Gaming the System - Random Number Generator Exploitation

This is question is intended for MichaelBlueJay or anyone other WoV with knowledge of the matter. In a prior post regarding betting systems, it was suggested that the system had to properly simulated/tested to avoid any exploits of the RNG program.

Well an RNG program is just a pseudo set of random numbers. I'm assuming if you had the coding for the RNG, you could theoretically predict all future numbers based on the code. So what kind of system would actually exploit flaws in the RNG? I'm guessing it would have to be some kind of system that places a series of on and off bets based on certain preconditions being triggerred.

And if there are systems that are able to exploit flaws in the RNG, why is it so hard to win at slot machines? Is it because professional RNGs are more advanced?

Does anyone remember the story about the former NGC member that exploited flaws in the RNG of Keno machines before getting caught? He "claims" to have uncovered a flaw in the RNG and was successfully able to beat the system at least before getting caught.

You're talking about Ron Harris, and he didn't uncover a flaw, he created it. When he went to "inspect" the machines under the auspices of the GCB, he gaffed the ROMs and made them beatable (by him).

Backing up a bit, no, just the code of an RNG won't allow you to predict future outcomes based on a running program. You'd need to know the initial conditions too - seed values, etc. Most good RNGs use an entropy source for that, like thermal noise or some suitably random time element, and if you don't know that precisely even knowing the algorithm won't help. If it's a *bad* RNG, sure, it's beatable. But even then, for something like dice rolls, even (int)RNG() % 6 + 1 is very hard to predict in the short run. It may be biased, but it's not predictable using a betting system.

The exploit done with the online poker deck worked because they were using a deck of cards, and once you can map RNG outcomes to shuffle orders you only need a few cards to know what the rest of the deck is. That doesn't work with dice.

A problem with simple RNGs is that there is a cycle of numbers. For example, with Microsoft Visual Studio the outcomes repeat every 2^31 random numbers draw. That is about 2.1 billion. It repeats odds and evens every 2^17, or 131,072.

When I first issued my challenge I was not aware of this, so had no rule to guard against it. So somebody accepted my challenge. His system was to remember the first 10 outcomes in roulette, and then wait for the first 9 of them to occur, in order, and then bet on the 10th one. That was clearly an abuse of bias in random number generators.

The one I use now I don't think would fall victim to that. However, if I redid the challenge, somebody could read on my site that I now use a Mersenne' rel='nofollow' target='_blank'>https://wizardofodds.com/ask-the-wizard/randomnumbers/]Mersenne Twister. Such a person could get the same sequence of random numbers I get, and could easily exploit my challenge. People have been known to beat Internet casinos by using various common RNGs until they figured out the which one the casino was using by the pattern of results.

Quote: MathExtremist

You're talking about Ron Harris, and he didn't uncover a flaw, he created it. When he went to "inspect" the machines under the auspices of the GCB, he gaffed the ROMs and made them beatable (by him).

Ron Harris did in fact mess with slot machines,

but also proved that the rng of Keno at the time could be cracked. That is what tripped him up in Atlantic City, hitting a jackpot with a friend on day one.
Here is a google preview of a very nice book to read.

License to Steal: Nevada's Gaming Control System in the Megaresort Age
HERE

I knew Larry Volk, the person that was murdered at the beginning of the book.
The NGC and the State of Nevada did NOTHING to protect him from the knowledge he had about casinos "fixing" their slot machines.
A must read book for what happened back in the 80s.
I will not say too much about the book, but Ron Harris was in charge of that case and new Larry Volk was in danger for his life by the things he knew.
I do not condone what Ron Harris did, remember he knew as much also, he only was in jail 3 years as I recall, and he probably feared for his life also. What made him "crack" ? only he knows.He did not steal lots of money as compared to Wall Street thieves, but he was and is labeled a criminal to this day.

Quote: Asswhoopermcdaddy

Well an RNG program is just a pseudo set of random numbers. I'm assuming if you had the coding for the RNG, you could theoretically predict all future numbers based on the code. So what kind of system would actually exploit flaws in the RNG? I'm guessing it would have to be some kind of system that places a series of on and off bets based on certain preconditions being triggerred.

There was a case of this in Montreal. A Keno player noticed that when a particular machine was unplugged - so that housekeeping maintenance (vacuuming) can be done on the casino floor, the machine started up with the same numbers as its last "cold start." Apparently, that particular machine did not re-seed it's start. He played the numbers (10 out of 10 Keno or something like that) and won big. He was the only player at that time of the overnight, and noticed this.
Then they discovered the re-seeding problem, and tried to apply criminal charges to him. They failed on that.
I lost the link to this case, but is was a pure RNG case.

As for sequences now repeating, theirs a random re-seed inserted between generations, to break up statit sequences.