Useful or accurate?
We have a tendency to discard a model because it is not entirely accurate. In doing so, we ignore the fact that it may still be useful.
To throw out the useful because it’s not accurate assumes accuracy is absolute. But even science is not absolute. Even the great Newton’s theories have proved inadequate. If even the seemingly soundest scientific theories are approximations, and need correcting, what sense does it make to discard our imperfect but useful model?
Take fiction. It is a person’s imagination. It is not a scientific way to understand human behavior. Yet, most would agree that reading Shakespeare reveals a lot about human behavior. It is useful.
When the world didn’t have Daniel Kahneman and Amos Tversky, they made do with the Bard. In fact, most of the world still does. Have you heard your mum say, It’s Danny or nothing?
Sometimes, the usefulness of a model emerges much later with time. Take the case of alchemy. Debunked as a pseudo-science for centuries, it led to the birth of chemistry. Even what’s obviously wrong may contain the seed of a transformative truth.
We often don’t know the flaws in our seemingly perfect model. But when we do, we can’t help but wait for perfect data. And opportunity passes us by while we wait.
Or we overfit data for our model to show us what we want it to. Anyone who has ever hired for their employer knows it is not uncommon for debriefs between interview stages to bias their opinion. Your mind is half made up. Then you go through the interview and back-score the candidate in a way that matches the initial assessment shared with you.
When we don’t have perfect information, we manufacture meaning and relevance, especially where what we see doesn’t support our theories. When we are so flawed ourselves, how can we aspire to perfection?
Tell me a useful model you use in your life that you know is not accurate but works for you.
