Rules versus systems

A rule-based challenge is something you learn: the spike is at this position, the gate opens on this beat, this enemy charges when you cross its line. Learning it is genuine skill, and once learned it stays learned.

A system-based challenge is something you understand, and understanding transfers. Once you know how momentum works in Rolling Gummy Quest, you can solve hills you have never seen. Once you know how thrust works in Candy Space Run, an unfamiliar asteroid field is a new instance of a problem you can already solve rather than a new problem.

Both are valid and a good library has both. What physics buys you specifically is generalisation: one understood system produces an unlimited number of solvable situations, which is why physics games tend to stay interesting longer than pattern games of comparable complexity.

Gravity: the baseline

Every platformer on this site shares one gravity value and one jump velocity, held in a single module. That is a deliberate constraint rather than a convenience: it means a jump arc learned in Candy Jump is correct in Candy Jump Challenge, Candy Sky Run, Candy Castle Hunt and Marshmallow Hopper.

Fixed gravity with a fixed jump also makes the arc computable, which is what lets collectible placement trace it exactly. The whole teaching mechanism described in our platformer guide depends on gravity being a constant rather than a per-game tuning knob.

The actual numbers

Gravity is 1,750 units per second squared and jump velocity is -720 units per second, in a play area 600 units tall. That produces a jump roughly 148 units high and 210 units long at full walk speed - about a quarter of the screen height and a third of its width on a typical landscape display.

Advertising keeps Candy Arcade free. Ads never cover gameplay or controls.

Momentum: earning speed before you need it

Rolling Gummy Quest replaces a jump with accumulated momentum. You apply torque, speed builds over time, and terrain either adds to it on a downhill or takes it away on a climb. There is no jump button at all: you leave the ground by crossing a crest with enough speed.

This produces a specific and unusual kind of planning. The decision that determines whether you clear a hill is made on the downhill before it, often several seconds earlier. Players who think in terms of position - am I going fast enough right now - stall constantly. Players who think in terms of momentum budget arrive at the bottom of each climb with what they need.

The failure state follows from the physics rather than being bolted on: if your speed reaches zero on a slope you roll backward, and rolling off the screen ends the run. Nothing about that is an arbitrary rule. It is what would happen.

Friction: one parameter, two games

Friction determines how long momentum persists after you stop applying force, and changing it alone is enough to make a different game. Gummy Run and Ice Cream Dash share lanes, obstacles, controls, difficulty thresholds and escalation rates. The only meaningful difference is that Ice Cream Dash overshoots lane changes and settles back over about a third of a second.

Every consequence flows from that one number. Inputs must be made earlier. Overcorrection becomes the dominant failure. A new technique appears - cancelling overshoot with a brief opposite tap - that does not exist in the higher-friction game. Speed power-ups become dangerous rather than useful, because more speed with the same grip is close to a straight downgrade.

This pair is the clearest argument we can make for tuning before adding. Before introducing a mechanic to make a game harder, check whether an existing parameter would do more. It usually would, and the result is more coherent than a bolted-on addition.

Thrust: acceleration instead of position

Thrust games remove drag entirely. You apply upward force against constant gravity, and because nothing damps your velocity, momentum persists much longer than players expect. Holding thrust does not move you up at a speed; it accelerates you upward continuously.

The skill this demands is thinking in acceleration rather than position, which is genuinely difficult and is why both our thrust games are rated Hard. New players hold thrust until they are where they want to be, by which point they are travelling upward fast and will overshoot badly. Experienced players tap in short bursts and let momentum carry them, which looks lazy and is correct.

Two details make it learnable rather than merely punishing. Candy Space Run draws a trajectory ghost predicting a full second ahead, which is longer than most reaction times - so the information you need arrives before the decision. Rocket Candy Run gives you three boost charges as an explicit recovery mechanism, which converts a physics problem into a budgeting problem you can reason about.

Buoyancy: gravity with the sign flipped

Underwater Candy Quest inverts gravity and adds heavy drag. Doing nothing makes you drift slowly upward; swimming downward requires continuous effort; and momentum dies quickly in every direction.

The feel is completely different from the thrust games despite being the same integration with different constants. High drag means direction changes are easy but slow, so the game rewards deliberate planning rather than quick correction - which suits a game about exploring a reef with a limited air supply. Currents then reuse the drag: because your own momentum dies quickly, a current can carry you a long way, which makes riding one a free method of travel rather than a hazard.

It is worth noticing that buoyancy, thrust and platformer gravity are the same three lines of code with different numbers. Almost all of the perceived difference between these games is parameter choice, not implementation complexity.

Ballistics: the fully solvable challenge

Candy Cannon Challenge uses genuine projectile physics - gravity, restitution on bounce, angle and power - and limits you to a small number of shots. Because the physics are deterministic, every level has a correct answer, which makes it the most puzzle-like game on the site.

The interesting design decision there is the trajectory preview, which deliberately stops before the apex. Showing the full arc would reduce every level to executing a line the game had already drawn. Showing nothing would make long shots guesswork. Showing the first portion gives you enough to judge angle and requires you to estimate range, which is exactly the skill the game is about.

Bounce restitution then turns aiming into planning. A ball that loses about forty percent of its energy per bounce can be banked off a wall to reach collectibles that no direct shot can touch, and most levels are built so that a single well-planned bank clears several sweets at once. The chain bonus for collecting three or more with one projectile makes that explicit.

When not to use physics

Physics is not automatically better. It costs predictability, and predictability is sometimes exactly what a game needs.

  • A timing game should not have physics. One-Tap Candy Collector measures a tap against a fixed window; adding momentum would make the same input produce different results.
  • A grid game should not have physics. Cell-based movement in a maze is precise because it is discrete; adding sliding would make corridors frustrating.
  • A deterministic pattern game should not have physics. Candy Factory's whole premise is that every section is exactly repeatable, and variable momentum would undermine that.
  • Any game where the challenge is reading a pattern is usually worse with physics, because physics adds a second source of failure and makes the first harder to diagnose.

The test is whether the physics creates the interesting decision or merely sits between the player and it. In Rolling Gummy Quest the momentum is the game. In a maze it would just be friction in the other sense.

Common mistakes

  • Holding thrust until you reach the targetWith no drag you arrive travelling fast and overshoot. Short bursts and coasting is the correct technique.
  • Thinking in position rather than momentumIn Rolling Gummy Quest the decision that clears a hill is made on the downhill before it, not at its base.
  • Overcorrecting in low-friction gamesThe correction costs more time than the original error did. Small inputs, made early, beat large ones made late.
  • Expecting a direct shot to work in a ballistics puzzleMost levels are built around a bank shot. Planning bounces is the intended solution, not a workaround.
  • Adding physics to a timing gameIt makes the same input produce different results, which destroys the precision the game is measuring.

Practical tips

  • In thrust games, tap rather than hold, and use the trajectory ghost - it predicts further ahead than you can react.
  • In momentum games, build speed on the downhill before the climb that needs it.
  • In low-friction games, cancel overshoot with a brief opposite input rather than fighting it.
  • In buoyancy games, releasing all inputs is a free way to gain height while you read the environment.
  • In ballistics puzzles, look for the bank shot that collects several items at once.

Key takeaways

  • Rule-based challenges are learned; system-based challenges are understood and transfer.
  • All our platformers share one gravity value, which is what makes jump arcs teachable.
  • Friction alone is enough to turn one game into a meaningfully different one.
  • Thrust demands thinking in acceleration rather than position, which is why it is rated Hard.
  • Buoyancy, thrust and platformer gravity are the same integration with different constants.
  • Physics is wrong for timing games, grid games and deterministic pattern games.

Frequently Asked Questions

What makes a physics-based challenge different?

You solve it by understanding a system rather than memorising a layout, and that understanding transfers to situations you have not seen before.

Why are thrust games harder than platformers?

Thrust games have no drag, so momentum persists much longer than players expect and you must think in acceleration rather than position. Both our thrust games are rated Hard for this reason.

Do all the platformers on Candy Arcade use the same physics?

Yes. Gravity, jump velocity, walk speed and coyote time come from one shared module, so an arc learned in one platformer is correct in all of them.

Why does the cannon trajectory preview stop early?

Showing the full arc would reduce each level to tracing a line the game had drawn. Showing the first portion lets you judge angle while still requiring you to estimate range.

Is physics always better than fixed rules?

No. Timing games, grid-based maze games and deterministic pattern games are all worse with physics, because it adds a second source of failure and makes the first harder to diagnose.