World 15

Making it

From grey boxes to gold: how a game actually gets built, tested, cut and shipped.

Most games that fail do not fail for lack of ideas. They run out of time and money, they find out too late that the fun is missing, or they ship the bug that everybody meets. These words are how teams talk about that risk: what to build first, what to test when, what to cut and when to stop adding. Knowing them lets you ask for the cheap version of a thing before you pay for the expensive one.

GrayboxA playable version of a level or game made only of plain untextured shapes. The hero of this site, Noun, begins as one: a square with the physics values the code calls GRAYBOX.
PrototypeA quick, throwaway version built to answer one question, such as "is this fun?" or "can the engine do this?". You expect to bin it.
Find the funPrototyping until you discover what is actually enjoyable in an idea, and then building towards that rather than towards the original plan.
Game jamAn event where people make a whole small game in a fixed short time, often 48 hours, usually to a theme. Ludum Dare has run since 2002 and the Global Game Jam since 2009.
Game design documentThe written description of what a game is and how it works: its rules, systems, content and goals. On a good team it is a living reference that gets changed, not a contract.
One-pagerA single page that says what a game or a system is: what the player does, who it is for and why it is exciting. If it won't fit on one page, you probably haven't decided yet.
Vertical sliceA short part of the game finished to final quality in every discipline at once: art, sound, code, design and interface. It is one thin slice through the whole cake.
Pre-productionThe phase in which a small team works out what the game is and how to make it, with prototypes, tools and plans, before a large team is hired to build it.
ProductionThe main phase of making the game, when the design is settled and the full team builds the levels, art, code and sound. It is mostly a matter of keeping work flowing.
MilestoneAn agreed date when a defined piece of work must be delivered and checked. With a publisher, passing one is commonly how the next part of the money gets paid.
AlphaThe point when every feature is in the game and can be played, however buggy and however ugly. After it, the rule is that no new features go in. Each studio draws this line slightly differently.
BetaThe stage when all features and content are in and the work is finishing and fixing. A public beta opens this to outside players, usually to test servers and balance.
Gold masterThe final approved build, the one that goes to the disc plant or the storefront. "Going gold" is named after the gold master discs that used to be sent for pressing.
CrunchLong stretches of forced overtime to hit a date, sometimes for months and often unpaid. It is common in the games industry, it damages people, and it is usually the result of a planning failure.
ScopeHow much game you are trying to make: how many features, levels, characters and hours. It is the one project variable you fully control, and beginners nearly always set it too high.
Scope creepThe project slowly growing during development, through a long series of small additions that each seemed reasonable, without the deadline or the budget moving to match.
Feature creepThe game collecting more and more features, because each sounded good, until they crowd each other out and the original idea is hard to find.
Kill your darlingsCutting something you love because the game is better or more finishable without it. The advice comes from writing: Arthur Quiller-Couch told students to "murder your darlings" in a 1914 lecture.
PlaytestWatching other people play your game to see what really happens. You say nothing and take notes, and you do not explain or help.
Think-aloudA playtest in which you ask players to say what they are thinking as they play. It comes from psychology and from usability research, where Clayton Lewis applied it at IBM in the early 1980s.
TelemetryData the game records and sends back about what players do: where they quit, how long a level takes, which weapons they use. It should be disclosed and it should not identify people needlessly.
HeatmapA map of a level coloured by how often something happened at each spot: where players died, walked, stopped or looked. It is telemetry you can see at a glance.
Bug triageDeciding which bugs to fix first, and which never to fix, using how bad each one is (severity) and how often players meet it (frequency).
Technical debtThe future cost of a quick fix or a messy shortcut in the code. Like a loan, it buys you speed now and charges interest on all later work until it is paid. Ward Cunningham coined the comparison in 1992.
QAThe work of finding what is broken before players do, by testing every system, edge and combination on purpose and writing bugs up so they can be reproduced. It is skilled work and often underpaid.
LocalisationAdapting a game for other languages and places: translation, but also longer text, right-to-left reading, fonts, voice acting, date formats and culture. Internationalisation is the engineering that makes it possible.
CertificationThe check a console maker runs before a game can be released on its machine. Sony calls its list TRC, Microsoft XR and Nintendo Lotcheck. They test things like save handling, button names and what happens when a controller is unplugged.
Day-one patchAn update that is ready when a game launches, holding the fixes made in the weeks after the final build went off to be manufactured or certified.
Post-mortemA look back after a project, or a failure, at what went right, what went wrong and what to change next time. The aim is to fix how you work, and it is pointless if it becomes a search for someone to blame.
Art bibleA shared document that sets how the game looks and how to get that look: palette, shapes, materials, proportions and line, with examples of what is right and wrong.