RIP Doomsday Model RCP8.5 – A Quiet but Profound Shift in Climate Science

The long run of RCP8.5 has come to an end

For more than a decade, anyone paying attention to how climate science has been communicated, not just conducted, has seen a peculiar pattern. A single scenario, RCP8.5, steadily migrated from a deliberately extreme modeling construct into something else entirely: a default storyline.

Originally, RCP8.5 was never intended to represent a likely future. It was a high-end boundary case—useful for stress-testing models, exploring sensitivities, and asking “what if?” questions under aggressive forcing assumptions. There’s nothing inherently wrong with that. Science needs bounding cases.

The problem arose in how it was used…and abused.

Somewhere along the way, RCP8.5 was relabeled—quietly at first, then routinely—as “business-as-usual.” That phrase carries weight. It implies plausibility. It suggests continuity with observed trends. And once that label took hold, it became a convenient foundation for a vast amount of downstream work.

From there, the amplification cycle was predictable:

• Academic papers used RCP8.5 as a central or default scenario
• Media outlets highlighted the most dramatic findings in doomsday tones
• Journalists reported those findings with minimal context
• Policymakers cited the results as evidence of likely terrible futures

The distinction between exploratory scenario and probable trajectory blurred, then largely disappeared.

By the late 2010s, RCP8.5 had become the backbone of countless headlines—sea level rise projections, extreme heat claims, agricultural collapse scenarios, economic damage estimates—all frequently tied to a pathway that required assumptions about coal use, population growth, and technological stagnation that were increasingly at odds with real-world data.

This wasn’t a fringe concern. Researchers themselves began raising flags. Papers in the peer-reviewed literature pointed out that RCP8.5’s fossil fuel requirements were difficult to reconcile with known reserves, that its demographic assumptions were outdated, and that emissions trends were already diverging from its trajectory.

Yet the scenario persisted—not because the evidence supported its centrality, but because it was embedded. In models. In datasets. In institutional frameworks. In narratives.

It proved useful, for the “useful idiots.” And usefulness, particularly when it aligns with compelling storytelling, has a way of outlasting scrutiny.

What we are now seeing with CMIP7 is not a sudden discovery, but a delayed acknowledgment that climate science was wrong. The modeling community has finally moved on from treating these extreme pathways as central cases. That shift is significant, even if it arrives years after the underlying issues were widely understood.

The question now is not whether RCP8.5 was misapplied—it clearly was—but how long it will take for that realization to propagate through the layers of research, reporting, and policy that were built upon it.

Here are salient points about what has occurred.

---

Roger Pielke Jr.’s recent analysis, “RCP8.5 is Officially Dead,” highlights what may prove to be one of the most consequential course corrections in modern climate research: the formal abandonment of the highest emissions scenarios—RCP8.5, SSP5-8.5, and SSP3-7.0—by the ScenarioMIP team underpinning CMIP7.

This is not a minor technical adjustment. It directly affects the boundary conditions used in climate models, which in turn propagate through impact studies, risk assessments, and policy frameworks worldwide.

Let’s walk through what has changed—and what hasn’t.

1. The End of “Business-as-Usual” High-End Scenarios

The central scientific development comes from Van Vuuren et al. (2026), who state plainly: “the CMIP6 high emission levels (quantified by SSP5-8.5) have become implausible”

That word—implausible—is doing a lot of work.

For over a decade, RCP8.5 (and later SSP5-8.5) was widely described in the literature as a “business-as-usual” (BAU) trajectory. Yet the assumptions underlying it were extreme:

• Massive expansion of coal use (roughly 5× increase by 2100)
• Very high population growth
• Minimal technological change or policy evolution

As early as 2017, critiques began appearing in the literature. For example:

• Ritchie & Dowlatabadi (2017) argued RCP8.5 required “extraordinary” fossil fuel use inconsistent with resource constraints.
• Hausfather & Peters (2020, Nature) noted that RCP8.5 “should not be used as a no-policy baseline” and is better treated as a worst-case scenario.

Yet despite these warnings, the scenario became deeply embedded in:

• Impact studies
• Media narratives
• Government risk assessments

CMIP7 now quietly removes it from the official scenario set.

2. A Compressed Emissions Range

The chart compiled by Pielke shows the core quantitative shift.

Key numbers:

• SSP5-8.5: ~128 Gt CO₂/year by 2100
• CMIP7 HIGH: ~71 Gt CO₂/year by 2100

That is nearly a 45% reduction in projected peak emissions.

Cumulative emissions tell the same story (table on page 5):

• SSP5-8.5: ~7,676 Gt CO₂ (2020–2100)
• CMIP7 HIGH: ~4,629 Gt CO₂

That is a reduction of ~40% !!

This compression reflects:

• Updated energy system trends
• Slower-than-assumed population growth (in most datasets)
• Empirical emissions trajectories that have diverged from earlier worst-case pathways

The result is a narrower—and arguably more realistic—range of future climate forcing, assuming the forcing is actually there in the first place.

3. Temperature Projections: Downward Adjustments

Using the FaIR v2.2 emulator, Pielke shows:

• SSP5-8.5: ~3.9°C warming (updated) vs 4.4°C in AR6
• CMIP7 HIGH: ~3.0°C
• CMIP7 MEDIUM: ~2.5°C

Two points matter here:

(a) Updated baselines reduce warming

The shift from 2014 to 2023 emissions baselines lowers projected warming across scenarios. This aligns with broader findings that near-term emissions have tracked closer to intermediate scenarios than extreme ones (IEA, 2023; IPCC AR6 WG1 Chapter 4).

(b) The “worst case” is now substantially cooler

CMIP7 HIGH is:

• ~0.9°C cooler than SSP5-8.5 (like-for-like comparison)
• ~1.4°C cooler than AR6 headline values

That is not trivial—it materially alters downstream impact estimates.

4. The Role of Demographics: A Hidden Driver

One of the more technical insights comes from the population assumptions embedded in SSP3.

From the table on page 8 of Pielke’s post:

• SSP3 population rises to 14.5 billion by 2100 (2024 update)

This is well above most mainstream demographic projections:

• UN World Population Prospects (2022): ~10.4 billion peak
• IIASA median pathways: significantly lower than SSP3

Pielke’s sensitivity test shows:

• Holding per-capita emissions constant but using lower population trajectories reduces warming by ~0.6°C.

This highlights a key scientific point: Emissions scenarios are as much about socioeconomics as physics.

Climate projections are not purely radiative transfer problems—they are conditional on assumptions about:

• Fertility rates
• Economic growth
• Technology diffusion

If those assumptions are unrealistic, the resulting climate projections inherit that bias.

5. The “Plausibility Vacuum” Problem

A recurring critique—originally articulated by Ritchie and others—is that scenario construction lacks systematic plausibility checks. Pielke summarizes this as a “plausibility vacuum”:

• RCPs specified radiative forcing first
• Socioeconomic pathways were backfilled later
• Internal consistency was not rigorously enforced

Even in CMIP7:

• Scenarios still inherit SSP storylines
• No formal validation against real-world energy or demographic trends is performed

This is not a trivial methodological issue. It affects:

• Model boundary conditions
• Ensemble spread
• Policy-relevant conclusions

As IPCC AR6 acknowledges (WG1, Chapter 1), scenarios are “not predictions” but conditional pathways. The difficulty arises when conditional pathways are treated as likely futures.

6. The “Medium” Scenario: Not Quite “Current Policy”

The CMIP7 MEDIUM scenario is described as reflecting current policies. But comparison with IEA projections (page 11 table in Pielke’s post) shows divergence:

• IEA STEPS: emissions fall below 30 Gt CO₂ by 2050
• CMIP7 MEDIUM: ~41 Gt by 2050, rising thereafter

This suggests:

• CMIP7 MEDIUM behaves more like policy stagnation than actual policy continuation

On an apples-to-apples basis:

• CMIP7 MEDIUM produces ~0.1–0.2°C more warming than SSP2-4.5 (when aligned to the same baseline)

That may sound small, but in climate modeling terms it is meaningful—especially when used for impact assessments.

7. Policy Implications: Legacy Scenarios Still Embedded

Perhaps the most consequential issue is institutional inertia. As shown in the table on page 14, RCP8.5 and SSP5-8.5 remain embedded in:

• National climate assessments (U.S., UK, EU, etc.)
• Central bank stress tests (NGFS frameworks)
• World Bank and FAO datasets

This creates a lag between:

• Scientific updates
• Policy frameworks built on older assumptions

The result is that risk assessments may continue to rely on scenarios now considered implausible by the very community that produced them.

8. What This Means Going Forward

Several conclusions emerge:

1. A genuine scientific correction has occurred

The removal of RCP8.5-class scenarios reflects:

• Empirical mismatch with observed trends
• Growing recognition of unrealistic assumptions

2. The upper bound of projected warming has shifted downward

Not eliminated—but constrained.

3. Scenario construction remains methodologically incomplete

The lack of systematic plausibility checks persists.

4. Policy frameworks lag behind science

Legacy scenarios continue to influence decisions despite updated understanding.

Final Thoughts

Climate science is often described as self-correcting. This episode provides a concrete example. It took nearly a decade from widespread critique to formal adjustment. Whether that timeline is acceptable is a separate discussion. Personally I think it wasn’t, and it reveals the serious confirmation bias that pervades the climate science community.

What matters scientifically is that:

• The most extreme emissions pathways are no longer treated as central or likely
• The scenario space has narrowed
• The dependence of projections on socioeconomic assumptions is more visible than ever
• Media outlets can no longer cite RCP8.5 as a doomsday scenario, but they will try for awhile.

The next step is straightforward in principle, though perhaps less so in practice: align research, assessments, and policy tools with the updated scenario framework, and apply far more scrutiny to the assumptions that drive these models in the first place.

Because in the end, climate projections are only as robust as the scenarios that feed them. RCP8.5 is mostly dead, let’s hope it becomes truly dead.