What a randomizer has to do
Every game with chance needs a way to produce it. The three classic tools are dice, spinners and cards, and each can generate a chance outcome with a precise probability. What differs is how outcomes are spread over time, whether the tool remembers what it has produced, and how it feels in the hand. Those differences are the real design choice.
Dice: independent, with a built-in shape
A fair die has no memory. Each roll is independent, so past rolls tell you nothing about the next. That means streaks are possible: on a six-sided die, the chance of the same face twice in a row is 1/6, and three of the same face in a row is 1/36, which happens often enough over a long game that players notice.
Dice also offer shapes through combinations. One die is flat, with every face equally likely. Two dice summed give a peak at seven, and more dice give a smoother bell shape, as described under dice odds. Designers can choose the die size or the number of dice to set the amount of spread.
The limits are that probabilities come from the number of faces. Getting a 1 in 5 chance needs a five-sided or ten-sided die, and chances like 1 in 7 need unusual dice or a workaround. Dice also need a surface to roll on, can fall off tables, and can be slightly unfair if poorly made, which is where the question of dice fairness comes in.
Spinners: flexible and quiet
A spinner is a pointer on a wheel divided into sections. The chance of each outcome equals the share of the wheel it covers, so a designer can pick any probability by changing the wedge size: a quarter of the wheel for 25 percent, a tenth for 10 percent, or an odd-sized sliver for a rare event.
Like dice, a spinner has no memory. Each spin is independent, which means repeats and streaks occur just as they do with dice.
Spinners are friendly for young children, since they do not roll away and are easy to read. They are less suited to fine detail: small wedges are hard to read, a worn or loose pointer may favour certain spots, and in some designs the result depends on how hard it is flicked. Players also sometimes try to control the flick, which makes the result partly a skill action.
Cards: a randomizer with memory
A deck is different in a way that matters. When cards are drawn and not returned, the deck remembers what has come out. This is drawing without replacement, and it changes the odds with every draw.
Consider a deck of ten cards numbered 1 through 10. On the first draw, each number has a 1/10 chance. Once a card is drawn, that number cannot appear again until the deck is reshuffled. After nine draws, the tenth card is completely determined. A ten-sided die, by contrast, could show the same number ten times in a row. After the first roll, the chance of nine matches in a row is (1/10)^9, which is tiny, but it is not zero.
This memory has useful consequences:
- Guaranteed spread. If a deck holds exactly four “event” cards in twenty, a player will see four events per full pass through the deck. Streaks of bad luck are bounded.
- Controlled composition. Designers decide exactly what the deck holds and can adjust it by adding or removing cards.
- Information. Players can track what has been played, which turns the deck into a source of strategy as well as luck. Systems built on this idea are why card counting is possible in some card games.
- Dependent odds. After each draw, the remaining chances are conditional on what has already come out.
A deck can copy the shape of dice. A set of 36 cards, one for each ordered pair of two dice, gives the exact two-dice distribution on each pass through the deck. A game using such a deck would produce sevens at the right rate, but without the long gaps or streaks of a true dice roll, because the pass is limited to 36 cards.
Decks have costs. They need shuffling, and a poor shuffle leaves the order partly predictable, as discussed in how many riffle shuffles are needed. They also take table space, wear out, and can be marked or tracked by players over time.
Outcome spread and feel
It is helpful to compare the three on a few dimensions:
| Dice | Spinner | Deck | |
|---|---|---|---|
| Memory | None | None | Yes, until reshuffled |
| Probabilities available | Limited by faces | Almost any | Any, set by card counts |
| Streaks | Possible and uncapped | Possible and uncapped | Bounded by deck contents |
| Table feel | Tactile, noisy, dramatic | Quiet, simple | Hidden information, a sense of “due” |
The sense of “due” deserves a note. In a deck, a card really is more likely to appear later if it has not yet been drawn, so the feeling that something is due is partly justified. With dice and spinners it is not, which is why the gambler’s fallacy applies to them but not in the same way to cards.
Matching the tool to the game
A few rules of thumb follow:
- If you want drama, fast resolution and a built-in sense of tension, dice suit.
- If you want exact, unusual probabilities, or a toy for young players, a spinner suits.
- If you want evenness across a session, or a game where tracking what has happened is part of the skill, cards suit.
- If you want an average outcome but fewer extremes, a deck constructed to match the distribution, or a sum of several dice, suits.
Many games mix tools. A game might use dice for movement, a deck for events and a spinner for a special mechanic. Mixing keeps each source of chance in the role where its behaviour helps.
Whatever tool is chosen, testing is wise. Playing or simulating many rounds shows how often streaks and extremes actually occur, and a feel for that comes more reliably from numbers than from impressions.