Super Crunchers....

Thanks to the recession, Super Crunchers, published in 2008, now has an interesting taint.

The central thesis of the book is that analysis of huge databases (Super Crunching) is leading to computer models that make better decisions than people and that maybe, just maybe, people should hand more decision making to these models as time and time again these models yield more accurate predictions than people themselves.

And then came the mortgage melt-down, the collapse of hedge funds and the banking crisis. All caused, in no small part, by people trusting computer models more than their own judgement.

To paraphrase Allan Greenspan, When you take a risk, if you are not willing to live with failure, then you probably shouldn't take the risk, no matter the upside. How easy it is the forget this. The result is bubble think. People making unncessary risks when they get a big cut of the success and get to blame the failure on someone else.

And that's the problem. A computer may make more accurate predictions than a man. But a computer can't take responsibility for its actions. A computer doesn't have to live with failure. A great computer model may make more accurate choises than I. But, I have to make those choices and live with the consequences.