Anthropics – Fable 5

This deserves a straight answer up front, then the reasoning. The short version: no, collapse is not locked in, and it is not the single most probable outcome — but the most probable outcome is also not "things turn out fine." The modal trajectory for 2026–2080 is a rough, uneven, partially managed degradation: real losses in welfare, safety, and stability, distributed very unequally across regions, with genuine collapse occurring locally and the window for global collapse remaining open but not dominant. Roughly: I'd put a protracted global loss of industrial complexity of the kind you define at something like 5–15% by 2100, severe global degradation short of that at 25–40%, and uneven muddling-through — adaptation in wealthy regions, collapse-like conditions in vulnerable ones — at around 50%. A smooth transition with broadly rising global welfare is the least likely scenario, well under 10%. These are judgment calls, not measurements; treat the ranking as more reliable than the numbers.

Now the reasoning, because the ranking only means something if it survives the strongest version of the collapse argument.

The collapse case, stated properly

The strongest argument in your list is not any single risk. Almost every item, taken alone, is survivable: antibiotic resistance currently kills ~1.3 million people a year directly and is projected toward ~10 million by 2050 — a catastrophe, but not a civilization-ender. Kessler syndrome would degrade GPS, comms, and Earth observation severely, but ground-based infrastructure existed before satellites. High-frequency trading produces flash crashes, which markets have absorbed repeatedly; the real financial fragility is leverage and shadow banking, not trading speed. Phosphorus reserves, at current extraction, last centuries — the genuine problem there is geopolitical concentration (Morocco controls ~70% of reserves) and inefficient use, not geological exhaustion. Fossil depletion is increasingly a non-binding constraint because demand transition is outrunning supply exhaustion in electricity, where solar plus storage is now the cheapest new generation in most of the world — a market fact, not a policy hope.

The strong argument is correlation and compounding. The risks are not independent draws. Climate stress simultaneously degrades food production, drives migration, intensifies water competition, and raises pandemic risk; migration and food prices feed political polarization; polarization destroys the coordination capacity needed to address the original drivers; epistemic decay (deepfakes, institutional distrust) makes consensus on basic facts harder precisely when consensus is most needed; meanwhile buffers shrink — fiscal space consumed by debt and aging populations, supply chains optimized to zero slack, aquifers drawn down, soil capital spent, infrastructure capex deferred. Each shock arrives at a system with less recovery capacity than the last shock found. That is a ratchet, and it is the correct frame. Joseph Tainter's diagnosis fits: complex societies fail when the marginal return on additional complexity goes negative and they can no longer afford to solve problems by adding more complexity. The "polycrisis" framing is not hysteria; it is the right model.

And your observation about democratic constraint is empirically sound. No electorate has ever durably voted for less. Climate policy survives politically only when it is invisible (regulation, subsidy) or profitable; carbon prices that bite get reversed (gilets jaunes, Australia's carbon tax repeal). Mitigation therefore happens at the speed of cost curves, not at the speed of physics. This is why we will overshoot 1.5°C — we effectively already have — and very likely 2°C.

Why the collapse case nonetheless doesn't dominate

Four structural counterweights, none of them wishful.

First, the historical base rate cuts both ways. Tainter's collapses — Rome, the Maya, the Bronze Age — were regional, and other complex societies elsewhere continued or absorbed the survivors. There has never been a global collapse, partly because there has never been a global civilization, which is the novel-fragility argument — but the same coupling that transmits shocks also transmits redundancy: knowledge exists in millions of copies, productive capacity exists on every continent, and there are multiple independent centers of state capacity. A global system can degrade everywhere at once; it is very hard for it to simplify everywhere at once, because simplification in one region raises the relative returns to complexity elsewhere. The 20th century is the strongest evidence: two world wars, a depression, a pandemic, and the destruction of most of Europe's and Asia's industrial base produced not simplification but the fastest complexity growth in history. Industrial civilization has already absorbed shocks far larger than anything climate is likely to deliver before mid-century.

Second, the food system has enormous hidden slack, and food is the main transmission channel from climate to collapse. Roughly a third of global grain feeds animals, another large share goes to biofuels, and about a third of all food is wasted. A synchronized multi-breadbasket failure — the genuinely frightening climate scenario, and one whose probability rises nonlinearly with warming — would mean price spikes, export bans, political crises, and famine in import-dependent poor countries. It would not mean caloric collapse in industrial countries, because the buffer of feed, fuel, and waste can be reallocated under price pressure, brutally but effectively. This is exactly the unequal-degradation pattern: the same shock that is survivable in the OECD is collapse in Egypt or the Sahel. (This is already visible: Yemen, Haiti, Sudan are experiencing your definition of collapse now. Global averages hide it.)

Third, we act — late, reactively, expensively, and only after vivid damage, but we act. The honest historical record on coordination is not pure failure: Montreal Protocol, leaded gasoline, acid rain, smallpox eradication, the non-use of nuclear weapons for 80 years, the COVID vaccine timeline (a year from sequence to deployment — unprecedented). The pattern is consistent: humans do not prevent; they respond once the cost of inaction becomes visible and concentrated. This is a terrible property for problems with long lags and tipping points — it guarantees we eat the overshoot — but it is a strong property against protracted collapse, because a decades-long decline continuously generates the vivid damage that triggers response. Collapse over decades, your definition, is precisely the timescale on which reactive systems perform least badly. Prevention will not look like prevention; it will look like belated, partial, costly adaptation — seawalls after floods, grid hardening after blackouts, antibiotic incentives after the death toll spikes.

Fourth, the energy transition has crossed from politics to economics. This is the single biggest change since the 2010s collapse literature. Solar, wind, and batteries are on learning curves that politics can slow but no longer stop; China alone is deploying at a rate that bends the global emissions curve regardless of Western polarization. This does not save the climate — transport, industry, agriculture, and the existing capital stock guarantee decades more emissions — but it caps the plausible warming range. The 4–5°C scenarios that anchored the strongest collapse arguments a decade ago now look like tail risk rather than baseline; current-policy trajectories cluster around 2.5–3°C by 2100. That is a damaged, dangerous world with real tipping-point exposure (AMOC weakening, ice-sheet commitment, regional habitability loss in wet-bulb terms). It is not, on the evidence, a world incompatible with industrial society in most regions.

What you left off the list, and what actually matters most

Your list omits the largest single risk: great-power nuclear war. Nothing on your list rivals it for capacity to produce your defined outcome quickly and globally — a NATO–Russia or US–China exchange, plausibly triggered through exactly the polarization and miscalculation dynamics you describe, followed by nuclear-winter agricultural failure, is the one mechanism that converts "uneven degradation" into "global simplification" within years rather than decades. Probability per year is low; integrated over 75 years with rising multipolarity and eroding arms control, it is a substantial fraction of the total collapse probability mass.

Also missing or underweighted: engineered pandemics (AI-assisted bioweapon design makes a worse-than-natural pathogen increasingly accessible — more dangerous than any climate-induced disease emergence); demographic inversion (every industrial society is aging fast; this drains fiscal capacity, shrinks militaries and workforces, and makes the infrastructure-renewal capex problem you mention much worse — it is a slow ratchet on state capacity everywhere, including China); the Taiwan semiconductor chokepoint (a single island's foundries underpin the entire digital economy — a blockade or war would be a supply shock dwarfing COVID); insurance retreat (insurers withdrawing from Florida, California, and Mediterranean coasts is the concrete mechanism by which climate damage becomes financial-system damage — watch this, it's the canary); and the interaction of AI with the epistemic crisis, which cuts both ways: it can finish off shared reality, or it can deliver the productivity and scientific capacity that relaxes several other constraints. AI is the highest-variance term in the whole system, and anyone confident about its sign is overconfident.

The intertwining, judged coldly

The cascading-failure argument is right about direction but tends to overestimate coupling tightness. COVID was a live experiment: a synchronized global shock to supply chains, labor, and finance simultaneously — and the system reorganized within roughly two years, at great cost, with the costs dumped disproportionately on the poor. That is the realistic template. Cascades in human systems differ from cascades in power grids because humans re-route, hoard, ration, substitute, and coerce at every node. The result of compounding shocks is therefore most plausibly not system death but system hardening with welfare loss: more autarky, more stockpiles, more authoritarianism, more borders, more inequality, lower trust, lower growth — a meaner, poorer, more fortified industrial civilization. That outcome satisfies neither the optimists nor the collapse theorists, and it is the most probable one.

The ratchet that genuinely worries me is not material but political-epistemic: each crisis handled through emergency powers and each year of polarization degrades the machinery needed for the next response, and unlike soil or aquifers, nobody has a costed plan for restoring institutional trust. Of all your items, this is the one where "we don't know what to do" is most literally true.

So: is there still a chance?

Yes, and more than a residual one — but be precise about what "preventing collapse" will look like, because it will not look like rescue. The realistic best case still includes: overshooting 2°C; losing significant coastal and agricultural territory; recurrent food-price and migration crises; the effective collapse of several states in the global tropics and their abandonment by the rest of the world; more authoritarian governance in more places; and a decades-long fight over distribution inside rich countries as adaptation costs land on electorates that, as you correctly say, have never accepted less. Industrial civilization persisting through that is not a comforting outcome. It is simply the most probable one, because the system's demonstrated capacity to absorb punishment, reallocate slack, and respond reactively exceeds the punishment most plausibly scheduled — with two exceptions that could falsify everything above: nuclear war and runaway-scale AI misuse, which together carry most of the genuine global-collapse probability, and neither of which is governed by the slow-degradation logic of the rest of your list.

The means to avoid collapse exist physically and technically across nearly every item you listed. The binding constraint is, and will remain, coordination — and the realistic forecast is that coordination will arrive late, partially, and after preventable damage, every single time, because that is what it has done every single time before. That is not a reason for hope or despair. It is the operating characteristic of the machine we are riding, and on current evidence the machine bends badly without breaking globally — while breaking locally, repeatedly, for the people with the least say in any of it.