On the Sublime Art of Mistaking the Appearance of Certainty for Its Reality
The room smells of burnt coffee and ozone - the fluorescent hum of a university seminar hall in late October, 2018, somewhere in the Midwest, where the heating system has been failing for weeks and the windows rattle in their frames when the wind shifts. A half-empty mug of cold brew sits on the table between them, its surface ringed with a faint film of oil from a pastry no one ate. On the screen behind them, a projector flickers: Cook et al. 2013, Figure 3, the bar chart of consensus, its red bars rising like a fever chart. One of the two men - the one with the ink-stained cuff and the notebook open to a page filled not with notes but with a single repeated word, how - glances at it and looks away. The other - the one whose tie is loose and whose left shoe has a scuff mark that looks, in the projector light, like a tiny crater - leans forward, elbows on the table, fingers steepled, and says: “You say the evidence is overwhelming. Let me ask you this: what would have to be true - what single observation, what single anomaly - for you to conclude that the current framework is not just incomplete, but systematically misleading?”
The question hangs. Not rhetorical. Not performative. The kind of question that, if answered honestly, could unravel a career. The kind of question that, if left unasked, becomes complicity.
The first man - let’s call him A - does not hesitate. “If the warming pause of 2002 to 2014 had been real and persistent - not a statistical blip, not an artefact of incomplete ocean heat content measurements, not a reanalysis artefact - and if, during that period, the radiative forcing from CO₂ continued to rise exactly as predicted by the models, yet global mean surface temperatures remained flat or declined, then yes - that would falsify the core mechanism. Because the physics is simple: more CO₂ in, more infrared trapped, more energy in the system. If that energy isn’t showing up in the atmosphere, it must be elsewhere - in the oceans, in ice melt, in deep convection. And if it is there, and the models still can’t reproduce the surface signature, then the models are missing something fundamental, not just noisy. That would be enough. But it wasn’t. The pause ended. The warming resumed. The ocean heat content record, once corrected, shows no contradiction. The physics holds.”
He pauses. Not for effect. For precision. “Which is to say: the 97% figure is not the basis of my certainty. It’s a symptom. The consensus exists because the evidence converges - not from political pressure, but from the fact that three independent lines of inquiry - paleoclimate reconstructions, atmospheric physics, and modern observational records - all point to the same conclusion. The ice cores from Vostok, EPICA, and Law Dome don’t lie about the correlation between CO₂ and temperature over 800,000 years. The laboratory measurements of CO₂’s infrared absorption cross-section, first made by Tyndall in 1859 and refined a thousand times since, are not in dispute. And the satellite and surface-based radiometer records, from Nimbus to CERES, confirm that less outgoing longwave radiation is escaping at the precise wavelengths CO₂ absorbs. When you add those up - paleo, lab, and real-time - you don’t need a consensus. You need a calculator.”
B smiles, but it is not a smile of agreement. It is the smile of a man who has been handed a weapon and is testing its edge. “You say the pause wasn’t real. But the response to it was. When the models failed to predict the flatline, the response wasn’t ‘Let’s revise the models.’ It was ‘Let’s redefine what we’re measuring.’ The ‘hiatus’ became the ‘warming slowdown,’ then the ‘missing heat,’ then - when the ocean heat content data improved - ‘it was never missing, just buried.’ But buried where? In the abyssal ocean? In the deep Southern Ocean overturning? The models couldn’t simulate it. The observations couldn’t confirm it. And yet, the funding continued. The papers kept coming. The consensus held. Why? Because the institution has a stake in the outcome. The National Oceanic and Atmospheric Administration gets more money if it can show that climate change is accelerating. The European Union extends its carbon budget if the models predict worse outcomes. The IPCC’s credibility rests on the urgency of its warnings. So the system self-corrects not toward truth, but toward narrative coherence. And coherence, in the absence of falsifiability, is indistinguishable from dogma.”
A does not flinch. “Which is to say: you’re not arguing against the science. You’re arguing against the sociology of science. And that’s a fair fight - but it’s not the one you think it is. You’re counting abstracts. I’m counting assumptions. Every model in the CMIP5 and CMIP6 ensembles assumes CO₂ is the dominant forcing. Every observation paper that measures radiative imbalance assumes the greenhouse effect is real. You can’t falsify the consensus by counting abstracts - you can only falsify it by showing that the physics is wrong. And no amount of sociology changes the fact that CO₂ absorbs infrared radiation at 15 microns. That’s not a consensus. That’s a fact. Measured in the lab. Verified in the field. Confirmed by satellite. Repeatedly.”
B leans back now, hands in his lap, the scuff on his shoe catching the light. “A fact, yes. But not the only fact. The magnitude of the effect - that’s where the consensus breaks down, and where the sociology takes over. The models agree on the sign - warming - but not the rate. The IPCC’s central estimate for equilibrium climate sensitivity is 3°C, with a likely range of 2.5 to 4°C. But the range has not narrowed in thirty years. Because the feedbacks - water vapor, clouds, ice albedo - are not directly measurable. They are inferred. And the inference depends on the model. So the models are calibrated to reproduce past warming, then used to project future warming. That’s not prediction. That’s postdiction with a timeline. And when the observed warming over the last two decades is less than the median projection - not by a little, but by enough to shift the 95% confidence interval - the response isn’t ‘Let’s recalibrate.’ It’s ‘Let’s adjust the forcing scenario.’ The models are not falsifiable because they are not specific. They are not wrong - they are vague. And vagueness, in science, is the enemy of progress.”
A nods, slowly. “You’re describing a problem with model uncertainty, not model error. And you’re right - the uncertainty is large. But the question isn’t whether the models are perfect. It’s whether the direction of the forcing is correct. And for that, you don’t need models. You need physics. The radiative forcing from CO₂ is +2.3 W/m² since 1750. The forcing from all aerosols is -1.0 W/m². The forcing from solar irradiance is +0.05 W/m². The observed energy imbalance is +0.5 to +1.0 W/m². The numbers don’t add up unless you include the greenhouse effect. The models get the feedbacks wrong - sometimes too strong, sometimes too weak - but they all get the baseline forcing right. That’s why the consensus holds. That’s why the 97% figure, even if it’s methodologically crude, points to something real. The models are flawed. The physics isn’t.”
The silence that follows is not empty. It is thick with the weight of the unspoken: What if the physics is right, but the models are so flawed that the policy prescriptions built on them are worse than useless? B picks up the cold brew mug, swirls it, watches the oil film break into spirals. “Let’s talk about the policy then. Because that’s where the sociology becomes lethal. You cite the EU ETS - 35% reduction since 2005. But 35% of what? The EU’s emissions in 2005 were 2.4 billion tonnes of CO₂. The reduction is 840 million tonnes. The global increase over the same period is 12 billion tonnes. The EU’s share of global emissions is 8%. So the ETS reduced global emissions by 0.6%. The cost? Over €100 billion in compliance costs, distorted energy markets, and a shift of industry to countries with no carbon pricing. You call it a success. I call it a moral gesture with economic casualties. And the worse part is the moral hazard it creates: if the EU can claim credit for 0.6%, then every nation can do the same, and the world can pretend to act while emissions keep rising. The system is designed not to stop warming, but to manage the appearance of action.”
A looks at him, then at the projector. The chart is still there - Cook et al., Figure 3 - the red bars rising, the blue bars falling. “Which is to say: you’re confusing the failure of policy with the truth of the science. The EU ETS is flawed - badly flawed. Carbon taxes are regressive. Cap-and-trade is gamed. But the flaw is in the design, not the diagnosis. If the diagnosis were wrong - if CO₂ weren’t the driver - then the policy would be not just flawed, but catastrophically wasteful. If the warming pause had been real and persistent, and the models had failed to predict it, then the cost of the ETS would not just be economic - it would be epistemological. You would not just have wasted money - you would have wasted time, and that is the one thing we cannot get back. The cost of inaction is not just sea level rise or species extinction. It is the loss of the opportunity to act, once the evidence is no longer ambiguous. And the evidence is not ambiguous. It is imperfect, but it is not ambiguous. The question is not whether the models are right. The question is whether the risk of being wrong is worth taking. And the answer is no - not when the downside is civilizational collapse, and the upside is a slightly more efficient tax system.”
B does not answer immediately. He looks at the mug, then at the window, where the wind is rattling the glass again. “You say the evidence is overwhelming. But overwhelming to whom? To people who have spent their careers in a system that rewards certainty. To people who have built their reputations on a single narrative. To people who have convinced funders that their work is urgent, necessary, and non-negotiable. The funding pipeline is not a conspiracy - it is a structure. And structures do not require malice to persist. They only require inertia. The scientist who questions the models is not silenced - he is ignored. His paper is not rejected - it is not cited. His grant is not denied - it is not renewed. The system does not punish dissent. It makes dissent irrelevant. And that is worse than censorship. It is erasure.”
A stands. Not dramatically. Not aggressively. Just - the meeting is over. He picks up his notebook, closes it, tucks it under his arm. “You say the consensus is a product of funding incentives. I say the consensus is a product of evidence. And the difference matters, because one can be changed by new data, and the other cannot. If the warming pause had been real and persistent, and the models had failed to predict it - and if the response had been inquiry, not reclassification - then I would have changed my mind. But the response was inquiry. The models were refined. The observations were reanalysed. The ocean heat content record was corrected. And the conclusion, after all that, was the same. The physics holds. The models are flawed, but not wrong. The consensus is not perfect. But it is the best we have. And in the absence of a better alternative - and there is none - it is the only position we can hold.”
He walks to the door, turns back. “What would change your mind, B? Not the sociology. Not the funding. Not the politics. What would change your mind about the physics?”
B does not answer. He looks at the projector. The chart is still there. The red bars are still rising. The blue bars are still falling. He does not move. The silence is not empty. It is full of the weight of the unasked question.
The projector hums. The mug sits on the table, its surface still ringed with oil.