Improper shuffling - how to exploit?

[Pocket Trips]

if 2 cards that were in a players hand are still next to each other after the deck is shuffled they will be dealt seperately to 2 consecutive people. If you recieve one of the cards you recieved last hand it might be possible that the person on your left or right has your other card from last hand. However the remote possibility of this happening is rare and even if you knew FOR SURE that the person next to you had a speciic card, say the jack of diamonds for example, the times when that information would not otherwise be obvious by just reading his hand would happen so infrequently that it would be just a huge waste of energy that could best be spent watching your opponents. Rather than watching cards being shuffled/dealt.

[callydrias]

[ QUOTE ]
I had thought about that too but I wonder how good that mix really is. My guess is that many 2 card combinations will still be together from the previous hand before the shuffle. Thanks for the answer though. It's really more of a theoretical question but it would be interesting if there was something to it.

[/ QUOTE ]

The answer is that the mix is usually not very good. I talked to a dealer about this a few months ago and he told me that all the wash really does is mix up some of the cards on the top of the pile and make everyone feel better. Many of the cards that aren't on the top layer still overlap each other when they're all spread out so when the pile is restacked, they retain their original relative ordering. Of course this process certainly introduces some randomness, but I think the cards have an overall nonrandom probability of ending up where they started (by deductive reasoning, not by empirical evidence).

[DrPublo]

The wash originates in stud games, where MANY cards would be exposed by the river. If someone's entire board (or hand, even) were put into the deck sequentially before a few riffles , players thought that cards, or a suit (in the case of a flush in the previous hand) could be generally tracked through the deck. So the practice of doing a wide mix of the cards was started, so that the riffles start with a generally random sequence of cards.

The Doc

[IsaacW]

[ QUOTE ]
Sorry Joshua, but you're absolutely wrong in what you said. The study proves 7 shuffles create randomness. Also you can't work backwards saying that 7 shuffles will now be predictable since 8 shuffles returns the deck to its original.

[/ QUOTE ]
Nice try, boss, but your trap fails. Joshua is exactly right. You said in the OP:
[ QUOTE ]
The famous card shuffling study said you need 7 perfect rifle shuffles to properly randomize a deck.

[/ QUOTE ]
This is where you are wrong. Perfect out-shuffles (in which the top card remains the top card) reproduce the original deck after 8 shuffles. Therefore, given that we know the original distribution, we can predict the 7th shuffle. The perfect shuffle is not a random operator.

The study to which you refer no doubt looked at imperfect random shuffles (in which the two half-decks are not necessarily exactly interleaved when the halves are recombined). In this case, 8 or 9 shuffles are required to produce an ordering in which all permutations of the cards are actually possible (reference). 7 of such shuffles will produce a deck that is almost random.

You fail it.

[GuitarMarc]

[ QUOTE ]
You fail it.

[/ QUOTE ]

Good Lord, what is your problem? Why even bother with some of you anti-intellectual types.

[ QUOTE ]
The study to which you refer no doubt looked at imperfect random shuffles

[/ QUOTE ]

Oh, you mean you're guessing because you don't really know.

[ QUOTE ]
Therefore, given that we know the original distribution, we can predict the 7th shuffle

[/ QUOTE ]

Yes, a computer simulation can tell you the location of any card in the deck based on 7 perfect rifle shuffles. A computer simulation could also tell you the location of any card after 10,000 perfect shuffles. Do you understand that's not the point? Apparently you do not. The point is in a live atmosphere 7 rifle shuffles will randomize the deck sufficiently from its original state whereas 4, 5 or 6 will NOT.

Nice try, know it all.

YOU FAIL to misunderstand the main point of the study. It's not as if this is controversial or anything.

[IsaacW]

[ QUOTE ]
[ QUOTE ]
The study to which you refer no doubt looked at imperfect random shuffles

[/ QUOTE ]

Oh, you mean you're guessing because you don't really know.

[/ QUOTE ]
That's right. I'm at a conference away from school right now and I cannot look up the paper to which you refer. Therefore, I do not know what type of shuffle they studied. However, a perfect riffle shuffle is well-defined and is not a random operator. No matter what the original ordering of the deck is, the ordering after the 7th perfect riffle is exactly known. I doubt that any serious mathematician would describe the order of a deck after 7 perfect riffles as a "random" ordering. This page seems to indicate that Bayer and Draconis did not investigate perfect riffles:
[ QUOTE ]
The researchers simulated shuffling on a computer, using the same idealised kind of riffle shuffle as Bayer and Diaconis, in which the pack is divided roughly in two and the cards are dropped on top of one another more or less alternately from each half.

[/ QUOTE ]

[ QUOTE ]
Yes, a computer simulation can tell you the location of any card in the deck based on 7 perfect rifle shuffles. A computer simulation could also tell you the location of any card after 10,000 perfect shuffles. Do you understand that's not the point? Apparently you do not. The point is in a live atmosphere 7 rifle shuffles will randomize the deck sufficiently from its original state whereas 4, 5 or 6 will NOT.

[/ QUOTE ]
You have changed your statement. In your original post you stated that 7 perfect riffles would randomize the deck. This is not true, as I and many others have pointed out. Now you say that 7 riffles (without the modifier "perfect") will randomize the deck. This is almost true. The reference I provided explains that 7 riffles will produce an ordering that is close to random, but in fact it takes between 8 and 9 riffles to "sufficiently randomize [...] a deck of 52 cards."

As it turns out, there is more than one way to define "sufficiently randomize." If you define it as when the total variation norm drops below half it's original value (as per Bayer and Draconis), then 7 riffles are required. If you define it as when the deck has lost 99% of the information contained in the original ordering, it only takes 5.

[ QUOTE ]
YOU FAIL to misunderstand the main point of the study. It's not as if this is controversial or anything.

[/ QUOTE ]
I agree with you 100% on this. Grammar nits will know what I'm saying [img]/images/graemlins/grin.gif[/img]

This discussion has gotten away from your original point, I'll agree. The answer to that is that if you are capable of tracking cards through the shuffle used at your favorite live cardroom, then do so and reap the benefits. If you can't, then realize that B&M poker is clearly rigged and stop playing [img]/images/graemlins/grin.gif[/img]

And anti-intellectual? Come on, if I were anti-intellectual I wouldn't be e-swordfighting with you about this.

But seriously, my Inter-wang is HUGE.

[GuitarMarc]

[ QUOTE ]
YOU FAIL to misunderstand the main point of the study. It's not as if this is controversial or anything.


[/ QUOTE ]

You're right, I wrote that incorrectly. Obviously it should be that you fail to understand the point which you still do. Where do you get off saying I am changing my statements. If you were here I would slap you silly. Why do aholes like you exist.

It was a theoretical question to begin with. I'm not going to argue the merits of the study for nitwits like you who are just trying to create problems.

[GuitarMarc]

Let me add that you're getting too caught up in the minutia and not even addressing the original question. We're not even talking about 7 perfect or imperfect rifle shuffles. We're talking about the way cardrooms do it which is only about 4. This does not create good randomness from the original.

Getting back to your side point - yes, a computer can keep track of the location of the cards if it is 7 PERFECT rifle shuffles vs. 7 imperfect rifle shuffles. Either one will simulate randomness for a human in a live environment. This is a side issue and not important since cardrooms don't do either one.

Sorry I called you an ahole earlier but I thought you were being inane. You still have a lot to learn about certain things, junior.

[GuitarMarc]

Cally, thanks. Yea, my suspicions were that the mix didn't really do much.

[SpaceAce]

[ QUOTE ]

You're right, I wrote that incorrectly. Obviously it should be that you fail to understand the point which you still do. Where do you get off saying I am changing my statements. If you were here I would slap you silly. Why do aholes like you exist.


[/ QUOTE ]

Wait, did you just accuse Isaac of anti-intellectualism then call him a name and threaten to kick his ass?

SpaceAce