Saturday, December 29, 2012

Super Crunchers

Host's Note: This is the first time, I am hosting a "Guest Post" in my blog and I must confess that I am only too happy to do so. Mukund's comments about a book he finished recently. 
After reading his comments, I am tempted to think innumeracy may equal illiteracy very soon in future.

Comments are more than welcome in the blog.



In this day and age, number crunching is no longer restricted to the class room or the text books. Unknown to thee, there is a customization happening at every walk of life - whether one likes it or not. That, in essence, is what "Super Crunchers" by Ian Ayres is all about. 

Before getting into the crux of the matter (or data), I have to commend the author on one dimension. In this era of being politically correct to further one's own interests, the author comes across brutally honest. Ayres has no problem laying the truth as is - quoting Sun chief's 90's statement of "there is no privacy, just get over it" - or taking on his fiercest of academic critics and laying behind-the-scenes events threadbare: like the exchange between the infamous Mary Rosh, and Ayres-Donahue combine. 

Now, onto where the real fun is....the book opens to an interesting wine collection story, to entice the audience, like a good on-stage presenter would with a nice, personal anecdote. Then on, all the way to the "distraction techniques" at Duke, in the end game of the book, the author keeps the audience hooked. Ayres also seems to have a way with anecdotes. Having heard the term 'Standard Definition' over two-dozen years in a variety of contexts and contents, I have not seen a simpler way of describing than he does in the words of his daughter Anna. Not just that; for working class readers that want to be hands-on, there's an exercise on what the mean and standard deviation are, for adult males in the US. In the interest of not providing a spoiler alert, I will not provide the answers here!

Another area where the author excels is delving into the human psyche. To quote: "humans are not only prone to make biased predictions; we are also damnably overconfident about those predictions and slow to change them in the face of new evidence". How very true! The example mock test he makes you take on the verge of explaining this concept and how we play "safe" when defining boundaries of error are truly eye opening. 

The beauty of number crunching, for even the non-believers, is that it is for real. Behavioral modeling or targeted messaging has never been in greater prominence as it is, today. Starting from a professional context, to routine tasks like shopping, watching / following sport, there is a set of lurking machine-eyes behind the scenes - to make that experience better, improve retention, etc. Privacy was/is always a concern, but when statistics show that folks apparently were willing to share social insurance number for a 50-cents off coupon, you know that collective opposition does not translate to individual behavior. Again, an instance of data debunking a 'myth'!