Modeling Like an Experimentalist

More and more of modern biology relies on models — whether simulational or analytical, mathematical or statistical. In my role as editor at Ecology Letters and Proceedings of the Royal Society B, I read a great many modeling papers. Over time, I’ve come to notice that while computational workflows have become commonplace, the way we think about modeling has not always kept pace with practice.

This led me to ask a simple question: how might we write better modeling papers? And behind that, how might we become better modelers?

My answer is that we should think more like experimentalists.

In my new paper in Ecology Letters, “Modelling like an experimentalist,” I suggest that modeling be understood as a kind of experiment — one with treatments, levels, and responses, just as in bench or field work. Framing computational work in this way can clarify design, reduce drift, and make our communication more transparent.

The paper also discusses a few habits of what I call the “experimental modeler”: validating with simulated truth, using sensitivity analysis to understand robustness, and employing deliberate perturbations to detect artefacts.

Thinking this way, I argue, doesn’t constrain creativity. It structures it — helping us to ask clearer questions and to build cumulative science.

You can read the paper open access here: https://doi.org/10.1111/ele.70251