Contextually Speaking

Reading is simply the fuel of thought. We eat to get energy for life. We read to get energy for thinking. Of at least to have something to think about. I just finished reading Signal and the Noise. Prediction. I have often noticed that fantasy football or simply trying to pick the winners of the games offers some pretty good insights into trying to figure out what's going to happen next in all kinds of different arenas. That book is a very big expansion on the idea that looking at how predictions work in very select cases offer general insights into making other kinds of predictions.

While reading this book, I realized that a good chunk of what I do professionally is about prediction. We expend all kinds of resources to develop a process that can produce something with very predictable behavior. Nate Silver is a big proponent of Bayesian thinking. My self-definition of this idea is that data has context. Abstracting a series of observations as some small sample of a much larger series and drawing predictions from that information, the standard statistical approach taught in the typical statistics class, robs data of meaning. In Bayesian thinking, the more you understand the context of your data, the better your prediction.

In thinking about making predictions, I thought about what kind of things I could do to train myself to think more probablistically. I looked in a statistics textbook that I have from one of my MBA classes. The discussion of the Bayesian theorem and conditional statistics in general was about 5 pages. I'm not surprised. That textbook was a piece of highly conventional crap (which reminds me of another post I mean to write about whether more education makes you more conventional in your thinking) so I'm not surprised that it's pretty free of discussion of anything interesting. But as I was looking through this book, I realized that I'm already pretty good at making predictions in the context of existing information.

One of the biggest decisions that I've made in my career over the last couple of years was to lead the development of a few new products. I made this decision after participating in one of the first conversations that we had with a potential partner company. Having been part of the team that dealt with the company that we were looking to replace, I had a good understanding of the risks involved in this type of project. When I asked myself if the situation was likely to repeat itself with this new company, I estimated that the probability of that would be pretty low. I was warned about the risks of taking on this project, but I trusted my assessment of the situation (a situation where I had better information than the people giving me advice).