Signal and Noise

A Reflection. Week 12

We make forecasts and predictions in every aspect of our lives; from predictions to how our lives will chart out to forecasts on quarterly profits for the company. Nate Silver, author of The Signal and the Noise: Why So Many Predictions Fail — but Some Don’t, seeks to understand the nuances of predictions. Why do some people make the large claims of the future backed by only their opinions/knowledge, while others take a more meticulous look at the data and try to distinguish between signal and noise.

We live in an age where we have data on everything. This data helps us understand what is happening on a massive scale, and we are able to take this information, sort out between what is fluff (noise) and what is significant (signal).

There are two types of people when it comes to predictions. The first is what Silver calls Hogs, are those people who you see on Fox News who are loud, obnoxious, and make these grand claims of how an event will turn out. The second are the foxes, which are people who seek to understand every piece of the problem, figure out the nuances, and sift through the data they have in order to come up with a probability that some future event will take place.

The thing with predictions is that they are probabilities. There is a 70% chance that X will happen. Statistics is used heavily in big data in order to take a massive amount of information and make sense with it. The problem is that many times the models that are used are incorrect. It is very likely that there is a bug among that 10,000+ lines of code that is used to build the model, and thus you cannot take the model for certainty. Even then, the datta gathered can be biased. The use of data is up to the user to make conclusions about. It is very easy to gather specific data on an issue so that your point is proven, and leave out other types of data that will weaken your argument.

Nate Silver, in his research and work on forecasting, discusses various examples in different fields where predictions are correct/incorrect. His discussions range from baseball to poker, the financial crisis to global warming. In each area, he clearly articulates how predictions can go wrong while others are more likely to be correct.

What is refreshing about this book is the importance on human judgement. Technology is a tool. Data is a tool. Human judgement and analysis of the tools is what is truly important. A lot of fields are too nuanced to simply leave it up to data-driven models to give us answers. Economics is deeply nuanced and far too complicated to have models make conclusions. During the financial crisis, many of the economics models forecasted that the housing market will continue to rise. Companies that relied strictly on these models were hurt as a result. Economists who used these models in conjunction with their deep understand of the nuances of the economy, were able to pick up where the models left off, making much better predictions of the downfall of the economy.

I definitely recommend the book. I think it was a bit long, but the stories are worth it. It will definitely give you a fresh perspective and quick-eye when people make prediction of what will happen in the future. Check out my collection where i am reading 52-books this year. Next up: Mind of the Maker, by Dorothy Sayers.