From Watts Up With That?

By Charles Rotter

A scenario the modelers built as a stress test became, for fifteen years, the scenario the press treated as your future. It has now been quietly retired. The reckoning has not yet begun.

---

There is an old joke in physics. A dairy farmer hires a theoretical physicist to improve milk production. The physicist disappears for six months, returns with a 200-page report, and announces: I have a solution. It works for spherical cows in a vacuum.

The joke endures because every physicist has met the spherical cow. To make a hard problem tractable, you strip the real world down to a geometry the math can handle. A round, frictionless object floating in nothing is something you can actually compute. A real cow is not. The simplifying assumption is a tool. useful inside the laboratory and the seminar room, useless the moment it walks out the door.

RCP 8.5 was a spherical cow. For fifteen years it walked out the door, into headlines, into government reports, into court filings, into bank stress tests, and into the climate fears of hundreds of millions of people who had no idea what it was. As of April 2026, the international committee that produces the official climate scenarios for the United Nations has quietly walked it back into the barn and shut the door behind it.

The retirement, in the form of a footnote

On April 7, 2026, Geoscientific Model Development published van Vuuren et al. 2026, the design paper for the next round of climate model scenarios, the inputs that will feed the IPCC’s Seventh Assessment Report (AR7) and the climate research of the rest of the decade. Buried inside is the verdict that closes the book on a decade and a half of climate communication: in the authors’ own characterization,

the CMIP6 high emission levels (quantified by SSP5-8.5) have become implausible.

That is the formal obituary, written by the same scientific community that wrote the birth certificate. The new CMIP7 framework eliminates the three extreme scenarios, RCP 8.5, SSP5-8.5, and SSP3-7.0, that have dominated climate research, climate journalism, and climate policy since the early 2010s.

The announcement was met, in the English-language press, with a silence so complete that Roger Pielke Jr., who has been documenting the misuse of these scenarios in the peer-reviewed literature for years, spent the first two weeks of May simply cataloging which mainstream outlets had not yet covered the story. The list was nearly all of them. Coverage finally arrived, more than a month after publication, only when the President of the United States posted about it on social media. The New York Times, the Washington Post, the AP, Bloomberg, and Carbon Brief then all ran stories within forty-eight hours, most of them framed around the President rather than the science, and most repeating the same factual errors about what RCP 8.5 had ever been.

To understand why that matters, you have to understand what RCP 8.5 actually was, and what it was not.

What it was: a high-signal input for a noisy machine

From a constructive discussion elsewhere with a leading climate modeler, a comment that helps us to better understand how we got into the scenario mess. I agree 100% with his comment:

“From a modeling side, most of us really do not care much (or do not know about) the…— The Honest Broker (@RogerPielkeJr) May 21, 2026

The cleanest description of what RCP 8.5 was built for came from an unnamed climate modeler quoted recently by Pielke. Asked why the extreme scenario became so dominant in model intercomparison runs, the modeler explained that most modelers do not pay close attention to the socioeconomic assumptions underlying any given scenario. What they need is an input that drives a climate model. In the modeler’s words:

a higher signal is better from a signal/noise perspective.

At ten to a hundred thousand dollars per model run, a higher-emissions scenario produces a cleaner climate signal you can analyze. Lower scenarios require many more runs to pull the signal out of the natural variability. Choices have to be made.

In other words, RCP 8.5 was useful to physical climate modelers in roughly the same way the spherical cow is useful to a graduate student studying angular momentum. It made the math tractable. It was never meant to be a forecast. It was a stress test, a deliberately extreme number you plug in so you can compare model outputs and study the climate system’s response to a large radiative forcing.

The original 2011 paper that introduced RCP 8.5 (Riahi et al. in Climatic Change) said as much: a baseline at the upper end of the published literature, assuming a roughly sevenfold expansion of global coal use through 2100, a population trajectory at the high end of UN projections, and effectively no technological progress in low-carbon energy. Even at the time, this was understood to be an outlier. Keywan Riahi, the lead author, now tells Seth Borenstein of the Associated Press:

it was never a likely case.

Zeke Hausfather, on a recent panel hosted by Andrew Revkin, conceded essentially the same thing, that the 2011 energy modeling literature had been clear about what RCP 8.5 was: a high-end statistical bookend, not, as he put it,

a particular likely outcome.

If that is what Riahi and Hausfather believe now, and what the literature said in 2011, it raises a question. Why was Riahi’s scenario quietly promoted from “outlier baseline” to “business as usual” in the IPCC’s Fifth and Sixth Assessment Reports, in the U.S. National Climate Assessment, in central bank stress tests, and in fifteen years of climate journalism?

And why did nobody, at any of those institutions, including the modeling community itself, shout?

The cow escapes

The transformation from stress test to forecast happened by gradual institutional drift, not by deliberate decision. The IPCC’s Fifth Assessment Report in 2013 designated RCP 8.5 as the reference scenario. Once a generation of impact studies, vulnerability assessments, and policy papers had been written under that assumption, the IPCC’s Sixth Assessment Report in 2021 inherited a body of literature so saturated with RCP 8.5 that disentangling it was no longer possible. The Sixth Assessment included mealy-mouthed language acknowledging the scenario was unlikely while continuing to feature it prominently throughout.

By that point, RCP 8.5 had become the implicit baseline for almost everything readers were being told about the climate future. It was the source of nearly every “by 2100” headline of the past decade. According to Pielke’s accounting, RCP 8.5 accounted for over half the scenario references in the 2018 U.S. Fourth National Climate Assessment, around sixty percent in the IPCC Special Report on the Ocean and Cryosphere, and roughly a third of all RCP references in the IPCC AR5. By early 2020, peer-reviewed studies invoking it were appearing at a rate of roughly twenty per day. So far in 2026, even after the formal retirement, they are appearing at thirty per day.

This was not an obscure technical artifact. It was the engine of the climate-alarm story for a generation.

It also became the basis of policy. Pielke’s tabulation runs to dozens of pages. National climate impact assessments in the United States, the United Kingdom, Germany, Canada, Australia, Japan, and the Netherlands all use RCP 8.5 or SSP5-8.5 as a reference scenario. The Network for Greening the Financial System framework, used by more than 140 central banks, built its “Hot House World” stress test on it, which then flowed into the bank capital tests run by the European Central Bank, the Bank of England, the U.S. Federal Reserve, and others. The World Bank’s Climate Change Knowledge Portal, which produces the climate inputs for Country Climate and Development Reports in more than 100 nations, defaults to SSP5-8.5 and SSP3-7.0. The pending climate-damages lawsuits against U.S. energy producers, including Multnomah County’s $51 billion claim, are built on damage projections derived from RCP 8.5 modeling.

And during all of this, prominent voices in the climate establishment were not quietly hedging. In a 2021 forum reply published in Issues in Science and Technology, Chris Field, co-chair of IPCC Working Group 2 for the Fifth Assessment, and Marcia McNutt, then president of the National Academy of Sciences, defended the description of RCP 8.5 as a “business-as-usual” pathway as something that, in their words,

remains 100% accurate.

That sentence, by the official position of the scenario community that designs these things, is now false.

Where were the modelers?

The deepest question this retirement raises is not how RCP 8.5 was built. Its construction was always documented in plain sight; anyone willing to read the 2011 Riahi paper could see what it assumed and why. The question is why the people who built it, and the people who used it, sat in silence for fifteen years while it was bastardized.

The modelers were not deceived. They knew what they had made. They knew the difference between an exploratory high-forcing case and a forecast. Hausfather himself, in the same panel discussion cited above, acknowledged that the 2011 energy modeling literature was clear about what RCP 8.5 was. The information was never hidden. It was simply ignored, first by the impact community, then by journalists, then by central bankers, then by federal courts. And the people who knew better said nothing.

Consider the moments at which speaking up would have mattered.

When the IPCC AR5 made RCP 8.5 the reference scenario in 2013, the modeling community had standing to object. They did not. When the U.S. National Climate Assessments of 2014 and 2018 built their headline projections on it, the scenario authors had standing to clarify. They did not. When central banks began constructing stress tests around a “Hot House World” calibrated to RCP 8.5, the scientists whose work was being repurposed for bank capital regulation had standing to push back. They did not. When the IPCC AR6 in 2021 again featured RCP 8.5 throughout its assessment, while including buried language warning that it had become unlikely, the lead authors could have insisted on a clearer correction. They did not.

When Roger Pielke Jr. and Justin Ritchie published their formal diagnosis of scenario misuse in Energy Research & Social Science in 2021, the response from senior figures in the climate establishment was not gratitude for a needed correction. It was, in the case of Field and McNutt, an active defense of the indefensible. Pielke and Ritchie spent the next five years being characterized in mainstream coverage as marginal critics rather than as the people who had been correct all along.

The few who did speak up, Pielke, Ritchie, Matthew Burgess, Steven Koonin, Richard Tol, and a small handful of others, paid the social and professional costs of speaking up. The much larger group that knew, and stayed quiet, did not pay those costs. They are now, in May 2026, beneficiaries of a course correction they did not lead and a quiet retirement that contains no admission of error.

This is not a story about scientific fraud. It is a story about institutional cowardice and the sociology of a discipline whose career incentives, funding pipelines, and journal access were aligned with letting a false framing run unchallenged. The technical scientists who could have stood up at any point during the past decade and said this is not what our scenario means, and made that clear in the public square, not just in cautious footnotes, did not do so. The professional cost of being branded a “denier” was apparently too high; the cost to the public of letting the bastardization continue was paid by someone else.

Pielke writes that science is self-correcting. On a long enough timeline, that is true. But science self-corrects through the work of individuals willing to take the personal risk of contradicting a consensus, not through the passive accumulation of new data inside a silent profession. The reckoning here is not just that RCP 8.5 was wrong. It is that the modeling community knew, and waited, and is now being credited with a correction that arrived only after the costs of denial had become higher than the costs of admission.

The reckoning that hasn’t started

What does it mean, in practice, that the foundation has now been declared unsound?

Every reader of this site has spent fifteen years being told, in increasingly dire language, what the climate of 2100 will look like. A great deal of that language was downstream of RCP 8.5. The 2026 retirement does not erase those projections from the journalistic record, but it does change how a thoughtful reader should weigh them on the way back through the archive. The “by 2100, coral reefs collapse” story; the “by 2100, the U.S. economy shrinks ten percent” story; the “by 2100, ice-free Arctic, climate refugees, mass extinctions, civilization-threatening heat” stories, a substantial fraction of these were the spherical cow’s footprints across the cultural landscape.

The new CMIP7 framework brings the high end of the scenario set down considerably. Pielke’s apples-to-apples comparison using the FaIR climate emulator finds the new HIGH scenario produces about 0.9°C less warming than SSP5-8.5 by 2081–2100. Compared to the IPCC AR6’s projection of SSP5-8.5, the gap widens to about 1.4°C. The implausibility of the upper-end legacy scenarios, in his phrase, means

the future is not what it used to be.

But the cleanup is not complete, and on this point the WUWT reader should pay close attention to where Pielke and the official scenario community part company. Three observations matter.

First, the new HIGH scenario is itself still implausible by the standards Pielke and his coauthors have proposed in the peer-reviewed literature. It inherits the SSP3 storyline, including a 2100 global population the IIASA recently revised upward to 14.5 billion, a number that exceeds every contemporary demographic projection and is driven primarily by an assumed 35% upward revision in Africa’s 2100 population.

Second, the new MEDIUM scenario, which van Vuuren et al. describe as a current-policy trajectory, produces higher cumulative emissions through 2100 than the IEA’s current-policy scenario produces by 2050. On the IEA’s own numbers, today’s policy trajectory delivers roughly 2.5°C of warming by 2100. The new MEDIUM delivers more. Calling it a “current policy” scenario is, charitably, a stretch.

Third, the mainstream coverage that finally arrived in May was not a reckoning. The first wave of New York Times, Washington Post, AP, Bloomberg, and Carbon Brief coverage uniformly described RCP 8.5 as the “worst-case” scenario (it was a baseline scenario), attributed its retirement to the success of renewable energy policy (it became implausible because its core assumptions about coal expansion and demographic trends were always wrong), and framed the story around President Trump rather than around fifteen years of institutional error. Pielke’s verdict on the coverage, in his most recent post: the outlets that built years of stories on RCP 8.5 chose to

minimize and deflect.

The lesson

The lesson here is not that climate scientists were stupid or dishonest. The modelers who pulled RCP 8.5 off the shelf because it gave them a clean signal were doing what physical scientists routinely do. It is their subsequent lack of action to correct the record as necessary that screams monumental cowardice. The spherical cow has its place. The trouble begins when the spherical cow is wheeled out of the lab and presented to the public as if it were a real cow, and then wheeled into the Federal Reserve and presented to bank regulators as the cow you should be valuing your loan book against, and then wheeled into a federal courthouse and presented to a judge as the cow whose damages an energy company should be required to pay.

That is what happened. The official scenario community has now admitted, in the formal scientific literature, that the cow was not a cow. They have done so in a footnote, with no apology, after a decade and a half of silence from the broader profession that knew, or could have known, all along. The press that built a generation of front-page stories on it has so far chosen not to notice.

The next time you read a headline about climate change and the year 2100, whether it appears in the New York Times, in a central bank report, or in a complaint filed in federal court, the first question to ask is which scenario the projection rests on. If the answer is RCP 8.5 or SSP5-8.5, you are reading about a spherical cow.

The barn door is shut. The cow has gone back inside. What remains is the trail of damage it left while it was out, and an open question about why so many people in a position to stop it kept walking past the open door.

---

Footnote: At this point it is safe to say the climate models are moooot. [Anthony]