Time-to-first-input < 5s. No splash chain, no settings, no mode select.
Time-to-core-verb < 30s. The player performs the signature verb (jump, shoot, plant) for real, with consequences, in the first half-minute.
One input to start — "press any key / tap" over a live game scene. Show the actual game behind the title (attract mode), not a static logo.
First input produces a juicy result: sound + motion + particles. The brain needs proof inputs matter before it will learn the pattern.
Goal in one sentence or zero: "Survive." "Reach the flag."
Detect input device: show touch hints to touch users, key hints to keyboard users — never a scheme the device can't use. If the core input is exotic (webcam, gamepad, mic), hint it explicitly and ship a plain mouse/keyboard fallback that's announced, not silent. Gimmick inputs (webcam/face/pose/hand, 3D presentation) are the product's identity — never drop them in a rebuild; standard inputs are fallbacks alongside. Hand tracking is undiscoverable on touch — hint it.
Browser permission prompts are modals. Camera/mic/motion requests are the most common cold-open modal in browser games — make them opt-in behind a button, or defer until after first play. Never fire getUserMedia on mount, and don't fetch the model before consent: the MediaPipe bundle pulled pre-consent from CDN is the biggest per-game load lever (6.1 MB → 0.15 MB and 2.5 MB → 0.14 MB on two games once deferred).

Teaching without a tutorial (George Fan's PvZ rules)

Doing > reading. Never explain in text what a situation can force the player to discover once. Gate progress behind doing it once — done = taught.
≤ 8 words on screen at a time. If an instruction needs more, redesign the situation, not the sentence.
Safe first exposure (Half-Life 2 pattern): observe the hazard eat a barrel before dodging it yourself.
Ramp: teach → repeat with one twist → combine. Never two new ideas at once; roughly one new element per level.
Objects self-describe: appearance telegraphs function; piggyback on real-world knowledge (red = damage, coins = money, arrows = direction).
Hints are adaptive, not up-front: "Try planting further left!" only after the relevant failure. No modal popups that pause play.
Bias the first decision: make the correct opening move the cheap, shiny, obvious one (Fan made sunflowers cheap and sparkly).
Never label anything "Tutorial".
Control hints live on the start screen only — never persistent chrome during play (allowed exceptions in game-ui).

Difficulty curve

Start below the floor: the first challenge is nearly unfailable — it teaches the verb and pays out a win.
Sawtooth, not ramp: spikes then recovery valleys; mastery registers in the valleys. (Shape details in level-design.)
Re-test old skills while teaching new ones — that combination is where the fun lives (Koster).
Embed difficulty choice in play (Chen's flOw): risk/reward verbs — dive deeper, take the shortcut, bank or gamble — beat an Easy/Normal/Hard menu. Optional hard routes + collectibles let novice and expert share one build.
Hidden mercy after repeated failure (Crash Bandicoot pattern): after 3–5 consecutive deaths at one spot, silently soften — extra checkpoint, slower hazard. Never announce it.

Failure & retry

"Death" means any failure event — a conceded point, a wiped wave, a failed puzzle attempt all follow the same rules.

Death → retry: one input, < 2 seconds. No confirmation dialogs, no unskippable death animations, no score screens between the player and the retry (Celeste/Super Meat Boy convention).
Failure must be legible (Juul): the player answers "what killed me?" instantly, and the counter-move is visible in the death itself. Fair failure motivates; opaque failure is a broken promise.
Lose little while learning: early deaths cost seconds. Punishment can scale after investment.
Keep something across failure: best-score deltas, unlock progress — a failed run becomes a deposit ("one more run").
Celebrate near-misses: "Best: 412 — you got 398" reframes failure as approach.

Assist mode (Celeste model)

For any skill-gated game, ship granular, judgment-free assists: game speed 50–100%, extra HP/invulnerability, skip-section. Available from the start, reversible mid-game, named neutrally ("Assist Mode", never "Easy/Cheat Mode"), no "intended experience" framing, no score asterisks unless competitive integrity truly demands it (Thorson). Baseline accessibility is onboarding too: readable text, never color alone as a signal, remappable or multiple input schemes, pause anywhere.

Session shape

First full loop ≤ 60–90s — a complete win-or-lose cycle with a score in the first minute.
Result screen = 1 glance, 1 button: score, best, delta, play-again.
End every session with an open loop: next unlock, almost-beaten score, teased content — adjacent to the retry button.

Audit checklist (every "no" is a defect)

Meaningful input within 5s of load?
Core verb performed in real play within 30s?
Every instruction ≤ 8 words; first-minute total < ~25 words?
Mute all text — still discoverable through doing?
Each mechanic introduced alone, failure cheap or impossible?
First failure shows what killed you + hints the counter?
Death-to-retry one input, < 2s?
Session ends with a visible reason to replay (delta, unlock, tease)?
Every object's function readable from its appearance?
Difficulty has recovery valleys after spikes?
Zero modal popups/cutscenes/settings before first play?
Native input works immediately (touch on mobile, keys on desktop)?
Skill-gated game has a neutral, mid-game-reachable assist option?
Something softens after 3–5 failures at the same spot?
Score/goal visible at all times?

Related skills: design-lenses (finds the problems this skill fixes), level-design (teaching through geometry), game-feel (the juicy first input), game-playbook (build order).

onboarding

Onboarding & difficulty

The first 30 seconds