Turning “What If” into “How Many”: The Rhetorical Alchemy of Climate Modeling

Charles Rotter

A recent Nature paper, “Projected impacts of climate change on malaria in Africa,” provides a textbook example of how layered model uncertainty can be transformed—through careful framing—into quantified predictions that appear far more authoritative than the underlying evidence warrants. The headline figures are stark: 123 million additional malaria cases and more than 500,000 additional deaths by mid-century, attributed to climate change. These numbers are already circulating as if they describe a measurable future risk. A close reading of the paper itself tells a very different story.

Abstract

The implications of climate change for malaria eradication this century remain poorly resolved^1,2. Many studies focus on parasite and vector ecology in isolation, neglecting the interactions between climate, malaria control and the socioeconomic environment, including disruption from extreme weather^3,4. Here we integrate 25 years of African data on climate, malaria burden and control, socioeconomic factors, and extreme weather. Using a geotemporal model linked to an ensemble of climate projections under the Shared Socioeconomic Pathway 2-4.5 (SSP 2-4.5) scenario⁵, we estimate the future impact of climate change on malaria burden in Africa, including both ecological and disruptive effects. Our findings indicate that climate change could lead to 123 million (projection range 49.5 million to 203 million) additional malaria cases and 532,000 (195,000–912,000) additional deaths in Africa between 2024 and 2050 under current control levels. Contrary to the prevailing focus on ecological mechanisms, extreme weather events emerge as the primary driver of increased risk, accounting for 79% (50–94%) of additional cases and 93% (70–100%) of additional deaths. Most increases stem from intensification in existing endemic areas rather than range expansion, with significant regional variation in impact. These results highlight the urgent need for climate-resilient malaria control strategies and robust emergency response systems to safeguard progress towards malaria eradication.

https://www.nature.com/articles/s41586-025-10015-z

This is not a study that reports new empirical discoveries about malaria transmission. It is an exercise in scenario construction. Its results emerge from a long chain of assumptions, models, and parameter choices, each, possibly, defensible in isolation, but collectively producing an impression of precision that the authors’ own caveats do not support.

The paper opens by positioning itself as a corrective to earlier work, arguing that prior studies focused too narrowly on ecological mechanisms while neglecting social and infrastructural disruption. That framing sets the stage for what follows: a shift away from biology and toward modeled institutional fragility.

“Nearly all existing projections share a central limitation: although they explore climate effects in isolation, they do not adequately account for non-climate determinants of malaria trends.”

This sounds reasonable, but what replaces that “limitation” is not observation. It is an expanded modeling framework that incorporates still more uncertain components.

The foundation of the analysis is future climate. The authors rely on downscaled CMIP6 global climate model outputs under the SSP 2-4.5 scenario, described as a “middle of the road” pathway. These models are known to diverge substantially in regional precipitation and extreme weather projections across Africa. The paper attempts to address this by using ensembles, but ensembling disagreement does not eliminate uncertainty—it simply averages it.

“Between-GCM uncertainty and variability were accounted for using an ensemble of CMIP6 members…”

That sentence does a great deal of rhetorical work. “Accounted for” sounds reassuring, but an ensemble mean is not a validation. It is a compromise among conflicting model structures, all of which share common assumptions and biases.

Those climate projections are then downscaled to a 5×5 km grid and fed into mechanistic models that translate temperature, rainfall, and humidity into mosquito and parasite suitability indices. These models are highly nonlinear and sensitive to thresholds. Small errors in climate inputs propagate into large swings in transmission suitability. This is a property of the models—but it is rarely emphasized when results are summarized.

The suitability indices are then used as predictors in a stacked statistical framework combining linear models, generalized additive models, boosted regression trees, and a Bayesian geostatistical smoother. The machinery is complicated, but it is also opaque. Even the authors acknowledge that the uncertainty behavior of such systems is poorly understood.

“There is currently limited precedent for fully characterizing uncertainty in stacked models, particularly when applied to spatially correlated data.”

That admission should fundamentally constrain how the outputs are interpreted. Instead, it is buried in the Discussion, far removed from the headline numbers.

Up to this point, the results are relatively modest. When ecological effects alone are considered, the authors find that climate-driven changes in malaria transmission are small and mixed—some regions increase, others decrease, and the continent-wide average effect is close to zero.

“We project that, considered in isolation, the ecologically driven impacts of climate change on malaria transmission would lead to minimal overall change in Africa by 2050…”

This result is rarely highlighted, because it does not support the sense of urgency implied by the paper’s title.

The dramatic figures emerge only when the authors introduce a second layer: disruption from extreme weather events. Floods and cyclones are modeled to damage housing, disrupt vector control, and reduce access to treatment. According to the paper, these disruptive effects dominate the outcome.

“Extreme weather events emerge as the primary driver of increased risk, accounting for 79%… of additional cases and 93%… of additional deaths.”

This is the pivotal move. The paper is no longer primarily about climate and malaria ecology. It is about assumed disruption to institutions and infrastructure, projected decades into the future.

Here the evidentiary basis becomes especially thin. The magnitude and duration of disruption are not derived from large datasets or controlled studies. They are assembled from a literature review of heterogeneous case studies and 34 expert interviews, then translated into parameters describing how much housing is damaged, how many clinics close, and how long recovery takes.

“The paucity of data necessitated a more heuristic approach to quantifying likely impacts.”

“Heuristic” is doing a lot of work here. These parameters are not measured. They are judged to be plausible. Because the data are sparse, the authors apply uncertainty ranges spanning 50% to 150% of their central values.

“A broad uncertainty range was considered appropriate because the scarcity of observational data precluded a more formal quantification…”

These heuristic disruption parameters are then applied continent-wide, month by month, under simulated future floods and cyclones generated by climate-driven storm models. At this point, the model is several layers removed from anything that could reasonably be called an observation.

Yet the final outputs are presented as cumulative totals of cases and deaths, with specific numbers and ranges.

“Our findings indicate that climate change could lead to 123 million additional malaria cases and 532,000 additional deaths in Africa between 2024 and 2050…”

What is rarely emphasized is that these figures depend critically on assumptions that freeze nearly all non-climatic progress. Malaria control coverage, housing quality, healthcare infrastructure, and socioeconomic development are held constant at present-day levels—except when they are damaged by climate events.

“We deliberately hold constant present-day levels of transport and healthcare infrastructure, housing quality and malaria control…”

This is not a neutral assumption. Historically, malaria burden has declined primarily because of improvements in drugs, vector control, infrastructure, and economic development. By suppressing those trends while allowing disruption to accumulate, the model structurally biases results toward worsening outcomes.

The authors also acknowledge that their projections are not forecasts.

“Our projections allow exploration of climate change effects but are not intended as forecasts of future conditions.”

They go further, stating explicitly that the uncertainty ranges are not statistical.

“The projection ranges are not… formal statistical intervals, providing indicative measures of uncertainty rather than probabilistic statements.”

These are not minor footnotes. They directly contradict the way the results are likely to be interpreted by policymakers, journalists, and advocacy groups.

This is where the rhetorical sleight of hand becomes apparent.

Deep in the paper, uncertainty is emphasized, limitations are acknowledged, and projections are framed as exploratory. In the abstract, figures, and conclusions, those same exploratory outputs are converted into quantified impacts with an unmistakable air of urgency.

The issue is structural. When multi-layered scenario models produce precise numbers, those numbers take on a life of their own. Caveats fade. Assumptions harden into facts. What began as “what if” quietly becomes “this will.”

The paper does not demonstrate that climate change will cause hundreds of millions of additional malaria cases. It demonstrates how easily such numbers can be produced when uncertain climate projections, sensitive ecological models, heuristic disruption parameters, frozen socioeconomic baselines, and long time horizons are combined in a single framework.

The appropriate way to read this study is not as a prediction, but as a thought experiment—one that is highly sensitive to its assumptions and explicit about its limitations, if one reads far enough. The problem is not what the authors say in the fine print. It is how far that fine print is removed from the numbers that will be remembered.

In the end, this paper tells us less about the future of malaria than about the current state of climate-related modeling. It shows how uncertainty can be multiplied, smoothed, and translated into apparent precision. It shows how scenarios can be mistaken for forecasts. And it shows how, once numbers are published in a journal like Nature, they are treated as evidence—even when the authors themselves say they are not.

That is the sleight of hand worth paying attention to.