Fair Dice

spb

Recruit
Joined
Jan 28, 2003
Messages
10
Reaction score
1
Location
Silicon Valley
Country
llUnited States
FuraxVZ said:
I contend the problem is with the rnd() function in small sets of data. In other words, each throw of a die is NOT 1/6. Given a result, the following result is influenced by the previous one. This is inherent in the rnd() function.
There are tests for this sort of dependency. I've run them. They don't support your
claim.

And you seem to be suggesting that you can find odd variations from expected values
if you grab small datasets. This is no surprise at all. If you're suggesting that the data
only support your contention if the dataset is kept sufficiently small, then I'd suggest
that the data do not support your contention at all. If indeed there was a dependency
present, then more data would make that dependency more apparent, not less. Granted,
there can be PRNG failures that don't effect the uniformity of the PRNG output given
a large enough output stream---this is true, but it is also irrelevent. Most of the tests
I've run on the dicebot output are designed to find exactly the sorts of dependancies
you describe. None of them indicate that the dicebot output varies from random output
in any statistically significant manner.

If you have any data or analysis which suggest this is not true, then feel free to
offer it. In the absence of such evidence, and in light of the analysis that suggests
that the dicebot is `fair' in this context, then it seems to me that continuing to argue
for the opposite is just superstition. I.e., ignoring evidence in favour of maintaining
some a priori belief.

If you didn't see my analysis on the ASLML, you can find two of the messages I sent
to the list here:

http://www.meshuggeneh.net/dicebot.txt
http://www.meshuggeneh.net/dicebot2.txt

If you find any problems with my methodology (or wish to question the validity of the
underlying techniques), feel free. I'd honestly be interested in discovering any
real failure of the dicebot.

But---again---just suggesting that the numbers look suspicious (particularly if you
grab small samples) just doesn't mean anything.
 
Top