Game Loops

Reference Notes

GD 2000 — Game Design

2026-11-04

Roadmap

The Loop

1. The core loop: action → feedback → reward → motivation
2. The engine loop (the code sense)
3. Loops nest; tight vs loose

Mechanics That Modify the Loop

4. Decisions & agency
5. Chance & information
6. Skill & mastery
7. Resources, risk & economy
8. Progression & escalation
9. Feedback loops & balance
10. Time, juice, people & story

Application

11. The seven games side by side
12. Reading, balancing, and designing a loop

A condensed handout version of the lecture. One slide per mechanic family.

The Core Loop

Action → Feedback → Reward → Motivation → repeat

Stage	Meaning	Fails when…
Action	the core verb the player performs	the verb is unclear
Feedback	what the game shows / sounds / feels back	it is slow or illegible
Reward	what the player gains (power, progress, information, story)	it feels unearned
Motivation	why they go again (curiosity, mastery, goal)	there is no reason

The decision is what turns a loop into a game. Ask of every pass: what did the player decide?

Everything else is a mechanic wrapped around these four stages.

The Engine Loop

The loop in the code — a different sense of “game loop”:

while (running) {
  poll_input();
  update(dt);
  render();
}

	Design loop	Engine loop
Who runs it	the player	the computer
Timescale	seconds → hours	milliseconds
Failure	boredom, confusion	lag, jank, bugs

The engine loop is the substrate; the design loop is the point. Feedback later than ~100 ms feels disconnected. Non-digital games have no engine loop at all — only the design loop.

The name “game loop” refers to both; keep them distinct. Everything else is the design loop.

Loops Nest; Tight vs Loose

Loop	Timescale	Digital example	Tabletop example
Micro	seconds	aim → shoot → the enemy reacts	move a piece → see the reply
Meso	minutes	clear a room → reach a checkpoint	win a hand of poker
Macro	hours	level up → finish the campaign	win the tournament

• Tight loops — short, immediate (Tetris, Jenga, a poker hand).
• Loose loops — long, deferred (chess, Monopoly, a D&D campaign).

Each loop feeds the next; designers tune the handoff. “One more turn” and “one more hand” are loose loops that feel tight.

Flat-feeling games usually have only one timescale designed.

Decisions & Agency

	Chess	Snakes and Ladders
Decisions per loop	every move	none
Skill	deep	none
Chance	none	pure

A loop with no decisions repeats but isn’t played. Design test: for each pass, what did the player decide?

Snakes and Ladders is the baseline control; Uno is the same simplicity plus choices.

Chance & Information

• Input randomness — dealt before you decide; you plan around it (Tetris, Hold’em hole cards, Uno draw).
• Output randomness — settles the outcome after you decide (Snakes and Ladders). This is where “that’s unfair!” comes from.
• Perfect information (Chess) — the loop is calculation.
• Hidden information (Hold’em) — the loop is inference and deception; bluffing and reading live between players.

Push randomness toward the input side to give players control.

Skill & Mastery

• Execution — precise, under pressure (Tetris, TF2).
• Knowledge / calculation — the right move (Chess).
• Judgment under uncertainty — (Hold’em, TF2 positioning).

A loop is only as interesting as the skill it tests. Low floor, high ceiling: easy to start (Tetris, Uno), hard to master (Chess, Hold’em, TF2). Mastery keeps the macro loop alive.

If a machine or spreadsheet can solve it, the loop is over.

Resources, Risk & Economy

• Economy — a currency with sources and sinks (chips, material, cards, metal, time). No sink → meaningless; no source → frustrating.
• Risk — the loop can cost you (Hold’em bets, chess sacrifices, TF2 pushes). Push-your-luck: reward and danger scale together.

Without risk, decisions are just optimizations.

The loop is the pump that moves value between sources and sinks.

Progression & Escalation

• Progression — a second loop that changes you (BG3 levels, TF2 unlocks; chess rating = player progression). It gives the loop a memory.
• Escalation — raise the stakes, not the rules (Tetris speed, rising poker blinds, TF2 sudden-death).
• Session vs career loops: minutes–hours vs days–years. Align them.

Difficulty is the shape of the escalation curve.

Feedback Loops & Balance

• Positive feedback — rewards the leader, so advantage compounds (Chess material, Hold’em big stack, Monopoly rent, TF2 snowball). Risk: runaway leader.
• Negative feedback — helps the loser (Uno Draw Two, Mario Kart items, rising blinds). Risk: the lead stops mattering.
• Balance — no dominant strategy, or the decision disappears and you’re back to Snakes and Ladders.

Balance is the mixture of the two — momentum plus hope.

Time, Juice, People & Story

• Time — turn-based (Chess, Hold’em, Uno) vs real-time (Tetris, TF2).
• Juice — feedback bigger than it needs to be; makes the loop legible and satisfying (~100 ms rule).
• People — a social loop where players are each other’s feedback; often the strongest retention mechanic.
• Story — reward as meaning; the slowest and often stickiest loop (Baldur’s Gate 3).

Narrative turns the loop’s reward into story rather than numbers.

The Seven Games Side by Side

Game	Core verb	Key mechanics	Loop length
Chess	move	perfect information, skill, no chance	long
Snakes & Ladders	roll	pure chance, no decisions	short
Tetris	place	real-time, execution, escalation	tight
Texas Hold’em	bet	hidden information, risk, resources, social	medium
Uno	play a card	chance, take-that, catch-up	short
TF2	shoot	real-time, team, roles, economy	medium
Baldur’s Gate 3	choose	narrative, progression, exploration	long

Same four stages in every row. What differs is which mechanics surround them.

Seven very different games, one shared skeleton.

Reading, Balancing & Designing

Read — five steps: the verb; the feedback and reward; the motivation; the mechanics (which stage each touches); the nested loops.

Balance — keep every pass worth playing: a goal in view; skill matched to challenge; positive and negative feedback; a real choice each pass; escalation that tightens.

Common failures: no decisions (Snakes and Ladders trap); dominant strategy; runaway leader; illegible feedback; stale loop.

Design — five questions: the core verb; what they get; why they return; which mechanics modify the loop; where it could feel stale, unfair, or slow.

Every failure is a mechanic problem — cheap to fix early, expensive to fix late.