SCAMMING SHUFFLEMASTERS
Quote: EvenBobReminds me of comparing a real roulette wheel to
a computer RNG. If you don't have a winning
method, how can you compare the two in
the short term, where the actual bets are made.
You don't need to make any bets to determine whether a real wheel is behaving equivalently to a computer RNG. The only thing that matters is the numeric outcomes and their distribution.
Quote: AyecarumbaIs there a "perfect" or "optimal" random state? Every PRNG is going to produce sequences of "clumps" (same values or sequenced series). It would be expected, and suspicious if not included.
If there is not an "optimal" state to compare to, then how do you measure the "randomness" of a particular shuffle?
A particular shuffle is just an ordering of cards. The algorithm for ordering those cards is what needs to be assessed for randomness. It's like asking "is 5 a random number"? It depends on how you arrived at 5. If you rolled a 6-sided die and it showed 5, then yes. If I'm answering the question "what's 3 + 2" then no.
Focus on the algorithm for generating the outcomes, not the outcomes themselves. An RNG generates numbers randomly -- it does not generate "random" numbers. Technically there is no such thing as a "random" number even though that phrase is in common usage. Cases like this show why it's misleading.
Quote: WizardYou could pass that test and still have a very non-random shuffle. For example, just moving the top card to the bottom.
The deck would still have a correlation in that case, though I don't know the exact maths to test for it. I would suggest if you did a series of cyclical tests (off set the new deck by 0, then 1 etc) you cover the cases where there was just a cyclical change.
I see that you could also go to several orders (look at the correlation between two cards that were next to each other in the new deck, then 2, then 3).
I understand that you can white box test any randomizer by looking at how it generates it's random effects, but this is proposing a black box test. Look at the outcomes and see if they conform to the expected result (in this case a shuffled deck), using a limited number of trials.
As for the roulette case, the betting system IS just a expression of a mathematical test. They are just different versions of the same thing. If there is a short term pattern (3 red in a row means there's more likely to be black in the next two spins, for a random example that is probably way off base from EvenBob's) you can test it mathematically by looking at short term correlations, or you can go ahead and bet it.
Quote: WizardThis makes me wonder how to set up a statistical test for a random shuffle. I thought about this a little and the best idea I can some up with would be to count how many cards remained consecutive and in the same order (reversing would be acceptable). You should expect one set in a random shuffle, but I think a shuffle was too short would result in more. One downside is that it is not very robust to test on something where the mean outcome is one. Then again, I could just perform the test lots of times, but that would get time consuming. Any other ideas?
Follow up: Since posting this I found the academic paper that concludes that seven riffle shuffles are needed to randomize a deck: Trailing the Dovetail Shuffle to its Lair (PDF). The math is rather heavy, but it looks at the number of "rising sequences." It may take some time for me to digest it.
Wiz, you might get some ideas Here
Quote: WizardThis makes me wonder how to set up a statistical test for a random shuffle. I thought about this a little and the best idea I can some up with would be to count how many cards remained consecutive and in the same order (reversing would be acceptable). You should expect one set in a random shuffle, but I think a shuffle was too short would result in more. One downside is that it is not very robust to test on something where the mean outcome is one. Then again, I could just perform the test lots of times, but that would get time consuming. Any other ideas?
What if the measure of randomness was not how far the cards were from "perfectly mixed", but how far the cards were from their fresh from the factory, starting sequence (Kc-Ac, Kd-Ad, Kh-Ah, Ks-As, or whatever is the typical, "factory" sequence)?
For example, every Rubik's cube can be solved in 20 moves or less, so "20 moves from solved", can be considered the most "random", "19 from solved" the next and so on.
Similarly, if each card in a deck was assigned a, "starting" value (As=1, 2s=2,...Qc=51, Kc=52). Decks could then be summarized by totaling the squared differences between each card's starting and current positions. Let's call this sum, "P".
Something would have to be done to factor in sequencing. I'm thinking that you could go through the deck comparing the starting value for each card to the one following. Total the squares of these differences and call it, "S".
Would the comparison of P and S be fair measures between decks? Could the P/S ratios be compared?
Quote: WizardI have never seen anything other than anecdotal evidence like this that the machines are not fair.
The same for the single-deck shufflers in Three Card Poker. Debates have raged for years among hole carders about those being fair. I've heard lots of APs singing the blues about their results against those, but nobody has proven anything non-random as far as I know.
In neither case I'm saying the machines are fair, but in America one is presumed innocent until proven guilty.
Would you ever consider analyzing the double-deck Shufflemater Deckmate?
For what? An extra chamber where only cards 2 through 7 are held and dealt to the dealer but only when APs are present? This kind of conspiracy would require a whole lot of Black Jack dealers to remain silent. I'm not feeling it. If however, you still think that hundreds of dealers can keep their mouth shut about the secret button then knock yourself out.Quote: JuniorWizWould you ever consider analyzing the double-deck Shufflemater Deckmate?
Those dealers already remain silent about the secret Oxygen-Pumping button and the secret signal that keeps CWs away from a table.Quote: s2dbakerFor what? An extra chamber where only cards 2 through 7 are held and dealt to the dealer but only when APs are present? This kind of conspiracy would require a whole lot of Black Jack dealers to remain silent.
BUT : Just when I thought I was out... they pull me back in.
Quote: Calty23So here is my take.. the Shuffle master machine works in a RNG way just like a slot machine.
I don't claim to know how the Shufflemaster shufflers work, but I have seen inside the single deck machines and it's like a wheel with a bunch of chambers. I can't think of any reason why there would need to be a RNG involved. The machine could take the cards from the same chambers in the same order and the cards would still be random enough. Why? Because (at least where I work) when a new deck are put into play they are washed, and riffle/striped/riffled twice before going into the shuffler, and when the cards have been used they are riffled before being put in the shuffler. This combined with the shuffler not knowing how many people are playing. or how they will play (in the case of blackjack) means that the cards get changed every time they're used.
Quote: Calty23In PaiGow Poker the machine knows exactly where the dealer is at all times.. they are #1 or #8.
The shuffler doesn't know what number is going to be generated so it can get the dealer hand in the right position. The cards are shuffled and ready to go before the number is generated.
Quote: Calty23with that said the machine can also rearrange scrambled cards and place them in complete order.
Yup, the single deck shufflers do have a "sort" mode. If you think it's doing it when not in sort mode I don't think there's much else to say, except there's a reason talk radio hosts hang up on conspiracy theorists.
Quote: Calty23On top of that most casinos have now installed Verify which is a LCD ITouch that can read any card that comes in its proximity..
I have no idea what you're talking about, and either does google, I checked.
Quote: Calty23How they get around the Illegal side of things is that since Information Technology from up stairs can control the randomness you can never catch them at this time.. but I am trying..
So you're saying that in addition to surveillance, the casinos have IT people watching the game and "goosing" the shufflers to make you lose? Interesting.
Ya know, casinos haven't always used automatic shuffling machines, but they have always had a house edge. It's hard for me to understand why anyone would think they would risk their gaming license to make a few extra bucks (minus the cost of those IT people, of course) when they're already going to beat the majority of the players in the long run. Perhaps another look at statistical mathematics is in order, specifically how standard deviation and law of large numbers works. The house edge works because the casino plays a single players lifetime of hands everyday. No rigged shuffler required.
Great to know when in a casino that deals down to the last card. I forget the names of the ones on the strip that do that . Anybody else know the names ?
Idiots right in my nick and there is always room for two.
Quote: IdiotWheelSorry I bumped it, just adding my thoughts on the difference between what might be called an ideal/perfect reordering of set/deck and what you might call random reordering.
Idiots right in my nick and there is always room for two.
There's nothing to be sorry for.
