Search

Book Review: Supercrunchers by Ian Ayres


In Blink, Gladwell presented a view that in order to make an effective decision an expert’s decision is far better than market studies and statistical analysis. Supercrunchers by Ian Ayres is the perfect antithesis to this view.

Ayres opens Supercrunchers with two anecdotes, one where an expert statistician outperforms wine tasting gurus as to the price of wine for several years. The other story is of a statistician who outperformed experienced talent scouts when it came to pick rookie baseball hitters.

He goes on to talk about regression and its superiority over expert opinions in predicting future outcomes. Unlike expert opinions regression also predicts the accuracy of the outcomes it predicts.The book transplants the reader into a magical world of large quantities of data(Terra bytes and Peta Bytes).

Here are some of his examples of companies using regression.
Farecast is a site used to predict when to buy an airline ticket.
EHarmony is one of many sites used to match singles based on several compatibility criteria.
Google works though petabytes of data and has an ambitious plan to scan every book ever published and offer search functionality on them.
A casino uses regression based on its customers’ profiles to determine the pain-point for its customers. When a customer reaches this point, an employee comes to the aid to the customer and pulls him or her away from the slot machines aby offering perks. The result is that the visit to the casino a pleasurable experience for the customer and will make him or her come back to play again.
Statistics can also be used to predict how a judge may vote based on a certain case.
Credit card companies use spending patterns to predict the stability of marriages.
The IRS holds so much information on us. This information could be used by the population for beneficial purposes.


He tries to make a case for EBM or evidence based medicine where a doctor can use statistical data to make decisions on patients instead of mere theory he or she learnt in med school.

He also believes that Hollywood could use analysis to predict the probability that a movie will be a hit, long before a single scene is shot. The analysis is made purely on the screenplay.

Ayres also talks about the use of Neural Networks for evidence based medicine. Neural Networks unlike regression cannot predict the accuracy of their decisions.

In the last part of the book, Ayres tries to win votes for his theories by showing the user the value of standard deviations and how they can be used in decision making. He also describes Bayes’ Theorem in great detail reminding me of the ERBA(Enterprise Risk Benefit Analysis) class I took at MIT.

I loved this book and think it, like Blink must be required reading for the SDM program as it offers another perspective on decision making, this time in favor of decision making techniques taught at MIT. Big surprise, as the author is a PhD from MIT. Reading this book was time well spent. The best part of this book is that it is only 217 pages long, though its content at times was a bit ‘heavy’.

blog comments powered by Disqus