HN Debrief

Seven Perfect Shuffles Randomize a Deck of Cards. But How Many Sloppy Ones?

  • Mathematics
  • Probability
  • Games
  • Security

The piece explains a new result that extends the classic Bayer-Diaconis riffle-shuffle work. The old headline result says around seven riffle shuffles mix a 52-card deck when the deck is cut according to a very specific random model and then interleaved in a similarly specific way. The new paper keeps the same interleaving model but relaxes the cut, showing that if the cut is sloppy rather than near-perfect, the deck still mixes in about 14 shuffles.

If you care about real randomness in cards, tabletop play, or any process that depends on repeated mixing, do not carry over the famous “seven shuffles” rule without checking the model. The practical lesson is that small deviations in how you split before each shuffle can double the mixing time, while some common habits like pile shuffling barely help at all.

Discussion mood

Interested but nitpicky. People liked the underlying result, but a lot of the energy went into fixing the article’s terminology and explaining that the famous seven-shuffle result is narrower and less magical than the writeup made it sound.

Key insights

  1. 01

    Seven shuffles is not the cutoff

    For a 52-card deck, the famous result only says total variation distance drops below 0.5 after seven riffle shuffles. That is a benchmark, not a sudden phase change into full randomness. Using the usual cutoff estimate of 3/2 log2 n, the mixing point for 52 cards lands closer to nine shuffles, which makes the article’s dramatic framing overstated.

    If you use the seven-shuffle rule in games, demos, or simulations, treat it as a rough threshold for this exact model, not as a universal guarantee. If you need stronger mixing, budget more shuffles than the headline implies.

      Attribution:
    • possiblywrong #1
  2. 02

    The proof changes the cut, not the riffle

    The new paper does not make the whole shuffle model more human. It keeps the Gilbert-Shannon-Reeds interleaving rule intact, where each next card comes from a pile with probability proportional to the number of cards left there. What it relaxes is the initial split, which matters because the classic model already allowed surprisingly uneven cuts through a binomial distribution.

    Do not summarize this result as 'messy human shuffles are solved.' It covers one source of sloppiness. If your use case depends on how cards are actually released from the hands, this paper does not answer that part.

      Attribution:
    • possiblywrong #1
    • zahlman #1
    • jtbayly #1
  3. 03

    Weak shuffling preserves exploitable deck structure

    Magic: The Gathering surfaced a concrete consequence. After play, decks often end up partially sorted into lands and spells, and players sometimes interleave them by hand or rely on pile shuffles. Without enough real riffle-style randomization afterward, that structure survives and can amount to mana weaving or outright cheating. In larger formats like Commander, the problem gets worse because 100-card decks need even more work to mix.

    If your game or product depends on fair random draws, define acceptable shuffling more tightly than 'presented deck was cut.' Tournament rules, onboarding, and UX should account for how easily humans preserve hidden structure.

      Attribution:
    • recursivecaveat #1
    • th0raway #1
  4. 04

    Some card games benefit from residual order

    Several comments pointed out that imperfect mixing is not always a bug from the player’s point of view. Games like gin rummy, bridge redeals, and even tarot readings can feel more textured because human shuffling lets fragments of prior sorting survive. That is mathematically bad randomization, but it can create patterns players notice and interpret as personality, momentum, or just more interesting play.

    When you design physical games or digitize them, decide whether you want statistical fairness or familiar human texture. Players may experience perfect randomness as colder and less legible than the biased shuffles they grew up with.

      Attribution:
    • RNanoware #1
    • brookst #1
    • bombcar #1
    • PaulHoule #1

Against the grain

  1. 01

    Perfect faros are a different phenomenon

    The title invited people to conflate random riffle shuffles with deterministic faro shuffles. A true in-shuffle or out-shuffle can cycle a deck back to its original order after a fixed number of repetitions, which is exactly why magicians care about it. That does not contradict the seven-shuffle theorem. It shows the article used overloaded vocabulary in a way that confuses two different ideas.

    If you explain this topic publicly, separate 'random riffle shuffle' from 'perfect faro shuffle' immediately. Otherwise people will walk away thinking the theorem is self-contradictory.

      Attribution:
    • vessenes #1
    • zahlman #1
    • have_faith #1

In plain english

asymptotic ↩
Describing how a formula or behavior works in the limit as the size of the system becomes very large.
binomially distributed ↩
Following a probability pattern where outcomes cluster around the middle but can still land noticeably off-center, like the number of heads in many coin flips.
Commander ↩
A Magic: The Gathering format that usually uses 100-card singleton decks.
faro shuffle ↩
A highly controlled shuffle that interleaves two equal halves exactly one-for-one in a deterministic pattern.
in-shuffle ↩
A type of faro shuffle where the top card changes because the first interleaved card comes from the lower half.
mana weaving ↩
A card-game practice, especially in Magic: The Gathering, of deliberately spacing resource cards through a deck before insufficient shuffling.
out-shuffle ↩
A type of faro shuffle where the original top card stays on top after the interleaving.
riffle shuffle ↩
A common card shuffle where the deck is split into two packets and the cards are interleaved as they fall together.
tarot ↩
A larger specialized card deck used for divination or reflective readings rather than standard card games.
total variation distance ↩
A measure of how far one probability distribution is from another, used here to compare the shuffled deck to a perfectly uniform random deck.

Reference links

Paper and technical references

Videos and explainers

Card history and personalities

  • John Scarne Wikipedia page
    Shared as background on a famous card expert, magician, and anti-cheating instructor mentioned in the side discussion