Are the rolls stacked in the computers favor, absolutly not.

[Jonas]

The cup in effect is just like adding a lever to a digital slot machine, purely decoration. EVEN if evey wiggle of the cup was factored in to the calculation, its still just the same thing. It might feel good, but its not going to change the statistical outcome.

What was the purpose of the cut deck/shuffle feature in the card games? Was that to randomize the deck or was that purely cosmetic?

thats kinda a philosophical question. Its does what it claims and does effect the outcome, but it doesn't effect your chances. If it was flipping a coin, you still have a 50/50 chance of getting tails. The fact you can call it, doesn't change your odds.

It sounds like what you're saying is random numbers are being "constantly" generated and when you click the board you get what you get.... Is this not an event driven (board click) game (number generator)?

I just had a chat with our chief coder, and basically the last change we made was that what effectively happens now is that there is a stream of random numbers, and when you click to roll the dice, you snag the next one in the stream.

So right now your human element is technically picking the dice rolls. So think of it it this way, your playing Lets make a deal, and Monty Hall gives you 10,000 doors to pick from, and they are all whizing by. When you say "now" it stops a whatever door was next to you, and you get the dice behind that door.

So the sequence is entirely changed at each time the die is rolled. (ie the rolls are not predetermined, you determine them by rolling the dice)

Of course, it really doesn't change the outcome statistically, as you have no way to manipulate the outcome, just like calling 'tails" doesn't increase the odds that your coin will land with tails. Its was just added to eliminate any possible prerolled weirdness. It's as random as it gets.

[Dust In The Wind]

I feel that the "cosmetic" change would be beneficial whether or not it changes your odds of getting the roll you desire and the opinion of players would result in better trust in the randomous of the roll. IE: I shook 3x and got the dbl 6's but next time I got the 2/3, it left me a only choice of leaving myself "open" and had I let it go on the 4th shake I would have gotten a better roll....

This leaves the player in control of how many shakes to determine when it picks from the random universe of possible rolls and would be a positive step in a direction that players would appreciate since they essentially get to choose when to choose.

I hope you in my explaination how the perception factor would actually be a reality check in choice, Yes odds still holds that the random number could be any of the possible rolls BUT I choose when to choose!

JUST DUST

PS - I really feel this would be the best solution to the perception problem in BG of rolls not being random and takes it away from the server and puts it back into the players hand. Hopefully I got my point across and DRow I am supporting your position (LOL).

[Death_Row]

I just had a chat with our chief coder, and basically the last change we made was that what effectively happens now is that there is a stream of random numbers, and when you click to roll the dice, you snag the next one in the stream.

This sounds like like (from Random.org):

Pseudo-Random Number Generators (PRNGs) : As the word ‘pseudo’ suggests, pseudo-random numbers are not random in the way you might expect, at least not if you're used to dice rolls or lottery tickets. Essentially, PRNGs are algorithms that use mathematical formulae or simply precalculated tables to produce sequences of numbers that appear random.

The basic difference between PRNGs and TRNGs is easy to understand if you compare computer-generated random numbers to rolls of a die. Because PRNGs generate random numbers by using mathematical formulae or precalculated lists, using one corresponds to someone rolling a die many times and writing down the results. Whenever you ask for a die roll, you get the next on the list. Effectively, the numbers appear random, but they are really predetermined. TRNGs work by getting a computer to actually roll the die

PRNGs are not suitable for applications where it is important that the numbers are really unpredictable, such as data encryption and gambling (Dice).

[Jonas]

I'll read over that page Death.

Effectively, the numbers appear random, but they are really predetermined. TRNGs work by getting a computer to actually roll the die

Predetermination really isn't a problem in this case unless the players had access to the future rolls, or a way to calculate them which they wouldn't. In fact, I think that it's your human involvement of clicking to roll the dice that picks from the stream of predetermined numbers, to the opponent would not only have to know the predetermined numbers, but then also know exactly when you we're going to click to know what you we're gonna roll. So it's like a computer shuffling a deck of cards, and you picking the card you want, just like the cut deck feature.

So you are an added unpredictable function of the die rolls.

If there is a problem it wouldn't be predetermination, it would be flaws in what was in generation of the rolls. So lets say that we are generating 3x the amount of doubles than we should, you or your opponent might grab one out of the luck of when you rolled the die. So the chance to get one would be the same for the both of you, but if we we're generating too many, then well there would be too many.

So here is a quick report, based off of actual game rolls from the server logs:
Games: 31970
Rolls: 2801476
Doubles: 457472

Doubles/Rolls: %16.3
This basically says that ~16% of the time a roll with be a double. Which is right on.... see you have 2 dice, the 1st one doesn't matter, then you have a 1/6 chance of your 2nd die matching right? so 1/6 = 0.166 So we're good there overall that we are getting the the number of doubles we should get.

It looks like the average game sees 87 rolls per game.
and that works out that we would see about 14 doubles in that game.
I think a game in this sense would be all the games in an average match. (not absolutely sure on that, so don't hold me to it)

anyhow back to the actual rolls...

Doubles distribution amongst the values:
1s: 76249 %2.7 of the time or 1 out of 36 times. (any dice combo in a particular order will happen )
2s: 75897 %2.7 of the time
3s: 76501 %2.7 of the time
4s: 76266 %2.7 of the time
5s: 76094 %2.7 of the time
6s: 76465 %2.7 of the time

So this basically says that which type of double (1s, or 4s for example) are evenly distributed. (keep in mind this is real data, so you are gonna see a little fluctuation)

Now onward and upward to doubles in a row (clumping) (note I'm not a statistician, so there could be some errors , but I think this is good enough to make my point)

Run of 0 81%
Run of 1+ 16.01%
Run of 2+ 2.22%
Run of 3+ fizzles off into hardly ever happening.

So I think the statistical odds are as follows:
No double: 83% (5/6)
1 double: 16.6% (1/6)
2 doubles: in row 2.7% (1/36)

So really pretty close to what we'd expect. So what the real world game data is showing is that there isn't an strange amount of Doubles. And it's also saying that there isn't a strange amount od runs of doubles in a row. (yes we checked in 3 ways, a seat 1, seat 0 and together, its was effectively the same for every seat position or both together)

Its practically useless to look at any single game, because we would expect it to be too low a sample to fit the statistical odds. (ie, if you file a coin 4 times, it would be entirely possible for it to be heads 3 out of the 4 times, and hence look unfair)

[Death_Row]

Hrmm, ya Not sure why the Bot would resign. See the AI is a Neural Net, so it trained it self how to play, so nobody really knows how it thinks

If this is indeed true (your words)... There's that PRNG which produces a list of number in advance...

From what I've read about PRNG and legal gambling sites (of which this is not... a gambling site that is)... if there is a complaint and PRNG is or was being used for randomization... the site is shut down. From the post in the beginning of this thread (and at least one other thread), there is a "feeling" of something not right (stronger words were used). You have shown from your link posted at the start of this thread... there maybe a problem with other backgammon games...

If you have a game where dice are/is used... why not use the technology (TRNG) that digitally simulates dice. And if you have a game that gives the player the choice as to when to throw the dice why take that away.

I'm not attempting to disputing your stats (I can't prove them either)...

All I'm saying (as well as Dust and others) is you've removed certain elements (1 being the cups) of the game that maybe contributing to the "effect or feeling" that the game may have an advantage.

By implementing the cups, TRNG and calling for new RNs each time the cup is moved (as in the physical game), and maybe even calling for several RNs as the dice are rolling out of the cup (this is all about brain storming)....This may go a ways in removing the "bad feeling".

Doing nothing may make feelings worse in different ways.

Put the choice & dice back in the hand of the player...

[Sailing_Away]

Death Row,

This discussion has been interesting to follow as I have a programming background and work in IT. Also, I was a math major and had taken several stats courses. I studied neural nets in grad school and recall the topic of random number generation from some comp sci courses. I am all over this from an interest perspective.

Jonas replied to you with data supporting the fact that the distribution as the sample reaches statistical significance is completely in line with the percentage odds. Samples smaller than that show that in real life any series of events can and sometimes will seem to defy the statistical norm. That's just life. Jonas proved nothing is wrong with the outcome of the random number generator used behind the scenes (which really has nothing to do with the neural net logic bots use to determine how to play the results of those rolls or deals from prior experience).

I am interested in how the methods of PNRG and TNRG differ, and perhaps Jonas might be too, but from a pragmatic point of view, I am doubting that implementation of TNRG will feel any different to the player or that the statistical results will be any different. In fact, TNRG might "feel" worse to up to half of those playing and "feel" better to the others.

There is also the question of how feasible is it to deploy TNRG. From a business perspective, it's an investment that has no ROI when resources could be dedicated to something like.... 7 ups in euchre instead (I just had to put that in).

I am curious about what TNRG might look like though as most of the comp sci courses really taught how to closely simulate randomness. Languages that have a randomization function even just use things like the microsecond value of the system clock, and it was believed there really was no such thing as a true random number generator.

[Dust In The Wind]

Being in the business of random generation (lottery) in which we deal with a set "universe" of possible outcomes to fit within a set percentage of variables and other rules of randomnise....

generally a "universe" consist as 1000, 10,000 or some other numerical value or in this case possible variations of a dice roll

a lottery has rules such as:

1) big winners will not fall within a certain number of each other and if selected the random generator will kick that generation out (IE: rolling dbl 6's within a set number of rolls)
2) same set of winners or losers will not occur consecutive more than x amount of times within a "universe"
3) patterns to x will not occur within a universe
4) etc.... there are lots of rules when it comes to a lottery and varies state to state and country to country

There are also those that say the universe has to be complete with all the possible variables to occur within said universe and the universe can be 1,2,3,x times the number of all possible combinations of the varibles (good example of this is using a single deck of cards to play blackjack or using 4 decks in a single/deckstack).

Also rules go by "exact" and "approximates" to say a 3/4 or 4/3 must occur 10% of the time can be 10% ONLY or 9.5 to 10.5.

So randomist even in a lottery can be said to be unfair since player #1 goes in and buys 2 tickets every other day and gets no winners and player #2 buys 5 tickets every day and gets several "winners" (not big ones mind you, LOL) and puts the "winnings" back into more tickets but each one has the same chance that the big winner will be the one they bought, the perception of player #1 is that it is "rigged" (I can assure you it isn't odds HAVE to be what they are) and player #2 is still somewhat happy with his play but he still doesn't hit the "big one" so again they could feel it is "rigged". Of course this continues until one does hit the big one.

So again we come to the game of BG and dice rolls... how the possibilites of the roll is generated can be many and from my "perception" of Jonas's explaination it is a data stream basically and when you hit the roll button it selects from that stream where ever it may be at that time (and also must figure the "ping time" or connection speed in msec. as a factor in from when you hit the button and the server reacts).

Bottom line is that I feel that the cup would help with the perception issue of a "fix" or "rigged" operation. I believe what you say on your data Jonas and the numbers are what they are supposed to be and being from this type of enviroment understand how and why some will feel the deal or rolls are not on the up and up. Just like the cutdeck option (some like it other don't but that is also true with a real set of cards) the cup gives the player an option to choose when to choose and shake those dice for luck before the throw and roll....

JUST DUST

PS - Look at your odds and read what it says and play to those odds.

PS - Before ya ask, no I can't play the lottery and no I couldn't give you winners or tell when and where to buy. Lottery is like printing money and you wouldn't believe the security that goes into it.

[Death_Row]

I'm really surprised (baffled is more like it) at the direction this thread is taking. I can trout out my degrees, minors, my 6 IT certifications (going for another one). I can run out all the stats on random numbers I've found on the internet. I can talk about those stats classes that kicked my butt (managerial stats and engineering stats) in college. We can get a PhD holder as well as someone with an MBA. But doing all that would not address the problem/s at hand.

1. There is a purported "feeling" the game (bot) has an advantage.
2. There is a purported "feeling" the game gives far to many doubles.

There is a reason (I believe) why the digital game houses these issue and the physical game does not. Everything is hinging on (around) the DICE. Either the random implementation of the dice or the way the player is allowed to interact (cups/roll timing) with the dice

As I previously stated... I'm not (can't) disputing Jonas's numbers.

from random.org
PRNGs are efficient, meaning they can produce many numbers in a short time, and deterministic, meaning that a given sequence of numbers can be reproduced at a later date if the starting point in the sequence is known. Efficiency is a nice characteristic if your application needs many numbers, and determinism is handy if you need to replay the same sequence of numbers again at a later stage. PRNGs are typically also periodic, which means that the sequence will eventually repeat itself. While periodicity is hardly ever a desirable characteristic, modern PRNGs have a period that is so long that it can be ignored for most practical purposes.

(dice are not periodic) in no form or fashion

Sailing_Away wrote
I am interested in how the methods of PNRG and TNRG differ, and perhaps Jonas might be too, but from a pragmatic point of view, I am doubting that implementation of TNRG will feel any different to the player or that the statistical results will be any different. In fact, TNRG might "feel" worse to up to half of those playing and "feel" better to the others.

Sailing go to http://www.random.org/randomness/ to get an understanding of PRNG/TRNG (or google'em both)

Sailing you brought up ROI (I do understand what ROI is)... how does it come in to play here being that this is a product that's currently being sold to the public?

[Dust In The Wind]

I too post many certs in "IT" including CCNP and have written many programs for various reasons and various purposes (I cannot get into detail). I work some very high-tech equipment (try working with equipment that will support an ink jet printer at 1000+fpm, yep 1000+fpm, imagine the data transfer rate) and deal with the impossible every day, but find a way to make it happen.

I firmly believe solutions are out there... even for "Global warming"

just hang on...

JUST DUST

[Jonas]

(dice are not periodic) in no form or fashion

I don't think the Periodic aspect of a RNG would play out in this scenario Death. 2 fold.

1) the periods they are talking about are not something that would happen during a course of a game, I think we'd be talking a period that might reveal itself over the coarse of many thousands of games, perhaps billions. That is an issue for encryption, but in this case the concern is the even feel of the rolls, which the period wouldn't rear it's head.
2) the periods would only be an issue for attempting to find out what your next roll would be. Not a factor of statistical likelihood of ones roll.

[Sailing_Away]

Sailing you brought up ROI (I do understand what ROI is)... how does it come in to play here being that this is a product that's currently being sold to the public?

Thanks for the link, and yes I could have googled the same as you mentioned.

I do also have my MBA with technology mgmt specialization, so no need to go find one.

ROI hinges on the evidence supporting the theory that if you build it, they will come. Are there similar sites as Hardwood that implemented TNRG over PNRG and found more people purchasing licenses or subscriptions as a result of it being a more enjoyable experience? Do you have figures showing the cost of deploying TNRG vs the quanitified benefits of having done so? I think this is where I am having the most difficultly... grasping the idea that TNRG will really affect in any positive way the Hardwood experience. That is not my final stance on it until I have done some research.

Then where would such a project fit in compared to all the other ones on Jonas' to-do list? Project prioritization isn't entirely ROI-based, but it plays a role. I have to think 7-up euchre is higher up on that list.

[Death_Row]

ROI hinges on the evidence supporting the theory that if you build it, they will come. Are there similar sites as Hardwood that implemented TNRG over PNRG and found more people purchasing licenses or subscriptions as a result of it being a more enjoyable experience? Do you have figures showing the cost of deploying TNRG vs the quanitified benefits of having done so? I think this is where I am having the most difficultly...

I must apologize I wasn't aware that you were/are the Chief Financial Officer/Chief Investment Officer for Silver Creek. You shoulda said (in the beginning) that you were speaking on behalf of the company. You must have inside information regarding this matter. Again please accept my most humblest apology.

[Dust In The Wind]

LOL

JUST DUST

[Sailing_Away]

I must apologize I wasn't aware that you were/are the Chief Financial Officer/Chief Investment Officer for Silver Creek. You shoulda said (in the beginning) that you were speaking on behalf of the company. You must have inside information regarding this matter. Again please accept my most humblest apology.

LOL wish I were... I could stand to have a change in career.

[Death_Row]

The logic that controls the die rolls has no information of what the computer needs. Nor does the AI have any knowledge of the Rolls prior to it's own roll.

We would have had to put in that logic for it to be able to a) roll what it needs, or b) take advantage of rolls before their rolled, and neither has been done.

If it appears the dice are rolling what it needs, it's either dumb luck, or its just more memorable than the times it doesn't.

One thing to keep in mind though, the AI does not play like people. It taught it self how to play, so even though it doesn't have access to the rolls, it may be able to play the odds better than most people. that is, it might be moving it's pieces to better take advantage of good luck than humans might.

Jonas... since there is a list "PRNG".... has there been any testing done to verify the AI isn't somehow utilizing (gotten access) those pre-defined numbers? There should be some way of exposing the list for testing purposes to find out how the numbers are being pulled off (by the bot) the list. Since no one knows what or how the AI/neural-net is doing things.

I still believe the implementation of the cups (giving control back to the players: single and server player) would help the perception of the game.