| name | encounter-design |
| description | Encounters as the intersection of space, adversaries and resources — across genres, not just action games. Covers the translation table for horde survival, ARPG, RTS, card, CYOA and autonomous sims; space vocabulary, enemy archetypes and composition, AI architecture choice, telegraph timing floors, the four-beat structure for introducing anything new, wave pacing, boss phases, environmental flow and landmarks. Use when laying out a fight, level, arena or world, placing enemies, designing a boss, or when encounters feel repetitive, tactically flat, unreadable or unfair. Reach for this on 'every fight is the same' or 'I died and don't know why'. |
Encounter Design
Purpose: Tools for designing how players experience space and adversaries. An encounter is not a stat block dropped into a room — it is the intersection of spatial geometry, enemy behavior, and player resources that produces a tactical experience. This skill covers the full chain from world structure down to individual enemy placement.
Core insight: The same enemies in a different space feel completely different. The same space with different resources creates different strategies. Encounter design is combinatorial — small changes to any axis reshape the entire experience.
Sources: The four-beat teaching structure is Koichi Hayashida's, developed on the Super Mario Galaxy / 3D World team. Landmark and wayfinding principles derive from Kevin Lynch, The Image of the City (MIT Press, 1960) — legibility, landmarks, paths and nodes are his. Environmental storytelling follows Don Carson, "Environmental Storytelling" (Gamasutra, 2000) and Harvey Smith & Matthias Worch, "What Happened Here?" (GDC 2010). Reaction-time floors derive from standard psychophysics (simple vs. choice reaction time; Hick's law, 1952). Enemy archetype names are common industry vocabulary with no single origin. The Encounter Triangle is this plugin's own framing.
Core Framework: The Encounter Triangle
Every encounter is shaped by three forces. The experience emerges from their intersection, not from any single axis.
Space
(WHERE it happens)
/ \
/ \
/ encounter \
/ experience \
/ \
Adversaries ———————— Resources
(WHO they face) (WHAT they have)
| Axis | What It Controls | Design Levers |
|---|
| Space | Geometry, cover, elevation, sight lines, movement options | Room shape, verticality, choke points, flanking routes, hazards |
| Adversaries | Enemy types, behaviors, group composition, timing | Archetypes, AI complexity, wave structure, spawn placement |
| Resources | Health, ammo, abilities, environmental tools, information | Scarcity, placement, risk/reward of acquisition |
Translate the triangle before using it
Most of this skill's vocabulary — cover, sight lines, flanking routes — comes
from third-person action games and reads as inapplicable everywhere else. It
isn't. Space is whatever constrains reachability. Resources are whatever the
player spends and cannot instantly replace. Translate the three axes into your
game's terms first, one line each, then use the rest of the skill.
| Genre | Space is… | Adversaries are… | Resources are… |
|---|
| Horde survival | Density gradients and escape lanes; the safe pocket that keeps shrinking | Spawn composition and pressure rate | Cooldowns, position, the level-up pause |
| Isometric ARPG | Pack spacing and door-to-door pathing | Pack composition, affix rolls | Potions, cooldowns, gear state |
| RTS | Map control, chokes, expansion distance | Enemy composition and timing pushes | Supply, build time, unit count, player attention |
| Turn-based / card | Board state and reachable positions | The threat clock and the enemy's hand | Actions, cards, health spent as currency |
| CYOA / narrative | Which options exist at this node | The pressure the fiction applies | Time, standing, irreversible commitments |
| Autonomous sim | Adjacency in the world graph | Faction posture and escalation state | Whatever the agents spend — see simulation-first-design |
Test: write your three lines. If the Resources line comes out empty — nothing
is spent, nothing is scarce — you do not have an encounter, you have an event.
That is a legitimate thing to have, and nothing below will improve it.
The combination is the design. Changing one axis changes the experience:
| Same Enemies, Different Space | Same Space, Different Resources |
|---|
| Open arena → skill-based dodging | Full ammo → aggressive push |
| Tight corridor → claustrophobic pressure | Scarce ammo → cautious resource management |
| Elevated platforms → vertical tactics | New ability just unlocked → mastery testing |
| Multiple entrances → flanking anxiety | Environmental tools available → creative solutions |
Diagnostic: When an encounter feels wrong, identify which axis is underdesigned. Most "boring" encounters have interesting enemies in uninteresting space, or interesting space with no resource tension.
Combat Space Composition
Space is not a container for combat — it is a participant. Every spatial element shapes player behavior.
Space Elements
| Element | Behavior It Creates | Design Consideration |
|---|
| Cover | Approach decisions, safety management | Must be destructible or flanakable to prevent camping |
| Sight lines | Long = ranged advantage; short = close-quarters tension | Mix lengths to reward weapon/ability switching |
| Elevation | High ground advantage, verticality adds decision layers | Ensure both high and low positions have trade-offs |
| Choke points | Force commitment, create risk/reward | Powerful but can feel unfair if enemies use them without counter |
| Flanking routes | Reward spatial awareness, punish tunnel vision | Must be readable — players need to know routes exist |
| Environmental hazards | Add urgency, create area denial | Should be readable before they activate; punish carelessness, not ignorance |
| Interactables | Creative problem-solving, environmental mastery | Explosive barrels, bridges, switches — reward observation |
Space Vocabulary
Different room shapes produce fundamentally different encounters:
| Space Type | Geometry | Behavior It Encourages | Best For |
|---|
| Arena | Open, symmetric, few obstacles | Skill-based movement, dodging, spacing | Boss fights, duels, skill tests |
| Corridor | Long and narrow, limited lateral movement | Funneled intensity, resource attrition | Building tension, gauntlets |
| Hub | Central area with multiple exits/entrances | Multi-directional awareness, choice of engagement | Exploration pivots, ambush setups |
| Gauntlet | Linear with sequential challenges | Sustained pressure, endurance testing | Climactic sequences, escape scenes |
| Balcony | Split-level with overlook positions | Ranged/melee asymmetry, vertical tactics | Mixed-range encounters |
| Labyrinth | Interconnected paths, limited visibility | Paranoia, ambush potential, navigation challenge | Horror, stealth, cat-and-mouse |
| Killzone | Inescapable area with escalating threat | Survival pressure, crowd management | Wave defense, holdout scenarios |
Design rule: Vary space types across a level. Three arenas in a row feel repetitive regardless of enemy variation. Alternate space vocabulary to keep encounters feeling fresh.
Enemy Design as System
Enemies are behavior systems, not stat blocks. A slow enemy that blocks a doorway is more interesting than a fast enemy with more HP, because the first one changes what the player does.
Enemy Archetypes
| Archetype | Role | Player Response It Forces |
|---|
| Aggressor | Closes distance, applies pressure | Reactive play, retreat decisions, priority targeting |
| Defender | Holds position, blocks progress | Approach planning, finding weak points, patience |
| Supporter | Buffs or heals other enemies | Target prioritization, "kill the healer" decisions |
| Disruptor | Breaks player patterns, denies abilities | Adaptation under pressure, fallback strategies |
| Artillery | Ranged area denial, long-range threat | Positioning awareness, closing distance under fire |
| Swarm | Overwhelms with individually-weak numbers | Crowd management, area-of-effect decisions |
| Elite/Boss | Multi-phase, tests multiple skills | Full mastery demonstration, pattern recognition |
Composition Rules
Encounters become interesting through archetype combinations, not individual enemies:
- Aggressor + Artillery = the player is forced to close distance (artillery threat) while managing melee pressure (aggressor). Classic push-pull.
- Defender + Supporter = the player must find a way past the defender to reach the supporter, or the encounter becomes attrition.
- Swarm + Disruptor = the player's crowd-clearing tools get disrupted, forcing improvisation.
Minimum viable encounter: Two archetypes that create a tension the player must resolve. A single archetype is a puzzle with one answer; two archetypes create a dilemma.
Enemy Readability
Players must identify enemy type and predict behavior from visual design alone.
Design rule: If the player dies to an enemy they couldn't read, that is a design failure, not a skill failure.
Telegraph windows have a floor
A telegraph the player cannot physically react to is decoration. Human simple
reaction time to a visual stimulus is roughly 200–250 ms; a choice reaction
— see the tell, pick between dodge and block — runs 400 ms and up, and rises
further with the number of options (Hick's law). Add the animation time of the
player's own response before the window closes.
| Response required | Practical telegraph floor |
|---|
| React at all (any input) | ~250 ms |
| Choose between 2–3 responses | ~450 ms |
| Choose and reposition | ~600 ms + traversal time |
These are floors, not targets. An attack telegraphed at exactly the floor is
unblockable in practice, because the player is also tracking other things.
Sourced for the underlying reaction times only; the per-row totals are working
values — measure yours with the Numbers Policy test-plan format.
The exception that proves it: telegraphs below the floor are a legitimate
design choice when the intended counter is prediction rather than reaction —
learn the pattern, act before the tell. Fighting games and soulslikes do this
deliberately. It only works if the pattern is learnable and consistent, and if
the first several encounters are survivable while learning.
Teach a new element in four beats
Koichi Hayashida's four-step structure, developed on the Super Mario Galaxy /
3D World team and widely adopted since, is the most reliable pattern for
introducing anything new — an enemy, a hazard, a mechanic:
- Introduce — the element alone, in a safe context, where failure costs
nothing and the behaviour is fully observable.
- Develop — the same element under mild pressure, or with one variation.
- Twist — the element combined with something already known, or inverted.
- Conclude — a final application that assumes mastery, often at scale.
Test: for each enemy or hazard in the game, find its four beats. If an
element's first appearance is already step 3, players will experience it as
unfair rather than difficult — and will report it as a difficulty problem, which
sends you tuning numbers that were never the cause.
Behavior Design Patterns
Enemy AI architecture determines what behaviors are possible. Choose the simplest architecture that supports the design intent.
| Architecture | Behavior Quality | Design Control | Best For |
|---|
| State Machine | Predictable, readable | High — every transition is explicit | Basic enemies, tutorial enemies, bosses with clear phases |
| Behavior Tree | Flexible, modular | Medium — composable but designable | Mid-complexity enemies, group behaviors, enemies that adapt to context |
| Utility AI | Context-sensitive, emergent | Lower — harder to predict exact behavior | Complex enemies, allies, sandbox/simulation behaviors |
| GOAP | Plans multi-step actions, strategic | Lowest — emergent plans can surprise designers | Strategic enemies, simulation games, enemies that use the environment |
Complexity rule: Start with state machines. Only add architectural complexity when the design requires behaviors that simpler systems cannot express. A state machine with 5 states is easier to tune, debug, and balance than a behavior tree with 30 nodes that produces the same behavior.
Difficulty Through Behavior
Good difficulty scaling changes what enemies do, not just their stats.
| Lazy Difficulty (avoid) | Behavioral Difficulty (prefer) |
|---|
| More HP | More aggressive — shorter hesitation before attacking |
| More damage | Better coordination — enemies flank, cover each other |
| Faster movement | Smarter target selection — prioritize vulnerable player states |
| More enemies (only) | More varied tactics — use cover, retreat when low, call reinforcements |
| Reduced player resources (only) | Faster adaptation — enemies change strategy when their approach fails |
Stat scaling is sometimes necessary, but it should never be the only difficulty lever. When the player's experience of an encounter doesn't change between easy and hard mode, difficulty is being faked.
Encounter Pacing
An encounter is not a single moment — it has internal rhythm.
Wave Structure
How enemies arrive shapes the encounter's emotional arc:
Intensity
| ___
| / \ ___
| / \ / \
| ___/ \_/ \___
| / \___
+--------------------------------→ Time
Intro Escalate Climax Breathe
- Introduction — Low-threat enemies establish the space and rules. Players orient.
- Escalation — New archetypes arrive, combinations emerge, pressure builds.
- Climax — Maximum challenge. The combination that tests everything introduced so far.
- Breathing room — Threat subsides. Resource collection, exploration, narrative beats.
Design rule: Every encounter needs a readable beginning and end. If the player cannot tell when an encounter starts or when it is over, the pacing is broken.
Rest Between Encounters
Rest is not absence of content. Rest is a different type of content:
- Exploration and discovery
- Narrative and environmental storytelling
- Resource management and planning
- Skill/loadout customization
- Anticipation of what comes next
Cognitive recovery is real. After high-intensity encounters, players need time to encode what they learned. Skip this and they feel exhausted without feeling accomplished. The relationship between encounter pacing and session pacing is covered in experience-design.
Boss Design
Bosses are multi-phase encounters that test accumulated mastery:
Environmental Flow
How players move through space without explicit markers. The environment itself should communicate where to go and what matters.
Guidance Tools
| Tool | How It Works | When to Use |
|---|
| Light | Players move toward light, especially in dark environments | Guiding through caves, corridors, ambiguous paths |
| Color contrast | Warm colors attract, cool colors recede; high contrast draws the eye | Highlighting interactables, goals, important paths |
| Geometry | Wide paths invite, narrow paths signal danger or importance | Controlling pace, signaling transitions |
| Motion | Anything moving draws attention over static elements | Drawing the eye to important elements in busy scenes |
| Audio | Sound sources guide exploration; volume implies proximity | Guiding when visuals are obscured or ambiguous |
| Vista/reward | Visible goals in the distance pull players forward | Motivating traversal, communicating progress |
Landmark Theory
Players build mental maps using distinctive landmarks. Effective landmarks:
- Are visually unique within their surroundings
- Are visible from multiple locations and angles
- Have a stable position (don't move or disappear)
- Help players orient without a minimap
- Create a sense of place — "the area near the broken tower," not "zone 3"
Design rule: If players need a minimap to navigate, the environment is not communicating well enough. Minimaps are a crutch for spaces that fail to self-document. Sometimes that crutch is necessary (open worlds, complex interiors), but the goal is environments that are navigable on their own.
World Structure Patterns
How the overall game world is organized affects encounter pacing, player agency, and narrative delivery.
| Pattern | Structure | Pacing Implication | Best For |
|---|
| Linear | A to B to C, one path forward | Tight authorial control, curated experience | Story-driven games, cinematic set pieces |
| Hub-and-spoke | Central area with branching paths | Player-directed pacing, return-to-safety rhythm | Adventure games, RPGs with a home base |
| Metroidvania | Interconnected, gated by abilities | Mastery-gated exploration, revisitation rewards | Progression-focused, ability-unlock games |
| Open world | Free traversal, distributed content | Player-paced, risk of aimlessness and content drought | Sandbox, exploration-driven experiences |
| Roguelike/procedural | Generated layouts per run | High variety, low per-room handcraft | Replayability-focused, emergent narrative |
Hybrid Structures
Most shipped games combine patterns:
- Linear corridors connecting hub areas (action-adventure)
- Open world with linear mission interiors (open-world action)
- Hub-and-spoke with Metroidvania gating (Zelda-style progression)
Design rule: Choose the structure that serves your encounter pacing. If encounters need careful tuning, linear gives you control. If encounters need variety, procedural gives you breadth. The structure is not the game — it is the delivery mechanism for encounters.
Environmental Storytelling
Space tells story without dialogue. The environment is a narrative medium.
Techniques
| Technique | How It Works | Example |
|---|
| Found objects | Items in context tell a story | A campsite with scattered belongings suggests hasty departure |
| Environmental details | Wear, damage, decoration reveal history | Claw marks on walls, faded posters, overgrown paths |
| Spatial narrative | The journey through a space IS the story | Ascending a tower shows civilization giving way to decay |
| Contrast | Juxtaposition creates meaning | A child's toy in a battlefield; a garden in a fortress |
| Absence | What's missing tells a story | Empty shelves, missing portraits, abandoned defenses |
Integration Principles
- Show, don't tell. The environment should make the player curious. If a narrator has to explain what happened, the environment failed.
- Reinforce mechanics. Environmental storytelling should support, not contradict, what the game systems are saying. If the story says "this place is dangerous" but enemies are trivial, the fiction undermines itself.
- Reward observation. Players who look carefully should find more story, more resources, or more tactical information than players who rush through.
- Layer, don't gate. Environmental narrative should enrich the experience for those who notice it without blocking those who don't.
For deeper narrative structure (quest design, branching dialogue, story arcs), see narrative-design.
Encounter Health Check
Evaluate every encounter before shipping:
Anti-Patterns
| Anti-Pattern | Why It Fails | Fix Direction |
|---|
| Stat-sponge enemies | HP as difficulty doesn't change player behavior — same fight, longer | Difficulty through behavior changes, not stat scaling |
| Empty arenas | Open space with no tactical features reduces combat to stat checks | Add cover, elevation, hazards, interactables |
| Spawn closets | Enemies appearing from nowhere breaks spatial reasoning | Enemies should be visible or their arrival should be telegraphed |
| Kill box repetition | Same rectangular room every fight creates encounter fatigue | Vary space vocabulary — alternate arenas, corridors, hubs, balconies |
| Behavior-less enemies | Enemies that just walk toward the player are mobile health bars | Give every enemy at least one behavior that changes what the player does |
| Unfair ambushes | Player punished for information they cannot have | Ambushes must be survivable on first encounter; telegraph before lethal |
| Difficulty cliff | Sudden spike with no ramp or new tools to match | Introduce the challenge element in a safe context before the hard version |
| Single-solution encounters | Only one valid approach kills replayability and expression | Ensure space and enemy composition support at least two viable strategies |
Cross-References
Three tight links. Everything else routes through the map, so adding a skill
touches one file rather than twenty: references/routing-map.md (in
game-design).
- systems-design — Encounters are where systems meet the player
- game-feel — Moment-to-moment feel inside the encounter
- playtest-design — Whether it reads the way you think it does