Same average, different feel
Variance measures how widely results spread around the average. Two systems can have an identical average and feel nothing alike. Rolling a single twenty-sided die and rolling three six-sided dice both average 10.5. The single d20 gives every number from 1 to 20 an equal chance. The three dice produce totals from 3 to 18, clustered near the middle, so extreme results are rare.
A designer choosing between these is not choosing the average. They are choosing how much of the result is left to chance on any given roll.
What high-variance designs feel like
A high-variance game has large swings. Outcomes are widely spread, and surprising results are common.
Typical effects:
- Tension and drama. Comebacks are plausible, and a long shot can succeed. Stories write themselves.
- Accessibility. A newer player has a realistic chance against an experienced one on any given game, because luck can outweigh skill in a single result.
- Short-term unreliability. Planning is harder. A well-judged decision can still turn out badly, and the result alone says little about whether the choice was good.
- Slower skill signal. It takes many games to tell who is better, which is one reason ratings and results from small numbers of matches should be read cautiously.
What low-variance designs feel like
A low-variance game is steadier. Results track decisions more closely.
- Skill is easier to see. The better player usually wins, and the same player usually wins again.
- Planning pays. Long-term strategy is meaningful because the plan is unlikely to be derailed by one roll.
- Less drama per game. There are fewer wild reversals, and a lead is often hard to overturn.
- Higher barrier for beginners. If skill dominates, newer players may lose repeatedly and find it discouraging.
Chess has no dice at all, and it sits at one end of the range. Pure dice-racing games sit at the other. Most games land in between, which is why discussions of board game luck often come down to degree, not a simple split.
The ways designers tune variance
Variance is not a single switch. Several design levers move it.
- Dice size and count. More dice summed together make results cluster; a single large die spreads them out. The arithmetic behind this is explained through game variance with dice sums.
- Frequency of random events. Many small random events tend to average out over a game, while a few large ones decide it. This is the law of large numbers working inside a single match.
- Size of payoffs. Big swings from one event raise variance even if the event is rare.
- Mitigation mechanics. Re-rolls, hand management, drafting and choices after seeing the roll let players soften bad luck. Systems like pity timers do something similar in video games.
- Game length. Longer games give luck more chances to even out, so the better side tends to win more reliably. The effect of series length in competitive play is the same idea applied to matches.
- Hidden information. Cards drawn from a shuffled deck add variance that interacts with player choices.
Who each style suits
There is no better choice in general, only a better fit.
- Casual and family groups often enjoy higher variance, because anyone can win and the mood stays light.
- Competitive players often prefer lower variance, since they want results to reflect their decisions.
- Spectators may favour swings, because unexpected moments are exciting to watch.
- Designers of repeat-play games, like roguelikes, often use high variance for freshness while giving players tools to manage it.
Mixed groups can be tricky. A game that delights a casual table may frustrate a competitive one, and the reverse. Adjustable rules, such as a handicap or an optional re-roll, can bridge the gap.
A caution on judging variance by feel
People often misread randomness. A run of bad rolls feels like evidence of an unfair game even when it falls within normal spread, and a win can be credited to skill when luck explains much of it. Before blaming a design, consider how many games you have actually played. Measuring a game’s real luck level takes a lot of results, and impressions from a handful of sessions are weak evidence.