The history of physics in games is really the history of a computational budget growing. Every major step in the genre happened when a class of calculation became cheap enough to run in real time, and each one unlocked a family of mechanics that had been impossible the year before.
This is an account of those steps in terms of what became affordable, rather than a list of titles. It explains why the genre looks the way it does and why browser physics games arrived so late and then so quickly.
Stage one: velocity and reversal
The earliest arcade games computed almost nothing. An object had a position and a velocity, and when it reached the edge of something, the relevant component of that velocity was negated.
That is genuinely all of it. No mass, no acceleration, no contact normals — just a sign flip. And yet it was enough to produce games that people found compelling, because a ball that bounces predictably is already a system you can learn.
What it made possible
Angle-based play. Hitting a ball with the edge of a paddle rather than the centre changes its trajectory, and suddenly there is skill. This is the ancestor of every bounce game, including Bounce Lab.
What it could not do
Anything involving mass. Objects could not push each other, stack, topple or transfer momentum, because none of those quantities existed.
Stage two: gravity and the arc
Adding a constant downward acceleration was the next step, and it cost almost nothing — one addition per object per frame.
The payoff was disproportionate. Once objects accelerate downward, trajectories become parabolas, and a parabola is something a player can learn to predict. Artillery games built an entire genre on exactly this, with angle and power as the only inputs.
What it made possible
- Aiming as a skill. Estimating an arc is learnable in a way that reacting to a ball is not.
- Jumping with weight. A jump under gravity has an arc, which makes timing meaningful.
- Wind and environmental variables. Once there is one acceleration, adding a second is free.
Cannon Curve is a direct descendant, and deliberately so — angle, power, wind, and an arc you have to read.
Try it: Cannon Curve
Set an angle and a power, then drop an energy orb onto the targets.
Play Cannon CurveStage three: mass and collision response
The step that created the genre proper was giving objects mass and resolving collisions between them properly, rather than against fixed geometry.
This is considerably more expensive. It requires detecting collisions between arbitrary moving objects, computing contact normals, and applying impulses that respect both objects' masses. For a long time it was simply unaffordable at interactive rates.
What it made possible
- Momentum transfer. One object can now push another, which is the basis of every chain-reaction game.
- Stacking. Objects can rest on each other, which requires persistent contacts and an iterative solver.
- Toppling. With rotation and centre of mass, structures can fall over in ways nobody scripted.
- Emergence. For the first time, the game could produce outcomes its designers had not anticipated.
That last point is what distinguishes the genre. Stack Physics has no level design at all beyond a foundation and a crane — everything interesting about it emerges from rigid bodies resting on each other.
Try it: Stack Physics
Drop crates from a moving crane and build a tower that stays standing.
Play Stack PhysicsStage four: constraints and soft systems
Once rigid bodies were affordable, the next expansion was relationships between them: joints, ropes, springs and cloth.
Constraints are cheaper than they look, because they are solved positionally rather than through forces, and iterative relaxation converges quickly. What they cost is design attention — a constraint system introduces a whole new category of instability.
What it made possible
- Swinging and grappling. A rope that pulls but does not push gives a whole movement vocabulary.
- Vehicles. Suspension is a spring-damper pair, and a wheel is a revolute joint.
- Deformable objects. A grid of masses and springs approximates cloth, rope and jelly.
- Structures. Bridges, cranes and machines built from jointed parts.
Rope Swing and Bridge Builder Physics sit here — though the latter deliberately removes joints entirely, which is a design choice rather than a technical limit.
The browser arrives late and then all at once
Web games were held back for years, and not by any single missing piece. Three things were inadequate simultaneously.
- JavaScript was interpreted. Numeric loops ran orders of magnitude slower than native code.
- There was no drawing surface. Animating by moving DOM elements does not scale past a handful of objects.
- There was no frame callback. Timer-based loops fight the compositor and produce judder.
The workaround was plugins — a separate runtime embedded in the page, with its own renderer and its own performance characteristics. It worked, and it came with security problems, mobile incompatibility and an installation barrier.
What changed
Just-in-time compilation made JavaScript fast enough for physics. The canvas element provided a GPU-backed drawing surface. requestAnimationFrame provided proper frame synchronisation. Pointer events unified mouse and touch. Within a few years the plugin had no remaining advantage, and browser gaming moved to the open platform.
Why physics games benefited most
Most genres need assets — art, audio, level data — and those have to be downloaded. A physics game needs computation, and computation is free once the code has loaded. A complete physics game can be a few tens of kilobytes, which is why the genre thrives on the web.
How the design language changed
Each technical step changed what designers could ask players to do, and the genre's vocabulary grew with it.
| Capability | New design verb | Example here |
|---|---|---|
| Velocity reversal | Angle a bounce | Bounce Lab |
| Constant gravity | Estimate an arc | Cannon Curve |
| Mass and impulse | Transfer momentum | Chain Reaction Lab |
| Rotation and torque | Balance a load | Box Balance |
| Persistent contacts | Build a structure | Stack Physics |
| Constraints | Swing and tether | Rope Swing |
| Directional gravity | Change which way is down | Gravity Switch |
| Force fields | Steer without touching | Force Field |
Reading that table downwards is a reasonable summary of the genre's history: each row became affordable at a different time, and each one produced a family of games.
Where it goes next
Predicting specific games is a mug's game. Predicting which calculations are about to become affordable is more tractable, and a few are visibly close.
- Real-time fluid simulation. Particle-based fluids are already feasible at small scales. Grid-based solvers are not yet viable in a browser, which is why Fluid Drift is explicitly a simplified model.
- Soft-body deformation as a core mechanic. Currently mostly decorative, because deformable collision is expensive.
- Much larger body counts. WebGPU and compute shaders make thousands of interacting bodies plausible in a tab.
- Deterministic multiplayer physics. Hard, because floating-point results differ subtly between machines, but solvable with fixed-point arithmetic.
The pattern holds: each of these will produce a family of games that is currently impossible, and in retrospect it will look obvious that they were next.
Try it: Physics Arcade Ultimate
Six zones, one run: gravity, bounce, balance, motion, collision and force.
Play Physics Arcade UltimateWhat has not changed
For all the technical progress, a few things about physics games have been constant since the very first bouncing ball, and they are probably the reason the genre persists.
The appeal is comprehensibility
A physics game is one of the few kinds of game where you can fully understand the rules. There is no hidden stat, no random damage roll, no behaviour you have to look up. Everything that happens follows from things you can see, and that is unusually satisfying.
Failure is always legible
When you lose, you know why. That has been true since the earliest bounce games and it remains the genre's strongest property. A loss that teaches you something is a fundamentally different experience from one that feels arbitrary.
Emergence beats authored content
A physics system produces more interesting situations than a designer can author, and it does so for free. That economics has not changed: Stack Physics has no level design at all, and every run is different.
The best mechanics are one rule deep
Flip gravity. Draw a surface. Let go of the rope. The most enduring physics mechanics are all single verbs whose consequences are complicated — not complicated verbs. That has been true at every stage of the genre's development and shows no sign of changing.
The technology decides what is possible. The single-rule design decides what is good.
Which is why the games that hold up best from each era are rarely the most technically advanced ones of their time. They are the ones that found one clear rule and followed it honestly.
Key takeaways
- The genre's history is the history of which calculations became affordable in real time.
- Velocity reversal produced bounce games; constant gravity produced arcs and aiming.
- Mass and impulse-based collision produced momentum transfer, stacking and genuine emergence.
- Constraints produced swinging, vehicles, cloth and structures.
- Browser physics waited on fast JavaScript, canvas and requestAnimationFrame arriving together.
- Physics games suit the web particularly well because they need computation rather than assets.
Frequently asked questions
When did physics games start?
In a sense with the earliest arcade games, which reversed a ball's velocity on contact. But the genre as we understand it — objects with mass that push, stack and topple — needed rigid-body simulation to become affordable in real time.
Why did browser physics games take so long to arrive?
Three things were inadequate at once: JavaScript was too slow for numeric loops, there was no efficient drawing surface, and there was no frame callback synchronised to the display. All three were solved within a few years of each other.
What made rigid-body physics possible in games?
Enough computational headroom to detect collisions between arbitrary moving shapes, compute contact normals and apply mass-weighted impulses many times per second — plus iterative solvers that made stacking stable.
Why do physics games suit the web so well?
Because they need computation rather than assets. A complete physics game can be a few tens of kilobytes, whereas most genres need art and audio downloaded before anything can start.
What is emergence in a physics game?
Behaviour the designers did not script. Once objects have mass and collide properly, dropping something at an unusual angle produces outcomes nobody anticipated — which is the defining property of the genre.
What is next for physics games?
Most likely real-time fluids at larger scales, soft-body deformation as a core mechanic rather than decoration, and much larger body counts as GPU compute becomes available in browsers.
Why are old physics games still fun?
Because the appeal was never graphical. A game built on one clear rule with legible consequences does not age the way one built on presentation does, and the earliest bounce games still teach angles as effectively as anything modern.
Did physics games get harder or easier over time?
Neither, really. They got deeper. More capable simulation meant more mechanics rather than steeper difficulty, and the most approachable games in the genre — sandboxes and untimed puzzles — are largely a recent development.
Why do so many physics games use the same handful of mechanics?
Because each one corresponds to a class of calculation that became affordable at a particular moment, and there are not many such classes. Bouncing, arcing, stacking, swinging and field-steering between them cover most of what real-time rigid-body physics can express.
What made plugin-based web games disappear?
The open platform caught up on every axis at once — fast JavaScript, a GPU-backed canvas, proper frame timing and unified pointer input — while plugins retained their security problems, their mobile incompatibility and their installation barrier.
References
Physical concepts in this guide were checked against the following educational sources. None of them are affiliated with Physics Arcade.
