If you find the information in this post too sensitive or useful, please don't quote me while posting a reply so that I can delete this post later. I am especially interested in the second opinion (observation?) of Sonny.
First, let's revisit the rules. 1) Last in first out. Any card inserted last would be dealt first before one or more preceding cards. This simply means, if we were to memorise a slug of rich cards (in this order of ranks --> AKQJ109)), and by chance if two of these cards go into the same slot, we know for sure that rank 9 would be dealt first. And one of the rich card would follow within roughly 9 cards. 2) This because muck is likely to consist 9-10 cards or else it doesn't get drop off for dealing. 3) If there are always 15-20 cards loaded in the chute for dealing, two more rules to take notice of: 4) a slot must have 7 cards in order to be dropped 5) we always have about 2 slots emply in the machine to keep the chute filled due to rule 3. As more cards get dealt and more slots get emptied in the machine, the probably of at least two cards - from one or more slugs that we have just memorised in last couple of rounds, going into same slot - should increase. Why is that? Because vast majority of other slots are already full. One possible brute force approach is more clearly illustrated below.
So we observe a table during a brief slow down of the action. Dealer scoops in all discards back into shuffler. It moves and reinserts all discards while 2 slots get emptied into the chute for the next action. We hear that the machine is quiet. But 2 slots are already empty in the machine and atleast couple more are going to be empty soon. When cards are dealt, we begin to memorise slugs of 7 to 10 cards preferably of all rich cards. We do it for 2-3 rounds accumulating a total number of cards - anywhere from 28-40 - in our memory. Note that we already anticipate reversal of any two adjacent cards wherever they have stayed in the same slot. Rest are useless to us if they have gone to other random slots, one by one, not preserving any noticeable reversal especially if these random slots were almost full each needing only one additional card. This is a situation of total breaking up of our slugs where almost all cards from our slug go into different slots - one by one. But be assured, this would happen less times than you think it would happen.
Now, in accordance to rule 1, if we see one of our two cards that may have gone into the same slot, we can predict the other. And the number of random cards inserted between these two would be upto 8. Refer to all rules again to visualise this situation. We can discuss real monetory/numerical issues (bet spread, advantage, frequency etc.) later on as these issues would still remain a challenge.
So what do you guys think?
PS: As of today, CSMs are NOT profitable. But Rome was not destroyed in a day. If you ponder over a problem long enough, the puzzle itself provides a solution.
Idk what you guys are seeing, but to me it looks like the top card in the “face up” pile gets pushed into a slot first. The card that gets pushed into a slot is being loaded such that its face is touching the nearest card’s back (as opposed to the inserted card’s back touching the nearest card’s face) in the slot it is inserted into. The card that is being inserted into a slot, if no more cards get inserted into that slot, is going to be the first card to be played once that slot is dropped into the “chute”.
At least that’s what I saw.
"Everyone wants to be rich, but nobody wants to work for it." -Ryan Howard [The Office]
Deviru thinks that the machine does.
38 slots with 10 cards each. 15-20 cards are dropped in chute every or every other round. So at any given point, only 3-4 slots get emptied while rest of 34 are full. So if 15-20 discards are shuffled back every round, they would be shuffled between 5 slots only as the rest are still full. This creates a sort of exploitable latency.
For simplicity purpose, let assume a rich slug of discards. From bottom to top --> King of spades, King of hearts, Queen of spades, Queen of hearts, Jack of spades, Jack of hearts, 10 of spades, 10 of hearts, 9 of spades, 9 of hearts, Ace of spades, Ace of hearts. 12 cards in tray with kings at the bottom and aces at the top.
During insertion, the order is upside down. So kings are visible on top while aces are at the bottom not clearly visible. Hence, let's assume that aces would be sucked in first followed by 9s, 10s, pictures cards upto the king of spades.
Note that most of 38 slots are full. So these 12 cards are going to get shuffled between 4-5 slots that are mostly or entirely empty. Since target cards are key cards and vice versa, we are not too concerned about the reversal or breaking of this slug sequence. Let's see how some cohesion would be maintained provided our assumptions about the said rules are correct.
So the bottom most cards not clearly visible, the aces, are inserted into slot 3 and 5 respectively. Next, 9s and 10s would be sucked in by slot 12 and 15. Jacks go to slot 20. Boom. No other slots are emply. So Queens and Kings come back to slot 12 or 15. Aces may be inserted back in slots 3 or 5. So in slot 3, we are going to see at least two cards out of 12. Same way, slot 5 may have other two or three, and so on.
So every time we identify one rich card in a given round, we know that atleast one more is to follow within next 8 cards.
To eliminate false keys and targets, we try to memorise multiple slugs, say 24-30 discards, instead of 12, with in two round or so.
The important issue is consistent rules. How exactly the cards are stacked into a given slot? Which cards is inserted first from the pile? Is there any consistency? I can only confirm one rule which is already disputed by RC - the top most card visible on the rear would be inserted last. In my example above, it was King of spades which was sitting at the bottom of the discards.
You are correct, at the least you have worded what you saw quite well.
Nevertheless, we do notice a consistency - even though the wheel spins are totally random, the cards are inserted and stacked in particular order as you observed.
I am waiting for Sonny's input to see if he wants to add something. Or may be, he is already raking in thousands beating CSMs!!!
Inserting numbered cards (1-52) of 2-3 decks in their numerical order (2,3,4,5,6,7,8,10,J,Q,K,A) as they are in a pack would shed much more light on Tassie's theory.
Which also begs the question; has anyone actually seen the new (?) decks being put into a CSM? Or do they just keep them in it?
Someone has already done it (scroll through Francis Salmon's posts on bj21)
https://bj21.com/boards/free/sub_boa...o-6-csm?page=2
It doesn't matter whether they put in new decks or not.
If 38 slots, each having 10 cards maximum, and dropping 15-20 cards on chute every round, what does it mean?
That means roughly 2 slots get emptied every round, and when these 15-20 cards are played and discarded back into the shuffler, there are roughly 4 slots that are empty. Rest 32-34 slots are entirely or mostly full. This means that discards only go to 4-5 slots whenever other slots are full. This seems to be happening not-so-rarely. Hence there is an exploitable latency.
Secondly, it is indeed beneficial to casinos to deal 2-3 decks. Why? If 2 decks are in discards, more slots are empty in CSM machine. So when these discards have more slots to go to into, more randomisation is achieved.
My point is, CSMs are exploitable from multiple pronges: Deal too less cards, you have less of complete randomisation in next rounds and more 'windowed counts' for APs - all because only few slots are active for shuffling in next rounds. Deal too many cards, APs get higher counts, but more randomisation achieved for next rounds because half empty CSMs would have more slots empty and cards would get distributed more evenly after 2 decks of discards being pushed back into 20-30 slots.
Have you visited CC last weekend?
Bookmarks