Modeling data to win an argument or solve a problem
Modeling data to win an argument motivates us to make assumptions that are often baked into the modeling process.
There is a better way: focus on solving the problem. It starts with “I don’t know”, and it takes creativity and an open mind to find out. The data may or may not be there. We may need an experiment to get the data. The method we use to model the data doesn’t matter anymore. Methods become tools. More importantly, focusing on solving the problem limits our assumptions to those we have to make to get from the data to a model for decision making. So we focus on data centricity.
The pleasure of winning an argument will always be there, but perhaps we can avoid it in favor of better decision making and problem solving. And even if we can’t avoid it, we’re probably better off making an argument to learn, not to win.