Mystery Card Games: Hidden Information, Tracking, and the Reveal

Three cards lie face down. One is a key; the other two are blank. You choose the left-hand card. Before you turn it over, the host reveals a blank somewhere else and offers you the remaining hidden card instead.

There are two cards left. It is tempting to call that an even chance. But something is missing from the picture: how did the host decide which card to show?

A host who knows where the key is and must reveal a blank creates one puzzle. A host who picks a card at random and happens to uncover a blank creates another. The cards can look exactly the same at the moment you decide, while the probabilities differ.

That is what makes mystery card games interesting. A card back conceals an identity, but the game around it determines what you can discover. Sometimes you are drawing from a shuffled deck. Sometimes you are remembering a location, following a moving object, or interpreting a carefully chosen reveal. Sometimes the hidden card simply gives a shared story its next unexpected turn.

The useful starting point is to work out what is known, who knows it, and what happens between choosing a card and seeing its face.

What kind of mystery is on the table?

Imagine four cards labeled on their faces: Fox, Hare, Owl, and Moth. Their backs are identical.

If someone mixes them without looking and you choose one with no clues, a random-selection model may fit. If you watch Fox being turned face down at position B, you have information. If B then visibly swaps with D, that information can travel with the card. If someone secretly rearranges everything afterwards, the earlier observation may no longer locate Fox.

The same small deck supports several different activities:

These families overlap. A matching game begins with a hidden arrangement, then gradually supplies information. A storytelling game might use a shuffled deck or fixed face-down positions. The labels describe useful differences in play, rather than mutually exclusive genres.

Before a round, establish the contents of the pool and how they reach their positions. During movement, establish whether the cards retain their identities. Before a reveal, establish whether the host knows the layout and has a choice about what to expose. Before a final decision, establish whether the player may switch.

Something can be unknown to the player without being unknown to everyone. A keeper who placed the key knows more than the chooser. In a deck mixed without anyone looking, neither person needs know the order. Those arrangements create different opportunities for clues and decisions.

Hidden does not automatically mean equally likely

Suppose a host always puts Fox at C. Turning the cards face down does not give each fixed position a one-quarter chance of containing it. C contains Fox every round, whether you know that or not.

There are different ways to introduce a probability model. The host could assign the cards through a uniform random shuffle. Alternatively, you could choose a position uniformly at random: against any fixed single-Fox arrangement, that guessing strategy succeeds one time in four in the mathematical model.

The two explanations put the randomness in different places. They also leave different room for learning. If you discover that Fox is always at C, choosing C becomes useful. If an independent uniform shuffle determines its location every round, last round's location does not predict this round's.

For the equal-chance examples below, assume identical backs, no marked edges or other clues, and the stated random-selection procedure. Mixing physical cards is a practical way to conceal their arrangement; its appearance alone does not prove that every arrangement is equally likely.

What changes when a card leaves the deck?

Start with a standard 52-card deck, without jokers, in a uniformly random order. There is one ace of spades, four aces, and 13 hearts. On the first draw:

Counting favorable cards works here because each card is equally likely to occupy the draw position. The word ace groups four distinct outcomes; it does not describe one special card.

Now draw an ace and leave it out. There are 51 cards remaining, including three aces. The next draw has probability 3/51 = 1/17 ≈ 5.88% of being an ace. If the first card was instead a non-ace, all four aces remain, giving 4/51 ≈ 7.84%.

These are conditional probabilities: the answer uses something you have learned about the first draw. Removing an ace makes the remaining deck poorer in aces; removing a non-ace makes it slightly richer.

If nobody tells you what the first card was, the probability that the second card is an ace remains 4/52. In a uniformly shuffled deck, the second position has the same ace probability as the first. There is no contradiction. The composition physically changes after removal, while the probability you assign depends on the information available to you.

Returning a card changes the next draw

With replacement, put the drawn card back and randomize again so the next draw is independent and uniform. The deck is restored to four aces among 52 cards. Merely returning an ace to the top and immediately drawing it again would not satisfy those assumptions.

Without replacement, leave the card out. The draws are linked by what has been removed. Python's official sampling documentation distinguishes these operations explicitly; the NIST statistical glossary describes the hypergeometric model for sampling without replacement.

For two draws, the distinction produces small but real differences:

For the second row, count the opposite event first. Getting no aces means drawing a non-ace twice. Subtract that probability from one to obtain the chance of at least one ace.

The apparently surprising comparison makes sense: removing a non-ace improves the next opportunity to find an ace, but removing an ace reduces the opportunity to find another. “At least one ace” and “two aces” ask different questions.

A remembered location is information too

In a matching game with four pairs, eight cards begin face down in a uniformly random arrangement. Turn over one card. If you have no information about the others, exactly one of the seven remaining cards matches it. Choosing a second card without clues gives a match probability of 1/7 ≈ 14.29%.

After several turns, that calculation may no longer describe your choice. If you correctly remember where the matching card was exposed, you can select it directly. If you remember that C held a different symbol, you can rule C out. Cards turned face down again retain their identities unless the rules call for a new shuffle.

For example, an earlier miss showed a Moon at B and a Tree at F. Later you uncover the other Moon at D. If B has not moved and you remember it correctly, choosing B completes the pair. The card back hides the face, but it does not erase what you learned.

Following a card is different from guessing

A tracking game starts by showing you the target. After that, its movements supply evidence.

Suppose Fox begins at B. The host swaps B with D, then A with C, then D with A. Fox travels from B to D, stays at D during the second swap, and finishes at A. There is one answer implied by those operations.

Visible movement does not guarantee that a player can follow it. Crossings, pace, interruptions, and the player's abilities can make the task difficult. What matters mathematically is that the movement preserves an identity that could be followed. Randomness might have chosen the swaps, but once the full sequence is known, the target's final location follows from it.

Pylyshyn and Storm's original tracking study investigated observers maintaining designated targets among moving, similar-looking objects. That supports treating tracking as a studied perceptual task. It does not establish a universal success rate for card games, show that every player can follow every sequence, or demonstrate that playing these activities improves memory generally.

For four cards containing one target, a person who ignores the evidence and chooses uniformly at random has a 1/4 chance of success. An observer's actual performance is a separate matter. A systematic mistake could even lead them to choose the wrong location more often than that benchmark suggests.

Now imagine that, after the visible swaps, the host applies a new, independent uniform permutation behind a screen. Fox has a 1/4 chance of occupying each final position, regardless of its tracked location before that operation. The earlier observations no longer predict its new position. An unspecified hidden movement would not justify the same conclusion: the independence and uniformity are doing the mathematical work.

That boundary deserves an explicit rule. “Follow the card” promises that the displayed movement matters. “The cards are randomly reassigned after the lights go out” announces a new chance event.

The host's reveal is part of the puzzle

A reveal can tell you more than the identity of one card. It may also tell you something about the choices the host was able to make.

Return to the key and two blanks. If the keeper knows the layout and must expose an unchosen blank, seeing a blank is guaranteed by the rules. If the keeper chooses an unchosen card at random regardless of its face, a blank is only one possible result. Sometimes the key itself appears.

Those procedures leave different sets of possible histories behind the same-looking table. To interpret a reveal, you need the rule that produced it. Richard Gill's mathematical treatment of the Monty Hall problem carefully distinguishes an overall switching strategy from a probability conditional on the particular option a host opened.

The Museum's Last Key below makes the host's knowledge, mandatory reveal, tie-breaking, and switching opportunity explicit. First, two other activities show what happens when the hidden cards have quite different jobs.

Three games to put the ideas to work

The following activities use original themes, prompts, and illustrative rounds. Their instructions are practical designs, not tested learning interventions. The switching puzzle uses established Monty Hall mathematics. All story contributions and sample outcomes are fictional examples, and the suggested pace can be adjusted for the people playing.

The Parcel That Changed the Ending

The idea: Build a short shared story through three card reveals. The pleasure comes from making an unexpected ingredient matter. There is no winning card and no score.

Players and materials: One to six storytellers, with one person also facilitating. A solo player handles both roles. Prepare six cards with identical backs, paper or a shared text area, and six table positions labeled A–F from left to right. Cards move; position labels stay on the table.

Write these faces, including the full instructions:

Review all six ingredients together. Mix the cards face down without looking, then deal one to each position. Neither the facilitator nor the storytellers need know the mapping. After dealing, each face is fixed to its position until revealed; the facilitator must not exchange cards to improve the story.

Write or read the opening:

“The last parcel at an empty station has no address.”

Agree on a light fictional tone and a clockwise order of offered turns. Nobody has to perform a voice or disclose anything personal. One person may contribute to all three beats, and anyone can pass.

Play the story:

1. Offer the next storyteller a turn. They choose any unrevealed position, naming it or asking a helper to indicate it. A tentative choice can change until they say “Reveal D,” or the corresponding label.
2. Turn over that actual card, read its complete ingredient aloud, and leave it visible at its position. No further switching occurs after the reveal instruction.
3. Invite one to three sentences that connect the ingredient to the unfolding story. A written line, a dictated line, or a pass is equally usable. If the storyteller passes, invite a volunteer; if nobody volunteers, leave the ingredient as an unresolved detail.
4. Move to the next offered turn and choose from the still-hidden cards. Keep revealed cards face up and unavailable. Do not reshuffle between beats.
5. After the third reveal, offer one voluntary closing sentence. End the story even if some details remain mysterious. If nobody wants the next offered turn at any point, close early rather than requiring another selection.

An illustrative story: For this example, the hidden layout is A Key, B Bell, C Seed, D Ticket, E Map, F Note. This description is for the reader; the players do not know it in advance.

Mina chooses D and says, “Reveal D.” Ticket appears. Mina contributes:

“The station worker finds a ticket caught beneath the string. It says the parcel will arrive tomorrow, although it is already sitting on her desk.”

Oren chooses F. Note appears:

“Inside the ticket is a folded note in her own handwriting: ‘Please leave this unopened until I get there.’ She checks the empty platform again.”

Mina chooses A. Key appears:

“A small key slips from the fold. Across the waiting room, a keyhole has appeared in a plain brick wall, with a chalk outline showing where a door will stand at dawn.”

Oren offers the closing sentence: “She puts the parcel beside the wall and makes two cups of tea.”

The reveal has changed what the storytellers can build on. They have made creative decisions after each card; no probability calculation determines the best ending.

Under an ideal uniform starting shuffle, each of the six ingredients has probability 1/6 of appearing at any specified first-choice position, provided the chooser has no positional clues. After Ticket is revealed, there are five hidden cards left. If nothing else identifies their locations, a specified remaining slot has probability 1/5 of containing Note. That is a conditional calculation about the remaining arrangement, not a fresh draw from all six cards.

Reset and interruptions: A reveal followed by a pass still counts as one of the three reveals. During a pause, preserve the cards and current story. If the hidden layout is disturbed, close this story with what is already known. To begin another, restore all six cards, mix them afresh, and start from the opening again; nobody needs to reuse an earlier contribution.

Adaptations and access: Read every ingredient aloud and keep its text available. Use large print, simple drawings alongside words, or tactile position labels. A player can choose by speaking a label and have the facilitator turn the card. Younger players can contribute a phrase or picture instead of sentences. In remote play, display the revealed ingredient in chat and accept typed contributions. A digital adaptation needs stable card identities and readable results; the story does not require a moving shuffle animation.

The Clockmaker's Courier

The idea: Follow a parcel through five visible swaps. Observation determines what information you have; a new lottery does not take place when you answer.

Players and materials: One mover and one or more observers. The mover runs the cards and does not compete in that game. Prepare four identical-backed cards labeled Parcel, Clock, Boot, and Umbrella; fixed table positions A–D from left to right; a fair six-sided die; and a private move record. Observers need a way to record an answer privately, such as paper, a message to the mover, or a helper.

Starting information: Place Parcel at A, Clock at B, Boot at C, and Umbrella at D, all face up. Everyone may inspect the arrangement until ready. Identify Parcel as the target, then turn all four cards face down without changing positions. The mover knows the identities; observers have seen the same starting layout.

Generate five moves by rolling the die privately five times. Use this mapping, recording the results before moving any cards:

A repeated roll is valid and means repeating that swap. A swap always refers to the cards currently occupying the named positions. The paper position labels stay still.

Play a round:

1. Perform the five recorded swaps in order. Keep every movement visible, use at least two seconds per swap, and let both cards settle before beginning the next. Agree on a slower pace before starting if needed. The two-second allowance is a house rule, not a research-based threshold.
2. Preserve each physical card throughout. Do not turn faces up, exchange identities, add a secret shuffle, or alter the sequence in response to an observer.
3. After the fifth swap, leave the cards still. Each observer records the position believed to hold Parcel. Tentative answers may change until the observer locks; there is no response timer. Keep answers private until everyone has locked, or declined to answer.
4. Collect the locked answers, then reveal all four faces and read the final A–D arrangement aloud. Check any disputed position against the recorded moves.
5. Give one point for a correct locked answer and zero for a miss or declined answer. There are no speed bonuses or penalties. Complete six valid rounds, then end. Highest score wins; equal scores remain tied.

An illustrative round: The five die results are 2, 5, 6, 3, 1. They specify A–C, B–D, C–D, A–D, then A–B.

Read just Parcel's journey as A to C, remaining at C, then D, then A, then B. Nia locks B. Eli loses track during the third movement and locks D. When all faces are revealed, B holds Parcel: Nia receives one point and Eli receives zero. The mover can replay the written sequence slowly to explain where the answers diverged.

This fictional round illustrates the rules, not a measured difference between two players. A perfect logical tracker reaches B. A uniformly random choice among A–D has probability 1/4 of succeeding against the final layout. That benchmark does not require the five-swap procedure itself to produce every possible final layout equally often.

Reset and interruptions: Restore the same known face-up starting arrangement before every round, let everyone inspect it, and record five fresh rolls. Do not carry the final layout into the next round. If movement is interrupted, a card is exposed, or a swap cannot be reconstructed, void that round without points. Restore the starting arrangement and generate a new sequence for the replacement round. Completed scores remain.

Logical-tracking alternative: Use the same starting layout, five moves, answer lock, reveal, scoring, and reset, but announce each swap in words or provide it as text. Permit notes and repeat an instruction on request. A player who cannot watch the movements can work out the destination from the operations. This is a sequence-reasoning activity; it does not assess the same perceptual skill as following continuous movement. Keep scores separate from the visual version.

Large face labels, a clear playing surface, and slower movements can help some visual players. Stable tactile labels identify the table positions without marking the target's back. A helper can record a spoken answer privately. None of these adjustments makes rapid visual tracking equally usable for everyone, which is why the logical version is a separate, complete way to participate.

A separate blackout demonstration: To show what destroys the value of the earlier tracking, announce a different rule before starting. After the five swaps, the mover works behind an opaque screen and rolls until obtaining 1–4; discard and reroll 5 or 6. Put Parcel at A for 1, B for 2, C for 3, or D for 4. Fill the other positions, from left to right, with Clock, Boot, and Umbrella in that order. Keep the accepted roll hidden until answers are locked, then reveal everything.

With independent fair rolls, each accepted number has probability 1/4. The target's new location is independent of its earlier journey. Only the target placement needs to be uniform for this demonstration; the decoy arrangement follows the stated fixed rule. Use the ordinary reset afterwards, and record blackout scores separately. Watching the first five swaps perfectly cannot identify this newly assigned location.

The Museum's Last Key

The idea: Choose a hidden key after the keeper reveals a blank. The keeper's instructions determine what that reveal tells you.

Players and materials: One keeper and one chooser; observers may make unscored predictions. Prepare three identical-backed cards, one labeled Key and two labeled Blank, three fixed positions A–C from left to right, a fair die, a fair coin, and a score sheet.

Setup and information: Before each round, the keeper secretly rolls the die. Results 1–2 put Key at A, 3–4 put it at B, and 5–6 put it at C. Place the blanks in the remaining positions. Under the fair-die assumption, each key location has probability 1/3. The keeper knows the whole layout. The chooser knows the composition and procedure, but has no clue to the placement.

Explain the complete reveal rule before the first choice: the keeper must always reveal an unchosen Blank, must never reveal the chosen card, and must always offer a switch. When both unchosen cards are blank, a fair coin selects which to expose: heads means the alphabetically earlier eligible position; tails means the later one. If only one unchosen Blank is available, no coin is needed.

Play six rounds:

1. The chooser names a tentative position and may reconsider until saying “Lock A,” or the corresponding position. Do not reveal anything before that lock.
2. The keeper exposes an unchosen Blank according to the rule above, leaves it face up, and announces its position and face.
3. Offer “Stay with your original card, or switch to the other hidden card?” Both remaining positions must be named. The cards do not move.
4. Allow an untimed decision. The chooser may reconsider until explicitly locking Stay or Switch. If they decline to decide, void the round without points, reveal the layout, and end the game early without requiring a replacement.
5. Reveal all cards. Give one point if the final selected card is Key, otherwise zero. Record the original choice, revealed position, final choice, and result.
6. Restore all three cards and make a fresh secret placement for the next round. End after six completed rounds and report the chooser's score out of six. Roles can be exchanged for a separate game with a fresh score.

An illustrative round: The keeper rolls 3, so Key is at B. A and C contain blanks. The chooser locks A. The keeper must show C: it is the only unchosen Blank. The chooser switches to B, locks Switch, and finds Key. Award one point.

For contrast, suppose Key was at A and the chooser again locked A. Both unchosen cards would be blank, so the keeper would flip the coin. If C were exposed, switching to B would lose. A switch is a strategy with a particular success probability, not a promise about this individual round.

Why switching wins two-thirds of the time in the model: Your original choice has probability 1/3 of finding Key. It has probability 2/3 of being wrong. Whenever it is wrong, the keeper has to reveal the other Blank, leaving Key as the only alternative hidden card. Switching wins precisely in those initially wrong cases.

For an initial choice of A, the complete key-location comparison is:

Each row begins with probability 1/3. Staying wins one row; switching wins two. The same reasoning applies to an initial choice of B or C. The host is not simply removing a random option: the rule protects Key from being exposed.

The fair tie-break also makes the conditional probability the same whichever blank position you see. With an initial A choice and a C reveal, the histories have weights 1/6 for Key at A, where the coin must select C, and 1/3 for Key at B, where revealing C is forced. Key at B therefore has conditional probability (1/3) / (1/6 + 1/3) = 2/3.

That last detail matters if you change the keeper's preferences. Suppose the informed keeper always prefers C when both blanks are available. After choosing A and seeing C exposed, Key at A and Key at B would then be equally likely, giving a 1/2 switching chance for that particular reveal. Always switching across all rounds would still win with probability 2/3. The overall strategy and the probability after one named reveal are different quantities.

This is the host-protocol distinction discussed in Gill's analysis. The arithmetic here was independently enumerated under the stated rules; it is not inferred from how persuasive the reveal looks.

The random-reveal variation: Announce this as a separate game and start a new score sheet. Keep the same uniform key placement and initial choice lock. This time, the keeper flips the coin to choose between the two unchosen positions, regardless of what they contain: heads reveals the alphabetically earlier position and tails the later one. The keeper may know the layout, but that knowledge must not influence the choice. No fresh coin flip is allowed because the first would expose Key.

If Key is revealed, announce “Key exposed; this attempt ends,” record that outcome separately, award no point, and show the remaining cards. There is no switching decision. The attempt counts toward the game's six attempts; it is not replaced with a safer reveal. If Blank is revealed, offer Stay or Switch and complete the round normally. A procedural error can be marked void and replaced, but an exposed Key is a valid outcome under this variation.

For an original choice of A, the possibilities are:

The six cells are equally likely: each key location has probability 1/3, followed by either coin result with probability 1/2. Two cells expose Key. Four expose Blank; within those four, staying wins twice and switching wins twice.

Thus Key is exposed with probability 1/3, and a blank is exposed with probability 2/3. Conditional on a blank being revealed, staying and switching each succeed with probability 1/2. This is a different model from the informed keeper's mandatory-blank procedure, even when the visible table looks identical.

Keep separate counts for attempts started, exposed-Key endings, blank-reveal rounds, and wins after each decision. If every attempt with a blank is followed by switching, the overall winning probability per started attempt is (2/3) × (1/2) = 1/3. The one-half figure uses only blank-reveal rounds as its denominator. Six attempts need not resemble any of these theoretical proportions.

Reset, interruptions, and access: Never move a card after an initial lock. If the layout remains intact during a pause, restate the chosen and revealed positions before resuming. If identities, the locked choice, or the procedure cannot be recovered, mark the round void and use a fresh placement; do not improvise a favorable reveal. Preserve completed records and distinguish voids from valid exposed-Key endings.

Use written Key and Blank labels as well as any symbols. Read every reveal and both remaining choices aloud, leaving the revealed face visible. A chooser can respond by speech, text, pointing, or a helper; choices have no time limit. A nonvisual player needs the position labels, the host protocol, and the spoken state, not an image of the card backs. An optional prediction sheet can compare what staying and switching would have done on the same informed-host round, without pretending those counterfactual results were two independently played trials.

What a digital shuffle can and cannot tell you

Physical cards make one relationship easy to grasp: turning over the card at C reveals the face already at C. Software can follow that same relationship, but a card-shaped button does not establish it.

A digital game might shuffle persistent card identifiers into positions, then return the identifier at the chosen position. It might choose an outcome before playing an animation. It might wait for a click and make a new random draw, with every card back acting as an equivalent button. Or it might keep card identities through visible movement specifically so that tracking matters.

A random choice among identifiers is also a distinct design decision. If several identifiers share the label Moon, equal chances for the identifiers do not necessarily give equal chances to the displayed labels. If outcomes are weighted, counting the visible backs does not establish their shares.

These mechanisms can serve different purposes. A preselected story ingredient can work perfectly well when the animation is simply its introduction. A tracking challenge needs the displayed operations to preserve the target's identity and determine where it ends. A hidden reassignment is a new operation that should be disclosed.

NIST describes Fisher–Yates as a random-permutation algorithm. Its uniform-permutation model requires the appropriate uniform random choices at each step. Smooth motion, elaborate crossings, or a convincing flip cannot establish that the selection procedure uses that model.

For a digital adaptation, establish when the outcome is fixed, what a selected position means, when a choice becomes final, and whether reset restores the pool. An animation ending should not silently create a second draw when the rules promise to reveal an existing card. A failed visual effect should not erase a result already committed and reported. If no authoritative result or recoverable state exists, cancel the unresolved round under the game's interruption rule.

These activities need no particular application. Paper cards and a clear explanation of the procedure are enough.

Make the result available to everyone playing

A reveal should leave people with information they can use. A persistent line such as “Position B: Parcel. Your answer was B; correct” communicates more than a brief sparkle or victory sound. Read it aloud when playing together, and keep the text available long enough to revisit.

Use stable position labels alongside color or decoration. W3C's guidance on color recommends conveying meaning through additional information such as text or shape. A red outline alone is a poor way to explain that a choice is locked or incorrect.

For digital adaptations, players should be able to select, confirm, stay, switch, and reset using the keyboard. Give controls meaningful names and state labels. W3C's keyboard guidance addresses operation without requiring precise pointer use or timed keystrokes. Its status-message guidance explains how updates can reach assistive technologies without moving focus. Concealed controls should identify their positions without announcing hidden faces prematurely.

Decorative motion can be removed from the storytelling and key games while retaining their choices and probability models. W3C's interaction-animation guidance concerns disabling nonessential triggered motion. A visual tracking challenge needs a different response: slower motion may help some players, while spoken or written swaps provide a distinct logical activity for others. Neither should be advertised as an identical test of ability.

Before inviting people to play, rehearse a complete round. Check the starting arrangement, an ordinary selection, an attempted change before and after locking, the actual reveal, scoring, and reset. Stage an interruption and verify that the restart rule is clear. For the keeper puzzle, check an initially correct choice, an initially wrong choice, both coin results, and a winning-card exposure in the random-reveal variation.

A mechanical rehearsal checks that the instructions can be followed. It does not establish how enjoyable, accessible, or educational the game will be for every group. Ask players what they need to read, hear, choose, and revisit, then adjust the format accordingly.

The best moment in a hidden-card game is often the instant the face appears and the earlier choices become understandable. In one game, a ticket changes the story. In another, the parcel finishes exactly where the swaps sent it. In the key puzzle, the revealed blank matters because of the rule that put it on display. Clear procedures let players enjoy the surprise and understand what produced it.

Sources and research notes

The references below support the mathematical models, research summaries, or practical guidance discussed here. Worked scenarios and suggested activities are editorial examples unless explicitly identified as recorded calculations.

The three activities, their themes, prompt wording, and fictional rounds are original editorial designs. Their underlying probability principles are established mathematics. Exact calculations and exhaustive checks verified the draw examples, informed and random reveal protocols, story-card counts, and tracking arrangements. The four-card tracking check covered all 24 starting permutations and all 6^5 five-swap sequences, or 186,624 cases. These checks establish consequences of the modeled rules, not human tracking accuracy. No participant study or physical playtest of these complete games was conducted.

• Sampling and shuffling: Python's random-module documentation distinguishes choice, sampling, and shuffling. NIST's glossary describes sampling without replacement; NIST's Fisher–Yates entry identifies a random-permutation procedure. These sources do not audit a particular physical shuffle or digital application.
• Host-dependent reveals: Richard D. Gill, The Monty Hall Problem is not a Probability Puzzle, accessible author manuscript, distinguishes overall switching success from probabilities conditional on named host actions. This article uses explicitly stated protocols and independent enumeration rather than assuming every reveal puzzle has the same answer.
• Tracking evidence: Zenon W. Pylyshyn and Ron W. Storm, Tracking multiple independent targets: evidence for a parallel tracking mechanism, 1988. The research account is limited to the accessible original abstract. The supplied research also cites Fehd and Seiffert's eye-movement study; its full text was not reliably retrievable, and no gaze-specific findings from it are asserted here. Neither paper tests the activities in this article.
• Access guidance: W3C's WCAG 2.2 and the linked explanations of color, keyboard operation, status messages, and interaction animation inform the recommendations. They are design guidance, not evidence that a particular implementation meets every player's needs.

Digital mechanisms are described generically. No product-specific behavior, runtime verification, or application accessibility certification is claimed.