According to Stafford Beer, the purpose of a model is primarily to assist in making predictions.
I share his perspective. Various models are constructed to somewhat accurately represent the system they model. This representation is only needed for the purpose of cheaper forecasts for changes in the actual system.
For example, engineers make a small version of an airplane and test its aerodynamics in a wind tunnel. This helps to predict how the real airplane will act in different wind conditions. This method is cheaper and easier than testing a real airplane.
Neuroscientists develop a biological model of human behaviour, they see how the brain works under stress, and can predict that cognitive performance typically declines in such conditions. Using this model is cheaper and more ethical than checking if your employees will be productive and stay healthy under stress conditions.
If a model fails to predict the represented system behaviour, then the model is useless. Using such a model is either worthless or even harmful. Imagine what would happen if the wrong model was used by airplane engineers to predict how the plane would operate in the sky?
Using a flawed model is worse than having no model at all simply because having no model means we are not blinded by the wrong model!
Several examples illustrate the pitfalls of employing non-functional or inappropriate models:
1️⃣ The adoption of pseudoscientific behavior models such as phrenology, astrology, numerology, and human design.
2️⃣ The application of a group model in team settings, focusing on individual efforts and responsibilities while overlooking the collective dynamics essential for a team’s success