70 articles on how candy arcade games work, organised into ten content clusters. Every article is written and tested by the person who built the games, and every claim about our own mechanics is verified against the running code rather than the design notes.
20 pillar articles 50 supporting guides 10 clusters
If you are new to the genre, read what candy arcade games are. If you want to score higher, read best strategies for candy collection games and then how arcade combo systems work. If you are building a game of your own, the design and HTML5 clusters are where the technical material is.
What the genre is, how its games are structured, and the vocabulary the rest of the library uses. Play the games in this cluster.
Candy arcade games are easy to recognise and surprisingly hard to define. This guide sets out what actually makes a game belong to the genre, which mechanics do the work, and why the candy theme is more than decoration.
Both descend from the same 1980s cabinets, and they have drifted apart in specific, describable ways. This compares them on the things that actually differ rather than on how they look.
Picking up a virtual sweet achieves nothing real, and it still feels good. The reasons are structural and describable, and they have direct consequences for how a collection game should be built.
The single reference for everything on this site: what the genre is, how every system works, what each category asks of you, and where to go next. Written to be read once end to end or dipped into.
If you have just found this site and want to stop dying in the first thirty seconds, this is the shortest useful path. Ten specific things, in the order they will help most.
Having built thirty of these, the difference between a good one and a forgettable one is consistent and has almost nothing to do with how it looks.
A collectible is about six numbers and a drawing function. This explains what each of those numbers does and why the drawing matters more than you would expect.
Most game sites categorise by theme, which tells you nothing about whether you will enjoy something. Ours categorise by demand, which does.
How collection mechanics work and how to make better decisions about what to gather. Play the games in this cluster.
Collection looks like the simplest mechanic in games until you try to build one that stays interesting for more than a minute. This guide takes the mechanic apart and explains what each piece is doing.
Most players improve at collection games by getting faster. That works for a while and then stops. This guide covers what to do after it stops, which is mostly about deciding what not to collect.
Scattered collectibles communicate nothing. Arranged ones are a language with about a dozen words in it. This is the vocabulary and what each word means.
Most players try to collect faster by moving faster. The players who actually collect faster move less, and this explains how.
Improving scores is a process rather than a talent. This is the process, in the order that produces results fastest.
Collection pulls you toward things and hazards push you away from them, usually in the same place. This is how to manage both without abandoning either.
A line of collectibles is the clearest directional instruction available, costs nothing, and is remarkably easy to get wrong. This is how to place one properly.
A balanced challenge is one where the right answer is not obvious and is nevertheless knowable. This is how to tune toward that.
Multipliers, combo tiers, break rules and the arithmetic that decides which play scores more. Play the games in this cluster.
A combo system is the clearest statement a game makes about what it wants you to do. This guide explains how they are built, what each design choice does to play, and why the rule that breaks a combo matters more than the one that builds it.
Two multipliers, a base value and a bonus can combine in several different ways, and the order matters. This guide works the numbers so you can stop guessing which play is worth more.
Every game here has a combo, and they all work slightly differently. This is how to read an unfamiliar one in about thirty seconds.
There are two kinds of multiplier in these games and they combine in a way that makes patience unusually profitable. This is the short version.
Building a combo is easy; keeping one is where the score is. These are the specific habits that protect a chain, in order of how much difference they make.
A broken combo takes nothing away from your score. It still feels like a disaster, and understanding why makes it easier to play well afterwards.
What each power-up actually does, when it helps, and when it is a trap. Play the games in this cluster.
A good power-up changes your decisions, not just your numbers. This goes through all six standard power-ups, what each one actually does to play, and why two of them are frequently a mistake to collect.
The magnet is the most useful power-up in the genre and the easiest to waste. This covers what it actually does, what it cannot do, and when to spend it.
Most players use multipliers the moment they collect them, which wastes roughly three quarters of their value. This is the timing that does not.
Speed is the only power-up on this site that is routinely correct to refuse. The rule for deciding is simple and most players have it backwards.
Everyone understands a shield as an extra life. In several games here it is worth considerably more than that, and in a few it covers less than you think.
Jump arcs, platform design, precision movement and why difficulty lives in level layout. Play the games in this cluster.
Candy in a platformer does two jobs at once: it rewards you and it tells you where to jump. This explains how the second job works, and why most players never notice it happening.
Platformer collectibles do a job beyond scoring: they tell you where to jump. This is how that works and how to read it.
If a jump feels impossible in one of these games, your launch point is wrong rather than your timing. This explains why, and what to do instead.
One of the most useful and least visible techniques in platformer design. It costs nothing, nobody notices it, and its absence makes a game feel broken.
One of the simplest level elements available and one of the most transformative. It changes not how hard a section is but what kind of attention it demands.
Distance-based difficulty, lane design, friction and how endless games stay challenging. Play the games in this cluster.
An endless game has no level design to lean on, so its whole difficulty curve is a set of dials being turned while you play. This explains which dials exist and which ones should be avoided.
Endless games have no finish line, so improvement is entirely about surviving longer at the same skill. These are the habits that do that.
The hard problem in endless design is not the first run, it is the hundredth. This is what keeps a game interesting once the player is good at it.
Three lanes is the most common structure in the genre for good reasons. This explains what lanes do to the design and how to play them well.
Two games on this site are identical apart from one number. The differences that follow are the clearest argument we can make for tuning before adding.
Beginners read one obstacle ahead and react. Experienced players read three and plan. This is how to make that transition.
Procedural generation, route planning and the design of fair timed puzzles. Play the games in this cluster.
A maze game is the only kind of arcade game where you can win by thinking before you move. This explains how the mazes are built, why a perfect maze is a worse game than an imperfect one, and how routing actually works.
The gap between a good route and a bad one is usually twenty seconds. These are the specific techniques that close it.
Candy Maze has no hazards at all. Its entire difficulty is a clock, which makes it an unusually clean case study in how to make time pressure fair.
Timing windows, one-input design, reaction time and precision scoring. Play the games in this cluster.
Raw reaction time barely improves with practice, and almost nobody who gets better at these games got faster. This explains what they did instead, and how to do the same.
A single-input game removes every variable except timing, which means the whole skill is accuracy. These are the techniques that improve it.
Every timing game is a window and a measurement. Understanding how both work explains why some timing games feel fair and others feel arbitrary.
One button that reverses which way is down. The rule takes a sentence to explain and the technique takes considerably longer to master.
You cannot get much faster. You can get considerably more accurate, and these are the specific methods that do that.
How these games are built: level design, balance, difficulty curves and playtesting. Play the games in this cluster.
A physics-based challenge is one you solve by understanding a system rather than memorising a rule. This goes through the five physics models used across the site and what each one demands of the player.
A fair level is one where you can name the reason you failed and do something about it. That is compatible with being extremely hard, and this explains how to get both.
Procedural generation is what lets an endless game stay endless without being repetitive. It is also the easiest way to ship an impossible level, and this is how to avoid that.
There is a finite list of ways to make an arcade game harder. Knowing the whole list, and which entries are traps, is most of difficulty design.
Getting the value hierarchy right means players never have to be told where the danger is. This is how to build one that does that work.
The instinct is to randomise so players cannot memorise a level. Memorisation is not the enemy, and preventing it replaces skill with attrition.
Two small asymmetries that nobody notices and that remove an entire category of perceived unfairness. This is how and why.
A hazard is fair when the information you need arrives before the decision you have to make. That is a constraint on timing, and it is stricter than most games realise.
Feedback is what makes a mechanically identical action feel good or flat. It is also the easiest thing to overdo, and overdoing it costs legibility.
Every collection game has one: the item worth a lot, placed somewhere awful. Getting it right is most of what makes a collection game interesting.
Good arcade chase behaviour is simple, stated, and exploitable. Sophisticated pursuit makes a worse game than three obvious rules.
Thrust is the hardest common control scheme in arcade games because it asks you to think in acceleration. This explains why, and what makes it learnable.
A momentum game makes the decision that matters happen long before the moment that matters. That displacement is the whole lesson.
Canvas rendering, touch input, local storage, performance and accessibility on the web platform. Play the games in this cluster.
Your high score lives in your own browser, on one device, and that has consequences most sites never explain. Here is exactly how it works, what breaks it, and what it can never do.
Most browser games are built on a desktop and then made to fit a phone, which is why so many are unpleasant to play on one. These are the differences that actually matter.
A canvas is a black box to a screen reader, so accessibility in a canvas game cannot be retrofitted with attributes. It has to be designed, and this is what that involves.
A browser arcade game is four things: a canvas, a loop, an input layer and some storage. This walks through all four and what each one is responsible for.
A busy collection game draws fifty or sixty objects a frame, sixty times a second. This is how that stays fast on a mid-range phone.
Touch input on the web works well once you have turned off about seven browser behaviours. This is the list, and the control layouts worth using.
Performance work on mobile has a short list of things that matter enormously and a long list that does not. This is the short list, in order.
There are no audio files on Candy Arcade. Every sound is generated in the browser, which turns out to solve three separate problems at once.
Getting a canvas to look right on every device is two separate problems: pixel density and aspect ratio. Scaling solves one of them.
The browser is a runtime everyone already has. That changes the economics of a small game more than any technical feature does.
Emulators catch layout problems and miss the ones that matter. This is what we actually check, and where.
Advertising keeps Candy Arcade free. Ads never cover gameplay or controls.