Quote:DJTeddyBearYes. At least with the brand machine I was watching.

When it is time to roll, the machine does little bounces to "shake" the dice. While this is going on, the players watch, and the one "stooter" eventually presses a button to "shoot" them. At that point, there is on huge bounce, with very unpredictable results. when the dice finally settle, the sensor reads them and pays the winners, etc.

FYI: I'm fairly certain that the "sensor" is just optical. I.E. The camera really does look at the dice and recognize the spot patterns. There isn't any electronics in the dice to "tell" the machine how they landed.

There isn't any electronics in the dice but there is electronics in the device to "launch" the dice. How about that.

Quote:100xOddsi've seen back to back 12s in e-craps.

i've also seen back to back 12's on a live table.

i hope the casino is tracking the #s rolled, and the gaming commission looking over these #s every one in a while to see if it's still random.

i also wonder how often they change out to a fresh set of dice?

It's just a little bubble. I just don't know how they can make it completely independent. As everyone knows, you only need a tiny bit of influence/bias in craps to overcome the house edge.

I hope this isn't coming off as a crazy conspiracy theory, because I don't think it is. I just wouldn't be quick to assume that these machines are completely random.

Quote:WizardIn my opinion class II games are rather silly and not very relevant to this topic. Nevertheless, I'd be interested to hear directly from Eliot about his point on the topic of autocorrelation.

Actually, it is very relevant to this topic because it has to do with pseudo-random generators and associated "random" outcomes. My question to the Wizard is do you test for randomness using standard tests for autocorrelation or the run test. Or, is your standard approach is to use the Chi-Square test like what the Class II machines have been doing in the old days?

In the cases regarding Eliot Jacobson, I hold him to a higher standard because he advertises he has a PhD. I was really shocked by his response and the fact he would remove his very strongly-worded comments.

Quote:WizardI'm still confused about what game we're actually talking about. Are we talking about what many call "bubble craps"? If so, is the theory that the dice are unequally weighted or somehow correlated?

Here's the website from Interblock that also makes an electronics craps game http://www.interblock.eu/products/G4/dice/ You can see that it has physical dice and it in a self-contained "bubble".

It would be nice if the other people could post a link to their version of e-craps.

What is autocorrelation?

Quote:WizardI'm still confused about what game we're actually talking about. Are we talking about what many call "bubble craps"? If so, is the theory that the dice are unequally weighted or somehow correlated?

I'm talking about bubble craps, and my theory is that the possibility exists that one roll may influence the results of the next roll. I believe the dice are weighted equally, and the results over time are distributed properly. However, that does not mean that one roll could not help predict the following roll.

Quote:Ardent1Here's the punchline -- software drives the launch speed of the ball and software tells how fast the wheel has to spin. Get my drift.

Nnnnnnnnnooooooooooooooooo comment.

Quote:FleaStiffOkay, I understand that the difference between RNG and Pseudo-RNG is something that fascinates programmers and mathematicians for hours on ends but is really nothing but senseless hair splitting for anyone else.

What is autocorrelation?

We don't have RNG, they are all pseudo-RNG, and we just call it RNG for ease or because we are lazy (so basically there was never any hair to split to begin with). Here's a good article about problems with pseudo-RNG http://vpgenius.com/articles/random-numbers.aspx

Autocorrelation is when the error-terms in a data series are correlated. If you have autocorrelation, you can't have randomness. Going back to the vpgenius article, you can make a strong case for autocorrelation in the "Old Method."

Quote:WizardNnnnnnnnnooooooooooooooooo comment.

Good answer.

I tried to clock the ball launch speed and I am positive there are at least 5 different ball speeds so clearly the software is discerning at what speed the ball is being launched at. I searched the internet and there is ample information about air pressure to launch the ball.

I also tried to clock the wheel, and again, I am positive there are at least 5 different speeds at which the wheel initial spins at and again this is also driven by software.

So basically, my position is there is like a 5x5 matrix or 25 cell of different combinations of ball speed and wheel speed. All my assertion is that software is picking one of the 25 cells right before the ball is released instructing the device on how fast to launch the ball in one direction and to set the speed of the wheel in the other direction.

The other form of randomness is due to the 16 or so deflector pins.

Quote:FinsRuleI'm talking about bubble craps, and my theory is that the possibility exists that one roll may influence the results of the next roll. I believe the dice are weighted equally, and the results over time are distributed properly. However, that does not mean that one roll could not help predict the following roll.

"exists that one roll may influence the results of the next roll" -- I would argue that is autocorrelation.

The big question that no one is addressing is "how" does the electronic device cause the the dice to come to its final outcome. I.e. how does the machine "simulate the "throw" of dice.