[Beginner] · The shape of the field, the real-time constraint that drives every technical decision, and an honest account of the industry.
Game development is software engineering under a hard real-time constraint, in service of a subjective goal. Both halves are unusual, and between them they explain nearly every way game code differs from the rest of this vault.
The 16-millisecond budget
At 60 frames per second, you have 16.67 ms to do everything: read input, update the world, run physics, run AI, cull, submit draw calls, and present. At 144 Hz it’s 6.9 ms. In VR you must hold 90 Hz on two eyes or people feel ill.
Miss the budget and the user sees it immediately. A web request that takes 200 ms instead of 100 ms is invisible. A frame that takes 20 ms instead of 16 is a visible stutter, and a stutter every second is a broken game.
This single constraint drives the differences:
| Most software in this vault | Games | |
|---|---|---|
| Latency target | p99, in ms or s | Every frame, in µs |
| Worst case | Rare, tolerable | Worst case IS the experience |
| Memory | Allocate freely, GC handles it | Allocation in the loop is a bug |
| Correctness | Must be right | Must be plausible, at speed |
| Architecture driver | Maintainability | Cache locality |
“Must be plausible rather than right” is the one that surprises people. A physics engine is not solving Newtonian mechanics; it’s producing something that looks convincing within the budget. Interpenetrating objects are pushed apart by an approximation nobody would defend as physics. Being visibly wrong is a bug; being actually wrong is normal → game physics.
And “worst case is the experience” reframes optimisation. In web work you optimise the average and watch the tail. In games the tail is the product — a 99th-percentile frame is one bad frame every 1.6 seconds at 60 Hz.
What’s actually in a game
The disciplines, and roughly how a team divides:
Engineering — gameplay, engine, graphics, physics, AI, networking, tools, audio programming. Tools programmers are the least visible and among the most valuable: everyone else’s productivity runs through them.
Art — concept, 3D modelling, texturing, rigging, animation, VFX, technical art (the people who sit between art and engineering, and are perpetually scarce).
Design — systems design, level design, narrative, balancing, UX.
Audio — composition, sound design, implementation.
Production — scheduling, scope, shipping.
Most of a game is not code. On a large title, engineers are a minority, and much of their work is building the tools and systems that let designers and artists work without them. That’s the structural difference from most software teams, and it’s why “content pipeline” is a first-class engineering concern → tools and production.
The two ways in
Use an engine. Unity, Unreal or Godot give you rendering, physics, audio, input, asset pipeline, editor and platform ports. You write gameplay. This is the correct default, and the reason is scope: the engine represents thousands of person-years you get for free → engines.
Write your own. You learn enormously and ship far less. Legitimate as education — and it’s the honest framing: a custom engine is a study project unless you have a specific technical need no engine meets.
The failure mode to name early: building an engine while believing you’re building a game. They are different projects with different outcomes, and conflating them is the most common way an ambitious first game dies.
The maths you actually need
Less than people fear for gameplay, more than they hope for graphics and physics.
Essential, immediately — vectors (add, dot, cross, normalise), and enough trigonometry to point things at other things. Dot product for angles and projection; cross product for perpendiculars and winding. If you understand the dot product, you can write gameplay code.
Essential for 3D — matrices and transforms, and quaternions for rotation. Not the theory; the operations and what they mean → transforms and rigid body transforms, which this vault already has because robotics needed exactly the same maths.
For physics — Newtonian mechanics, integration, and why Euler integration explodes → ODEs.
For graphics — linear algebra properly, plus the rendering equation → computer graphics.
You don’t need most of it to start. A 2D game needs vectors and nothing else.
The industry, honestly
Since you asked out of genuine interest, this is the part worth knowing before you invest years.
Pay is lower than equivalent software roles, often substantially — because supply exceeds demand. Many people want to make games; comparatively few companies need them. The same skills in fintech or infrastructure pay more.
“Crunch” is real, though improving under unionisation pressure and public scrutiny. Studio culture varies enormously; it is worth researching a specific studio rather than the industry.
Employment is volatile. Layoffs follow project completion as a structural pattern, not an anomaly, and 2023–25 saw very large ones across the sector.
Most games don’t make money. The distribution is brutally long-tailed — a small number of titles earn nearly everything.
The counterweight, which is also real: it is genuinely creatively satisfying in a way most software isn’t, the technical work is among the hardest and most interesting in the industry, and the skills transfer outward well — real-time rendering, performance work and simulation are wanted in visualisation, simulation, film, defence, automotive HMI, medical imaging and increasingly robotics.
The pragmatic route, and the one this vault would recommend given the one-active-course rule: keep game development as a side discipline while earning from general software. It’s a legitimate hobby, an outstanding way to learn graphics and performance, and it does not require you to accept the industry’s economics to benefit from its skills.
Related
- engines and the game loop — the technical starting point
- getting started — what to actually build first
- computer graphics — the rendering half, already here
- programming fundamentals — if this is where you’re starting from
Source: [reference] — cross-referenced against the roadmap.sh game developer roadmap.
Software engineering
Infrastructure & security
Computer science
Hardware & engineering