Inside the Flawed Model Behind the Club of Rome’s Earth for All

A sweeping peer-reviewed critique published in Frontiers in Sustainability Resource Management argues that one of the most influential sustainability books of recent years rests on a model that is not, by any established scientific standard, a functioning system dynamics model. The book Earth for All, presented to the Club of Rome by Sandrine Dixson-Declève, Owen Gaffney, Jayati Ghosh, Jørgen Randers, Johan Rockström and Per Espen Stoknes, has been translated into more than ten languages and promises a bright new world achieved through five policy turnarounds covering poverty, inequality, women’s empowerment, food and energy. According to the critique’s authors, Harald Sverdrup, Peter Schlyter, Hördur Haraldsson, Kristin Vala Ragnarsdottir and Michael Dorsey, the analytical basis for those recommendations was constructed after the recommendations were made, and a close inspection of the model code reveals a construct with severe and simultaneous technical flaws that invalidate it as a system dynamics model.

The workflow problem sits at the heart of the controversy. The authors reconstruct the sequence of events: a team of economists and Club of Rome members met and decided which parts of the economy needed adjustment, identifying the five turnarounds. The Earth for All book was then written around these proposals. Only afterwards was the Earth for All model, known as E4A, built, giving the misleading impression that the advised turnarounds, intended to achieve sustainability by the end of this century, emerged from scenario analysis of a system dynamics model. Background papers appear designed to strengthen the impression that the book and model were products of rigorous systems analysis, yet none of the documents are published in peer-reviewed journals or otherwise formally peer-reviewed in any traceable sense. The authors call this the reverse of the scientific method: results were presented before the model was constructed to support them.

The critique applies state-of-the-art peer-review methodology for integrated assessment modelling, drawing on the foundational system dynamics literature of Forrester, Senge and Sterman, and evaluates the E4A model against four criteria: content and relevance, internal consistency, validity, and system dynamics. The model and documentation were downloaded from the Club of Rome website and the source code was read in both STELLA Architect and Vensim versions. On content and relevance, the review finds that several central aspects discussed in the book do not exist in the model code at all. The turnarounds are not represented in the simulation routine, and the causal loop diagrams shown in the supporting documents contain multiple elementary mistakes and do not correspond to what is actually inside the model according to its code.

Internal consistency is where the technical flaws become fatal. Any model that violates the fundamental principles of thermodynamics is considered invalid, and the review finds that the economic and financial modules in the E4A model do not maintain accounting, mass or energy balances. The resource model has the same flaw, and the population module is similarly broken, producing negative population numbers during simulation runs. The model lacks several real-life stocks that cause delays in the physical Earth system, and there are no dynamic supply chains; delays are instead set by command rather than by system processes. Preset forced delays and smoothing functions are used to conceal discontinuities arising from the missing mass balances and jumps between forced input and output values. The labour market module yields more employed people than the working-age population simulated in the population module, making it inconsistent with the mass balance of that module.

The most striking revelation concerns what the authors find when they look inside the simulation code. The E4A model, they conclude, is not a high-level simple systems model for overview assessment, as its defenders claim, but a set of command functions securing the appearance of predetermined values in the outputs. Forcing functions are used in modelling when mechanisms or processes are unknown or ignored, commanding the model to write out a preordained outcome regardless of what happens inside the system. Where smoothed step changes are not used, if-then statements command outcomes. The construct has been wrapped in software used for system dynamics modelling, which gives it the outward appearance of a bona fide dynamic model. But the model has no feedback functions or dynamics independent of the preset output commands, and several stocks in the code have been short-circuited. By definition, the authors argue, it is not a system dynamics model but an entirely different type of construct.

Because the simulation outcomes are exogenously driven, the scenarios do not represent comparable outcomes of different policy choices working on the same system. Even so, both the Giant Leap scenario and the Too Little Too Late, or business-as-usual, scenario crash in the model after the year 2100, because the forcing functions were only set until 2100 and the code loses control of model output after that. The module-level failures compound the problem: the food and agriculture module lacks basic components such as food supply and does not constitute a valid agricultural model; the poverty turnaround is based on handouts and printing money; the inequality turnaround assumes women are 50 percent of the world population, while women’s empowerment is absent from the simulation code; there is no economic model representing sustainable finance or market-driven mechanisms; and no natural resources are represented as stocks or mass balances of any kind, with the energy turnaround assuming endless resources are available.

Validity testing is entirely missing. The review finds that the E4A model has not been tested on independent, observed data from the past that was not used during calibration, nor aligned with findings from prior research in the field. Assessing sustainability, the authors explain, depends on the ability to express sustainability boundaries quantitatively and to quantify the gap between the present system state and a sustainable state, a gap-closure approach used in implementing quantitative environmental policy, most successfully in the Convention on Long-Range Transport of Air Pollution in Europe. The E4A model lacks clearly defined sustainable system states as goals, which are essential for integrating qualified sustainability concepts into a well-defined numerical model. Since there are no quantitative sustainability goals in the model, all statements about what is sustainable or not are untested and unsupported. The book refers to the UN Sustainable Development Goals, but these remain conceptual rather than defined in the quantitative terms needed for policy implementation, and several of them conflict with one another.

The ethical dimensions of the critique are what give it its force. The authors note that the book is given the appearance of being the published analysis of results from the E4A model, presented as similar to the original Limits to Growth report of 1972, whose impact on the environmental debate is hard to overstate. Yet a scrutiny of the codes of both models leaves no doubt that nothing of the World3 model behind Limits to Growth was included in the E4A model. Public statements that the model is fully endogenous and that nothing is locked up are, the review finds, inconsistent with the actual model code. The authors argue that it is ethically dubious to purport to give science-based advice unless the modelling work is sound and consistent with good scientific practice, and that peer review is central to maintaining the quality, legitimacy and trust capital of the scientific community. Advanced modelling requires a whole-team adaptive learning effort, and ethical responsibility extends to the entire group.

The critique also carries a warning about policy influence. The authors observe that the Earth for All analysis is beginning to gain traction in actual policy development in the European Union and in national policies, citing a policy brief by Potočnik and colleagues that was influenced by the book, thereby perpetuating unsupported claims into advice for the European Commission. Given that the public must largely take model-based threat assessments and policy recommendations on trust, poor or scientifically flawed models and unsubstantiated claims put science-based advice and the legitimacy of the sustainable development field at risk. The authors presented a detailed critique to the Earth for All team and Club of Rome members in January and March 2024, but received no factual engagement or substantial response, a factor they describe as driving the publication of this article. They conclude that it would be inappropriate, and potentially hazardous, to base policy advice at the EU, regional or global levels on the E4A body of work, and suggest that the Club of Rome, whose internal review process obviously failed, needs to act decisively on these shortcomings to preserve its good name and standing in the scientific and policy fields.

Subject of Research: A peer-reviewed critique of the Earth for All book and E4A model finds that the policy recommendations preceded the model, which lacks feedback loops, mass and energy balances, and validation against observed data.

Article Title: Limits to accountability; on the ethics of policy advice based on the Earth for All book and E4A model by the Club of Rome

Article References: Sverdrup, H., Schlyter, P., Haraldsson, H., Ragnarsdottir, K. V., & Dorsey, M. (2026). Limits to accountability; on the ethics of policy advice based on the Earth for All book and E4A model by the Club of Rome. Frontiers in Sustainable Resource Management, 5, Article 1841381. https://doi.org/10.3389/fsrma.2026.1841381

Image Credits: AI Generated

DOI: 10.3389/fsrma.2026.1841381

Keywords: Limits, accountability, ethics, policy, advice, based, Earth, book, model, Club, Rome, scientific research