With the planet heating up, the urgent need to reduce uncertainty in climate projections could have consequences in the way Earth system models evolve. I argue that a period of confusion in the development of models can be beneficial and can lead to improved understanding and reduced uncertainty in due time. I have identified three phases of development in modeling—the illusion, the chaos, and the relief—through which models seem to evolve before their results can be said to approach the truth with some confidence.
Whereas these phases are based only on my limited observation of the field, they underline the fact that answers to scientific questions are usually ‘unknown’ before they become ‘uncertain.’ Realizing in which phase a model component may be could facilitate their development. I give examples from carbon cycle modeling and highlight the benefits of encouraging new and audacious approaches in modeling.