"Allie" wrote...
Thanks for your response, Harlan. Your link doesn't seem to be
the same issue I am talking about here. The link says that large
arrays can get non-random, and that there are repeat sequences
within 500 tries. The apparent bug I noticed is not that.
....

Not exactly the same, but another aspect of the poor quality of the
pseudorandom number generator in the ATP. In short, it's unwise to use the
ATP's random number generators for anything meaningful. Presumably Microsoft
has aware of the problem for years, but they haven't fixed it. That's not a
guarantee that they never will fix it, but it's pretty clear it's a very low
priority.

Another point made in the article I linked was Dave Braden's mention of
alternatives. You really can't use the ATP with any reliance, so you need to
find alternatives.