Play Patterns: A Design Vocabulary for Card Decks

It started with a flip.

I was designing a deck and caught myself doing the thing I always do with a card: reading the front, pausing, turning it over to see what's hiding on the back. And it hit me that this tiny gesture is the whole reason cards exist as a format. A card has two faces. You only see one at a time. Flipping is an act of revelation, and no book, no slide, no PDF can give you that moment.

Then I started listing all the other moves that belong to cards. Tapping one sideways. Stealing one from the table next to you. Holding one in secret and slamming it down at the same time as everyone else. Each move exploits something a card physically is. And each one, I realized, is a reusable building block for designing activities.

I call them play patterns. This article is my catalog of them.

Two warnings before we start. First, this connects to the docudeck problem: a deck designed with zero play patterns is a document in a box, and I wrote a whole rant about that. Second, this catalog is a living document. I'll keep adding, merging, and renaming patterns as I find them, and I'd genuinely love your help spotting the ones I missed. What you're reading is version one.

What counts as a play pattern

My test is simple: the move has to exploit a physical property of cards. Cards are two-faced, discrete, shuffleable, holdable in secret, and small enough to pass around. Every pattern below leans on at least one of those properties. Writing on a whiteboard is a fine activity, but it's not a play pattern, because a whiteboard can do it too.

I've sorted the moves into three families based on what actually changes: the card's angle, the card's owner, or the card's position relative to other cards.

Family one: orientation

The card stays where it is. Its angle changes. And here's a detail I find a little too satisfying: flip turns the card out of its plane, while reverse and tap both spin it within the plane, one a half turn and the other a quarter turn. Three moves, and between them they cover every way a card can rotate.

Flip is rotation over an edge, and it's the king. Front prompts, back answers. What makes it powerful goes beyond drama: when the front asks a question and you commit to an answer before turning the card, you're doing what learning scientists call retrieval practice, the single best-documented study technique there is. In the classic experiment, people who recalled material actively retained 61 percent of it a week later, against 40 percent for people who just re-read it (Roediger and Karpicke, 2006). A flip card with a question on the front is a retrieval machine you can hold in one hand. The design rule that follows: make people commit before they flip. A flip without a guess first is just reading with extra steps.

Reverse is rotation within the plane, half a turn. The card carries two readings on the same face, one per orientation, and turning it upside down switches which one is active. If that sounds exotic, it's actually the oldest trick in the catalog: Tarot has distinguished upright from reversed meanings for centuries. In a serious deck, reverse is how you encode duality on one card. A strength read one way, its shadow side read the other. A question and its counter-question. I use it less than flip, but when a deck is fundamentally about tensions and opposites, nothing else says it as physically.

Tap is rotation within the plane, a quarter turn, and it means the card's status changed. Active becomes exhausted. Available becomes used. Discussed becomes done. The card becomes its own little status display that the whole table reads at a glance. My favorite fact in this entire catalog: this move was considered so valuable that Wizards of the Coast patented it. US patent 5,662,332, filed in 1994 for Magic: The Gathering, covering the act of rotating a card 90 degrees to show it's been used. A quarter turn of cardboard, patented. It expired in 2014, so tap away. Just keep it to two states; I've watched four-orientation schemes collapse into "wait, which way means done?"

Family two: possession

The card changes hands, or changes zones. This family is the social one.

Draw is the fundamental move: take an unknown card from the deck. It injects randomness, and randomness is one of the oldest documented creativity techniques we have. Random stimuli knock thinking out of its ruts by forcing associations you'd never choose deliberately, and the classic low-tech implementation of the random word method is literally shuffling a pile of index cards. Draw also solves a facilitation problem nobody talks about: it picks the next prompt so you don't have to, and the deck can't play favorites.

Deal distributes cards to people, and its real product is ownership. The three cards in front of you are yours in a way a shared list never is. Play is the opposite direction: committing a card from your hand to the table, turning a private opinion into a public, physical object the group can point at instead of pointing at you.

Then comes my favorite pair in the whole catalog, because they're the same transfer with opposite souls.

Give means you choose a card to hand to someone. Steal means you take a card from someone for yourself. Give is choosing for another person: you scan your cards thinking about who they are and what they need, which is perspective-taking disguised as a game move. Steal is choosing for yourself at someone's expense, and it's the pattern nobody dares put in a serious deck. That's a shame. Send someone to steal one insight from the neighboring table and watch what happens: energy spikes, tables start watching each other, and what people choose to steal tells you exactly what they value. Game designers know this family well; they call these hostile moves "Take That." Facilitators just need to keep the stakes playful. Steal cards, never dignity.

Hide in hand is the quiet setup pattern: cards you see and others don't. Private information inside a shared activity. On its own it does little, which is exactly the point; hiding exists to make revealing possible, and we'll get to the reveal in family three. Discard (and its gentler cousin, setting a card aside) makes rejection physical. Choosing what you won't keep is a real decision, and the discard pile is a diagnostic goldmine that almost every activity forgets to debrief. And mulligan, a compound of discard and draw borrowed from card games: don't like your hand, exchange it once. It's a second chance built into the rules, and it quietly tells participants that the randomness serves them, never the reverse.

Family three: arrangement

Multiple cards in relation to each other. This is where cards become a thinking surface.

Order puts cards in a line where position means something: priority, chronology, intensity. Its power is brutality. A line has exactly one first place, so the trade-off conversations a group can dodge forever in discussion become unavoidable at the table. And order has a softer talent too: storytelling. A sequence is a storyline, and laying cards left to right is the fastest way I know to build a narrative together, one beat per card, rearrangeable until the story clicks. Group is its sibling with buckets instead of positions. I almost called it sort, and the research world does call it card sorting, but sorting suggests putting a list in order, and order already owns that job here. Grouping is about belonging: distribute the cards into piles by what they have in common. It's also the most academically respectable pattern in the catalog: card sorting is a formal UX research method, and psychology has used it since the Wisconsin Card Sorting Test. The open variant, where participants invent the categories themselves, shows you how a group actually thinks. The contested cards, the ones that fit two piles, are the payload. Fight for them.

Combine places two or more cards together to mean something none of them means alone. The researchers who study design cards name combinatorial creativity as one of the core things cards afford (Lucero et al., 2016), and the math is on your side: a modest 30-card deck contains 435 possible pairs. Pairing, matching two cards that belong together, is combine's most common special case. My one rule: never let a combination land without a task attached. "These two cards, now pitch the product" works. "These two cards, interesting, right?" dies on the table.

Spread lays the cards out so everything is visible at once, and the deck stops being a sequence and becomes a landscape. People stand up, walk, browse, and pick the card that speaks to them. The name comes straight from Tarot, and picture-deck facilitators have used the pick-what-resonates opener forever. Shuffle manufactures the randomness that draw spends, and it performs something too: a visible shuffle is a public promise that nothing is rigged. Handing the deck to a participant to shuffle transfers that trust to them.

And then there's simultaneous reveal, the pattern I'd defend in a knife fight. Everyone commits privately, everyone reveals at once. What it kills is anchoring, one of the most robust biases in psychology: the first number spoken in a room drags every following estimate toward it, and the anchor works even when it's meaningless (Tversky and Kahneman, 1974). Lock all choices before any is visible and the anchor never exists. This is exactly why Planning Poker is played with cards face-down and flipped together. One honest warning from the field studies, though: teams that get sloppy about the reveal discipline, letting early birds show their card first, lose the entire benefit without noticing (that finding shows up in Planning Poker case study research). The pattern works when the synchronization is real. Police the countdown.

The second level: card design patterns

Everything above is a move. There's a second level: ways the card object itself is designed. The two levels meet in the flip, which only exists as a move because someone designed the card with two meaningful faces.

The two-faced card is that design decision made deliberate: deciding what belongs on the face everyone can see and what belongs on the face you earn by flipping. Partial information designs the card to withhold: a blank to fill, a sentence stem, an ambiguous image that each viewer completes differently. Incomplete cards turn readers into authors, and projective decks like OH Cards have built their entire genre on images with no official meaning. My favorite version is the simplest: a word without its definition. A docudeck prints the word and the definition on the same card, and all you can do is read it. A real deck splits them across two cards, and suddenly matching words to definitions is the activity. Same content, one design decision apart, and one of them is a game. Blank cards take that to its conclusion: empty cards in the box, inviting participants to extend the deck in its own format.

And one pattern I've never seen documented anywhere, so I'm naming it here: card mirrors. Divide the card face in two halves carrying the same content, one half rotated 180 degrees. Place the card flat between two people facing each other, and both read it right-side-up at the same time. No more tilting, no more "wait, let me turn it toward you." One card, two readers, perfect symmetry.

How to use this catalog

Please don't try to cram ten patterns into one deck. The principle I argued in the docudeck article holds: one primary play pattern, maybe a secondary to support it. The primary pattern is your deck's identity. Everything past three is a complexity smell.

Patterns also chain, and a few chains come up again and again:

• Shuffle, draw, flip: the discovery chain. The engine of most great prompt decks.
• Deal, hide in hand, simultaneous reveal: the anchoring-free chain. Use it any time honest individual opinions matter more than fast consensus.
• Spread, group, order: the convergence chain. From landscape to categories to priorities in one session.
• Draw, combine, play: the creation chain. Random inputs, forced association, public commitment.

Help me grow this thing

Fifteen moves, five design patterns, one made-up insult for decks that use none of them. That's version one.

I have a backlog of candidates I haven't developed yet: stacking cards into piles, covering one card with another, passing the whole deck between tables, playing cards as votes. Some will graduate into the catalog, some will turn out to be variants of what's already here.

So here's my question, and I'm asking it for real: what move do you make with cards that isn't in this catalog? Find me a pattern I missed and I'll add it, with your name on it.

Sources

• Roediger, H. L., & Karpicke, J. D. (2006). Test-enhanced learning: Taking memory tests improves long-term retention. Psychological Science, 17(3), 249–255.
• Tversky, A., & Kahneman, D. (1974). Judgment under uncertainty: Heuristics and biases. Science, 185(4157), 1124–1131.
• Lucero, A., Dalsgaard, P., Halskov, K., & Buur, J. (2016). Designing with cards. In Collaboration in Creative Design (pp. 75–95). Springer.
• Roy, R., & Warren, J. P. (2019). Card-based design tools: A review and analysis of 155 card decks for designers and designing. Design Studies, 63, 125–154.
• Chi, M. T. H., & Wylie, R. (2014). The ICAP framework: Linking cognitive engagement to active learning outcomes. Educational Psychologist, 49(4), 219–243.
• US Patent 5,662,332 (1994, expired 2014). Trading card game method of play, including the tap mechanic. Wizards of the Coast.
• Haugen, N. C. (2006). An empirical study of using Planning Poker for user story estimation. Proceedings of AGILE 2006.