Thursday, September 20, 2007

The End of Intuition?? Oh really?

This week's Economist features a book review of Super Crunchers. The book purports to explore how massive data banks automate decision making in such a way that the human element will become less important.

EVERY time a world-class chess player loses to a computer, humans die a little. In this book Ian Ayres, a professor of law and management at Yale University, explains how in many less high-profile endeavours, human intuition and flair are more easily beaten. The sheer quantity of data and the computer power now available make it possible for automated processes to surpass human experts in fields as diverse as rating wines, writing film dialogue and choosing titles for books.

The author originally intended to call his book “The end of intuition”. He changed his mind after a Google AdWords campaign which randomly chose which of two advertisements for the book to display: “Super Crunchers” garnered 63% more clicks than his original choice. He tells of credit card companies that are using similar randomised trials to see which combination of offers and advertising make for the most successful mailshots.

And so begin the wild predictions that doctors will cease to be needed for their diagnoses, teachers for their teaching. The world will go the way of the automaton. We'll all be plugged in and data slaves.


Pardon my skepticism here. Technologists have been playing this game for quite some time. But data, no matter how crunched and sliced and diced, is only as good as the use it is put to. One of my Rotary colleagues says "Information without action is wasted" I might modify that in this scenario "Good information without right application is wasted."

Take for instance, the above scenario. The title "Super Crunchers" got 63% more clicks. And it is likely that google was even able to give some demographic data about those clickers: income, interests, etc (all of course anonomized to protect individual identity). That's nice. However, what if it still turns a bunch of people off (frankly, I would have preferred "The End of Intuition" as a more interesting title. Super Crunchers sounds like a new brand of potato chip). What if the people it turns off are really the types of people who the author wants to reach?

More to the point, can we really automate teaching or doctors visits? We may be able to design a general program that works very well for the broad populace, but what about children who are exceptions. Our oldest spent Kindergarden in a Montessori School and now she is in a Classical Christian School -- two opposite models, both of which claim to be tested and superior to all comers. In my experience and observation, I can say that each model works for some kids and not for others. Each model also produces very different end results, and part of the choice lies in what end results parents want from their children's education. I don't think parents want an algorithm deciding for them the ultimate purpose of education.

Neither does this take into account people's adaptability (most models never do). Banner ads were ubiquitous, until people simply started to disregard them. The big wave in advertising now is "product placement" -- every time you see a Pepsi or an Apple computer in a movie, you can bet that some cash was exchanged to make sure that it was a Pepsi, not an RC cola. However, once consumers become aware of it, they become partly innoculated, and it becomes something of a joke.

Maybe I just need to read the book to get it, but I'm so turned off by the title....

From a more positive angle, another quote from the article:

Even the occasional government is accepting that properly analysed data trump ideological conviction. Mr Ayres sings the praises of Mexico's Progresa/ Oportunidades programme, which gave assistance to poor people only if their children attended health clinics and schools. It was tried out on 506 randomly selected villages. The results were so convincing that the programme was expanded 100-fold despite a change of government.

Here we see a good use of data. Rather than breathlessly extolling how data will make human decisionmaking obsolote, here we have data that actually improves human decisionmaking. That's where we will be going. Access to crunched data will not diminish humanity, it will however become a powerful tool in the hands of decision makers.

What they do with that data will make all the difference. And let us remember that decision makers are sinners like the rest of us. Once again, we're back to the orientation of the human heart. Good data in the hands of Stalin is frightening. It all boils back to Sin and Redemption.

And now we're back on the theologians turf...

Soli Deo Gloria