Cloth of Gold: Competitive Cellular Automata as a Strategy Game

February 27, 2025

A game-theoretic analysis of a modified Conway's Game of Life where two players compete for survival and territory through strategic pattern placement and cellular evolution.

Cloth of Gold: Competitive Cellular Automata as a Strategy Game

Abstract

Cloth of Gold presents a two-player strategy game built on a competitive modification of Conway’s Game of Life. By introducing player ownership and majority-rule dynamics to cellular evolution, the game transforms a zero-player mathematical curiosity into a contest of strategic depth. This framework proves that the competitive rules fundamentally alter the mathematical properties of the cellular automaton, create complex game-theoretic equilibria, and give rise to emergent territorial dynamics. The game features simultaneous hidden placement, economic feedback through pattern recognition, and multiple victory conditions, creating a rich strategic space at the intersection of discrete dynamical systems and competitive game theory.

Interactive Demonstration

Explore competitive cellular automaton dynamics below. Click cells to place Player A (blue) or Player B (red) populations, then watch them evolve according to competitive Game of Life rules. The colored background shows discretized territorial control via metaball influence fields.

Speed:10 gen/sec

Generation

Player A

Territory: 0

Player B

Territory: 0

Click cells to cycle: Empty → Player A (blue) → Player B (red)

Territory colors (background):

Player A territoryPlayer B territoryContestedCompetitive death

1. Introduction

1.1 From Conway to Competition

Conway’s Game of Life, introduced in 1970, demonstrates how complex behaviors emerge from simple rules. It operates as a “zero-player game”—once initial conditions are set, evolution proceeds deterministically. The competitive variant transforms this into a two-player system by introducing:

Cell ownership: Each live cell belongs to one of two players
Majority dynamics: Cell fate depends on the relative strength of neighboring players
Strategic placement: Players add new cells each round
Economic incentives: Specific patterns generate resources

The name “Cloth of Gold” references both the Conus textile (cloth of gold cone snail) whose shell patterns resemble cellular automata, and the historical Field of the Cloth of Gold—a diplomatic summit that became a competitive display of wealth and power.

1.2 Game Overview

The game proceeds in rounds, each representing one week (168 generations of cellular evolution)

Placement Phase: Players simultaneously place up to 10 units (plus bonus units from patterns)
Evolution Phase: The cellular automaton runs for 168 generations
Scoring Phase: Pattern recognition awards bonus placement units for the next round

Victory can be achieved through:

Extinction: Eliminating the opponent’s population
Territory: Controlling 75% of the board for sustained periods
Economy: Accumulating sufficient resources over time
Population: Having the larger population after 52 rounds (one year)

1.3 Theoretical Contributions

This work provides:

Formal analysis of competitive cellular automaton dynamics
Proof that competitive rules break fundamental properties of Conway’s Game of Life
Game-theoretic characterization of strategic equilibria
Complexity analysis of optimal play
Framework for understanding territorial emergence in discrete systems

2. Mathematical Framework

2.1 State Space Definition

Definition 2.1 (Game State)

The state space consists of:

Cellular configuration: $W \in \{0, a, b\}^{m \times n}$ $W \in {0, a, b}^{m \times n}$ where:
- $0$ = empty cell
- $a$ = Player A’s cell
- $b$ = Player B’s cell
Territory map: $T: \mathbb{R}^2 \to [0,1]^2$ via metaball influence fields
Resource state: $R \in \mathbb{N}^2$ tracking accumulated placement units
Time: $t \in \mathbb{N}$ marking the current generation

2.2 Neighbor Counting Functions

For cell position (i,j), define:

Player A neighbors:

$N_A[w_{i,j}] = \sum_{k=i-1}^{i+1} \sum_{l=j-1}^{j+1} \mathbb{1}[w_{k,l} = a], \quad (k,l) \neq (i,j)$

Player B neighbors:

$N_B[w_{i,j}] = \sum_{k=i-1}^{i+1} \sum_{l=j-1}^{j+1} \mathbb{1}[w_{k,l} = b], \quad (k,l) \neq (i,j)$

Total neighbors:

$N[w_{i,j}] = N_A[w_{i,j}] + N_B[w_{i,j}]$

2.3 Evolution Rules

The competitive cellular automaton evolves according to:

Rule 1 (Death by isolation)

$w_{i,j}(t+1) = 0 \text{ if } N[w_{i,j}] < 2$

Rule 2 (Death by overcrowding)

$w_{i,j}(t+1) = 0 \text{ if } N[w_{i,j}] > 3$

Cells that die by overcrowding while outnumbered by the opposing team (i.e., when $w_{i,j} = a$ and $N_B > N_A$ , or $w_{i,j} = b$ and $N_A > N_B$ ) are marked as competitive deaths and briefly displayed as purple indicators.

Rule 3 (Competitive ownership)

For cells with $N[w_{i,j}] \in \{2, 3\}$ (the survival range):

w_{i,j}(t+1) = \begin{cases} a & \text{if } (w_{i,j} \neq 0 \text{ or } N[w_{i,j}] = 3) \text{ and } N_A[w_{i,j}] > N_B[w_{i,j}] \\ b & \text{if } (w_{i,j} \neq 0 \text{ or } N[w_{i,j}] = 3) \text{ and } N_B[w_{i,j}] > N_A[w_{i,j}] \\ w_{i,j}(t) & \text{if } N_A[w_{i,j}] = N_B[w_{i,j}] \end{cases}

This rule unifies birth and conversion:

Living cells convert ownership based on neighbor majority
Empty cells with exactly 3 neighbors birth as the majority team
Tied situations preserve current state (living cells stay their color, empty cells stay empty)

The competitive ownership rule enables dynamic territorial conquest - patterns can be infiltrated and converted rather than merely destroyed. This creates fluid frontlines where cells constantly flip allegiance based on local competitive pressure.

2.4 Territory Definition

Territory is determined by discretized metaball influence fields:

Definition 2.2 (Influence Field)

For player $P$ with cells at positions $\{x_k\}$ , the influence at cell $c$ is:

$I_P(c) = \sum_k f(\|c - x_k\|)$

where $f(r) = \max(0, 1 - r^2/R^2)$ for influence radius $R = 10$ cells.

Definition 2.3 (Discretized Territory)

Each cell has a discrete territorial state:

\text{Territory}(c) = \begin{cases} \text{A's territory} & \text{if } I_A(c) > 1.0 \text{ AND } I_A(c) > I_B(c) \\ \text{B's territory} & \text{if } I_B(c) > 1.0 \text{ AND } I_B(c) > I_A(c) \\ \text{Contested} & \text{if } I_A(c) > 1.0 \text{ AND } I_B(c) > 1.0 \text{ AND } |I_A(c) - I_B(c)| < 1.5 \\ \text{Neutral} & \text{if } \max(I_A(c), I_B(c)) < 1.0 \end{cases}

This discretization creates territorial quantization - territory expands in discrete cell-by-cell jumps rather than continuous gradients.

3. Theoretical Properties

3.1 Breaking Conservation Laws

Theorem 3.1 (Non-conservation)

Unlike Conway’s Game of Life, the competitive variant does not conserve any local quantity.

Proof :

In Conway’s Game of Life, patterns can be designed to conserve various quantities locally (e.g., glider guns maintain population through cyclic behavior).

In Cloth of Gold, ownership conversions irreversibly transform cellular identities. Consider a Player A cell surrounded by Player B neighbors:

Initial:  bbb     After 1 gen:  bbb
         bab                   bbb
         bbb                   bbb

The center ‘a’ cell has 2-3 neighbors (survival range) but $N_B > N_A$ , so it converts to ‘b’. This conversion is irreversible - there’s no local operation that can determine the cell was originally ‘a’. The competitive rules inject new information each generation, breaking time-reversibility and preventing conservation laws.

■

3.2 Territorial Pressure Dynamics

Definition 3.1 (Pressure)

The competitive pressure at position $(i,j)$ is:

$P_{i,j} = N_A[w_{i,j}] - N_B[w_{i,j}]$

Theorem 3.2 (Pressure-driven flow)

Cells flow down the pressure gradient.

Proof :

From the evolution rules (for cells with 2-3 neighbors):

If $P_{i,j} > 0$ : Cell becomes/remains type a
If $P_{i,j} < 0$ : Cell becomes/remains type b
If $P_{i,j} = 0$ : Cell maintains current state

This creates a flow from high to low pressure regions, analogous to fluid dynamics. The boundary between territories acts as an interface with surface tension proportional to the pressure differential. At tied boundaries ( $P = 0$ ), cells stabilize and preserve their current allegiance, creating natural demarcation lines.

■

3.3 Pattern Stability Under Competition

Definition 3.2 (Competitive Stability)

A pattern $P$ is competitively stable if it maintains structure when surrounded by opponent cells.

Theorem 3.3 (Stability Criteria)

A pattern $P$ is competitively stable iff:

All cells have 2-3 total neighbors (survival condition)
No cell has more enemy than friendly neighbors (no conversions)

Proof :

Necessity: If any condition fails:

Cells with $\neq 2$ - $3$ total neighbors die by isolation or overcrowding
Cells with enemy majority convert to enemy ownership, breaking pattern structure

Sufficiency: If all conditions hold:

All cells survive (2-3 neighbors)
All cells maintain ownership ( $N_{\text{friendly}} \geq N_{\text{enemy}}$ )
Pattern structure is preserved

■

Corollary

Most Conway stable patterns (blocks, beehives) are NOT competitively stable.

3.4 Chimeric Patterns and Emergent Cooperation

Definition 3.3 (Chimeric Pattern)

A stable structure containing cells from both players where $N_A = N_B$ at each position.

Theorem 3.4 (Chimeric Stability)

Mixed-ownership patterns can achieve stability impossible for single-player patterns.

Proof :

Chimeric patterns achieve stability when each cell has equal friendly and enemy neighbors. For example, consider this pattern:

aaabbb
abbbaa

At the boundary cells, each has neighbors from both teams creating local equilibrium. When $N_A = N_B$ , the tie-preservation rule maintains current state, preventing conversion while meeting survival conditions.

In contrast, a pure pattern at a contested boundary faces pressure $P \neq 0$ , causing erosion or conversion. Chimeric patterns achieve $P = 0$ at key positions, creating stable demarcation lines.

Strategic Implications:

Demilitarized zones: Stable boundaries neither player can unilaterally breach
Peace treaties: Mutual investment in shared structures
Prisoner’s dilemma: Cooperate to maintain or defect to destroy

This emergence of cooperation from competitive rules represents a fascinating phase transition from pure competition to strategic symbiosis.

4. Game-Theoretic Analysis

■

4.1 Strategy Space

Definition 4.1 (Strategy)

A player’s strategy consists of:

Placement function: $\sigma: (W, T, R) \to \text{Distribution over placements}$
Pattern library: $\Pi \subseteq \{\text{all constructible patterns}\}$
Temporal plan: $\tau \in \{\text{Rush}, \text{Boom}, \text{Turtle}, \text{Adaptive}\}$

The strategy space has cardinality:

$|\Sigma| = |\{\text{placement distributions}\}| \times 2^{|\text{patterns}|} \times |\text{temporal plans}|$

For an $m \times n$ board, this exceeds $10^{mn}$ possible strategies.

4.2 Boundary Conditions

Definition 4.2 (Boundary Treatment)

Cells outside the $m \times n$ board are treated as permanently empty (state $0$ ) for all calculations.

This creates edge effects:

Patterns near boundaries have fewer neighbors
Boundary cells are easier to kill (less support)
Corner positions are particularly vulnerable
Gliders reaching boundaries are destroyed

Strategic implications: Board edges act as natural barriers, creating “coastal” dynamics where patterns must be adapted for survival with fewer neighbors.

4.3 Nash Equilibrium Analysis

Theorem 4.1 (No Pure Strategy Equilibrium)

For non-trivial boards ( $m, n \geq 10$ ), no pure strategy Nash equilibrium exists.

Proof :

Suppose pure strategy $s^_$ is a Nash equilibrium. Then $s^*$ must be a best response to itself.

Consider the placement phase. If player A plays deterministic placement $s^*$ , player B can:

Identify A’s placement pattern
Design a counter-placement targeting A’s structures
Achieve advantage through targeted disruption

The chaotic evolution (Lyapunov exponent $\lambda > 0$ ) means small placement changes cascade unpredictably, preventing stable best responses for any pure strategy.

Therefore, no pure strategy can be a best response to itself.

■

Theorem 4.2 (Mixed Strategy Equilibrium)

A mixed strategy equilibrium exists.

Proof :

The game satisfies conditions for Nash’s existence theorem:

Finite action space per round (finite board positions)
Compact strategy space (probability distributions)
Continuous payoff functions (population counts)

By Nash’s theorem, a mixed strategy equilibrium exists.

■

4.4 Information Theory of Hidden Placement

Lemma 4.1 (Information Deficit)

With simultaneous hidden placement, each player faces uncertainty:

$H(\text{opponent:placement} \mid \text{my:placement}) = \log_2\binom{\text{available:cells}}{10}$

For a $100 \times 100$ board with $50\%$ available cells:

$H \approx \log_2\binom{5000}{10} \approx 33 \text{ bits}$

This information deficit forces probabilistic reasoning and prevents perfect play.

5. Economic System Analysis

5.1 Resource Generation

Definition 5.1 (Pattern Values)

Beehive + 2 blocks = 1 bonus unit
Honey farm + 8 blocks = 8 bonus units
Boat patterns = water survival probability

The resource function:

$R(t+1) = R(t) + 10 + \sum_{p \in \text{patterns}(t)} \text{value}(p)$

5.2 Economic Growth Models

Theorem 5.1 (Compound Growth)

Successful economic strategies achieve super-linear resource growth.

Proof :

Let $n(t)$ be the number of honey farms at round $t$ .

Best case (no disruption)

$R(t) = 10t + 8 \cdot \sum_{i=1}^t n(i)$

If $n(t)$ grows linearly: $n(t) = \alpha t$

$R(t) = 10t + 8\alpha \cdot t(t+1)/2 = O(t^2)$

Quadratic growth dominates linear strategies asymptotically.

■

5.3 Optimal Pattern Packing

Problem: Maximize resource generation in area $A$ .

Theorem 5.2 (Packing Density)

Honey farms achieve $2 \times$ better resource density than distributed beehives.

Proof :

Beehive $+ 2$ blocks requires $\approx 25$ cells, generates $1$ unit:

$\text{Density}_{\text{distributed}} = A/25 \text{ units/round}$

Honey farm $+ 8$ blocks requires $\approx 100$ cells, generates $8$ units:

$\text{Density}_{\text{honey}} = 8A/100 = 0.08A \text{ units/round}$

Ratio: $0.08A / (A/25) = 2.0$

Honey farms are twice as space-efficient when defended.

6. Dynamical Systems Analysis

■

6.1 Lyapunov Exponents

Definition 6.1 (Lyapunov Exponent)

Measures sensitivity to initial conditions:

$\lambda = \lim_{t \to \infty} \frac{1}{t} \log\left(\frac{|\delta W(t)|}{|\delta W(0)|}\right)$

Theorem 6.1 (Increased Chaos)

Competitive rules increase the Lyapunov exponent.

Analysis:

Conway’s Game of Life: $\lambda \approx 0.2$ - $0.5$
Competition adds stochastic elements through placement
Territorial pressure creates additional instabilities
Combined: $\lambda_{CG} \approx 0.3$ - $0.8$

Higher $\lambda$ implies:

Reduced prediction horizon
Greater strategic uncertainty
Increased importance of adaptability

6.2 Phase Transitions

Definition 6.2 (Density)

$\rho = (|A| + |B|)/(mn)$

Theorem 6.2 (Phase Structure)

The system exhibits distinct phases:

\begin{aligned} \rho \in [0, 0.1) &: \text{ Sparse - patterns evolve independently} \\ \rho \in [0.1, 0.3) &: \text{ Percolation - local interactions begin} \\ \rho \in [0.3, 0.5) &: \text{ Competition - territorial warfare} \\ \rho \in [0.5, 1.0] &: \text{ Overcrowded - mutual destruction} \end{aligned}

Critical point: $\rho_c \approx 0.35$ where largest sustainable populations exist.

6.3 Territorial Crystallization

Theorem 6.3 (Discrete Voronoi Emergence)

Discretized territories naturally approximate Voronoi cells.

Proof sketch: The evolution rules create pressure away from boundaries (conflict zones have higher death rates). The discretized influence field creates a step-function approximation of continuous Voronoi cells. Over time, territorial boundaries minimize interface length at the cell level, converging toward discrete Voronoi-like patterns which minimize perimeter for given areas. The quantization creates “jagged” boundaries that approximate smooth Voronoi boundaries at larger scales.

7. Computational Complexity

7.1 Decision Complexity

Theorem 7.1 (PSPACE-completeness)

Determining optimal placement is PSPACE-complete.

Proof :

Membership in PSPACE: Simulating $t$ generations requires $O(mn)$ space.

PSPACE-hardness: Reduction from Generalized Geography:

Encode graph vertices as board regions
Encode edges as possible cellular transitions
Winning Geography strategy $\leftrightarrow$ winning Cloth of Gold strategy

Since Generalized Geography is PSPACE-complete, so is optimal Cloth of Gold play.

■

7.2 Pattern Recognition Complexity

Lemma 7.1

: Identifying all valuable patterns requires:

Naive: $O(mn \cdot |pattern:library|)$
With hashing: $O(mn)$ expected time
Via convolution: $O(mn \cdot \log(mn))$ using FFT

Optimization via Quadtrees: The implementation uses quadtree spatial indexing:

Time complexity: $O(k log n)$ where $k =$ active cells
Space complexity: $O(\min(k log n, mn))$

Quadtrees provide significant speedup for:

Sparse boards (early game): $O(k) \ll O(mn)$
Localized conflicts: Only affected quadrants need updating
Pattern detection: Hierarchical search pruning

Real-time pattern recognition is computationally feasible even for large boards.

8. Strategic Dynamics

8.1 Rock-Paper-Scissors Structure

Theorem 8.1 (Cyclic Dominance)

The strategy space exhibits cyclic dominance:

$\text{Economic} \to \text{Defensive} \to \text{Aggressive} \to \text{Economic}$

Proof :

Economic > Defensive: Resource advantage eventually overwhelms static defense
Defensive > Aggressive: Attacks waste resources against fortified positions
Aggressive > Economic: Disrupts high-value economic patterns

No single strategy dominates all others.

■

8.2 Opening Theory

Conjecture 8.1: Optimal openings follow principles:

Claim territory near board center
Establish one honey farm by round 5
Maintain 60% economy, 40% military patterns

Empirical validation required.

8.3 Endgame Convergence

Theorem 8.2 (Convergence)

As $t \to \infty$ , one of four outcomes occurs:

Extinction of one/both players
Stable territorial division
Periodic oscillation
Chaotic coexistence

The 52-round limit ensures termination before full convergence.

9. Emergent Phenomena

9.1 Spontaneous Symmetry Breaking

Initially symmetric positions spontaneously break symmetry through:

Placement variations
Chaotic evolution
Competitive pressure

This mirrors phase transitions in physical systems.

9.2 Information Propagation

Glider velocities in Cloth of Gold:

Isolated: c/4 (standard Conway speed)
Through friendly territory: c/4
Through enemy territory: destroyed
Along boundaries: deflected/reflected

This creates directional information channels.

9.3 Evolutionary Arms Race

The game naturally generates an arms race:

Player A develops new pattern
Player B develops counter-pattern
Player A develops counter-counter-pattern
Cycle continues

This drives strategic innovation.

9.4 Chimeric Pattern Formation

The stalemate preservation rule enables spontaneous cooperation:

Observation: Under competitive pressure, players may discover chimeric patterns - stable structures with mixed ownership that neither can disrupt unilaterally.

Example formation:

Initial conflict zone:    After evolution:    Stable chimera:
aaabbb                    a.0.0b              ab
aaabbb                    .0.0.               ba

These patterns represent:

Emergent cooperation from pure competition
Nash equilibria in local spatial games
Phase transition from competition to symbiosis

The formation of chimeric patterns suggests that competitive cellular automata can spontaneously generate cooperation - a profound emergent property not present in the original Conway’s Game of Life.

10. Victory Condition Analysis

10.1 Victory Hierarchy

Definition 10.1 (Victory Precedence)

Extinction: Immediate win when opponent reaches 0 population
Territory: Control >75% for 10+ consecutive rounds
Economic: First to accumulate 2×base placement units
Population: Highest count after 52 rounds

10.2 Strategic Implications

Each victory type rewards different strategies:

Extinction → Aggressive early game
Territory → Expansion and control
Economic → Pattern mastery
Population → Balanced survival

This creates multiple viable paths to victory.

10.3 Expected Game Length

Theorem 10.1 (Game Duration)

Expected game length:

$E[\text{duration}] = 52 \cdot P(\text{no early victory}) + \sum_{t<52} t \cdot P(\text{victory at } t)$

With balanced players: $E[\text{duration}] \approx 35$ - $45$ rounds.

11. Connections to Other Fields

11.1 Biological Competition

Cloth of Gold models:

Competitive exclusion principle
Territorial behavior
Resource competition
Population dynamics

Similar dynamics occur in:

Bacterial colonies
Coral reef growth
Forest canopy competition

11.2 Military Strategy

The game embodies classic military principles:

Force concentration (honey farms)
Defensive depth (pattern redundancy)
Maneuver warfare (glider attacks)
Economic warfare (pattern disruption)

11.3 Computational Models

Connections to:

Agent-based modeling
Swarm intelligence
Distributed computing
Evolutionary algorithms

12. Open Problems

12.1 Theoretical Questions

Optimal Strategy Characterization: Does an approximate Nash equilibrium have a simple description?
Pattern Completeness: What is the minimal pattern set for strategic completeness?
Cooperation Evolution: Can mutual cooperation emerge from pure competition?
Complexity Bounds: Tighter bounds on computational complexity of k-round lookahead?

12.2 Empirical Questions

Opening Book: Can we develop chess-like opening theory?
Pattern Library: What patterns are competitively viable?
Skill Ceiling: How much does skill matter versus luck?
Metagame Evolution: How do strategies evolve with repeated play?

12.3 Extensions

Multiplayer: How do dynamics change with $3+$ players?
Fog of War: Limited visibility of opponent’s territory?
Terrain: Non-uniform board with obstacles/resources?
Evolution: Patterns that mutate over time?

13. Implementation Considerations

13.1 User Interface Challenges

Visualizing 168 generations efficiently
Planning complex patterns during placement
Tracking territory changes
Pattern library management

13.2 Computational Optimizations

Quadtree Acceleration: The cellular automaton uses hierarchical spatial indexing:

QuadTree structure:
- Root: entire board
- Internal nodes: board quadrants
- Leaves: active regions with cells

Update algorithm:
1. Only process leaves with active cells
2. Merge identical subtrees (hashlife-inspired)
3. Prune empty regions

Performance gains:

Sparse configurations: O(active cells) vs O(mn)
Local updates: Only affected quadrants recalculated
Memory efficiency: Empty regions require O(1) space

Boundary Handling: The finite board with empty boundaries simplifies computation:

No wraparound calculations needed
Edge cells naturally die from lack of support
Gliders exit the playable area cleanly

13.3 AI Opponents

Developing competitive AI requires:

Pattern recognition
Strategic planning
Opponent modeling
Real-time decision making

Monte Carlo Tree Search with pattern heuristics shows promise.

13.4 Balance Testing

Key parameters requiring tuning:

Generations per round (currently 168)
Base placement units (currently 10)
Pattern values
Victory thresholds

14. Conclusions

14.1 Summary

Cloth of Gold transforms Conway’s Game of Life from a zero-player mathematical curiosity into a strategic contest. The competitive modifications:

Break fundamental conservation laws
Create territorial dynamics
Enable complex strategies
Generate emergent phenomena

The game exists at the intersection of:

Cellular automata theory
Competitive game theory
Complex systems science
Strategic programming

14.2 Contributions

This work provides:

Formal framework for competitive cellular automata
Mathematical analysis of modified dynamics
Game-theoretic characterization of strategic space
Complexity results for optimal play
Open problems for future research

14.3 Future Directions

Cloth of Gold opens several research directions:

Empirical strategy analysis through tournaments
AI development for superhuman play
Extensions to other cellular automaton rules
Applications to biological/social competition modeling

The game demonstrates how simple rule modifications can create profound changes in system behavior, transforming a deterministic system into a strategic battleground where creativity, planning, and adaptation determine success.

References

[1] M. Gardner, “Mathematical Games: The fantastic combinations of John Conway’s new solitaire game ‘life’,” Scientific American, vol. 223, pp. 120-123, 1970.

[2] S. Wolfram, A New Kind of Science. Wolfram Media, 2002.

[3] E. R. Berlekamp, J. H. Conway, and R. K. Guy, Winning Ways for Your Mathematical Plays, 2nd ed. A K Peters/CRC Press, 2004.

[4] J. Nash, “Non-Cooperative Games,” Annals of Mathematics, vol. 54, no. 2, pp. 286-295, 1951.

[5] J. Maynard Smith, Evolution and the Theory of Games. Cambridge: Cambridge University Press, 1982.

[6] M. J. Osborne and A. Rubinstein, A Course in Game Theory. Cambridge, MA: MIT Press, 1994.

[7] T. C. Schelling, “Dynamic models of segregation,” Journal of Mathematical Sociology, vol. 1, no. 2, pp. 143-186, 1971.

[8] R. Axelrod, The Complexity of Cooperation. Princeton, NJ: Princeton University Press, 1997.

[9] M. Mitchell, Complexity: A Guided Tour. Oxford: Oxford University Press, 2009.

[10] C. H. Papadimitriou, Computational Complexity. Reading, MA: Addison-Wesley, 1994.

[11] R. A. Hearn and E. D. Demaine, Games, Puzzles, and Computation. Wellesley, MA: A K Peters/CRC Press, 2009.

Named after Conus textile, the cloth-of-gold cone snail whose shell patterns inspired this exploration of competitive cellular automata.