Tuesday, March 25, 2014


I recently read Creation, by Steve Grand, which came highly recommended by Amazon CEO Jeff Bezos. Grand is an expert in artificial life, but he tries to remove the "artificial" as much as possible by modeling his creations on real life, including evolution, learning and other features of real life that aren't natural to the way computers are as we know them.

To illustrate with an example, when Grand asks a group of intelligent people to create a computer simulation of a ball bouncing, they usually come back with a sinusoidal function's (the shape of which somewhat mimics a bouncing ball) absolute value (so it doesn't bounce through the floor) with a decaying co-efficient (to mimic friction and elasticity, which reduces the speed and height of the ball over time). This is the wrong approach. What Grand emphasizes about creation simulations is that they should be modeled on real life as much as possible. A better approach would be to model the ball's co-ordinates based on its current speed and the effects of gravity (e.g. g of 9.82 m/s acts on the ball's speed in the vertical direction, thereby changing the velocity of the ball's vertical, thereby altering the ball's y-co-ordinate position), friction and the floor.

Grand applies this same line of thinking to creating artificial life forms. He goes into an extensive discussion of the properties of life at their most basic, in order to model artificial life on a basis that makes it best mimic real life.

The book was definitely interesting, but for my taste it delved into the philosophical a bit too much. Grand has apparently written a very popular game in which players create artificial life, and so at times the book came across as a bit of an ad for the game.

I read somewhere that Amazon has based some of its services on some of Grand's principles for creation, which is kind of neat. Enjoy!

No comments: