3DD : 1 to 38 An algorythm question for anyone

Unlike my last post, this question is not just to brag about my own stuff, this really is a question. I will brag more, later.

OK, Ready?

Part A: I want to create a random numerical event in a linear range of 1 to 38. Obviously, it's to feed roulette wheel type games.

We need to create this 1 to 38 range randomnization from throwing three (6 sided) dice all at the same time, and each die is distinguishable from the others. Let's call this "AmeriDice" and color 'em Red, White and Blue.

To create the numeric range, we CAN NOT use any computer random number generators or any external randomizations.

However, we can use a long queue to record a history of rolls we have thus far, and it is OK to "Prime the pump", or in other words, fill up the queue with rolls for a while before it becomes required to produce a 1 to 38 range randomization upon every subsequent roll.

And to make it safe to use in gaming, we need to allow this queue length to varry, so anyone who knows the algorytm and is recording rolls, still won't be able to predict the next outcome.

Part B: Proove it.

I have a program to do this, but I'm not certain about the mathematical fairness of what I came up with.

thanks!

Quote: discflicker

However, we can use a long queue to record a history of rolls we have thus far, and it is OK to "Prime the pump", or in other words, fill up the queue with rolls for a while before it becomes required to produce a 1 to 38 range randomization upon every subsequent roll.

I don't think that'd be legal under gaming regulations. You should ask first before investing time and effort in this approach.

Its deja and his brother vu, all over again! Lets forget about the regulations for just a bit, and assume that we're trying to run an honest game; we need to run some roulette wheels and all's we got's are these stupid "AmeriDice".. HOW (Its an Indian Casino, get it?) ?

Here is my approach, and my question of whether it's statistically a fair way to do it....

Lets say I took the approach that there is already a fair 1 to 36 randomization going on, and its working great for the past 10 zillion years.

Now, all of a sudden, this thug comes along, about every 37th roll, on the averasge, and it says, hey, you arent a 1 to 36 range number, I'm takin' you hostage, and now you are a Green.

And while thats going on, another 1/37 rolls on average, the same thing happens, only these thugs tell you that you're a Double-Green.

Would this be a fair approach to generatine a 1 to 38 range number?


The cat will jump out of the bag next post... OK, clue: its why I'm trying to keep the 3d to 2-D NUMERIC ONLY... It leaves the COLOR HISTORY to act as an independent source of randomization!

Quote: discflicker

Its deja and his brother vu, all over again! Lets forget about the regulations for just a bit, and assume that we're trying to run an honest game; we need to run some roulette wheels and all's we got's are these stupid "AmeriDice".. HOW (Its an Indian Casino, get it?) ?

I get it, I just don't know what tribal jurisdiction allows dice but not wheels. Most allow either both or neither.

Quote:

Now, all of a sudden, this thug comes along, about every 37th roll, on the averasge, and it says, hey, you arent a 1 to 36 range number, I'm takin' you hostage, and now you are a Green.

And while thats going on, another 1/37 rolls on average, the same thing happens, only these thugs tell you that you're a Double-Green.

You need the 2nd thug to occur with p=1/38, and you need the 2nd thug to always beat the first one. But that begs the question - if you can already generate a probability of 1/38 some other way, why bother with the dice at all?

Hmmm. I’ll have to re-check how I handled the second thug.

I know I had to jump through hoops to handle tiebreakers, when both thugs showed up at the same time.
I think I set it up as 2 independent 1 to 37, one each thug, using the color history in 2 unrelated ways somehow.
I know I had some problems keeping the color history to remain independent from the numbers being rolled.

That’s why I wrote this little monitor to check and make sure it works OK….

The reasons why should be obvious by now. Imagine this big casino, shaped like a die propped up on one corner and slowly spinning around in the sunset light of the canyon carved by the Virgin river, somewhere near Mesquite (those washed out home properties will do just fine). Inside, there is no cash and there are no chips. Everyone makes their credit card deposits at the door, and some form of credit-id is used for everything that goes on inside The House Of Cube.

The inside looks like the inside of a die, as expected, so imagine it looks like Circus Circus, and there is a big walk-up stage suspended above, where players who want to be shooters line up like at the buffet downstairs. The entire casino is filled with terminals and kiosks everywhere and players can book bets anywhere in comfort. Up on stage, a single live dealer with a microphone announces the shooter, and the BIG GAME is played with an entire gaming staff of one unskilled stickman for the entire house.

If the shooter gets on a hot roll, and that applies to over 100 different kinds of games, so somebody’s got to be winning somewhere, players can tip the shooter.

Also, the shooter has SHOOTER'S CHOICE, meaning whatever randomization he wants to use, he can, as long as it feeds the appropriate games below (or on-line, as I expect this to be also televised world-wide, and of course I know I'm dreaming, but I'm trying to make it real, that's what I do).


OK here are my SOLID reasons for this conversation and I do appreciate your help, MathEx !!!!

Of COURSE I KNOW THE MAP IS STUPID.

OF COURSE I THINK PEOPLE WOULD NOT WANT TO LEARN IT.


Before "3DD to 1 to 38", ALL I HAD WAS THE MAP. Here is how I do things right now:

3-dice, distinguishable feeds:

3-DD games (1 to 216 range)
Sic-Bo (like a 6 x 2D)
3-D (3-18 3 dice 56 distinct)
1 to 36 range games, which in turn feed 2-D games
2-D (numerically mapped through the 1-36 to be in-synch with the 3D games)
A bunch of sub-ranges under 1-36 and under 2-D lots of stupid simple games can be developed, I offer "Coin-Flip" (1 to 2 range), "Rock-Paper-Scissors" (1 to 3 range), etc.

I need a separate RollStation exclusively for Roulette spins, but I really USE them:
->Roulette games (I got some really cool ones)
->Special combination games including Crapless-Craps WITH DONT BETS, etc
->Any 2D level game or below (Craps to Coin-Flip)

To see what going on, players have to learn the map. But with 3DD->1 to 38, I can now do it like THIS:

3-DD Feeds:

3-DD
Sic-Bo
3-D

AND NOW,

1-38 (From 3DD -> 1 to 38, Feeds a whole slew of Roulette based games)

->Roulette games (I got some really cool ones)
->Special combination games including Crapless-Craps WITH DONT BETS, etc
->Any 2D level game or below (Craps to Coin-Flip)
-> 1 to 36 range games can start from HERE, which in turn feed 2-D games, easy sub-range games, etc…


This up side is that I can now run the entire casino with one 3DD dealer, and there aren’t any 3-d to 2-d maps to remember because players will look at the 1-36 range as being the source for all 2-d dice games

The down side is that there will be some “Null Rolls” required and some of the pure craps players might object to it.

My systems, however, can be configured a whole lot of ways, and any program can run anywhere in the network. So, if players don’t like the big game,, it can be easily broken down into subsets of Roulette-only, pure 2-D craps only, whatever.. It all works in any way!


ANYWAY, getting back to the math discussion, do you think this is a fair way to generate a 1-38 range from 3DD? If you need to run the program you can get a copy, or, since you haven’t run anything yet (which I understand), perhaps you want to see the source code? I guess I’m looking for some kind of a blessing from a math wiz that says yup, this is a fair map.

I will pay in disc golf discs!!!

Thanks

I think you're overestimating the desire of the gambling populace to play with dice. You'll never have a successful casino, in the US anyway, where there are no card games. But more to the point, I agree that the modern casino will get to the point where it looks like this:

Quote:

The entire casino is filled with terminals and kiosks everywhere and players can book bets anywhere in comfort.

When that happens, however, gaming outcomes will be generated by a software-based RNG, not with dice.

So what about the validity of the method?

Quote: discflicker

So what about the validity of the method?

I don't know what sort of proof you are looking for. A proof that your method is random? There are statistical tests of randomness based on samples, but I am hard pressed to come up with a proof. There is Bell's Theorem in quantum mechanics which may have some application, but since I do not understand it, I cannot say.

The statistician in me is skeptical of your method because if you make some event conditional on another series of events, I do not think you have generated a random event by definition. I do not know for sure, but it doesn't pass my "smell" test. I doesn't seem right.

However, do you really need a random event or is an unpredictable event enough? They are not the same.

More to the point, why clone 38 number roulette in the first place. Base your "roulette" on 36 numbers, then you just roll the dice. The other problem disappears. You create the house edge by either shaving some off the payoffs or do the same as in La Boule and make one of the 36 act the role of 0 and/or 00, or have a rng pick a "spoiler" for each roll. Or you can charge a commission on wins as in baccarrat or Pai Gow. I think more would play this game than 00, which imho is a ripoff game.

Or, if you must keep 0 and/or 00 roulette you could make it a two step game. In the first stage, Use a rng to generate 0 00 or numbers with the required probability. If 0 or 00 occur, game over. If not, then roll the dice for the 1 to 36 outcome. This is really no different that the dealer getting blackjack. At that point the game is over. If no blackjack, the game continues.

All my games feature something called TheBoxer allowing them to box any two, three,.. events. If a player wants to play Traditional Roulette, for example, he can "box" the Red 15 times in a row, and it will pay out like a lottery.

But I am not trying to change any game at all. Traditional Roulette nor any of my own games, none of them change becauase of the randomization method that happens to feed it. That is the exact point of this whole thing and the basis upon which my RollStation works.

I need to create a 1 to 38 range because it is needed for Roulette. Any by the way, I also need to create a 1 to 37 range for European Roulette games. And while we're at it, why can't this invention concurently produce a 1 to 52 range for you know what. And hows' about doing that for every 1 to n range up til 52 so each of them can be used to seed an automatic shuffler of a virtual deck of cards dealing a live game of Hold 'em?

I am finding this site to be helpful, but everyone is also encouraging me to forget about what I'm asking about because they dont see what I see in it.

Can we stick to the question, please?


The question is: Using 3 distinguishable dice, can I generate a fair 1 to 38 range number on every roll. No random number generators are allowed, everything determined must result directly from the sequence of rolls. However, previous rolls can be used in any complex mathematical way required, and the queue of numbers may be pre-loaded before the 1-38 range numbers start to be produced, and then they must be produced on every subsequent roll. Also, a variation in processing must be available to alter the outcome so that anyone who knows the algorythm and is also tracking roll history, still can't make any predictions without knowning the present (secret) value of this variation.

And prove it.


Thanks!

Quote: discflicker

I am finding this site to be helpful, but everyone is also encouraging me to forget about what I'm asking about because they dont see what I see in it.

Can we stick to the question, please?

The question is: Using 3 distinguishable dice, can I generate a fair 1 to 38 range number on every roll. No random number generators are allowed, everything determined must result directly from the sequence of rolls. However, previous rolls can be used in any complex mathematical way required, and the queue of numbers may be pre-loaded before the 1-38 range numbers start to be produced, and then they must be produced on every subsequent roll. Also, a variation in processing must be available to alter the outcome so that anyone who knows the algorythm and is also tracking roll history, still can't make any predictions without knowning the present (secret) value of this variation.

And prove it.

You're getting the responses you're getting because this is a public discussion forum, not a client consultation. Many of the readers here are either in the industry to some degree or have been around it for a long time. I personally don't see a use for the machinations you're going through with three dice, especially if you envision the players not actually throwing them (which I took from your "kiosks and terminals" comment).

What you're proposing is a reasonably complex process and I doubt you'll get too many people offering to help you for free. Also bear in mind that if someone else invents a technique that solves your problem above, and there is no contractual duty to assign it to you, then you can't legally obtain a patent on it. Your IP counsel should be able to educate you on these matters.

Invention in the gaming space is not for the faint of heart. There are a few successful inventors on this forum, and a few others who are trying, but it's really, really, really hard to break through. I've seen a myriad of new dice games at G2Es in years past, only a small handful made it into live play -- and over half that do die out quickly.

I am not discouraging you, I like the basic idea. But we have a conceptual problem. What exactly are the characteristics you desire for the 1-38 numbers? "Fair" is not a well defined term. If you mean random that is one thing, if you mean unpredictable that is another. If you want a mathematical proof that a method is random, I don't know if it even exists in theory. However, is it not true that using prior data and arithmetically manipulating it to get a random number is in itself a random number(pseudo) generator and you have ruled out using such a generator. I think it was von Neumann who said:

"Anyone who considers arithmetical methods of producing random digits is, of course, in a state of sin."

I just lost some work ... let me start again, sorry.

The 1 to 256 randomness of the dice throws produce all of the randomness (if there is such a word) this method requires.

OK, I will spill the beans and show how I did it, but I humbly ask everyone to tell me if, at least in theory, this produces a fair mapping. Please see my previous post for a description of fair. that was very first thing I spoke abouit.

I am generating a 1 to 36 range number upon every 3 DD roll. However, I am doing so according to the numeric-only mapping described in the previous post. This occurs ONLY upon the 56 distinct combinations generated from INDISTINGUISHABLE dice. Because of this, the position of any number amoung dice within a roll by color, as well as the color of any dice within a roll, and the history of these positions and colors, HAVE NOTHING TO DO with the 1 to 36 range determinations. These just happen to be happening at the same time, and are independent of it.

So, that leaves color and number position free to act as an independent source of randomization, and how I justify using them as an outside influence.

Next, I keep track of a history of these numerical and color histories in a long queue, and map this into a count of 37 of them times the secret safety factor. Next, I use the MODULA function against this count, and on average, every 37 rolls, it equals zero and the "right" thug appears, and replaces the 1 to 36 range number with a green. I'm also doing this at the same time for color history changes unrelated to number position, and this becomes the "left" thug.

Because the 1 to 36 range number is always random, I can kick it out at any time and replace it and still maintain a random distribution over it.


wew!


What do you think? Is this fair?

thanks!

If you use a 'secret safety factor' rather than an RNG it sounds as though this could be vulnerable to getting cracked.

Also, not sure if this simplifies it at all, but you could just generate 1 extra outcome, using the color, and then use the total on that particular roll to signify either a '0' or '00'. This would only require '1 extra thug' rather than 2.

At last, an answer by someone not trying to talk me out of business!

Yes, the idea of the safety factor being tied into an RNG is OK with me, but the current roll also plays a part in the thug determination (the current roll is added into the queue before the mod functions are performed upon the summation through them), so even if you knew it all, you still wouldn’t be able to make any predictions other than if a 0 or 00 does come, which one it might be because of my tiebreaker history queue.

As far as two thugs, I'm thinkin’ of using 'em both in the 1 - 38 game but only one in the 1 37 game, not sure why, I supposed I can get it working with only one. My EXTENSIVE testing program (shown above) proves that “thuggin’” really works, providing uniform averages and deviations from aspects including each 1 to 216 number, each of the 56 distinct 3I die combos, from the 3d point numbers, the 2-d point numbers and from the synthetic 1 to 36 range created in-between. For each of these I took running deltas, and tried to find any significant deviations or repeating patterns among them after 100,000,000 rolls, and I found nothing outside of 0.000013 % accuracy.... I think it works, but want someone to certify it for me, and I will pay for it. Need any disc golf discs? Hehe!!

I do NOT understand how anyone can’t see any value in this. After throwing dice for centuries, nobody other than myself and another inventor named Gramera (see my patent) have attempted to obtain ranges not aligned within the 216/56-36/21-6/1 families? Nobody thought about running a roulette wheel from a random 3-Dice throw?

Seriously, I just don't like computers.

Well, I like programming 'em but I don't want to play any game where they determine anything because it IS fake any way you look at it. And just trust me on this one, I have put a lot of thought into what is fake and what is real in more ways than one.

If most people knew that slot machine reels aren’t actually spinning these days any more, that they're just fancy displays, I bet half of 'em wouldn’t want to play any more.

My ideas are pure in the sense that this is something that can be accomplished, actually, IS accomplished by a PHYSICAL occurrence RANDOMLY occurring in our real world.

I liked the comments by MathEx, but his ideas on gaming and where it’s going are negative and inconsistent. Of course were gonna automate everything I don’t understand why we don’t use chip less live gaming a lot more than we do. I don’t know why he kept confusing this with RNG gaming, why, when we go chip less, we must go RNG, or why, when we go BIG GAME, we must go RNG. I wouldn’t play any games like that for the same reason they still use real balls in huge live KENO games. It’s the same idea, and yes of course the randomizations could start in the keno room rather than on stage in the House of Cube, but then I would loose the Live Draw, as he called it.

Of course some of my plans and ideas are stupid and unreasonable to implement, but I have a whole system, top to bottom, right now, in my pocket. This discussion is just a tiny little part of it. MathEx has no idea of what I have but keeps discouraging me, and hey guess what, I am learning a lot from that too.

If MathEx needed to create a new game, I can do it in minutes starting from an amazing database I created that he knows nothing about. I can define new wager classes, actions and payouts, event create the button icon pointers all within the database. I can use a simple visual layout tool if needed to put the button where I really want to see it, and then its all done... a new game in MINUTES, it comes up in the list of games available within the currently defined dealer domains, also database defined and changeable ion-the-fly in real-time as the games are being played, along with everything else. The player simulation testing tool implicitly works on it as well, so MathEx could perform statistical testing of wagering results and obtain an EMPERICAL answer without even having to write a single equation.

Please check out game number 1 in the games thread.


Thanks for your time

Anybody else wanna check my arithmatic?

Marty

Quote: discflicker

Of course some of my plans and ideas are stupid and unreasonable to implement, but I have a whole system, top to bottom, right now, in my pocket. This discussion is just a tiny little part of it. MathEx has no idea of what I have but keeps discouraging me, and hey guess what, I am learning a lot from that too.

If MathEx needed to create a new game, I can do it in minutes starting from an amazing database I created that he knows nothing about. I can define new wager classes, actions and payouts, event create the button icon pointers all within the database. I can use a simple visual layout tool if needed to put the button where I really want to see it, and then its all done... a new game in MINUTES, it comes up in the list of games available within the currently defined dealer domains, also database defined and changeable ion-the-fly in real-time as the games are being played, along with everything else. The player simulation testing tool implicitly works on it as well, so MathEx could perform statistical testing of wagering results and obtain an EMPERICAL answer without even having to write a single equation.

I'm not trying to discourage you - I'm trying to suggest a practical, realistic view of the business. I've worked with far too many game inventors with inventoritis to think that every single one of them has the greatest gambling product ever. I'm not even sure what your product is, so I clearly can't comment on it more directly. I just think that the idea of rolling three dice and trying to use some complicated mental gymnastics to convert those results into other dice results won't be popular.

I don't know what you mean by creating a new game in minutes. Are you referring to some sort of online casino software?

Yes its all part of a package I been working on for the past 10 years full time, and the first game is shown in the Games threads.... Ricochet... didn't you say you played it? Check it out..., and that's just the tip of the burg. Actually, I know that your not trying to discourage me, only trying to save me some time.. I been to G2e twice now, I'm starting to learn the ropes. So thanks, and please check out the Ricochet description.


I'm still looking for some officlal certification for " 3DD -> 1 to 38"

Quote: discflicker

I'm still looking for some officlal certification for " 3DD -> 1 to 38"

The fastest way to get what you're looking for would probably be to offer to pay for it, if you're looking for an answer ASAP.

Just sayin'.

Im in no hurry for this because it is already specifically mentioned in my pending patent.