True Randomness in Shuffles

You shuffle until the deck looks scrambled. But looks can be deceiving: in shuffle math, 'random' is a precise claim about all possible card orders, not a vibe.

• A hand-shuffled deck can look random even when some orders are much more likely than others.
• Mathematicians define a truly random shuffle as one that gives every possible ordering the same chance.
• To decide what truly random means, we need to look at the whole list of possible outcomes.

One question before the evidence: After a truly random shuffle of a 3-card deck, which of these is true? Pick the statement you think is correct.

• Pick between 'Every exact order has the same chance' and 'Scrambled orders are more likely than the sorted order'.
• Your guess becomes the hypothesis we test next.

Run a perfect random shuffler with just three cards. It picks one of the six possible orders at random each time. Watch the tally for every order.

• There are exactly 6 orders of a 3-card deck.
• After thousands of shuffles, each order's count is close to the same.
• Seeing a 'sorted' order happen is normal; it is just one of the six equal outcomes.

For a 3-card deck, each of the 6 orders has probability 1/6. For n cards, each of the n! orders has probability 1/n!. The sorted deck is not cursed; it is just one of many equally likely exact outcomes.

• There are 6 orders for A, B, C; each gets probability 1/6.
• For n cards there are n! orderings; a true random shuffle gives each one probability 1/n!.
• 'Looks messy' is a category, not an exact outcome. Many messy-looking orders exist, so seeing one is common, but each exact messy order is just as likely as the sorted one.

The mathematical definition is an ideal model. With a real riffle shuffle, cards near the top tend to stay near top, and many perfect orderings never appear. So 'randomly shuffled' in everyday life is an approximation.

• Ideal random: every ordering must be possible and equally likely.
• Real riffle shuffles are biased: top cards are likely to stay near top.
• That is why a few thousand shuffles by hand are not truly random, even if the deck looks mixed.

Imagine a shuffle that always splits the deck perfectly in half and interleaves the cards. No matter how many times you do it, only a tiny cycle of orderings ever appears, so most of the possible orders never show up. That means it is not truly random, even if the deck looks mixed.

• A process can be unpredictable-looking and still miss most possible orders.
• The test is not 'does it look random?' but 'could every ordering appear with equal chance?'
• A deterministic repeatable shuffle fails the test because it only reaches a tiny cycle of orders.

Back to the opening: a deck is truly random not when it looks messy, but when each of the n! exact orderings has probability 1/n!. The sorted order is not more or less likely than any other exact order; it is simply one among a staggeringly huge set.

• Truly random shuffle = uniform distribution over all n! deck orders.
• Every exact order, including the original one, has probability 1/n!.
• A 'random-looking' deck can be biased; a truly random deck is defined by equal chances, not by appearance.