Why the Next Card Does Not Remember the Last One
A shuffled deck does not owe you a heart. What memory can mean after cards are removed, and what it cannot mean.
People talk as if a deck keeps score. A flush came last hand, so a flush is unlikely now. Four red cards appeared, so a black card is due. The talk feels like math because it mentions the past. It is not math. A fair shuffle does not remember a story. It only has whatever cards remain, and only when those cards have actually been removed from the possibilities in front of you.
This is the other half of a point made in Variance in Poker: short runs swing, and they do not schedule a correction. The plain conversion of counts into odds is in Odds and Probability Are Not the Same Number. No part of that correction is a betting plan. Adults can lose money believing a card is due.
Two different questions
The first question is about cards you can still be dealt because they are still in the deck. If the four aces are in the discard pile, face up, the next card cannot be an ace. That is not memory. That is subtraction. The second question is about a pattern in results that have already been shuffled back into a full deck. Last hand's flush was returned, the deck was shuffled, and the next hand is a new draw from 52 cards. The flush did not spend the deck's allowance of flushes. There is no allowance.
Confusing the two questions is the whole fallacy. Removal changes a count. A finished, reshuffled hand does not.
A small example with the cards still out
Inside a single hand, cards you have seen are simply absent from what can come next. Unknown cards keep their share of what is left. You update by counting the absentees. You do not update by deciding that a run of red cards has gone on long enough. If many red cards are among the cards you have actually seen, then fewer red cards remain, and a precise count would use that fact. A mood about the run would not. The deck is not trying to balance the picture.
Across hands, the slate is the shuffle
Between hands, the cards go back. A fair shuffle is meant to make each order of the deck equally likely, regardless of the previous order. If that is true, the chance of being dealt a pair now does not fall because you were dealt a pair ten minutes ago. The starting-hand counts apply again, to a full deck, or they do not apply because the shuffle was bad. "I am due for a pair" is not a third option. It is a story laid on top of a full deck that does not contain your history.
You can test the feeling with a coin, which has no removal at all. Five heads do not make the sixth toss more likely to be tails. The coin has no account to settle. A deck across hands is closer to the coin than to a dwindling pile, provided the cards were returned.
What you are allowed to count
Count cards you have seen if you are estimating what remains in this hand, and say that you are doing only that. Do not count a personal losing streak as if it removed cards from the next shuffle. Do not count a winning streak as evidence that the deck has entered a generous period. Streaks are what randomness looks like over a small number of trials. They are allowed to continue. They are allowed to stop. Neither event pays a debt.
If a game seems to produce patterns a fair shuffle would rarely sustain, the right suspicion is about the shuffle, the software, or the people, not about a cosmic balance. That suspicion still does not make the next card owe you anything.
Frequently asked questions
Does the turn depend on the flop? It depends on which cards the flop removed. It does not depend on whether the flop should have paired. The leftover deck is the whole influence.
If I have missed ten flops, is a set closer? No. Each new hand, after a shuffle, starts again. Ten misses are a hard evening. They are not a coupon.
Is there any sense in which cards remember? Only the physical sense: a card that is out of the deck is out. A deck that wants to balance your night is not a real mechanism.