On this page
A score is a piece of feedback
The purpose of a score in an arcade game is not to rank you. It is to tell you, at the end of a run, how that run compared to your last one — and ideally to imply why.
That reframing has a direct consequence: a score that does not distinguish between a clean run and a sloppy one is a bad score, however consistent it is. Plain survival time, for instance, cannot tell the difference between elegant flying and lucky flying, which means it quietly endorses both.
The four metrics
| Metric | Measures | Good for | Blind to |
|---|---|---|---|
| Events passed | Gates, rings, obstacles | Clear, countable progress | How well each was taken |
| Distance | Ground covered | Continuous fields with no events | Everything except survival |
| Points | Weighted events plus bonuses | Expressing design priorities | Nothing, if weighted well |
| Collection | Items gathered | Games about choosing lines | Survival, deliberately |
Choosing between them is a design statement. Gap Hunter records gates passed, because gates passed is the same thing as how narrow you got. Endless Wings records distance because there are no events. Sky Coin Run records coins because the game is about whether you take the greedy line, and recording survival would have rewarded ignoring it.
Make the metric match the design
The expensive mistake is scoring something other than what the game is about, because players optimise toward the score and away from the design.
The clearest example is accuracy. In ordinary flappy scoring, a clean centre pass and a desperate scrape are worth the same, so the game silently endorses the scrape — and a player can genuinely get worse at flying while their scores rise. Wing Master fixes this by scoring the centre band and recording perfect passes rather than points.
Players optimise toward whatever you measure. Measure the wrong thing and they will get better at the wrong thing, then conclude the game was shallow.
Score inflation
Any system with a multiplier eventually faces inflation: a long run multiplied by a large factor produces a number that makes every earlier run meaningless.
Three things keep it in check here:
- A multiplier cap. Ours is ×4, reached at a chain of 31.
- Linear base values. A gate is worth a flat 10 points; nothing compounds except the multiplier.
- Metric choice. Several games record something uncountable-by-accident — gates, perfect passes, coins — which cannot inflate at all.
This matters more than it sounds because best scores are stored permanently in your browser. A score you set during a run where a mechanic was mistuned would sit at the top of your list forever, and there is no server to reset it from.
Score display is part of the system
Where and how a score is shown changes how players relate to it, and it is usually an afterthought.
A score drawn large in the centre of the play area is impossible to ignore, which is right for a game where the score is the point and wrong for one where watching it costs you attention you need elsewhere. A score tucked into a HUD bar above the canvas is glanceable between obstacles and invisible during them.
We use both, deliberately. Most games draw the running score large and central because it is the one number that matters. Games with several metrics — gates, collectibles, combo — put them in the bar instead, because four large numbers in the play area would compete with the obstacles for the same attention.
The rule we follow is that anything drawn inside the play area must be worth looking at during play. Everything else belongs outside it.
Key takeaways
- A score is feedback about how this run compared to your last, not a ranking.
- The four common metrics are events, distance, points and collection, and each is blind to something.
- Players optimise toward whatever you measure, so the metric has to match what the game is about.
- Multiplier caps, flat base values and uncountable metrics are what keep scores from inflating.