There’s a very old joke I’d like to share with you, and it goes something like this:

A man is found repeatedly firing an elephant gun in the local park. “What are you doing that for?” he is asked. “It’s to scare away the elephants”, he opines. “But there are no elephants in this park”, comes the reply. “See,” he smiles, “it’s working already”.

Now, if you were to believe the man with the elephant gun, what you would be dealing with here is a perfect example of what is known as the Preparedness Paradox (sometimes ‘Preventative Paradox’). It is a paradox in which the removal of a problem is misinterpreted as evidence that the problem never existed in the first place. It’s an argument that has been made many times before in real-life situations. Take, for example, the Y2K or Millennium Bug. Widespread faults in date-handling code were thought to pose a significant threat, leading to a global IT initiative to find and correct the bugs in order to avert disaster. As it happens, the new millennium dawned with very little drama, causing many to declare that the whole thing had been hyped. Not so, said the IT managers. According to them, what we had witnessed was a classic case of the Preparedness Paradox. Similar arguments have been made regarding the necessity for lockdowns during the Covid-19 pandemic. A lockdown that avoids a surge in numbers invites the interpretation that the threat analysis had been flawed all along.

I don’t want to get drawn into a Y2K or Covid debate here, other than to point out that the reason why the elephant gun joke is supposed to be funny has nothing to do with the Preparedness Paradox. Instead, the humour stems from the alternative interpretation, i.e. there never were any elephants to scare away. So, with that interpretation in mind, I want instead to reflect upon RCP8.5 and the IPCC’s recent withdrawal of the scenario as a plausible worst-case upon which to formulate climate policy. Could it be that those who see this as a belated admission that it was always an implausible scenario are actually guilty of falling for the Preparedness Paradox? Did RCP8.5 really do a great job of scaring off elephants, or were there never any elephants to begin with?

The Case for the Elephant Gun

As I’ve just explained, decisive action has the disadvantage that it completely erases the visible evidence that it was ever needed. Many mainstream climate scientists are adamant that this is exactly what has happened in the case of RCP8.5 and that its recent removal belongs firmly in the Preparedness Paradox category. The premise is that when RCP8.5 was conceived in the late 2000s, global emissions were tracking aggressively upward. The scenario was never intended as a ‘business as usual’ scenario but a plausible stress-test deliberately used to map out the severe risks of a completely unmitigated fossil-fuel-heavy world. Supporters argue that the alarm raised by modelling RCP8.5 directly led to global treaties (like the Paris Agreement), national carbon mandates, and massive state subsidies for green energy. In this view, the measures were necessary and the resultant actions have flattened the curve, thereby causing the demise of RCP8.5’s plausibility. This is seen as a triumph of proactive global policy. It is a textbook Preparedness Paradox: the potential disaster described by RCP8.5 disappeared because the warning was successfully heeded.

The Case for Non-Existent Elephants

Critics, such as policy analyst Dr Roger Pielke Jr, argue that the flattening of the emissions curve is irrelevant because RCP8.5 was always built on highly unrealistic, extreme assumptions. Yes, it was a stress-test but it was never a credible one. As Justin Ritchie and Hadi Dowlatabadi demonstrated in 2017, for RCP8.5 to occur, the world would have to undergo a massive, unprecedented return to coal. The model required a five to eight-fold expansion of global coal use by 2100, involving the liquidating and burning of coal reserves that do not physically exist in extractable forms. In addition, it required the somewhat bizarre economic assumption that when oil ran out humanity would convert coal into liquid fuel for cars rather than using newer and cheaper alternatives. It is also worth noting that global coal use actually peaked around 2013 (long before major global climate policies like the Paris Agreement could even take effect), and so the RCP8.5 trajectory was impossible from the start. Whilst the downturn in coal use has been reversed by recent events, there is still not the slightest indication that the growth in usage necessary for RCP8.5 to regain any credibility could possibly materialise. Therefore, claiming that the RCP8.5 scenario had been averted by the measures it inspired is akin to the firing of an elephant gun in a public park many miles from the Serengeti.

The Greatest Pity

The debate boils down to who is pushing the historically accurate narrative. If the extreme scenario was an artificial, fundamentally flawed baseline then the policies justified by it were always built on what would ultimately prove to be a modelling error. Furthermore, it was an error that carried with it huge costs that could never be justified. If, on the other hand, the scenario was a perfectly possible trajectory that was actively averted by decades of climate activism and policy shifts (the Preparedness Paradox), then the logic holds up and those very real elephants have now fled.

Which side you take in this argument depends upon how one chooses to read the room, and so I’ll leave it to my readers to form their own opinions. However, I have to declare that my sceptical orc genes predispose me to take the side of those who believe that RCP8.5 was never plausible as a stress-test in the first place. Consequently, no matter how much one might argue that the curve has been flattened, such flattening can’t have altered the status of RCP8.5.

The greatest shame, however, is that RCP8.5 was not only an implausible stress-test from the outset but was also mis-sold by many as a business-as-usual scenario, fuelling endless doom-laden predictions that have led directly to environmental security theatres such as the UK’s Net Zero debacle. In engineering, an extreme boundary case is used to find a system’s breaking point. It is never meant to represent the expectation. By branding RCP8.5 as “business-as-usual” in thousands of peer-reviewed papers, climate scientists and activists perfected the measurement of a purely fictional environment, leading to a massive misallocation of academic effort and wholly inappropriate regulatory initiatives.

So, no I don’t believe the retirement of RCP8.5 is a story of policy triumph; it is a story of belated scientific adjustment. Yes, a worst-case scenario has its purpose in risk management insofar as it creates the boundaries within which mitigation policies have to operate. However, to achieve that purpose, the posited worst-case has to have credibility, and I don’t believe that was ever the case in this instance. The implausibility of RCP8.5 should have been admitted long ago. The scenario didn’t become implausible because the world changed; the world simply continued to operate within physical limits that the scenario had always chosen to ignore. Sooner or later, reality bites and I don’t think that RCP8.5’s demise has come a moment too soon.

20 Comments

  1. RCP8.5 was also based on a projected world population of 12 billion (never a number advanced by any serious demographer), continuous annual growth in world GDP, and directly linked increases in energy requirements

    Liked by 2 people

  2. Those consumed by climate fear and those rewarded by promoting climate fear and an energy transition, promoted RCP8.5. Investigative climate reporting by mainstream journalists and even scientists might destroy their careers, so it didn’t happen. Judith Curry and Richard Pielke, were among the few exceptions and deserve recognition for not caving in to the hysteria and malpractice that characterized academia and liberal politics, especially since they were being destroyed by members of their own liberal faction. For us traditional environmentalist, center left voters, this has been a frustrating and sorry time.

    Liked by 3 people

  4. Zeke Hausfather tries valiantly to rescue something from the ashes:

    https://skepticalscience.com/death-of-rcp85.html

    Meanwhile, it is indeed worth noting that the UK’s own Climate Change Committee has not been shy about invoking RCP 8.5:

    “How much more climate change is inevitable for the UK?”

    https://www.theccc.org.uk/2020/04/21/how-much-more-climate-change-is-inevitable-for-the-uk/

    This column is largely based on the RCP8.5 scenario, which represents a considerable re-carbonisation of the global economy. Emissions would be required to grow to about three times their current levels by 2100 to achieve a forcing level of 8.5Wm-2 by 2100.

    Even the much-hyped report from as recently as 20th May 2026 invokes RCP 8.5:

    https://www.theccc.org.uk/publication/a-well-adapted-uk/

    Description: Relative to 1981–2000, average sea level has been steadily increasing since 1970 in the UK. Sea level is projected to rise substantially by the mid to late century under global warming scenarios.
    Source: Permanent Service for Mean Sea Level (no date) Station 202: Newlyn – Tide‑gauge data and site information; Met Office Hadley Centre (2018) UKCP18 Marine Report.
    Notes: (1) All data are for the Newlyn area, with annual data expressed relative to 1981-2000. (2) The long-term trend is a loess smooth with a 50-year span. (2) ‘Average scenario’ is the change in average sea level expected based on a medium-emissions scenario (RCP4.5), using the mid‑range estimate (67th percentile). (3) ‘Extreme scenario’ is the change in average sea level expected based on a high-emissions scenario (RCP8.5), using the upper-end estimate (95th percentile).

    Liked by 1 person

  5. From March 2024:

    “Climate scenario analysis guidance for departments, ALBs and public bodies HM Treasury Financial Reporting Team”

    https://assets.publishing.service.gov.uk/media/66c850688b9aa088744f4335/FRAB_152__02__Appendix_2_-_Advice_on_Climate_Scenario_Analysis_Guidance.pdf

    Current advice for climate change adaptation is to consider impacts associated with 4oC temperature rise by 2100 as a way of capturing a worse case scenario. This is equivalent to RCP6.0 or RCP 8.5.

    Like

  6. Chris,

    RCP8.5 was also based on a projected world population of 12 billion (never a number advanced by any serious demographer), continuous annual growth in world GDP, and directly linked increases in energy requirements

    Indeed. I should have mentioned that.

    Like

  7. The “Preparedness Paradox” using RCP8.5 was much in evidence in the run-up to and during COP21 Paris in December 2015. But there was another element that I am unable to name.

    In the run-up to COP21, nearly every country submitted an INDC (Intended Nationally Determined Contributions). These were primarily to show emissions projections with and without policy through to 2030. Not the UK, as the EU did a joint INDC. Several organisations put forward projections on how the aggregate of the INDCs would play out through to 2100. One such group was Climate Interactive. Using RCP8.5 as a baseline, they projected that the effect of the INDCs would be to constrain 2100 warming to just 3.5oC. Just a degree less warming.

    The UNFCCC, using a different method, claimed the INDCs would constrain warming to 2.7 °C.

    As Climate Interactive C-Roads software allowed downloads, I had a little play with the data. C-Roads had extracted data from the RCP database. This has separate model outputs for each of the four RCPs and for each greenhouse gas. To keep things simple, RCP divides the world into five, rather odd, regions. C-Roads then split out the EU and about 20 major countries from these regions. I downloaded the RCP8.5 CO2 emissions data along with the UN population projections from C-Roads. The CO2 data is expressed in grams of carbon, instead of tonnes of CO2. After a bit of data manipulation, I produced the following chart

    In the actual data to 2012, CO2 emissions per capita have fallen since 1980 in both the USA and the EU27. This is due to various factors, but in general advanced economies grow both through services and through high-value manufacturing. There is no economic reason for either to rise to well beyond pre-1973 peaks. The projection in China is simply crazy. It is this way in part because China’s population is projected to fall to 1 billion by 2100. Indeed, all but one country confirms the “Preparedness Paradox”. That country is India.

    The projected CO2 emissions for India were so low that even the 2015 estimate was around 30% below the estimate. The figures for Africa (excluding South Africa) were even worse. Per capita CO2 emissions will barely increase. In other words, more professional BAU scenarios for India and Africa would generate the opposite of the elephant gun approach.

    So, whilst you would expect the inputting INDCs into the Climate Interactive C-Roads database to correct the crazy forecasts, these are partly offset by upward revisions elsewhere. Whether even a 50% increase in CO2 emissions in 85 years is realistic, even with emission reduction policies in some countries, is another story. As is my extrapolation from AR6 WG1 SPM table SPM.2 that constant global CO2 emissions 2020-2100 would produce ~3.5 °C of warming in 2100.

    Liked by 2 people

  8. Mark,

    RCP8.5 may not be receiving any future contracts but that does not mean it is now out of a job. Like the zombie arguments that we sceptics are supposed to indulge in, it will continue serving the cause for many years to come. AI was quick to reassure me that the physical science behind all the papers that were pushing RCP8.5 is still valid. So, I asked it this question:

    The physical science was never the issue. It is the damage caused by the flawed socioeconomic framing and the incorrect risk assessments. What can be done about redressing that error.

    This is what it had to say:

    Redressing the systemic damage caused by decades of treating a low-probability, extreme scenario as a guaranteed “business-as-usual” trajectory requires a targeted, structural overhaul across three main domains: corporate finance, infrastructure planning, and academic governance.

    Because the error was socioeconomic and operational, rather than physical, the remedy must focus on recalibrating how probability and risk are handled.

    1. Re-Audit Corporate and Banking Climate Stress Tests

    The Network for Greening the Financial System (NGFS) and central banks globally embedded RCP8.5 (and SSP5-8.5) deep into regulatory frameworks. This forced private banks and corporations to model extreme, implausible outcomes as their baseline financial reality.

    • Decouple “Baseline” from “Stress Test”: Financial regulators must force institutions to use realistic current-policy pathways (like the updated CMIP7 baseline models capping warming closer to 2.5–3C for standard asset pricing and financial forecasting.
    • Recalibrate Risk Capital Requirements: Forcing banks to hold excessive capital reserves against a fundamentally impossible “return-to-coal” scenario creates massive capital inefficiencies. Risk models must reallocate that capital based on highly probable physical changes combined with transition risks.
    • Isolate High-End Feedback Loops: Instead of running an impossible socioeconomic scenario, stress tests should evaluate plausible emissions tracks combined with extreme climate sensitivity (e.g., severe permafrost thaw) to measure true tail-risk.

    2. Overhaul Local Government and Infrastructure Planning

    Treating an extreme upper bound as a statistical certainty has caused a massive misallocation of public funds, leading to either over-engineered infrastructure or legal dead-ends.

    • Shift to Adaptive Pathways Engineering: Instead of building municipal infrastructure to withstand an immediate, implausible 4.5–5C world by 2100, engineers must design modular infrastructure. This means building for a realistic 2.5C trajectory while designing retrofittable footprints that can scale up only if the worst physical outcomes begin to materialize.
    • Correct Real Estate and Insurance Devaluation: In many regions, insurance algorithms and coastal real estate risk models have aggressively penalised property values based entirely on the assumption that RCP8.5 sea-level rise projections were a 100% certainty. Insurance regulators must force providers to discount these models and anchor premium adjustments to realistic baselines.

    3. Implement Disclaimers on Retrospective Literature

    With over 100,000 papers utilizing RCP8.5 improperly, the distortion in the academic record cannot be fixed by simply writing new papers. The existing catalog must be structurally contextualized.

    • Retroactive Journal Corrigenda: Major scientific journals should issue systematic, standardized addenda or metadata tags for all past papers that used the phrase “business-as-usual” alongside RCP8.5. The tag should clarify that the paper models a “low-probability, high-impact worst-case boundary,” not a central tendency.
    • Ban the Metric in Policy Summaries: In synthesis documents meant for prime ministers, governors, and CEOs, the use of original RCP8.5 economic damage functions must be explicitly barred unless accompanied by a mandatory probability discount factor.

    4. Depoliticise and Stabilize the Public Debate

    The socioeconomic misframing provided fuel for two damaging cultural extremes: severe climate anxiety/doomism on one side (by making mitigation look impossibly difficult), and complete climate skepticism on the other (when real-world data inevitably diverged from exaggerated models).

    • Anchor Public Communications to “Enacted Policy”: Media relations from major scientific bodies must lead with the delta between where we are actually heading based on real-world actions versus where we want to go.
    • Emphasize the Feasibility of Mitigation: Moving away from the “inevitable catastrophe” framing of RCP8.5 allows public discourse to focus on achievable, marginal economic gains in green tech, reducing the political paralysis caused by hyper-alarmist framing.

    End of response

    I have to say, I don’t see any of the above happening. Also, keep in mind that this is what is required to redress the misinterpretation of RCP8.5 as a business-as-usual scenario. It doesn’t address the more basic problem that it was supposed to represent a low probability/extreme impact risk when in fact the physical and structural impossibilities upon which it was premised had rendered it a zero-probability risk from the outset.

    Liked by 1 person

  10. Pingback: Preparedness Paradox and RCP8.5 | ManicBeancounter

  12. I was once proof-reading a paper on the effect of global warming on some environmental parameter or the other. It only refered to RCP8.5, so I suggested to the authors that they might want to justify that. The reply came back that they used RCP8.5 because it showed the clearest difference from the baseline.

    Liked by 2 people

  13. I asked AI “In how many papers or studies has RCP 8.5 been cited?“

    This was the reply:

    Estimates indicate that RCP 8.5 (and its successor, SSP5-8.5) has been mentioned or utilized in over 105,000 papers on Google Scholar. Its foundational paper alone (RCP 8.5—A scenario of comparatively high greenhouse gas emissions) has garnered over 2,600 citations on SpringerLink. [1, 2, 3]

    Because of its broad scope, this climate pathway has been widely used by researchers, central banks, and governments as a benchmark for severe climate risks. However, due to recent advancements in global decarbonization, major scientific bodies have begun formally retiring the scenario from the basket of plausible futures. [1, 2]

    Liked by 1 person

  14. ” . . . due to recent advancements in global decarbonization, major scientific bodies have begun formally retiring the scenario from the basket of plausible futures.”

    Wrong, global climate change mitigation policies have had zero effect upon the actual global climate and have barely altered the trajectory of coal use, and hence they have had no effect whatsoever upon future projections of climate change. RCP8.5 was always highly implausible.

    https://www.theguardian.com/environment/2025/oct/22/global-use-of-coal-hit-record-high-in-2024

    Liked by 2 people

  15. Mark,

    I think it is only fair to point out that if one believes that RCP8.5 was a plausible worst case, then it is perfectly consistent to argue that it was the flattening of the curve that now renders it implausible. Also, if you believe that the only reason that sceptics ever thought it was implausible is because they had latched onto the idea that it was a business-as-usual scenario, then one can convince oneself that they are now mistaken regarding the real reason for it now becoming implausible. My position is that it was never a credible worst case, and so the flattening of the curve is irrelevant. You can’t render incredible something that was never credible in the first place.

    Liked by 1 person

  16. Jaime,

    I think those who refer to the flattening of the curve are arguing a counterfactual: yes, coal usage is still going up but not by nearly as much as it would have done. My article grants them the benefit of the doubt; not because I think it is true, but because my argument dismisses any supposed flattening as being irrelevant.

    Liked by 1 person

  17. John,

    What is the curve that has been “flattened?” Carbon dioxide emissions? Please provide numbers (e.g. graph, table, etc.) that shows this flattening.

    William DeBoer

    Like

  18. William,

    It depends upon what you mean by flattening. As I suggested to Jaime, a flattened curve doesn’t have to look in the least bit flat, as long as one can claim that it is flatter than the curve upon which RCP8.5 is premised. But as I say, I’m not actually interested in proving this one way or the other. My argument rests upon the claim that RCP8.5 was never credible in the first place and therefore cannot be rendered incredible by any degree of flattening. And when I say ‘never credible’, I mean as a business-as-usual scenario or as a worst-possible scenario.

    Like

  19. Here’s the scenarios specified for use in CMIP6 climate models:

    It shows that SSP5-85 is virually the same as RCP8.5. It also indicates the preference for 5-85 including 8.5 Wm^2 radiative forcing by 2100 and temperature gain of 5 degrees at that same time. Pielke took particular issue with the nearly quadrupling of FF emissions and the tripling of CO2 in the atmosphere.

    It shows that SSP5-85 is virually the same as RCP8.5. It also indicates the

    Liked by 1 person

Leave a comment