Did you ever wonder why or how scientists use models for their research? Models are widely used in science – probably every field of research uses some kind of model – but different research fields think differently about them! This is a big challenge in interdisciplinary projects like HESCOR, where scientists from different disciplines try to bring together their different models. On the one hand, social scientists often use „implicit models“ that qualitatively describe phenomena, for example, the interaction of humans with their environment. On the other hand, natural scientists use „explicit models“ [Epstein08] that quantitatively formulate phenomena, for example, the movement of a particle due to an external force.
But what is a good model? Just looking at the huge variety of models developed by different scientific disciplines, it becomes clear that there is no such thing as „a good model“ or „a bad model“. The British statistician George Box said in his famous quote: “All models are wrong, but some are useful”. So we have to be more precise on what we expect from a model for a certain application: (1) Why do we want to model something? (2) What do we want to achieve? (3) What information do we have? And how can this information be useful? (4) What can we actually achieve? And (5) How do we determine what we cannot achieve?
Let’s take a look at the first two questions: Why do we want to model something? It sounds trivial, you might simply say “we want to model this because we want to answer this and that question”. But what exactly is that question? And what can the model contribute to it? There are plenty of possible answers, but you can group them to categories like: explore, specify, test, explain, generalize… It makes a big difference whether you want to use a model to specify the amount of humans in a certain region, or to explore different the reasons that humans migrated there in the first place!
Once this is determined, we can ask: What do we want to achieve? Accuracy? Complexity? Efficiency? Generality? Consistency? Diversity? Understandability? Falsifiability? … All of these aspects are desirable, but they often contradict each other. For example, there is the famous Bonini’s paradox saying: the more complex a model, the less it is understandable [Wikipedia]. A nice example for this is the map-territory-relation (maps are models of the actual landscape): The most complex and potentially accurate map has a 1:1 scale, but it would be completely useless. When you would unfold the map, it would reveal the same exact image as if you just looked at the landscape itself! So it would be neither understandable nor efficient to have a map of 1:1 scale. Instead, you would prefer a map that balances complexity and accuracy with understandability and efficiency, depending on its intended use.
Discussing these questions becomes particularly important in an interdisciplinary context because different scientists will give different answers! Therefore, in HESCOR, it’s critical that we are open to dialogue and honest conversations about why we are doing something and what we want to achieve in order to come up with the most effective model-based results – and that takes time!
References:
[Epstein08]: Epstein, J.M. (2008). ‚Why Model?‘. Journal of Artificial Societies and Social Simulation 11(4)12. https://www.jasss.org/11/4/12.html[Wikipedia]: https://en.wikipedia.org/wiki/Bonini%27s_paradox (last access in Nov. 2025)
Image Blog Teaser: Unsplash
