Charles Rotter
For more than a decade, some of the most frightening claims about future climate change came from a scenario called RCP8.5. Most people never heard that name on the evening news. Instead, they heard the results. Temperatures could soar by 2100. Sea levels could rise dramatically. Heat deaths could climb. Crop yields could fall. Wildfires could worsen. Electricity demand could explode. Economic damages could reach enormous levels.
Behind many such projections sat RCP8.5, or its newer cousin, SSP5-8.5.
We’ve noted previously on this blog that the scientific community that develops scenarios for climate models has made an important change. In April 2026, the scientists designing scenarios for the next generation of climate models published the ScenarioMIP-CMIP7 paper.
Buried inside the technical language was a remarkable sentence:
“the CMIP6 high emission levels (quantified by SSP5-8.5) have become implausible.”
That deserves to be read twice. The scientists who design these scenarios now say the emissions levels represented by SSP5-8.5 have become implausible. This does not mean global warming disappears. It does not mean carbon dioxide has no effect on climate. It means something much simpler.
The highest emissions scenario that played a huge role in climate research and climate communication is no longer considered a plausible future by the scientists designing the next generation of scenarios. That created a very large problem. What happens to government policies, risk assessments and public warnings that relied on it?
An August 2026 proposed amendment to the Fifth National Climate Assessment, or NCA5, attempts to answer that question.
Its answer amounts to: go back and check the work.
The proposed amendment says:
“NCA5 is amended to clarify that results dependent on RCP8.5, SSP5-8.5, or equivalent high emissions pathways should not be used as expected, baseline, business-as-usual, likely, central, or policy-relevant federal planning futures.”
It then gives these scenarios a label that would have been shocking to see in a federal climate assessment only a few years ago:
“implausible scenario outputs.”
That is a major change. To understand why, it helps to understand what RCP8.5 actually was. RCP stands for Representative Concentration Pathway. The number 8.5 refers to a radiative forcing level of about 8.5 watts per square meter by 2100. These pathways were created so climate scientists could run models under different future greenhouse-gas conditions. They were tools for experiments. RCP8.5 gave researchers a very high-emissions case. That can be useful because a strong climate signal is easier to see in a model than a weak one.
Think of an automobile crash test. Engineers might crash a car at unusually high speed because they want to see how the car behaves under extreme conditions. That experiment can provide useful information. But imagine if someone took the results and told every driver, “This is your expected accident.” The experiment has suddenly become something else.
The proposed NCA5 amendment recognizes this basic problem. It explains that the RCPs:
“were not originally developed as forecasts of the most likely socioeconomic future.”
It says RCP8.5 was included to cover the upper range of possible forcing for climate-model experiments. The controversy begins with what happened afterward. RCP8.5 became extremely popular. Researchers used it in impact studies. Governments used it in climate assessments. Journalists used studies based upon it to describe the future. Financial institutions used high-emissions scenarios for stress testing. In many places, RCP8.5 was described as “business as usual.” That description was being challenged years before the 2026 announcement. And one of the people making that argument was hardly a member of the climate-skeptic community.
Climate scientist Zeke Hausfather and climate researcher Glen Peters wrote a blunt article in Nature in January 2020 titled “Emissions – the ‘business as usual’ story is misleading.”
Their subtitle was even clearer:
“Stop using the worst-case scenario for climate warming as the most likely outcome — more-realistic baselines make for better policy.”
Hausfather and Peters argued that falling clean-energy costs and changing energy trends meant very high emissions scenarios were becoming increasingly unlikely. But there was another side to the criticism. Researchers such as Justin Ritchie, Hadi Dowlatabadi, Matthew Burgess and Roger Pielke Jr. had been examining the assumptions needed to produce RCP8.5.
A major problem was coal.
Producing RCP8.5 required an extraordinary expansion of coal use during the twenty-first century, along with a particular combination of assumptions about population, technology, energy demand and economic growth. These concerns were published in peer-reviewed journals years before the CMIP7 announcement. The proposed NCA5 amendment cites that research directly. This distinction matters because there are now two competing explanations for the death of RCP8.5.
One explanation says the world changed.
Renewable energy became cheaper. Technology improved. Climate policies reduced emissions. Natural gas replaced some coal. As a result, a scenario that might once have been possible is no longer realistic. The authors of the 2026 ScenarioMIP-CMIP7 paper themselves point to technological developments, climate policy and changing emissions trends.
But critics have made a second argument. They say RCP8.5 had serious problems as a normal planning scenario from the beginning because of the extreme combination of the assumptions required to produce it. The proposed federal amendment clearly takes this second concern seriously.
It states:
“The core deficiency is not merely that climate policy, renewable energy deployment, efficiency improvements, technological innovation, or natural gas substitution altered emission trajectories after these scenarios were developed. The deeper deficiency is structural.”
It then points to assumptions about coal expansion, fossil-fuel resources, energy use, population, economic development and technology. That makes the history considerably more interesting than the simple story that climate policy saved us from RCP8.5. Even Hausfather and Peters were warning six years ago that the business-as-usual framing was misleading. And that was in Nature, not exactly a secret clubhouse for climate skeptics.
Roger Pielke Jr. has been even more critical. After the 2026 scenario announcement, Pielke wrote a May 20 opinion article for the Washington Post titled “The climate apocalypse? Don’t count on it.” Pielke has argued for years that RCP8.5 was being misused in climate research and policy. The Washington Post article called the retirement of RCP8.5 an important correction. But Pielke also argued that problems remain in the new scenarios, including assumptions about population and emissions.
Getting rid of an implausible old scenario does not guarantee that every new scenario is realistic.
The proposed NCA5 amendment makes exactly that point. It specifically warns federal agencies against automatically accepting the new CMIP7 scenarios merely because they replaced the old ones. Future scenarios, it says, should be judged by their assumptions and by how well those assumptions match observations and reasonable expectations.
That is an excellent rule. Apply it to everything.
I published a May 21 article as part of this discussion. My article compared RCP8.5 to the old physics joke about a “spherical cow.” The joke involves a physicist trying to solve a difficult real-world problem by simplifying a cow into a perfect sphere. The simplified model can be useful for doing mathematics. The trouble starts when somebody forgets that the spherical cow is an artificial tool. My WUWT article argues that something similar happened to RCP8.5.
I said:
“A scenario the modelers built as a stress test became, for fifteen years, the scenario the press treated as your future.”
That captures the skeptical case in one sentence.
My article argued that RCP8.5 escaped from the modeling world and entered government reports, headlines, financial stress tests and public debate as though it described an ordinary future. The article also pointed to an important comment from Keywan Riahi, lead author of the original RCP8.5 paper. Discussing RCP8.5 with Associated Press reporter Seth Borenstein in 2026, Riahi said:
“it was never a likely case.”
That raises an uncomfortable question. If it was never a likely case, why did it become so important? RCP8.5 was useful to researchers. A stronger emissions pathway creates a stronger modeled climate response. That can make it easier to study the response and compare different climate models. Researchers studying impacts also had practical reasons to use it. Many climate-model runs were already available under RCP8.5. Using the same scenario allowed researchers to compare results across studies.
Then came institutional momentum.
Studies cited other studies. Government reports cited the studies. Journalists cited the government reports. New researchers used the same scenario because previous researchers had used it. Soon the unusual case had become normal. My WUWT article argues that this institutional process deserves much more scrutiny than it has received. Mainstream coverage in May 2026 took a different approach.
Much of the reporting focused on President Trump because he publicly commented on the retirement of RCP8.5.
Carbon Brief strongly challenged Trump’s interpretation. It stressed that RCP8.5 was only one scenario and that retiring it does not overturn the basic science of greenhouse warming.
They argue that removing RCP8.5 does not tell us that future warming will be zero. But there is a separate question that should not disappear:
How much did RCP8.5 affect the size of past claims about climate impacts?
That is where the proposed NCA5 amendment becomes important. It does not merely propose changing the label attached to RCP8.5. It calls for federal agencies to examine actual claims based upon it.
For example, the amendment says:
“Any NCA5 result based on RCP8.5, SSP5-8.5, or equivalent high forcing should be treated as an implausible-scenario output.”
It says such a result should not be used for federal policy, planning, regulation, public communication or grant-making unless it is reproduced under a plausible scenario and checked against observations. That could have large consequences. Consider electricity.
NCA5 includes projections about rising electricity demand and higher spending on transmission and distribution under RCP8.5. Utilities make expensive decisions that can last for decades. If a projection depends heavily on an implausible emissions pathway, planners should know that before spending billions of dollars. The amendment says these estimates should be reproduced using plausible scenarios and compared with real trends in electricity demand, technology, population and adaptation. That sounds like basic due diligence.
The same issue appears in public health. NCA5 contains claims about ozone, pollen and Valley fever that depend on RCP8.5 or similar high-emissions assumptions. Those estimates can affect public-health planning, regulations and worker-safety policies. The proposed amendment says federal officials should not present such numbers as expected future health burdens unless researchers can reproduce them under plausible scenarios and compare them with observations.
Again, the question is simple:
What happens to the projected harm when the implausible scenario is removed? Perhaps much of the effect remains. Perhaps it becomes smaller.
Perhaps the timing changes.
Perhaps some particular result largely disappears.
There is only one sensible way to find out.
Run the numbers again.
Economic damage estimates deserve even closer attention.
Long-term economic projections can contain layer upon layer of assumptions.
First comes an emissions scenario.
Then comes the modeled climate response.
Then comes a projection of local climate conditions.
Then researchers estimate how people, businesses, crops or infrastructure respond.
Then assumptions must be made about population, economic growth, technology and adaptation.
Finally, economists may have to decide how to value costs many decades in the future.
Each step adds uncertainty.
The proposed amendment therefore calls for clearer disclosure of discount rates, socioeconomic assumptions, adaptation assumptions and scenario dependence.
That is especially important when these calculations produce enormous dollar figures.
A computer can print a very precise number.
That does not make the assumptions behind the number equally precise.
The amendment also points to regional claims throughout NCA5.
It identifies projections involving agriculture and economic damage in the Southeast, rainfall in the Caribbean, health and infrastructure in the Midwest, drought in the Northern Great Plains, extreme heat in the Southern Great Plains, wildfire smoke in the Northwest, health and labor effects in the Southwest, sea ice in Alaska, and fisheries and migration in Pacific islands.
The proposal does not say all these problems disappear.
It says federal agencies should separate three different things:
what has actually been observed;
what models project under plausible scenarios;
and what models project under implausible high-emissions scenarios.
That distinction should be obvious.
Yet it can disappear remarkably quickly in climate communication.
A scientific paper might say:
“Under SSP5-8.5, our model projects X by 2100.”
A government summary becomes:
“Climate change could cause X by 2100.”
A headline becomes:
“Climate change will cause X by 2100.”
Three sentences can describe three very different levels of certainty.
The model did not change.
The language did.
The proposed amendment attempts to stop that process by requiring the scenario warning to travel with the result.
It even provides standard language:
“This result depends on RCP8.5, SSP5-8.5, or an equivalent high emissions pathway that is no longer policy-plausible as a central federal planning future.”
Imagine seeing that sentence under every RCP8.5 map or late-century damage estimate. Readers could then decide for themselves how much weight to give the result. There is another remarkable part of the amendment.
NCA5 already admitted that SSP5-8.5 had a plausibility problem.
According to the amendment, Chapter 3 of NCA5 says SSP5-8.5 is no longer plausible without major departures from trends in energy efficiency, technology and lower-carbon energy. But the proposed amendment says that warning was not consistently carried into regional chapters, captions, public summaries and impact claims. That is a serious communication problem. A warning hidden in one chapter does little good if dramatic numbers from another chapter are presented without it. The amendment would require the warning to follow the claim.
There is an important lesson here for climate skeptics as well. The retirement of SSP5-8.5 does not prove that every climate projection is wrong. The new CMIP7 framework still includes a high-emissions scenario. But there is a revealing difference. Even that new high scenario is expected to produce less forcing than SSP5-8.5. And the CMIP7 researchers reached that result even though their new High scenario includes very pessimistic assumptions, such as climate-policy rollback and continued heavy fossil-fuel use.
But something else in the CMIP7 paper may matter even more.
The authors acknowledge that deciding which scenarios are “plausible” involves judgment. They do not assign probabilities to their scenarios. So even a plausible scenario should not automatically be treated as a prediction. “Plausible” means researchers believe the future could reasonably develop that way. It does not mean they know that it will. This is where climate policy often gets ahead of climate science.
Governments want numbers.
Regulators want numbers.
Engineers want numbers.
Politicians definitely want numbers.
A range of conditional model experiments is less useful politically than a frightening number attached to the year 2100. But nature does not owe policymakers certainty simply because Congress has a budget deadline. The proposed amendment offers a better approach. Start with observations. Clearly identify what has been measured. Clearly identify what has been modeled. Show the assumptions. Show the uncertainty. Check whether the model reproduces the real-world quantity being discussed. Then present future projections.
The amendment specifically calls for observational metrics, periods of record, data sources, uncertainty and model-performance checks before future projections are used for policy. It should never have become controversial.
There is also a larger institutional question. Why did it take so long? Hausfather and Peters were publicly warning in Nature in 2020 that treating worst-case scenarios as the most likely future was misleading. Pielke and other researchers had been criticizing the use of extreme emissions scenarios for over a decade. The original assumptions behind RCP8.5 were publicly available. Yet the scenario remained deeply embedded in climate research and government assessments.
My WUWT article argued that the scientific community failed to correct the public use of RCP8.5 when it had repeated opportunities to do so and the institutional question is legitimate.
How did an extreme scenario become so widely used that, years after researchers began warning against treating it as business as usual, the federal government now needs an amendment telling agencies how to handle all the results built upon it? The proposed NCA5 amendment is therefore more important than the retirement of one climate scenario. It proposes a standard that should have been applied all along. A scenario should be judged by its assumptions. A model projection should be clearly separated from an observation. An extreme experiment should be identified as an extreme experiment. A result that changes greatly when unrealistic assumptions are removed should not quietly remain a government planning baseline.
And a replacement scenario should receive exactly the same scrutiny as the scenario it replaced. Researchers can still use extreme scenarios for experiments, model comparisons and sensitivity tests. The proposed amendment explicitly allows such technical uses. The important question is what happens when the result leaves the laboratory. If a federal agency wants to spend money, write regulations, redesign infrastructure or warn Americans about a future danger, it should be able to show that its assumptions are reasonable.
If a severe projection survives that test, then present it.
If the projection becomes much smaller, say so.
If the answer remains highly uncertain, say that too.
It means checking assumptions before turning a model result into a fact.
The strange history of RCP8.5 shows why that matters.
The spherical cow can stay in the laboratory. Just don’t build a farm around it.
