Wednesday, 20 April 2011


The seeds for the forest were sewn playing chess. When he was 8 year s old, Andreas came face to face with the great Chess Grandmaster Kasparov. Pawn in hand, he made the first move. But Andreas was not alone. With him were 5 other 8 year old german chess players too, all as eager as he to make the opening gambit. And so it was, that a team of 6 8 year olds beat the great grandmaster of the game – proving that six heads are definitely much better than one – even the biggest and best one.

“People say that there are no scientists like Einstein today. But I say that doesn't matter. Much better to have a crowd of scientists not as good as Einstein, than to have one Einstein. My experience with Kasparov showed this.”

So says the now 48 year old Andreas, who is one of the leaders in the field of probability at CERN. And when he starts talking, you realize that the public and social trend for crowd sourcing in fact has a truly mathematical as well as experiential basis for delivering results. That in fact it is maths which really has created and driven the crowd sourcing phenomenon of today. We are merely following the numbers.

“We have developed Boost Tree Decision Making. Basically, you plant a decision and then go down the decision tree, until you get to the leaf – the most probable and true answer,” says Andreas. It is truly as simple as that. Or so it sounds when an expert describes it so simply in words. The algorithms to get there are another matter.

But in fact Boost Decision Tree Making does not stop there. In fact it has been shown that you don't just plant one tree to get the right answers. You plant many trees, with short roots, leading to quicker multiple answers. It turns out that one tree does not a forest make because it doesn't give you the full complete picture or the true answer after all. This is because if you just plant a decision and follow the roots to get to the answers, they will only come from one source and one starting point. Thus they can only be a binary variant on the same tree, rather than creating a whole forest of trees, which gives you the whole landscape of a forest to look at and check the verity of your answers. At least that is how I understand it. It gives a whole new slant to that expression not seeing the wood for the trees.

This algorithmic path to truth also has social application and others too which are part of our daily lives.

“Take your iLife on your mac” says Andreas. “You put in your photograph and it will go through the internet looking for matches, following an algorithmic path. That’s boost tree decision trees at work.” It works like magic - and like magic, it is in the end, all about numbers.

But that is not all. The Boost Decision Trees method also has a military application too, becoming the tool for decisions about making air strikes and selecting targets – an altogether much less palatable application for what was conceived of as a tool to link not destroy humanity. The twin sides of technology – revealing itself in the forest of our psyche.


No comments:

Post a Comment