I received the book “Oscarmetrics: The Math Behind the Biggest Night in Hollywood,” by Ben Zauzmer. I liked it; it was a lot of fun, a good mixture of stories and graphs:
This one is my favorite:
Also this:
I also passed the book over to a student to review:
OscarMetrics is, first of all, written in a very interesting way. The format is that every chapter is a different category in the Oscars. The way that the graphs are set up shows a degree of professionalism in how the book is constructed, and they show data in a very informative way, though sometimes not completely clear. The questions used for each chapter get a bit repetitive, and each chapter ends up using an incredibly similar format. The book starts to drag near the end, and I think that the author shouldn’t have done the chapter gimmick, and focused more on the more impactful awards. He even mentions this in the section about the Popular Film category that never was.
I finished OscarMetrics. I don’t think that it is an especially great book, but it was interesting enough to keep me engaged for the time it took to read it. Some interesting bits were the facts that nominations for best editing are one of the biggest qualifiers for winning best picture, and the bits about the Oscar voters vs. regular critics and audiences. According to the book, the Oscars slightly prefer movies that others like, but every once in a while they buck the trend and choose a memorable bad movie compared to the others. The chapter on best pictures quantifies the greatest upsets in Oscar history, and it unsurprisingly is Crash (23.5% chance to win), beating Brokeback Mountain (73% chance to win), in 2005. Each chapter gave a bit of interesting information, but none of it was really memorable enough for me to be able to write about it without looking at the book again. The ending chapter that talked about the movies most deserving to get Best Picture was interesting, but it didn’t really have as much to do with his method, and more just him aggregating other sources about great movies. However, I had never heard of many of the movies written about, and I think that this book really can be used as a list of good movies.
It’s interesting to compare my impression with the student’s. I think we’re coming in with different expectations. The student sees a book and is expecting something to be read from cover to cover. I see an applied statistics (“analytics”) book and what I want to see is some new information or a new synthesis, ideally in graphical form.
Thanks for blogging about this book. While you have posted several of its most interesting charts and graphs, they are descriptive in a limited or naive sense. There is a significant hole in Zauzmer’s analysis…extreme value theory…which doesn’t even rate a dismissive mention. Arthur de Vany’s 2004 book, Hollywood Economics: How Extreme Uncertainty Shapes the Film Industry, was and, based on the evidence, still is the most deeply mathematically insightful book about the film industry.
What’s really amazing about EVT is the timidity with which it is approached and dealt with (or not) by otherwise well-informed people such as yourself. Another good example of this lack of acknowledgement of the high relevance of EVT is Anita Elberse’s 2013 book, Blockbusters. Elberse’s book is about ‘tentpole’ strategies in entertainment and film where one or two blockbuster box office hits relative to a string of mediocre performers can and will make a studio profitable for a year. One of de Vany’s key points is that any failure to consider the high uncertainty that extreme values (outliers) bring to any stats 101 analysis is not merely misleading, it’s erroneous.
This is happening despite EVTs high relevance, even ubiquity, for so many natural phenomena and human behaviors.
These pictures are nice examples of how far you can go in statistics just by being clever and inventive. Without need to learn any theory or subscribe to any paradigm.