Full House: The Spread of Excellence From Plato to Darwin

From Wikipedia, the free encyclopedia

Full House: The Spread of Excellence From Plato To Darwin is a book by zoologist Stephen Jay Gould, published in 1996.

In Full House, Gould demonstrates how one type of statistical misconception leads to misunderstanding of important phenomena. The misconception is paying attention only to the "high score" or extreme value, when a continuous distribution of values exists (what Gould calls a "full house") and is what actually drives the phenomena.

The book focuses on two main examples of this misconception: the disappearance of the 0.400 batting average in baseball, and the perceived tendency of evolution towards "progress" making organisms more complex and sophisticated.

In the first example, Gould explains that the decline of the top batting average does not imply that there has been a decline in the skill of baseball players. Quite the contrary: he shows that all that has happened is that the variance of the batting average decreased as professional baseball got better and better, causing the extreme value of the distribution—the best batting average—to decrease as well.

In the second example, Gould points out that many people wrongly believe that the process of evolution has a preferred direction—a tendency to make organisms more complex and more sophisticated as time goes by. Those who believe in evolution's drive towards progress often demonstrate it with a series of organisms that appeared in different eons, with increasing complexity, e.g., "bacteria, fern, dinosaurs, dog, man". Gould explains how these increasingly complex organisms are just one hand of the complexity distribution, and why looking only at them misses the entire picture—the "full house". He explains that by any measure, the most common organisms have always been, and still are, the bacteria. The complexity distribution is bounded at one side (a living organism cannot be much simpler than bacteria), so an unbiased random walk by evolution, sometimes going in the complexity direction and sometimes going towards simplicity (without having an intrinsic preference to either), will create a distribution with a small, but longer and longer tail at the high complexity end.