I saw a talk today with (for once!) some useful advice. It was by Bala Ravikumar and he discussed Benford’s law, which says that in “real life” the first digit of numbers tend to be small. (Benford’s law basically says that the distribution of miscellaneous data is uniform on a logarithmic scale. On a logarithmic scale, e.g., the range [1.0, 1.999…] is several times as large as the range [9.0, 9.999…]) More specifically, he noted that tax fraud has previously been detected in people whose tax numbers didn’t confirm to the law.
So… if you feel like faking a tax return, make sure about 30% of your numbers start with a “1” and only 4.5% start with a “9”!
In other news, I gave a talk today which I developed out of a previous “better know a theorem” post; the talk is here. I was punked in the Q&A question by Jim Geelen 😦 who pointed out that integer linear equality systems can be solved in P-time, essentially, using Gaussian elimination (although then he mumbled something about a Hilbert basis).
Filed under: math, probability, talks, tax | Leave a Comment