From Model Land to Reality: Navigating the Complexities of Net-Zero Modelling

Models are crucial tools for the transition to net zero, but can be misused in the fast-changing fields of energy systems and human behaviour. Author Erica Thompson joined the Transition Accelerator’s Normand Mousseau and James Meadowcroft to explore the challenges and pitfalls of applying models to decision-making in a wide-ranging conversation.

Here are our highlights:

• It’s a myth that models are objective and tell us the objective truth: “This is completely wrong,” says Thompson. “They don’t. They are social objects… the onus is really on the modeller to be able to say, ‘Yes, I believe that this model is useful. It’s useful in the following ways.’”
• Tracking north Atlantic storms didn’t lead Thompson to a career in climate science, but it prompted the statistician to question how models are used in solving climate challenges. During the COVID-19 pandemic, she observed, models became part of the “the machinery of debate” to frame a dialogue around public response. “It was a case study of how models are fed through to decision makers, policymakers and how they are used in practice… not just to support decision-making, but also to justify decisions that are made with many other pressures around them.”
• Building model relevance into real-world policies and decisions requires escaping from “Model Land”, according to Thompson. “Inside the model, everything works, but nobody cares. Nobody has any interest in what happens within the boundaries of your model unless you make a claim that it has something to do with the real world.”
• The ability of a model to successfully influence debates, policy discussions, and actions depends on answering a number of ethical questions first: Whose judgements are represented in your model? Who decides what is “politically feasible”? What is included in your model, and what isn’t?
• Effectively bridging science and policy requires a two-way dialogue between the modeller and policymaker, and training decisionmakers to ask the right questions. “It’s not just the case that the modellers are coming along and throwing this information over the fence”, says Thompson.
• Assumption-driven models using any kind of social component — pandemic behaviours, voting results, or the energy mix in 2100 — can be unpredictable. Modellers can help this process by asking what assumptions have been made, where they come from, and whether modellers have consulted widely enough to have confidence in these assumptions. We have to be careful in extrapolating, warns Thompson: “You move quite rapidly into a regime where the assumptions that you’ve made are at least as important as the data, and possibly a lot more important.”
• To get more diversity in modelling, we need more diversity in modellers. “The assumptions that we make… are very much framed by the backgrounds that we have as modellers,” says Thompson. “Many modellers come from very typical sorts of white middle-class backgrounds, and we have certain sets of experiences, and we don’t have other kinds of experiences. And so we have blind spots and biases.”