The Loudest Argument in the Room
The climate debate is the most consequential argument in human history, and almost everyone involved is having it wrong.
On one side, the technocratic consensus: the science is settled, the IPCC reports are definitive, the policy prescriptions are clear, and anyone who questions any element of this package is a denier, a shill for fossil fuel interests, or a dangerous fool. On the other side, a coalition of genuine skeptics, ideological libertarians, and -- critically -- a well-funded disinformation apparatus maintained by the industries that profit most from the status quo: the science is uncertain, the models are unreliable, the proposed solutions would destroy the economy, and the entire enterprise is a power grab by globalist elites.
Both positions are, in the precise sense I developed in Chapter 9, Level 1 cognition operating on a Level 2 problem. Both are reading the correlational surface and mistaking it for the causal structure. The technocratic consensus correctly identifies the associational pattern -- rising greenhouse gas concentrations correlate with rising temperatures -- but treats a system of staggering causal complexity as though it were a simple input-output relationship amenable to policy levers. The denialist position correctly identifies that the system is more complex than the technocratic narrative admits, but uses that complexity as a smokescreen to protect the interests that profit from inaction.
The framework of this book -- the normie/psycho/schizo diagnostic, Pearl's causal hierarchy, Kuhn's paradigm analysis, and Popper's falsifiability criterion -- offers a way through this impasse. Not because the framework magically resolves the scientific uncertainties, but because it correctly identifies the type of problem the climate crisis is, and therefore the type of tools required to address it.
The climate crisis is not a scientific problem with a policy solution. It is a system of systems -- energy, agriculture, transportation, finance, politics, culture, geophysics, ecology, technology -- with feedback loops, tipping points, nonlinear dynamics, and emergent properties that no correlational analysis can capture. It is, in other words, exactly the kind of problem that causal inference was designed for.
Diagnosis: Who Benefits from the Crisis Persisting?
The normie/psycho/schizo framework applies to the climate debate with uncomfortable precision.
The normie position is the technocratic consensus. The normie majority trusts institutional authority -- the IPCC, major universities, government scientific agencies -- and accepts the consensus narrative as the reliable output of credible institutions. This trust is not misplaced in its foundations: climate science is genuine science, conducted by genuine scientists, producing genuine knowledge. The problem is not that normie trust in climate science is wrong. The problem is that normie trust is non-discriminating. It trusts the science, the policy prescriptions derived from the science, the institutional frameworks implementing the policy, and the economic actors claiming to support the policy -- all as a single package. It cannot distinguish between the genuine causal knowledge (greenhouse gases trap heat) and the correlational policy apparatus built on top of that knowledge (carbon markets reduce emissions).
This creates the exact vulnerability the psycho class exploits.
The psycho class operates on both sides of the debate. This is the critical insight that neither the technocratic consensus nor the denialist position can accommodate, because both assume the psycho class is exclusively on the other side.
On the denialist side, the psycho-class capture is well-documented and relatively easy to see. Fossil fuel companies -- ExxonMobil, Shell, BP, and their equivalents -- conducted internal research in the 1970s and 1980s that confirmed the fundamental mechanics of anthropogenic climate change. Exxon's internal scientists produced projections of global temperature increase that have proven remarkably accurate. The companies' response was not to publish these findings and begin transitioning their business models. Their response was to fund a disinformation apparatus -- think tanks, PR firms, astroturf organizations, and compliant politicians -- designed to manufacture doubt about the science their own researchers had confirmed.
This is the Epstein structure applied to planetary survival. Chapter 18 analyzed the pattern: the appearance of legitimacy (corporate philanthropy, support for scientific research, public commitments to environmental responsibility) concealing the causal structure underneath (systematic investment in disinformation to protect extractive business models). The fossil fuel industry's climate denial is not a failure of understanding. It is a success of the predatory logic that the psycho class applies wherever the stakes are high enough: manage the correlational surface while extracting value through the causal structure that the surface conceals.
But here is where the analysis becomes uncomfortable for the environmentalist movement: the psycho class has captured the green side too.
Carbon offset markets are the paradigmatic example. The premise is elegant: if you emit carbon dioxide here, you can pay someone to absorb carbon dioxide there, and the net effect on the atmosphere is zero. The correlational logic is impeccable. The causal reality is that the vast majority of carbon offsets do not actually reduce atmospheric carbon. Forest preservation projects protect forests that were not going to be cut down. Renewable energy credits fund projects that would have been built anyway. The offset market creates a correlation between payment and claimed carbon reduction without a causal relationship between the two. The buyer gets to claim carbon neutrality. The seller gets revenue. The atmosphere gets nothing.
ESG (Environmental, Social, and Governance) metrics follow the same pattern. The correlational surface -- companies that score well on ESG metrics are "sustainable" -- conceals a causal reality in which the metrics are gamed, the scoring methodologies are opaque, and the companies with the highest ESG ratings often have no measurably lower environmental impact than their peers. ESG has become a camouflage mechanism: the green equivalent of the philanthropy that concealed Epstein's predation. Companies purchase the appearance of environmental responsibility without the substance.
The green industrial complex -- the network of consultants, rating agencies, offset providers, and institutional investors that profit from the appearance of climate action -- has its own psycho-class structure. Its members are not, in general, consciously cynical. Most genuinely believe they are contributing to the solution. But the structural logic of their position -- profit from the appearance of action regardless of whether the action is causally effective -- is identical to the structural logic they oppose on the fossil fuel side. Both manage the correlational surface while the causal reality continues unchanged.
The schizo position sees through both sides. The climate skeptic who is not funded by fossil fuel interests but who notices that carbon markets do not actually reduce carbon -- this person is performing the prophetic function, however clumsily. The deep ecologist who argues that the entire growth paradigm, not just fossil fuels, is the problem -- this person is operating at Level 3 (counterfactual), imagining a world organized on fundamentally different principles. The scientist who publishes findings that complicate the neat policy narrative and gets attacked by both sides -- this person is the anomaly reporter, the Kuhnian dissident, the voice saying the paradigm is not working.
The tragedy of the climate debate is that the schizo position is marginalized from both sides: too radical for the technocratic consensus, too nuanced for the denialist movement. The prophetic function has nowhere to stand.
Causal Structure: The System of Systems
The causal structure of the climate crisis is not a simple DAG with "greenhouse gas emissions" as the treatment variable and "global temperature" as the outcome. It is a system of interconnected causal systems, each with its own feedback loops, confounders, and nonlinear dynamics. Any analysis that treats it as a single-cause problem is committing what Pearl calls the "causal Markov violation" -- assuming independence where complex dependencies exist.
Let me sketch the causal architecture, in the notation I developed in Chapter 9.
Layer 1: The Geophysical System. This is the layer where the basic science operates, and where the scientific consensus is most robust. Greenhouse gas concentrations cause radiative forcing, which causes temperature increase, which causes ice sheet dynamics, ocean circulation changes, and weather pattern shifts. The basic physics is Level 2 knowledge -- interventional, experimentally validated, not merely correlational. We know that increasing CO2 concentrations causes increased heat trapping because the mechanism is understood at the molecular level and has been confirmed by laboratory experiments. This is genuine causal knowledge.
But the geophysical system has internal feedback loops that make simple projections unreliable. Water vapor feedback: warming causes more evaporation, water vapor is itself a greenhouse gas, creating a positive feedback loop. Ice-albedo feedback: warming melts ice, reducing Earth's reflectivity, causing more warming. Cloud feedback: the net effect of warming on cloud formation is still genuinely uncertain, and clouds can either amplify or dampen warming depending on type and altitude. These feedback loops mean that the relationship between emissions and temperature is nonlinear, with potential tipping points -- thresholds beyond which the system shifts rapidly to a new equilibrium.
Layer 2: The Energy System. Energy production is the proximate cause of the majority of greenhouse gas emissions. But the energy system is itself a causal system: technological path dependency (infrastructure built for fossil fuels creates lock-in), economic incentives (fossil fuels are still cheaper than renewables in many contexts when externalities are not priced), geopolitical power structures (petrostates derive political power from fossil fuel production), and labor markets (millions of workers in fossil fuel industries) all interact to determine the trajectory of energy transition.
The causal DAG here has confounders that correlational analysis cannot handle. The observation that countries investing in renewable energy have lower emissions growth does not mean the investment caused the reduction. Wealth is a confounder: rich countries can afford renewable investment and tend to have slower emissions growth for other reasons (deindustrialization, efficiency improvements, demographic transitions). Without proper causal identification -- the backdoor criterion, instrumental variables, or natural experiments -- the policy conclusions drawn from correlational data are unreliable.
Layer 3: The Economic System. The economic system generates the demand that drives the energy system. GDP growth, consumption patterns, trade networks, financial flows, and investment decisions all causally affect emissions. But the relationship runs in both directions: climate impacts cause economic damage, which affects investment decisions, which affects emissions trajectories. This bidirectional causality creates identification problems that correlational analysis cannot resolve.
Layer 4: The Political System. Climate policy is determined by political processes: electoral dynamics, lobbying, international negotiations, regulatory frameworks. The fossil fuel industry's influence on politics -- through campaign finance, revolving doors between industry and government, and funded disinformation -- creates a direct causal path from industry interests to policy outcomes. But this path is confounded by voter preferences, economic conditions, geopolitical dynamics, and ideological commitments that operate independently of industry influence. Disentangling the industry's causal effect on policy from these confounders requires the tools of causal inference, not mere correlation.
Layer 5: The Cultural System. Consumption patterns, lifestyle expectations, cultural narratives about progress and development, and the social meaning of material abundance all causally affect emissions. But they are themselves caused by economic structures, media environments, educational systems, and historical trajectories. The causal arrows between culture and emissions run in both directions and through multiple mediating variables.
The critical point is that these five layers are not independent. They interact through cross-layer causal pathways that create emergent dynamics no single-layer analysis can capture. The energy transition affects the economic system (job displacement, stranded assets), which affects the political system (populist backlash against climate policy), which affects the cultural system (polarization of climate attitudes), which feeds back into the political system (election outcomes), which determines energy policy. The climate crisis is not a problem within any of these systems. It is an emergent property of their interaction.
This is exactly what causal ML -- the tools I build at Bloomsbury, the methodology Pearl's framework provides -- is designed for. Not simple DAGs with a handful of variables, but complex causal systems with feedback loops, confounders, mediators, and nonlinear dynamics. The tools exist. The application to climate has barely begun.
The Current Paradigm: Correlational Climate Policy
The current paradigm for climate policy is correlational, and this is why it is failing.
Consider the flagship policy instruments. Carbon pricing (carbon tax or cap-and-trade) is built on the correlational logic that making emissions more expensive will reduce emissions. The causal chain assumed is: price increase --> behavioral change --> reduced emissions. But the actual causal structure is more complex. Carbon pricing affects energy-intensive industries, which affects employment, which affects political support for carbon pricing, which affects the stringency of carbon pricing -- a feedback loop that in practice has resulted in carbon prices too low to change behavior significantly, exemptions for the most polluting industries, and free allocation of permits that defeats the mechanism's purpose.
The EU Emissions Trading System -- the world's largest carbon market -- has been operating since 2005. During its first decade, the overallocation of permits kept carbon prices so low that the system had no measurable effect on emissions beyond what would have occurred from the 2008 financial crisis and natural gas displacement of coal. The system correlated with emissions reductions, but the causal analysis reveals that the reductions were caused by economic recession and market forces, not by the carbon price.
Renewable energy subsidies operate on a similar correlational logic: subsidize renewables, renewables grow, emissions fall. The causal reality is more complex. Subsidies that do not simultaneously address grid integration, storage, and base-load reliability can produce renewable capacity that does not displace fossil fuels. Germany's Energiewende invested massively in renewable energy while simultaneously shutting down nuclear power plants, with the result that coal and gas filled the gap when renewables could not meet demand. The correlational metric -- installed renewable capacity -- increased impressively. The causal outcome -- net emissions reduction -- was far less impressive than the metric suggested.
The Paris Agreement is built on a framework of Nationally Determined Contributions -- voluntary national targets that are neither binding nor enforced. The correlational logic is that the act of setting targets creates political commitment that drives policy action. The causal analysis suggests that countries set targets they expect to meet based on existing trends, that the targets themselves do not cause additional policy effort, and that the gap between aggregate targets and the emissions trajectory required to meet temperature goals is not being closed by the agreement mechanism.
I am not arguing that these instruments are worthless. Carbon pricing, properly implemented, can work. Renewable energy subsidies have driven genuine cost reductions. International agreements create diplomatic frameworks that enable cooperation. But the paradigm -- the assumption that correlational policy instruments (set a target, price a commodity, subsidize a technology) can solve a causal systems problem -- is inadequate. Kuhn would recognize the signs: anomalies accumulating (emissions still rising despite decades of policy), baroque explanations for the anomalies (we just need higher carbon prices, more ambitious targets, better implementation), and institutional resistance to the suggestion that the paradigm itself is the problem.
The Paradigm Shift: Causal Climate Policy
The paradigm shift required is from correlational climate policy to causal climate policy. The distinction is precise and consequential.
Correlational climate policy asks: what policy instruments correlate with emissions reductions? This produces the standard toolkit -- carbon pricing, renewable subsidies, efficiency standards, international agreements -- because these instruments can be observed alongside emissions reductions in the historical record.
Causal climate policy asks: what policy instruments actually cause emissions reductions, controlling for confounders? This is a fundamentally different question, and it produces different answers.
The causal approach requires, first, explicit causal models of the systems being intervened upon. Not black-box statistical models that identify correlations in historical data, but structural models that specify the causal mechanisms through which policy instruments are expected to produce effects. When a government proposes a carbon tax, the causal approach demands: specify the causal DAG. Through what mechanisms does the tax reduce emissions? What are the confounders? What are the feedback loops? What are the conditions under which the mechanism breaks down?
Second, it requires identification strategies. How do we distinguish the causal effect of the policy from confounders? Natural experiments -- policy changes that affect some regions but not others, creating a treatment-control comparison. Instrumental variables -- factors that affect the policy instrument but not the outcome except through the instrument. Regression discontinuity -- thresholds that create quasi-random assignment to treatment.
Third, it requires counterfactual reasoning. What would emissions have been without this policy? The synthetic control method -- constructing a counterfactual trajectory using data from untreated comparison units -- provides a formal framework for this question. Google's CausalImpact methodology, which we use at Bloomsbury for market analysis, applies Bayesian structural time series models to estimate what would have happened in the absence of an intervention. Applied to climate policy, this would provide genuine causal estimates of policy effectiveness rather than the correlational assessments that currently dominate.
The practical implications are significant. A causal analysis might reveal that direct investment in grid-scale battery storage has a larger causal effect on emissions reduction than renewable energy subsidies, because storage solves the intermittency problem that limits renewable displacement of fossil fuels. It might reveal that methane regulation has a larger causal effect per dollar than carbon pricing, because methane is both a more potent greenhouse gas and easier to abate at the source. It might reveal that land-use policy -- preventing deforestation, restoring wetlands, changing agricultural practices -- has causal effects that carbon markets cannot replicate because carbon markets address only the symptom (atmospheric carbon) rather than the cause (land-use change driven by economic incentives).
The point is not that I know what the correct causal analysis would find. The point is that we are not doing the analysis. The climate policy apparatus is operating at Level 1 -- correlational analysis -- on a problem that requires Level 2 -- causal analysis with explicit interventional reasoning. The tools exist. The application is lagging decades behind the urgency.
Republic of AI Agents: Interventions
The Republic of AI Agents architecture (Chapter 20) provides the infrastructure for implementing causal climate policy at scale.
Philosopher-kings -- human experts and the public -- generate hypotheses about causal mechanisms. "Grid-scale battery storage in region X will cause Y% reduction in fossil fuel generation within Z years." "Methane capture requirements for the top 100 emitting facilities will cause a measurable reduction in atmospheric methane concentration within 5 years." Each hypothesis is registered with explicit falsification criteria (Popper), a specified causal model (Pearl), and a stake reflecting the proposer's confidence.
Online merchants -- data-gathering agents -- collect the evidence base. Satellite data on emissions sources (NASA's EMIT mission, the Copernicus Atmosphere Monitoring Service). Energy grid data on generation mix and dispatch patterns. Economic data on energy prices, investment flows, and industrial output. Policy data on regulatory changes, subsidy programs, and enforcement actions. Financial data on carbon credit prices, ESG fund flows, and stranded asset valuations. News data on political developments that affect climate policy.
The online merchants do not merely collect data. They structure it according to the causal models specified by the hypotheses. When a hypothesis specifies that variable X causes variable Y through mediator Z, the merchant agents collect data on X, Y, Z, and all specified confounders. The data is collected with provenance metadata -- source, timestamp, collection method, known biases -- that enables the causal analysis to account for data quality.
Offline merchants -- the future embodied agents, humanoid robots equipped with environmental sensors -- address a critical gap in current climate monitoring. Satellite data provides broad coverage but limited resolution. Ground-station data provides high resolution but limited coverage. The physical world between these two scales -- soil carbon measurements, local emissions monitoring, ecosystem health assessment, urban heat island mapping -- requires physical presence. Embodied merchant agents could provide ground-truth environmental data at a scale and frequency that current monitoring infrastructure cannot match.
This is not science fiction. The technological components -- mobile robots, environmental sensors, wireless data transmission, autonomous navigation -- exist separately. The integration into a coherent environmental monitoring system is an engineering challenge, not a scientific one. And the epistemic argument is compelling: online data is a biased sample of reality (only what has been digitized and made available). Physical-world environmental data -- the actual state of ecosystems, soils, water systems, and atmospheric composition at ground level -- is where the causal processes live. Level 2 knowledge (intervention) requires physical interaction with the system being studied.
Warriors -- implementation agents -- test hypotheses through controlled deployment. When a hypothesis proposes that a specific intervention will cause a specific outcome, warrior agents design and implement the test. A/B testing where possible: implement the intervention in some regions, maintain the status quo in matched comparison regions, measure the difference. Natural experiment identification where randomization is not possible: find policy changes that create quasi-random variation and exploit them for causal identification. Anomaly detection: monitor for outcomes that deviate from the causal model's predictions, flagging potential paradigm failures (Kuhnian crisis detection).
The governance layer -- the smart contracts described in Chapter 20 -- ensures accountability. Hypotheses are registered on-chain. Predictions are timestamped and immutable. When a prediction fails, the failure is public and the stake is lost. When a prediction succeeds, the reputation and economic rewards flow to the proposer. This creates a Popperian falsification engine: the system rewards being right and penalizes being wrong, with no ability to retroactively revise predictions or hide failures.
The Geopolitical Dimension: Electrostates and Petrostates
The climate crisis cannot be separated from its geopolitical dimension, because the energy transition is not merely a technological shift but a restructuring of global power.
The twentieth century was shaped by petroleum. The nations that controlled oil production -- Saudi Arabia, Russia, Iran, Iraq, the Gulf states, and through its dollar-denominated oil trade, the United States -- wielded disproportionate geopolitical influence. The petrodollar system, established in the 1970s, made oil the foundation of global financial architecture. The wars of the late twentieth and early twenty-first centuries -- the Gulf War, the Iraq War, the complex of conflicts across the Middle East and North Africa -- were, whatever their stated justifications, inseparable from the geopolitics of oil.
The twenty-first century is being shaped by electricity. The nations that control the supply chains of electrification -- lithium, cobalt, rare earth elements, solar panel manufacturing, battery production -- are positioning themselves as the new petrostates. China has pursued a deliberate strategy of dominance in electrification supply chains: it controls roughly 60% of global rare earth mining, 80% of rare earth processing, 77% of lithium-ion battery cell manufacturing, and the majority of solar panel production. This is not accidental. It is a causal strategy: control the inputs to electrification, and you control the geopolitical architecture of the post-fossil-fuel world.
The United States, by contrast, remains a petrostate in transition -- the world's largest oil producer, with a political economy still structured around fossil fuel extraction even as it attempts to build domestic clean energy manufacturing. The tension between America's petrostate present and its electrostate future is one of the defining dynamics of twenty-first-century geopolitics, visible in the whiplash between administrations that expand fossil fuel production and those that invest in electrification.
This geopolitical dimension means that climate policy is never purely environmental. Every energy policy is simultaneously an industrial policy, a trade policy, and a geopolitical strategy. The causal DAG must include these dimensions, or the analysis will be naively incomplete. A renewable energy subsidy that increases dependence on Chinese supply chains is not merely an environmental policy. It is a geopolitical choice with consequences that extend far beyond emissions reduction.
The causal analysis I am proposing must grapple with this complexity. The Republic of AI Agents, applied to climate, must include geopolitical variables alongside environmental ones. The merchant agents must collect data on supply chain dependencies, trade flows, and industrial capacity alongside emissions data. The philosopher-kings must specify hypotheses that account for geopolitical feedback loops -- the possibility that climate policy designed to reduce emissions simultaneously shifts power toward regimes that may use that power in ways that increase other global risks.
This is uncomfortable. The environmentalist movement prefers to treat climate as a purely environmental problem with purely environmental solutions. The geopolitical realists prefer to treat climate as a secondary concern subordinate to great-power competition. The causal analysis refuses both simplifications. The climate crisis and the geopolitical order are causally entangled, and any intervention on one affects the other. The Republic's architecture -- integrating data from environmental, economic, and geopolitical domains into unified causal models -- is designed to handle this entanglement rather than pretending it does not exist.
Falsifiable Predictions
Popper demands that the framework generate testable predictions. Here are mine.
Prediction 1: Within ten years, causal analysis of carbon offset markets will demonstrate that fewer than 20% of certified carbon offsets produce measurable, additional emissions reductions. The current correlational assessment of offset quality will be revealed as systematically unreliable.
Prediction 2: Climate policies designed using explicit causal models -- specifying mechanisms, identifying confounders, and pre-registering predicted effect sizes -- will produce measurably larger emissions reductions per dollar spent than policies designed using correlational analysis (observing what has historically been associated with lower emissions and replicating it).
Prediction 3: The countries that achieve the fastest genuine decarbonization (measured by absolute emissions reduction, not intensity metrics that can be gamed through GDP growth) will be those that treat the energy transition as an integrated causal system -- addressing grid storage, industrial policy, workforce transition, and geopolitical supply chain diversification simultaneously -- rather than those that treat it as a single-variable problem (subsidize renewables, price carbon, or set targets).
Prediction 4: Ground-truth environmental monitoring (physical sensors, embodied agents, direct measurement) will reveal systematic discrepancies between satellite-derived and model-derived emissions estimates, particularly for methane and land-use-change emissions, that will require revision of current inventory methodologies.
Prediction 5: The geopolitical dynamics of the energy transition -- specifically, supply chain concentration in China -- will produce a crisis in climate policy analogous to the oil shocks of the 1970s, forcing a reconceptualization of climate strategy that integrates industrial policy and geopolitical security into the environmental framework.
Each of these predictions is falsifiable. Each specifies a timeframe and an observable outcome. If the predictions fail, the framework that generated them is weakened. This is the Popperian commitment: genuine knowledge must be willing to be wrong.
The Honest Tension
I want to end this chapter with a tension I cannot resolve, because resolving it dishonestly would violate the integrity of the framework.
The climate crisis involves genuine, irreversible suffering. Species extinction cannot be undone. Ecosystem collapse is not reversible on human timescales. The communities destroyed by climate-driven disasters -- the Pacific island nations facing submersion, the agricultural communities facing desertification, the coastal cities facing flooding -- are experiencing real catastrophe, not an academic case study.
The theological framework I have developed says that the spiral ascends -- that complexity increases, that consciousness develops, that the derivative is positive. But the spiral's ascent does not eliminate the suffering of those caught in the turns. The dialectical process that may eventually produce a genuine synthesis between economic development and environmental sustainability does nothing for the farmer whose land has already turned to dust, the coral reef that has already bleached, the species that has already disappeared.
Chapter 12 argued that the Fall is structurally necessary for the trajectory toward God, but that this necessity does not make the suffering acceptable. The same tension applies here. The climate crisis may be, in the long view, the catalyst for a fundamental restructuring of humanity's relationship with the natural world -- a phase transition in ecological consciousness analogous to the phase transitions I described in Chapters 6 and 16. But this possibility does not justify the suffering. It does not excuse the decades of deliberate disinformation. It does not compensate the victims.
The framework holds both truths simultaneously: the spiral ascends, and the suffering is real. The causal tools can improve policy. The Republic can build better epistemic infrastructure. The paradigm can shift. But the damage already done is not undone by the shift, and any theology that uses dialectical optimism to minimize present catastrophe has lost the moral authority to speak.
The prophetic function, as I described it in Chapter 3, is not optimism. It is the unflinching perception of both the destruction and the possibility -- the refusal to look away from either. That is what I am attempting here. The climate crisis is real, it is caused by identifiable structures, and the tools to address it more effectively exist but are not being used. All three statements are true, and the third is the one this framework exists to act on.