Poker Machine Shuffle Operation
Thread Rating:
The issue: Do the vp machines from the last 6 years or so use a continuous shuffle method until the draw cards are selected, or do they deal 10 cards as soon as the bet button is pushed, whereby 5 cards are shown on the screen from the deal, and cards number 6 through 10 lay in waiting for a sequential selection as soon as the draw button is pushed? In other words, if you're dealt K59J4 of different suits and you keep the KJ, cards number 6,7,and 8 will replace the 3 discards.
From the discussion, on a place called Vegasrex(?) someone posed this question to which someone else answered with the "10 cards are selected upon pushing the maximum bet button" solution. Then another poster came on and said it was required that all cards keep shuffling until DRAW is hit so that if anyone had some device that could "see" what 5 cards lay in wait, they'd be foiled. He did not provide any proof or documents, just said that it is required these days by the Nev. Gaming Commission or something like that.
Then in a surprise appearance R. Singer gave a reply, it seemed because of something he just experienced at the HardRock in LV. He was at one of the new machines at a bar in the new casino section of the HRH and had some kind of machine problem. A tech fixed it then showed him some type of cards deal/draw sequence where the 10 cards were dealt up front and there was no continuous shuffle, and he related that to a machine he said he had and tested at his home over the summer. Well naturally, his critics jumped on that statement of experience only they took it to a forum that he does not post on (I think he's banned there by the AP's who run it, LVA). Just like with what mkl does here he's called a liar, but what's interesting is he was just offering an input based on a recent experience. The funniest part is, those who do the criticism have absolutely no proof on how they say the machines operate. They just like to jump on him.
Does anyone here have a balls-on explanation with a documentation trail other than "Jean Scott says" or "I know it's true" that I can take to these other forums? Maybe an experience with the machines themselves?
My impression has been that video card/dice/etc. games are required to perform in a manner that replicates real card/dice/etc. games. (Maybe this is not true and they just must present the same probabilities.) If my impression is correct, then the order of all ten potentially-used cards should be established with an initial shuffle rather than with a re-shuffle after the first round. This would apparently agree with the comments by Singer and thlf. I think this method must be modified if there are multiple hands being played.
Please pardon the intrusion of an uninformed comment.
Quote: DocSorry, Jerry, I have no real knowledge of how these devices work nor relevant personal experience. I freely admit that the following comment is just based on how I think video games are supposed to work.
My impression has been that video card/dice/etc. games are required to perform in a manner that replicates real card/dice/etc. games. (Maybe this is not true and they just must present the same probabilities.) If my impression is correct, then the order of all ten potentially-used cards should be established with an initial shuffle rather than with a re-shuffle after the first round. This would apparently agree with the comments by Singer and thlf. I think this method must be modified if there are multiple hands being played.
Please pardon the intrusion of an uninformed comment.
I'm looking for honest assessments either way, and so far I've gotten them. Thanks! What you're saying seems to make perfect sense, but that type of logic doesn't go very far on forums that appear to pay posters to put up Singer-hating remarks.
My opinion is that either way, random is random. Who has a device and can use it, that "sees" the hidden 5 cards anyway? Plus what you said I can relate to. They don't shuffle the deck after the initial deal in live poker, so there's no point in doing it on a machine.
Quote: DocI think this method must be modified if there are multiple hands being played.
Quote: JerryLoganPlus what you said I can relate to. They don't shuffle the deck after the initial deal in live poker, so there's no point in doing it on a machine.
I don't play video poker, so I might have this screwed up royally. What I meant by my "multiple hands" comment was this. If you play thirty hands at a time, get five cards initially, choose your holds, and then get different draws to make the thirty hands, then there must be something different going on. I think the remaining 47 cards have to be replicated and re-shuffled for the draw to each of the hands in order to provide the different results. Or maybe I am missing something.
Oh, and if there are forums that pay posters to put up Singer-hating remarks, please let me know where they are. I know nothing about Singer and have no opinions on him, but for enough pay, I might be willing to post anonymous barbs. I have a lot of time on my hands, as evidenced by my presence here.
Aren't the techs in those situations looking at a history? I.E. If the history shows that the patron hit the hold button, then a mili-second later, un-hold got recorded, it would be obvious, even without looking at the video tape, that the person intended to hold it, but electronically, the machine took it as two button pushes, and didn't hold the card.Quote: thlfI have witnessed numerous times where somebody lost a jackpot because a card did not hold and the jackpot would have been won on the re deal.
The history might record all five cards that are at the top of the deck when the 'draw' button is pressed. But whether the cards were being shuffled in the interum is impossible to tell based on what was posted / observed.
Who has a device? Someone who really understands the guts of the program.Quote: JerryLoganMy opinion is that either way, random is random. Who has a device and can use it, that "sees" the hidden 5 cards anyway? Plus what you said I can relate to. They don't shuffle the deck after the initial deal in live poker, so there's no point in doing it on a machine.
Random Number Generators are NOT truly random. Therefore, someone with enough knowledge might be able to figure out, or design something that, based on the five cards showing, would tell him what is likely to be drawn.
For example, if you were dealt, 3H, 5D, JS, 4C, JH, conventional strategy says to keep the jacks. But if your knowledge/device told you that exact sequence meant that half of time the draw would be 6C and 7D, wouldn't you go for the straight every time?
Continuous shuffling is not needed on live poker for a different reason: The shuffler does NOT truly randomize the cards. But since they are fed into the machine in a mixed / unknown sequence, knowing how the shuffler works does not provide any advantage.
When the player decided to draw X more cards, I'd return X more unique numbers, mapping again to cards, and display them. I'd then log this yet again.
A simple review of the log would show something like:
1st Draw : Qh 2c 3d Kd 4s // H Ts Js H As // Pay out : 4
Or something similar.
The Random Number generator would have some sort of external effect that would make the draws non-deterministic (such that knowing the working would not allow me to tell that draw A B C D E means F or G is coming next). This would probably include time AND another factor to add noise into the proceedings. This means I'd not select the next cards until the 'Draw' button is pressed. I would be highly surprised if there was a single VP machine being used to do that had any sort of determinism with it's RNG.
However, I've read somewhere (hearsay) that certain types of machine for certain jurisdictions are really simulating a pull tabs, so the cards are already drawn 'behind', and not only that, the result is already drawn, so you ca't NOT get your Royal Flush if it's drawn for you.
I believe Washington State uses this type of machine for slots (Class-A or Class-B?).
Quote: DJTeddyBearI too have no intimate knowledge, but here's my gut feel:
Aren't the techs in those situations looking at a history? I.E. If the history shows that the patron hit the hold button, then a mili-second later, un-hold got recorded, it would be obvious, even without looking at the video tape, that the person intended to hold it, but electronically, the machine took it as two button pushes, and didn't hold the card.
The history might record all five cards that are at the top of the deck when the 'draw' button is pressed. But whether the cards were being shuffled in the interum is impossible to tell based on what was posted / observed.
The tech is indeed looking at the history, however thay cannot see if someone hits a hold button and inadvertently releases it or whatever. In other words there is no history of the button pushes. There is only a history of what made up the ten cards for that hand (as well as previous hands). What the tech is trying to do is determine if the player is believable first of all, and see if it was even possible that they could have scored the jackpot. Then it is up to the discretion of the casino as to whether or not they pay. Most strip casino's absolutely will not pay. Most locals will pay if the player is a regular and as I say believable. I have seen it happen both ways.
Quote: JerryLoganThe issue: Do the vp machines from the last 6 years or so use a continuous shuffle method until the draw cards are selected, or do they deal 10 cards as soon as the bet button is pushed, whereby 5 cards are shown on the screen from the deal, and cards number 6 through 10 lay in waiting for a sequential selection as soon as the draw button is pushed? In other words, if you're dealt K59J4 of different suits and you keep the KJ, cards number 6,7,and 8 will replace the 3 discards.
Hokay, where to start? I guess first, this all comes mostly from memory, and we all know how faulty that can be, but I think the following is accurate. That being said:
So, maybe you've heard of MAME. For those that haven't, it's an arcade machine emulator. Being an emulator, it has a few features you can't really perform on the original machines. One of these is a state save feature: the emulator saves a copy of the RAM, the registers, the pointers, the stack, everything that represents the state of the machine at that instant. You can then load that state back into the emulator, and things pick up from that point, just as if they had gotten there naturally (like they did the first time).
Anyway, MAME has had emulation support for IGT Player's Edge (Plus) for a while now, and a couple dozen ROM sets are dumped for it. One day, I decided to answer the same question for myself, so I conducted my own experiments using several ROM sets. What I found was this:
- Older ROM sets (pp0043, for one) would shuffle up (at least) ten cards and deal them in order. So if the first ten cards were 9c 8c 6c Jh Js 4h Ah Ad 8h 2d, you would get the first five (9c 8c 6c Jh Js), hold the jacks, and be dealt 4h Ah Ad for aces up. If instead, you held the 8c, you'd get dealt 4h Ah Ad 8h and still wind up with aces up, only with eights instead of jacks. By saving your state after the deal, you can dump all five cards to find out what the five draw cards are, figure out what your best play is (for example, if you need a 3 for the straight, but it's not the first draw card, you're out of luck), reload the state and make that best play. I think it used continuous shuffle between games, so that you don't get dealt the same five cards if you go back to a state between games.
- Newer sets shuffle up and deal your hand, then continuously shuffle the cards until the draw. If you save the state after the deal, you can go back to it over and over and you'll always get dealt different cards.
Granted, these tests were only performed on a small subset of Player's Edge machines, and Game King-level machines aren't in MAME yet, so I can't test those at all, but I think those work the same way as the newer PE machines. If you want, you can get MAME and the Player's Edge sets and test it yourself.
Quote: thecesspitIf I was designing a VP system, I'd not be shuffling the cards at all. I'd assign each card a number from 1 to 52. Then I'd draw 5 unique numbers from my random number generator. This would map to 5 cards. I'd then place this information in the log.
This is almost how a computerized shuffle works. It's called a Knuth shuffle or Fisher-Yates shuffle. It works like this:
- Choose a random number from 1 to 52
- Swap the card in that position with card number 52
- Choose a random number from 1 to 51
- Swap the card in that position with card number 51
- Choose a random number from 1 to 50
- Swap the card in that position with card number 50
- Etc. until you've shuffled as much of the deck as you want
If you need five cards, you shuffle up five cards. The advantage is that you only need to shuffle and deal as many cards as you need, and you can ignore the rest of the deck until you need more cards. It's called "continuous shuffle" because the effect is the same as if you simply made the random swap with the same end position every time until you need the card (swap random with 47, swap random with 47, etc.). You don't run through the algorithm while you're waiting, though (don't want to waste cycles). You just continuously generate random numbers.
BTW, easiest way to make it reasonably non-deterministic (IMO) is to put a RNG in hardware and make it generate random numbers as fast as it possibly can (the fastest are in the 50 megabits per second range, or about a million and a half 32-bit floating point numbers per second). Unless you can tap and analyze the outputs directly, I don't think you could reasonably figure out what's coming next, and even if you could, it'd be difficult to time it to (sub-)microsecond accuracy without a bunch of suspicious-looking equipment.
A good person to ask about this would be MathExtremist.
The Fisher-Yates would be better than my method as you would add some more "randomness" by having the current state as the entry to the next state, plus the random number generator, plus the entropy source.
Course, you have to be careful that you don't turn your F-Y into a Sattolo algorithm and end up with bias.
Quote: thecesspitMy only query is why would it 'continue shuffling' between hands... it seems pointless exercise to me. Just draw numbers when you need to rather than continouly doing useless work.
I believe the reason they do that is game security. It is easier to exploit a game if the next outcome is predetermined. I couldn't explain how, but that is what I'm told. Hopefully MathExtremist will chime in, this is right up his alley.
Quote: WizardI believe the reason they do that is game security. It is easier to exploit a game if the next outcome is predetermined. I couldn't explain how, but that is what I'm told. Hopefully MathExtremist will chime in, this is right up his alley.
It seems far, far-fetched that someone could compromise a machine in real time with what would have to be an electronic device connected into the motherboard plus relay the info to an external reading device in order to exploit a machine such as that. Plus that type of software change would require R&D funding along with other expenses, and it just doesn't seem to be worth the benefit when there really isn't any.
Quote: WizardI believe the reason they do that is game security. It is easier to exploit a game if the next outcome is predetermined. I couldn't explain how, but that is what I'm told. Hopefully MathExtremist will chime in, this is right up his alley.
Hmm, but the next outcome is not pre-determined. You select the next card when asked for. It's kinda akin to shuffling the blackjack when the player asks for a card, not standing there shuffling all the time.
I can kind of see it maybe helping to add the time factor I mentioned and the computation power is "free", so it hardly matters.
Quote: WizardI believe the reason they do that is game security. It is easier to exploit a game if the next outcome is predetermined. I couldn't explain how, but that is what I'm told. Hopefully MathExtremist will chime in, this is right up his alley.
As I recall, all the Silicon Gaming machines did was shuffle the deck at the beginning of the hand and then deal cards off the top. Of course, this was over a decade ago, so I'm not aware of precisely how the modern IGT machines work. As has been stated, it doesn't matter mathematically what you do under the assumption of randomness, and even with deterministic pseudorandom numbers, it's still virtually impossible to get a bead on predicting the future. It can be done, but it requires a computer and that makes it illegal:
How to hack Video Poker machines
You're right. It DOES seem far fetched.Quote: JerryLoganIt seems far, far-fetched that someone could compromise a machine in real time with what would have to be an electronic device connected into the motherboard plus relay the info to an external reading device in order to exploit a machine such as that. Plus that type of software change would require R&D funding along with other expenses, and it just doesn't seem to be worth the benefit when there really isn't any.
But that's exactly what Ron Harris did when he exploited a flaw in the RNG for a Keno game. How he did it was by watching the Keno in the hotel room, using his laptop, and relaying the next numbers to bet to his buddy down in the Keno area.
http://en.wikipedia.org/wiki/Ronald_Dale_Harris
Quote: DJTeddyBearYou're right. It DOES seem far fetched.
But that's exactly what Ron Harris did when he exploited a flaw in the RNG for a Keno game. How he did it was by watching the Keno in the hotel room, using his laptop, and relaying the next numbers to bet to his buddy down in the Keno area.
http://en.wikipedia.org/wiki/Ronald_Dale_Harris
Relaying machine information from a player to another person in a hotel room for further analysis on a computer is one thing. but attaching firmware up to a machine and reading what the next 5 cards will be on every hand is another. I knoiw people with criminal minds are innovative, but they can't do the impossible. And it's only "impossible" because of the risk factor of certain detection being involved.
Quote: JerryLoganIt seems far, far-fetched that someone could compromise a machine in real time with what would have to be an electronic device connected into the motherboard plus relay the info to an external reading device in order to exploit a machine such as that. Plus that type of software change would require R&D funding along with other expenses, and it just doesn't seem to be worth the benefit when there really isn't any.
Watching procedures at a table game, it often seems to me there are more safeguards preventing dealer cheating than player cheating. (More specifically, a player and dealer cheating in collaboration).
I have no idea about safeguards with machines under repair. But the people most likely to cheat a VP machine would be those who have access to the guts. See the previous intriguing post about save-states in MAME. I still don't know if an employee using a hacked machine to make valid cash-out tickets would be technically or practically possible. But it seems that continuously shuffling the 47 cards left in the deck would make the hacking a lot harder.
Quote: DJTeddyBearYou're right. It DOES seem far fetched.
But that's exactly what Ron Harris did when he exploited a flaw in the RNG for a Keno game. How he did it was by watching the Keno in the hotel room, using his laptop, and relaying the next numbers to bet to his buddy down in the Keno area.
http://en.wikipedia.org/wiki/Ronald_Dale_Harris
RNG cycle-timing a game that runs once every 5 minutes is one thing. That's feasible without a real-time interface (as demonstrated) if the RNG can be cycle-timed. It's not feasible for a VP game unless you're sitting in front of it with your laptop, or a concealed version thereof.
And while that's what got him caught, because his buddy was greedy, the far worse crime was the EGM gaffe. That's flat-out crooked. At least cycle-timing an RNG with your laptop isn't cheating per se (plus, it's clever), but it's still considered a crime by the GCB. But yes, that's why they don't let you use "any device to assist: 3. In analyzing the probability of the occurrence of an event relating to the game;" (NRS 465.075).
Quote: dudestupidWatching procedures at a table game, it often seems to me there are more safeguards preventing dealer cheating than player cheating. (More specifically, a player and dealer cheating in collaboration).
I have no idea about safeguards with machines under repair. But the people most likely to cheat a VP machine would be those who have access to the guts. See the previous intriguing post about save-states in MAME. I still don't know if an employee using a hacked machine to make valid cash-out tickets would be technically or practically possible. But it seems that continuously shuffling the 47 cards left in the deck would make the hacking a lot harder.
Agree that inside access is key to the ability to cheat a vp machine. It also makes sense for the continuous shuffle to make hacking more difficult, I think. But there's still the qestion of the connected hardware needed outside the machine and readily accessible on each hand by the player, with none of this apparatus being capable of being spotted.
I prefer the idea of creating valid cash-out tickets somehow. It eliminates all the hours sitting at the machines. There's got to be some way for an employee to successfully do that to supplement their income, as long as they button their lips and stay away from greed.
Quote: JerryLogan
I prefer the idea of creating valid cash-out tickets somehow. It eliminates all the hours sitting at the machines. There's got to be some way for an employee to successfully do that to supplement their income, as long as they button their lips and stay away from greed.
Thats what the bar code system on the cashout tickets is supposed to prevent. Each ticket is individually coded, and if a repeat or unrecognised code comes up when the ticket scanned, there is trouble. Thats not to say it cannot be done. It happened where I work. All it needs is someone sloppy with procedure and you can get paid.
