How to Predict Everything

Physicist J. Richard Gott of Princeton published a provocative article in Nature back in 1993, that described a simple method for the estimation of the likely lifetime of "things" on the basis solely of the length of their existence to date. Using the hypothesis that a current measurement is chosen from a random time sequence, one can use statistical methods to estimate the 95% confidence limits for its total period of existence. Journalist Timothy Ferris spent time interviewing the scientist, and explains his central idea in a way that most people can understand. Gott has applied the method to Broadway plays, restaurants, and species, including homo sapiens, with provocative results. He argues that his calculations give an additional justification for the colonization by human beings of other planets in our solar system! I think that the examples he chooses to use are ones in which a Poisson process is likely, and that he avoids cases where there are systematic mechanisms that would cause it not to work. For example, his method would predict that a person of age 100 would be likely to live to 110 or 120. Nevertheless, this is still an interesting article that shows how statistics can be creatively applied to "real life".

Publication information
Pick Attribution: 

Timothy Ferris

Publication Date: 
Monday, July 12, 1999