Nina's Reading Blog

Comments on books I am reading/listening to

Archive for October, 2013

The Signal and the Noise: Why So Many Predictions Fail–but Some Don’t

Posted by nliakos on October 20, 2013

(wondering if I should capitalize¬†so in the title. . . It isn’t a conjunction here. On the cover, everything is lowercase. Aha, inside they capitalize it. So will I.)

by Nate Silver (Penguin 2012)

This is the University of Maryland’s First Year Book for 2012 – 2013. It’s about predictions and probability and statistics. It’s kind of interesting, but there is much about math, statistics, and probability that went right over my head. Charts and tables leave me cold. The main takeaway for me is: Don’t trust other people’s predictions because they don’t know as much as you think they do (or as much as they think they do, either). This goes for predictions about baseball, politics, earthquakes, computer chess, climate change, economics, online poker, influenza, the stock market, gambling and enemy or terrorist attacks. There is a chapter about each of these, and more.¬†

Silver would like everyone to base predictions on something called Bayes’ Theorem. Thomas Bayes lived in the 18th century; his theorem is about approaching the truth gradually by gathering more and more evidence and revising our predictions based on this evidence. You start out with a “prior”–an initial guess about something. And you guess the probability of something being true or false. But essentially you are changing your prediction as you accumulate more facts that do (or do not) support your hypothesis. He spends the second half of the book saying that if we used this method, our predictions about almost everything would improve. I am probably over-simplifying this, but it seems to be an important part of his message.

I think there’s a lot of good information here, but it went into a lot of detail that I could have done without. All in all, it was not one of my favorite First Year Books.

Posted in Uncategorized | Leave a Comment »