Hamilton starts with a brief overview of the algorithms proposed for generating pseudorandom numbers with an exponential distribution. However, the majority of this paper focuses on a paper by Ahrens and Dieter [1]. Despite the fact that the Ahrens and Dieter paper was published in 1988, Hamilton claims to be the first to report errors in their algorithm that keep it from working. (He has also found a book that reproduces the algorithm complete with errors). Most of the errors are minor, but it is disappointing that a very short algorithm would contain three errors. Hamilton provides a copy of the software that implements the algorithm, but this was not included with the review copy. The body of the paper does include a pseudocode implementation of the algorithm. The algorithm runs when converted into a programming language, although I did not test to see if it produces the exponential distribution.
In order to reduce the risk of this happening in the future, perhaps journal editors should require authors to submit an actual program in addition to the pseudocode, as well as the first ten numbers generated by the algorithm when it is started with specified seed values as, for example, Brysbaert [2] did. People who need to generate random numbers that follow an exponential distribution should read Hamilton’s paper.