Implementations Of True Random

 There are various standards of randomness. We can distinguish between a square distribution and the bell curve. Beyond that, some tests calculate probability of a number repeating, two and three times, for cryptographically random. In the same family of tests, we can test for localized order, or surges of bell curve fitting data interspersed by data that displays higher entropy.

With these thoughts in mind, most people are willing to accept series of range 0 - 9, from ten sided dice when they are cast in groups. This rule is to compensate any tendency to repeat a “side,” like a similar problem that attends coin flips. 

To be vigilant against human interference, we can plan to cast five ten-sided dice repeatedly, and line the resulting group up in a row. This prevents the transcriptionist from choosing orders from the available five choices (sometimes in an effort to avoid that very failing). 

The strategy of lining the experiment up in a line is a good rule to have for all dice testing.

Another strategy is to roll twenty or twenty-five ten-sided dice in a large cup, analogous to a Yahtzee cup. It is sufficient to group these five at a time.

Ten-sided dice have a zero, so they can be used for ranges like 0 - 1000, or 0 - 100,000, by recording the results appropriately.

A 0 - 60 range (for sexagesimal notation) can be contrived by rolling two groups of five dice: a six-sided set, and a ten-sided set. The first digit of a 0 - 60 number is developed by decrementing the relevant six-sided observation by one, for a zero to five range. The second digit accepts the ten-sided instance that pairs with it.

Six-sided dice do give a contiguous list, if you record groups of five numbers as five digit base six. However, it obfuscates the range. To guarantee a range of 0 - 7775 (decimal), record the six-sided die casts, and decrement each die face (or recorded digit at a later moment) by one. That satisfactorily yields a 0 - 55555 (base 6) range.

We have noted that coin flips tend to become biased by the side facing upwards when it is flipped. We don’t usually think of it, but a six-sided die, cast fairly on a hard surface will yield a very satisfactory random distribution of odds and evens. One and zero are merely an alternative way to record the results.