[to determine what may be interesting about a possibly-random number]

The problem is easily stated, and easily come-across, but automation could be tricky. Given a suspicious number, output from a log or bad file, we should be able to quickly determine whether the number fits any "interesting" criteria. If it does, then the problem is more-than-likely to have its roots in code somewhere (the specifics of which may become obvious when the reasons-for-interest are reported).

Recently, VISA was in the news for charging several people seamingly random and very-high amounts, such as $23,148,855,308,184,500.00. Looking at the number in hex makes the error obvious:

20 20 20 20 20 20 12 50

The sequence x20 repeats six times at the beginning of the number. Noticing a repeating pattern in the hex representation of a number should be easy, as might noticing that a number is almost exactly equal to UINTMAX (most of the number's binary representation is "1", etc.