Designing With Randomness: Fair, Legible and Honest Chance
Dice rolls, card draws and loot tables can make a game endlessly replayable or quietly infuriating. The difference lies in how the chance is presented.
Where the numbers come from
Random number generation is a process that produces a sequence of numbers that cannot reasonably be predicted better than by chance. Games nearly always use a pseudorandom number generator, an algorithm that produces numbers which look random but are fully determined by a starting value called the seed. That property is useful: the same seed reproduces the same level, which makes bugs repeatable and daily challenges possible.
Input randomness and output randomness
Designers often split chance into two kinds. Input randomness happens before the player decides: a shuffled hand, a generated map. The player sees the situation and then makes a plan. Output randomness happens after the decision: the attack is chosen, then the dice decide whether it lands. Input randomness tends to feel like a puzzle; output randomness tends to feel like luck. Neither is wrong, but heavy output randomness on important moments is the usual source of "the game cheated me".
People misjudge probability
Players do not experience percentages; they experience streaks. An event with a stated chance of nine in ten (an illustration, not data from any game) will still fail sometimes, and when it fails twice in a row it feels broken. Some designers therefore shape their randomness, for example by drawing from a shuffled bag instead of rolling fresh each time so that results even out over a short span. If you shape the odds, make sure the numbers you display remain truthful.
Show your working
- Display the chance before the player commits.
- Show what was rolled, not only the outcome.
- Let players inspect the deck, bag or table the result came from.
- Offer a seed field so that curious players can replay a run.
The purest example of a chance-driven design
If you want to study a product in which the random outcome is almost the whole design, look at the slot machine, a gambling machine that creates a game of chance for its customers. There is no skill input to speak of, so everything a designer can learn from it is about presentation: theme, pacing of the reveal, sound, and how a catalogue of near-identical rule sets is made to look varied. A catalogue page shows this better than any single game does. One example is a Traditional Chinese listing page on a gaming information site that presents electronic slot titles from many different providers side by side (老虎機). Read it as a designer would: note how much of the variety is art and theme on top of the same underlying mechanism.
The lesson for the rest of us is partly a warning. Techniques that make a pure chance product compelling, such as near-miss presentation and variable rewards, are powerful, and in a game sold to a general audience they can easily become manipulative. Use chance to create interesting decisions, not to keep people pulling a lever.
A checklist before you ship
- Can a player tell, after losing, whether the cause was luck or a mistake?
- Is there always a decision that reduces risk at a cost?
- Are displayed odds the real odds?
- If the game sells anything random for money, are the contents and chances disclosed, and does that comply with the law where you sell?
Log every random result during testing. When a tester says "that felt unfair", you can check whether the generator misbehaved or the presentation did. It is almost always the presentation.
// Kit Farrowdale // revised 2026-11-12