Abstract Strategy Agent Skill

Abstract Strategy Game Design

Purpose

Design abstract strategy games—games with perfect information, no randomness, and strategic depth. Provides frameworks for ideation, design, and evaluation.

Core Definition

Abstract strategy games require:

• Perfect Information: All game state visible to all players
• No Randomness: Outcomes determined solely by player decisions
• Minimal Theme: Mechanics over narrative
• Player Agency: Success depends on strategic thinking

Quick Reference: Game Types

| Type | Core Mechanic | Examples |

|------|---------------|----------|

| Connection | Form paths/networks | Hex, TwixT |

| Territory | Control areas | Go, Othello |

| Capture | Eliminate pieces | Chess, Checkers |

| Pattern | Create arrangements | Gomoku, Pentago |

| Racing | Reach goal first | Chinese Checkers |

Design Principles

The Holy Grail: Depth-to-Complexity Ratio

Maximum strategic depth with minimum rules complexity.

How to achieve:

• Start with single strong core mechanism
• Remove anything that doesn't support the core
• Every rule should create multiple strategic implications
• Prefer emergent complexity over explicit rules

Meaningful Decision Architecture

Four components of meaningful choice:

• Awareness: Players understand options
• Consequence: Immediate and long-term effects
• Permanence: Decisions have lasting impact
• Reminders: Game state reflects past choices

Ideal Parameters:

• Branching factor: 20-40 moves/turn for human play
• Horizon: 3-5 moves ahead with effort
• Multiple paths: 3-4 viable strategies minimum

Core Mechanisms Toolkit

Board Topology

• Grids: Square, hexagonal, triangular, irregular
• Connectivity: How spaces relate
• Edges: How boundaries affect strategy
• Size: Larger = exponentially more complex

Piece Systems

• Uniform: All pieces identical (Go)
• Differentiated: Unique abilities (Chess)
• Transforming: Change during play (Checkers kings)
• Ownership: Fixed vs. capturable

Movement & Placement

• Placement only: Pieces don't move once placed (Go)
• Movement only: Pieces start on board (Chess)
• Hybrid: Both placement and movement (Hive)

Victory Conditions

• Elimination, Position, Pattern, Territory, Points, Stalemate

Balance Considerations

First-Player Advantage Mitigation

• Pie Rule: Second player can swap after first move
• Komi: Point compensation for second player
• Variable Setup: Randomized starting positions
• Simultaneous: Both move at once

Avoiding Degenerate Strategies

• No single dominant path
• Counter-strategies exist for every strong position
• Passive play punishable
• Aggressive play doesn't guarantee victory

Design Process

Three Starting Points

1. Mechanism-First

• Identify interesting core mechanic
• Build minimal game around it
• Add only what enhances core
• Remove everything else

2. Experience-First

• Define target player experience
• Identify mechanisms that create it
• Prototype and test rapidly
• Iterate on feedback

3. Constraint-Based

• Set specific limitations (components, time, space)
• Find creative solutions within constraints
• Often leads to elegant designs

When to Add/Remove Complexity

Add when:

• Core feels solved too quickly
• Players master in <10 plays
• Decisions feel obvious

Remove when:

• Rules take >10 minutes
• Players forget rules
• Strategies feel arbitrary

Scrap when:

• No tweaking fixes fundamentals
• Core mechanism isn't interesting
• Feels like inferior version of existing game

Brainstorming Techniques

1. Mechanism Extraction from Non-Games

Extract from physics, biology, economics, chemistry, social systems:

• Pieces that "decay" unless refreshed (entropy)
• Moves creating "waves" along patterns (physics)
• Pieces forming "bonds" limiting movement (chemistry)
• "Market" squares with fluctuating values (economics)

2. Extreme Property Isolation

Take one property to absolute extreme:

• Game where pieces visible only when adjacent to your others
• Every move must maintain rotational symmetry
• Pieces exist only one turn unless refreshed
• Board wraps in non-intuitive ways (Klein bottle)

3. Impossible Constraint Challenges

Start with seemingly impossible constraints:

• Game on a 1D line
• Pieces in probability clouds until observed
• Victory condition voted on by piece positions
• Pieces leave "trails" becoming new pieces

4. Anti-Pattern Starting Points

Design intentionally bad games, then invert:

• Always-draw game → Add accumulating positional advantages
• Pure calculation → Add pieces that change rules
• Dominant strategy → Make it vulnerable to specific counters

5. Mathematical Structure Mining

• Pieces move along Hamiltonian paths only
• Positions valued by prime factorization
• Fractal boards with repeating patterns
• Moves must preserve mathematical invariants

Evaluation Framework

Strategic Richness Indicators

Depth:

• Games last 20+ meaningful turns
• Opening, midgame, endgame feel distinct
• Multiple viable opening strategies
• Comebacks possible but not trivial

Complexity:

• New players grasp rules in <5 minutes
• Experts keep discovering patterns
• High-level play looks different from beginner

Common Failures

| Problem | Symptoms | Solution |

|---------|----------|----------|

| Analysis Paralysis | Excessive turn time | Limit options, clearer objectives |

| Solved Game | Same outcome always | Increase branching, add variety |

| Kingmaker | Loser picks winner | Simultaneous resolution |

Testing Protocol

Phase 1: Proof of Concept

• Test core mechanic in isolation
• Verify basic fun factor
• Identify broken strategies

Phase 2: Mechanics

• Test each subsystem
• Look for unintended interactions
• Measure game length

Phase 3: Integration

• Full game, all systems
• Different skill levels
• Quantitative data

Phase 4: Blind Testing

• Players learn from rulebook only
• Identify ambiguities
• Test learning curve

Testing Checklist

Mechanical

• [ ] All rule interactions verified
• [ ] Edge cases resolved
• [ ] Victory achievable but not trivial
• [ ] No unbreakable stalemates

Balance

• [ ] First player wins 45-55%
• [ ] Multiple strategies win regularly
• [ ] No dominant opening
• [ ] Skill affects outcome

Experience

• [ ] Games complete in target time
• [ ] Players want rematch
• [ ] Decisions feel meaningful
• [ ] Players improve with practice

Accessibility

• [ ] Rules learned in <5 minutes
• [ ] Rules fit one page
• [ ] No ambiguous situations
• [ ] Components distinguishable

Quick Evaluation Filters

30-Second Test: Can you explain core concept in 30 seconds?

Originality Test: Does it feel like variant of existing game?

Decision Test: Are there obviously interesting decisions?

Depth Test: Could this sustain interest for 50+ plays?

Session Structure (2 Hours)

• 10 min: Pick 3-4 brainstorming techniques
• 60 min: Generate 15-20 ideas per technique
• 20 min: Expand 5-10 promising ideas
• 20 min: Combine and explore hybrids
• 10 min: Apply filters, select for prototyping

Anti-Patterns

1. Complexity as Depth

Pattern: Adding rules, exceptions, and special cases to make the game feel "deeper."

Why it fails: Complexity and depth are different. Complex rules create burden; depth emerges from simple rules with rich interactions. Chess has simpler rules than many shallow games.

Fix: Ruthlessly remove complexity that doesn't add strategic options. If a rule requires explanation but doesn't create interesting decisions, cut it.

2. Solved Game Blindness

Pattern: Creating a game where optimal play always produces the same outcome—often draws or first-player wins.

Why it fails: Once players discover the solution, the game becomes rote execution rather than strategic exploration. No amount of polish fixes a solved game.

Fix: Test extensively with strong players. If games start converging on identical patterns, add asymmetry or increase branching factor. The pie rule helps but doesn't solve fundamental issues.

3. Decision Paralysis

Pattern: Every position has dozens of equally viable options with unclear consequences.

Why it fails: Strategic games need meaningful comparison between choices. When all options seem equivalent, decisions become random rather than strategic.

Fix: Reduce branching factor or create clearer evaluation heuristics. Players should be able to identify 3-5 promising moves without analyzing every possibility.

4. Theme Creep

Pattern: Adding narrative or thematic elements that don't connect to mechanical decisions.

Why it fails: Abstract strategy games work because mechanics are the content. Theme that doesn't inform decisions is decoration that slows play without adding depth.

Fix: Either commit to a themed game (different framework) or keep theme purely cosmetic. Don't let theme suggest mechanics that don't serve strategy.

5. Perfect Information Violations

Pattern: Adding hidden information, simultaneous resolution, or dice "for variety."

Why it fails: Abstract strategy games are defined by perfect information and determinism. Adding randomness or hidden elements creates a different game type with different design principles.

Fix: If the game needs variety, add it through board setup, victory condition selection, or piece starting positions—not through mid-game randomness.

Integration

Inbound (feeds into this skill)

| Skill | What it provides |

|-------|------------------|

| brainstorming | Ideation techniques for mechanism discovery |

| research | Historical game analysis and mathematical structure research |

Outbound (this skill enables)

| Skill | What this provides |

|-------|-------------|

| (playtesting) | Designs ready for player validation |

| (rulebook writing) | Tested mechanics ready for documentation |

Complementary

| Skill | Relationship |

|-------|--------------|

| brainstorming | Use brainstorming for raw idea generation; abstract-strategy provides evaluation and refinement frameworks |