Thanks for reading and commenting

Yes, it is a very intersting topic indeed.

The natural phenomenon was used in the product but the 0-1 distribution wasn't good enough. I came to know that unpredictability might not agree with good statistical characteristics. A "massage" process was employed to eliminate the bias.

The source of the unwanted bias was unknown. It might come from the natural phenomenon or from the acquisition process. The solution made the source irrelevant.