Lots of mathematics but the conclusions at the end are important..
http://statweb.stanford.edu/~susan/papers/casino.pdf
Lots of mathematics but the conclusions at the end are important..
http://statweb.stanford.edu/~susan/papers/casino.pdf
For people who are too lazy to read the whole article, the conclusion from the article is
The study above shows that a single iteration
of a 10-shelf shuffler is not sufficiently random. The president of the company responded
“We are not pleased with your conclusions, but we believe them and that’s what we hired
you for.”
Please read the study. Seriously, I hate to support casinos. But, I hate worse to libel anyone. And misleading info is not helpful to APs. Jumping to conclusions is a sure path to failure.
Last edited by Norm; 04-03-2017 at 03:53 PM.
"I don't think outside the box; I think of what I can do with the box." - Henri Matisse
This study was preliminary based on a design. There wasn't even a physical device to study. There wasn't even an alpha test. Do not suggest crimes without evidence.
"I don't think outside the box; I think of what I can do with the box." - Henri Matisse
Random = unpredictable
Sufficiently random = unpredictable for X% of the population.
A single deck BJ game can be played such that the cards are picked up from the previous round and dealt without shuffling, just a cut, and most people could not predict the next card. For those people the deal would be sufficiently random. The typical 3 passes and one strip single deck shuffle is sufficiently random for everyone. It takes 7-9 passes to completely randomize a single deck, FWIW.
You could probably find an even bigger discrepancy by looking at the correlations between pairs of cards instead of their absolute change in position. For a random shuffle the histogram of gap sizes between formerly adjacent cards would be linear because there are 51-k ways to have a gap size of k. Shelf shufflers would have zigzags in that histogram corresponding to the number of shelves. And human shuffles would be biased towards smaller gaps unless they do half a dozen perfect riffles, in which case they'd be biased towards larger gaps.
Bookmarks