antist.ai
all scenarios

Drawing set · L0–L3

make a game

Games break the usual rule that working software is enough. A game can run perfectly and still be worthless if it is not fun, so the first job is finding the fun, not building the engine. This drawing set maps the shapes a game project can take. Honest note: this site has not shipped a game of its own yet; its only first-hand evidence is the realtime matchmaking behind the Turing Test, and every engine below is marked as not verified here.

Design philosophy

A game is a promise of a feeling. Code that runs is only the floor; the product is whether a stranger wants one more round.

  1. Find the fun before the engine

    Prove one mechanic is fun with grey boxes before you build menus, art or systems. Most game projects die from building before playing.

  2. Small and finished beats big and forever

    A tiny game people can finish teaches you more — and reaches more players — than a large one that stays in development.

  3. Watch, do not ask

    Players cannot tell you why a game is not fun, but their hands can show you. Trust what they do over what they say.

First game? Use an existing engine, make one small mechanic fun in a weekend, and publish it free. Only think about stores, servers and multiplayer once strangers play it twice.

SheetL0

Jam prototype

Used byYou and a few friends at your screen
Per month$0

①Why move up

You want to know whether one mechanic is fun. Nothing else matters yet — not art, not menus, not a story.

②Architecture

One scene in an existing engine, grey boxes for everything, one mechanic, one way to lose or win. No saving, no settings, no server.

Δ New at this level

  • + One playable mechanic
  • + A win or lose condition

Shapes on the map: Browser game (HTML5 framework) · General-purpose engine (Godot, Unity)

③Tools

Use an existing engine; writing your own at this level is a way of avoiding the scary question of whether the game is fun.

④Code and tests

  • Watch two people play without explaining anything; write down every moment they hesitate.
  • Keep the whole prototype small enough to throw away without regret.
SheetL1

Playable by strangers

Used byStrangers who found your link
Per month$0–10

①Why move up

The mechanic survived your friends. Now you need people who owe you nothing to find it, finish it and tell you where they quit.

②Architecture

The prototype becomes a short complete game: a start, an end, saving if it is longer than a few minutes, settings for sound and controls. It ships as a web build and a downloadable build on a free indie storefront.

Δ New at this level

  • + A beginning and an ending
  • + Save and settings
  • + A web and a desktop build
  • + A public page to collect feedback

Shapes on the map: Browser game (HTML5 framework) · General-purpose engine (Godot, Unity)

③Tools

Publish where players already browse for small games rather than building your own store page. Keep the web build: a link people can click converts far better than a download.

Also worth knowing — not verified by this site

④Code and tests

  • Test the exported build on a weak laptop and a phone browser, not only your development machine.
  • Track where players quit — the level or minute — and fix the first drop-off before adding anything.
  • Make saving robust: a lost save loses a player for good.
SheetL2

Money and data

Used byPlayers who paid
Per month$10–50

①Why move up

People now pay for the game. A bug that corrupts a save is a refund request, the store page is your storefront window, and cloud saves mean you hold player data.

②Architecture

The game moves to a paid store with its own build pipeline, achievements and cloud saves. Releases become versioned patches with notes, and every save format change needs a migration from the old format.

Δ New at this level

  • + A paid store release
  • + Versioned patches with notes
  • + Save format migrations
  • + Cloud saves through the store

Shapes on the map: General-purpose engine (Godot, Unity)

③Tools

Use the store's own services — cloud saves, achievements, payments — instead of running your own servers. The store already handles refunds, taxes and currencies.

Also worth knowing — not verified by this site

④Code and tests

  • Every change to the save format ships with a migration, tested against real old save files.
  • Keep a set of saves from every released version and load all of them before each patch.
  • Build the store page and trailer before launch; a game nobody hears about has not launched.
SheetL3

Multiplayer and scale

Used byMany players, playing together
Per month$50–1000+

①Why move up

Players want to play with each other in real time, and once they compete, some of them will cheat. Latency, fairness and uptime become part of the game design.

②Architecture

An authoritative server owns the game state and clients only send inputs. Matchmaking groups players into rooms, and servers sit close to where players are. Client and server versions must match, so they deploy together.

Δ New at this level

  • + An authoritative game server
  • + Matchmaking into rooms
  • + Version checks between client and server
  • + Server-side cheat detection

Shapes on the map: Game + realtime server · Live service (servers, anti-cheat, live ops)

③Tools

Start with the smallest realtime layer that works — one room per match on an edge platform — before dedicated game servers. The Turing Test on this site runs its matchmaking this way.

④Code and tests

  • Never trust the client: validate every input on the server.
  • Test with artificial latency and packet loss; the office network is not the internet.
  • Refuse to match clients whose version does not match the server's.

The architecture map

Every shape this can take. Most exist so you know what not to build yet.

Browser game (HTML5 framework)

Jam prototypes and small games anyone can play from a link

$0 · Low

△ You need 3D, heavy physics, or a store that sells downloadable games

General-purpose engine (Godot, Unity)

Most indie games; 2D or 3D; export to desktop, web and consoles

$0 to start (check engine license) · Mid

△ Players need to play together in real time

Game + realtime server

Players meet each other live — the Turing Test's matchmaking shape

$5–100/mo · High

△ Cheating, many regions, or live events need a dedicated operations layer

Live service (servers, anti-cheat, live ops)

Large player base, ongoing content, competitive play

$1000+/mo + a team · Extreme

△ This is the last floor. A solo developer reading this should build something small and finished first

Concepts you'll actually need

One plain sentence and one metaphor each. Click any tool module for the same treatment.

Core loop
The few actions a player repeats over and over; if they are not fun, nothing built around them will be.
≈ The rhythm of a song — arrangement cannot save a beat nobody wants to tap along to.
Frame budget
At 60 frames per second, all logic and drawing for one frame must finish in about 16 milliseconds.
≈ A kitchen that must plate a dish every few seconds — any slow step stalls the whole line.
Authoritative server
In multiplayer, the server decides what really happened; players' machines only send inputs, so a modified client cannot cheat.
≈ A referee on the field instead of letting each team keep its own score.
Playtest
Watching someone play without helping them; where they get confused or bored is your to-do list.
≈ Watching a guest try to find the bathroom in your house without giving directions.