But the most consequential harms are often second- and third-order cascades: poor harvests, grain price rises, disease, famine, migration, unrest, political instability and long recovery times. A single eruption may be a 2–4-year shock. An eruption cluster can become a 7–15-year depression because societies do not get time to recover before the next forcing arrives and forcing can compound.

This is where analogy to the recent Hormuz Strait maritime chokepoint disruption is useful. This case showed how quickly a geographically concentrated shock can become an energy, fertiliser, shipping, inflation and security problem.

A major eruption could do something similar, but through ash, aviation closure, food-system stress, climate anomalies and trade interruption at the same time. For New Zealand, this matters because geographic isolation is both a resilience asset and a vulnerability. We may be food-producing, renewable-energy-rich and distant from many conflict zones, but we are also deeply dependent on imported liquid fuels, imported components, global markets and functioning maritime trade.

What is the global community still missing?

The VICS Conference suggested several gaps.

First, we still need better scientific reconstruction of past eruptions: eruption timing, source attribution, sulphur budgets, stratospheric injection, halogens, aerosol microphysics, plume height and eruption season. These details determine climate impact.

Second, we need regional impact modelling, not just global means. Adaptation happens in breadbaskets, ports, watersheds, insurance books and political systems.

Third, we need integrated food-trade-economy-public-health models for abrupt sunlight reduction scenarios. This is where our own Tambora work and our ASRS/nuclear-winter work should be brought together.

Fourth, we need serious scenario exercises. The VICS Conference included exercises around Kikai and Campanian Ignimbrite scenarios, two large magnitude 7 eruptions with likely associated climatic shocks, these represented not the worst-case event (e.g. a rare super eruption), but a ‘reasonable worst case’ given both volcanoes have endured these eruptions before, VICS asked what if these eruptions happened today?

Scenario exercises were run at VICS with interdisciplinary experts from academia (a range of different expertise from volcanologists, climate experts, historians) to policymakers. The aim was to provide a timeline of likely impacts from volcanic climate hazards, and the impacts to society. As part of the exercise, intervention points were added, where policy action and preparedness could mitigate some of the worst impacts.

Given the interdisciplinary range of the participants, the idea was to build a shared mental model and common language, to learn from each other and leave with the same picture of the threat and how everyone’s work connects. By laying out the timeline as a series of events, it also aimed to help to make an otherwise ungraspable large risk, somewhat tractable, while exposing the gaps and uncertainties in our knowledge.

As a test exercise, and with a limited window of time, the scenarios were a success. Discussions were clarifying and productive and helped some participants to fully understand the scale of the risk, when taken as a whole. The exercise created connections between policymakers and scientists and built an understanding of the limitations in governing the response during or after the event. VICS hopes to build on these scenario exercises at their next events, as they stimulated important conversations, which talks and posters alone could not achieve.

Fifth, the world needs to know what actually works to mitigate volcano climate risk. Cassidy’s “what worked and what didn’t” in past volcano climate shocks framing is crucial. Historical cases suggest that buffers, imports, social trust, functioning markets, local adaptation, transparent information, and competent governance can reduce harm. But panic buying, food export bans, conflict, brittle supply chains and poor communication can amplify it (see the figure below, which compares historical and modern context, and what worked and didn’t work to mitigate volcano climate impact).

Implications for action

For New Zealand and other remote nations, volcanic risk should be treated with anticipatory governance of catastrophic societal risk, not just geological hazard management. That means adding large distant eruptions to national risk assessments; stress-testing food, fuel, medicine and maritime logistics; building food plans for ASRS (abrupt sunlight reduction scenarios) that cover both nuclear and volcanic winter; modelling trade interruption alongside climate disruption; and creating a vulnerability register and mitigation register for vital functions.

Internationally, volcanic-risk work needs to move from “what is the eruption probability?” to “which societal functions fail, in what order, and what keeps them running?” That means whole-of-system exercises involving volcanologists, climate scientists, food-system experts, insurers, shipping experts, public-health planners, defence, finance, and policymakers.

The strongest lesson from VICS may be that the past is not past. It is an archive of natural experiments in stress, shock and response. The next large eruption will not by itself determine whether the outcome is difficult, disastrous or globally catastrophic. The outcome will depend on the  state and responses of the systems it strikes.

The eruption is the trigger. The catastrophe is the cascade. The work now is to make sure the resilience building initiatives exist to put the brakes on those cascades.

An election-year proposal, put in the open for debate

By Islands for the Future of Humanity

TLDR/Summary

• The global risk picture is serious and, for several risks, worsening: geopolitical turmoil, the risk of nuclear conflict, bioengineered pandemics, large volcanic eruptions, severe space weather, AI-enabled systemic failures, and the trade and supply-chain collapse that an isolated trading nation would feel most.
• New Zealand is ill-prepared. Useful work on planning and resilience exists, but it is fragmented, ad hoc, domestic, short-horizon, and largely silent on global catastrophic risk.
• We propose a Parliamentary Commissioner for Catastrophic Risk (PCCR), an independent Officer of Parliament modelled on the proven Parliamentary Commissioner for the Environment, and we’ve drafted a Bill to show it can be done.
• The public mood is shifting from efficiency and profit toward resilience and security. The data back this: a majority of New Zealanders, across the political spectrum, support institutional action on catastrophic risk.
• 2026 is an election year. Which party will pick this up and campaign on it?

The global situation is serious, and New Zealand is not ready

The world is not short of warnings. Across our recent work, including our two Policy Quarterly papers (2021, 2025) and our synthesis of the latest European risk-science conference, we have made the same case. Risk governance still systematically under-weights the low-probability, high-impact, cross-border catastrophes that a small, remote, trade-dependent nation is most exposed to. Great Power war, nuclear war and nuclear winter, extreme pandemics, large volcanic eruptions, solar storms, AI catastrophes: are each individually unlikely in any given year, but collectively close to certain over the decades, and capable of severing the flow of fuel, fertiliser, medicines and food on which NZ depends.

Our country has pieces of an answer, a National Risk Register, DPMC’s resilience work, the National Emergency Management Agency (NEMA), sectoral strategies. But these all remain fragmented, focused on familiar domestic hazards, after-the-fact response plans to international crises, and are quiet on the risk of global catastrophe. Some activity is real. The coordination, the scope, and the public conversation are missing.

Why central, independent oversight, and why now

This is not just an organisational gap. As one of us argued at the SRA-E conference, fragile systems are selected for. Where no agency faces real pressure to look across silos and over the horizon, none does, and individually rational behaviour (each department defending its own patch, each firm externalising its risk) compounds collective danger. Siloed risk assessment can even improve one entity’s resilience while degrading the whole. Correcting that requires a body whose explicit job is to suppress this drift: to aggregate across hazards, take the long view, a view of global reality, realistic futures, tail risk, and answer to Parliament rather than the three-year cycle.

That is the logic behind risk mitigation best-practice harnessing “three lines of defence”: operational agencies on the front line, DPMC providing managerial oversight, and an independent scrutineer, the PCCR, as the third line. The Commissioner would not run emergencies. Instead, the Commissioner would:

• Assess global, systemic, and catastrophic risk
• Critique national risk assessments and risk registers (including what they leave out and how little they say about mitigation options)
• Advocate for basic-needs continuity and regional cooperation with Australia and the Pacific, and, crucially,
• Open these questions to the public through deliberative democracy, such as Citizens’ Assemblies.

The aversion to “scaring the public” has to go; the evidence is that people are ahead of the institutions, and are seeking security, resilience, and sustainable wellbeing amid global turmoil.

We’ve drafted the Bill – in the open

To move from argument to action, we’ve prepared draft legislation for NZ: the Parliamentary Commissioner for Catastrophic Risk Bill. It borrows the durable architecture of the Parliamentary Commissioner for the Environment, Officer of Parliament status, a long appointment term insulated from electoral pressure, independence from ministerial direction, and adds a statutory definition of basic needs, oversight of national risk processes, and a mandate for transparent reporting. The draft has been peer-reviewed by a legal colleague, though no doubt it could be improved further in a robust Select Committee process with public input.

We envision a small office (perhaps 8–10 people, no operational role) against the scale of harm it would help avoid: the cost-benefit ratio could be overwhelming, and the resilience co-benefits, local food, energy security, better infrastructure, would help New Zealanders now.

We are publishing this openly, for discussion, inspiration, and debate. That is deliberate. We are not slipping it to a staffer’s personal email, or hand-delivering hard copies to leave no digital trail, as recently alleged in the lobbying over NZ’s climate legislation. Good risk governance is the opposite of regulatory capture: transparent, contestable, and built in public.

An election-year question

“Balancing the books” is not a national strategy for a century this turbulent, where it is unclear how business-as-usual energy and financial architectures will keep performing. We need a vision of serious, long-term resilience, de-risking while we still can in the face of catastrophes we can foresee and the ones we can’t. The UK is drafting law, the EU is building the system, the Nordics already live preparedness. 2026 is a NZ election year. Global instability is real and rising, and this should be front and centre.

So here is the question for every NZ political party: who will take this Bill, or one like it, and campaign on it?

Please consider filling out our webform with your thoughts and feedback.