Benford's law and fraud detection, or: Why the IRS should care about number theory!

Steven J. Miller, Department of Mathematics and Statistics

Faculty Lunch Talk, Tuesday, October 21st, 2008
 

This talk will be a 20 minute, less technical version of previous talks. The abstract below is just for illustrative purposes, as I haven't yet figured out exactly what I want to say (I know I'm slacking, but I only found out a few hours ago I was speaking!).

TENTATIVE ABSTRACT: Many systems exhibit a digit bias. For example, the first digit base 10 of the Fibonacci numbers or of 2ⁿ equals 1 about 30% of the time; the IRS uses this digit bias to detect fraudulent corporate tax returns. This phenomenon, known as Benford's Law, was first noticed by observing which pages of log tables were most worn from age -- it's a good thing there were no calculators 100 years ago! We'll discuss the general theory and applications, talk about some fun examples (maybe the 3x+1 problem or the Riemann zeta function), and discuss some current open problems suitable for undergraduate research projects.

For more information, email Steven.J.Miller AT williams.edu

Slides of a version given at the Boston IRS headquarters are available here (or click here for a no-tab version).

Slides of this talk are available here.