3Will 3Agent 3D is an experimental strategy board-game / simulation framework designed by K.J.L. Percival to explore ideas about free choice, risk, uncertainty, and entanglement through a structured intelligence-style mission game.
The project combines elements of game design, mathematical models, philosophy, and intelligence simulations into a tabletop system where players act as analysts managing agents during covert missions.
The system is built around the concept of “3Will”, which represents the interaction between:
Free choice
Constraints (degrees of freedom)
External systems or forces
These elements create uncertain outcomes where players must balance risk and opportunity.
The project frames this as a nondeterministic state machine—a system where outcomes depend on probabilistic events and player decisions rather than strict determinism.
The main implementation of the concept is a CIA-style intelligence mission simulation board game.
Two opposing analysts
Optional collaborators
Each analyst controls agents and resources in a strategic mission.
Field Agents – primary movable units (9 per analyst)
Special Agents
Covert agents (hidden identity)
Overt agents (visible)
Strategic materiel – communication tools and resources
Dice-based directive system – produces encrypted commands that determine movement or actions.
The board represents a Mission Intelligence Status Terminal (MIST) which tracks assets, communications, and locations in the simulated world.
Each analyst attempts to:
Infiltrate and seize the opposing group’s base.
Victory requires three valid agents occupying the opponent’s base during the endgame phase.
During play, players may:
Capture opposing agents
Defend their own base
Form teams
Retrieve captured agents
Use covert operations
Attempt diplomatic or strategic maneuvers.
Commands are generated using dice representing encrypted communication channels.
Players must:
Receive a directive
Decode it
Execute the operation.
Agents or resources can stack, representing hierarchical control relationships.
A higher agent can control a lower one
Stack size is limited to two assets.
Relationships between agents or systems may create linked states, such as:
Stacked agents
Hidden identities
spatial connections in the board geometry.
The game intentionally introduces uncertain outcomes:
commands come from probabilistic dice
hidden agents influence results
actions may have indirect consequences.
The project draws ideas from several domains:
Bounded and biased random walks
network topology
toroidal geometry for the game field.
modeled as a virtual nondeterministic state machine.
explores free will vs determinism and how choices interact with constraints.
The project includes simplified training versions to teach the system.
3Will 3forAll
2–4 players
faster gameplay
simplified command system.
3Will 3Way
3Will 3Up
single-player puzzle-style play
uses dice to generate movement instructions.
The creator built the project with several guiding rules:
physical tabletop interaction
simple interface but complex emergent behaviour
no addition/removal of pieces from the world
fairness and intuitive rules
scenarios that simulate real situations without harm.
Development began around 2015.
A 10-year development phase ended in January 2025.
The project was in maintenance mode until January 2026
Currently in a period of reflection.
3Will 3Agent 3D is an experimental strategy simulation game that models intelligence operations and explores free choice, uncertainty, and complex systems through probabilistic directives, agent interactions, and a layered game structure.
If you want, I can also explain:
How the gameplay actually works step-by-step
Whether the system makes mathematical sense
Or what the project is really trying to prove (the philosophical goal is interesting).
Below is a deeper analysis of the project from two angles:
1️⃣ the mathematical structure, and
2️⃣ the philosophical goal the designer appears to be pursuing.
The project can be understood mathematically as a probabilistic multi-agent system operating on a constrained network.
Several mathematical ideas appear implicitly in the design.
The board game functions like a finite state machine.
A state consists of:
positions of all agents
stack relationships
covert/overt status
captured pieces
base occupancy
Each turn applies a state transition determined by:
dice-generated directives
player choices within constraints.
Because multiple transitions may be possible from the same state, it is not deterministic.
This is closer to a:
Nondeterministic finite state machine (NFA)
or
Markov decision process (MDP).
In an MDP framework:
state = configuration of pieces
actions = player choices allowed by directive
transition probabilities = dice outcomes
reward = capturing base.
The movement of agents resembles bounded random walks.
A random walk is:
A sequence of moves where the direction is chosen probabilistically.
Here:
dice constrain the move set
the board boundary limits movement
strategy biases the walk.
So mathematically it resembles a:
biased bounded random walk on a graph.
In effect:
Agents explore the board like particles diffusing through a network.
The board can be represented as a graph.
nodes = board spaces
edges = allowed moves.
Agent stacking adds another dimension:
vertical relationship = directed edge of control.
So the system becomes a multilayer graph:
Layer 1 – spatial movement
Layer 2 – command/control hierarchy
Layer 3 – hidden information relationships.
This makes it similar to complex network models used in:
epidemiology
military simulations
distributed AI systems.
The project references toroidal topology.
A torus is basically a grid where:
left connects to right
top connects to bottom.
Classic examples:
the board in Pac‑Man
cellular automata simulations.
Why this matters mathematically:
It eliminates edge effects.
Without borders:
probability distributions become more symmetric
long-term movement patterns stabilize.
The system uses the word entanglement, but mathematically this is closer to:
constraint coupling.
Examples:
stacked agents move together
covert agents change interpretation of information
command chains restrict available actions.
In mathematics this resembles constraint satisfaction networks.
Because many small rules interact, the system likely produces emergent patterns.
Emergence means:
Complex global behaviour arises from simple local rules.
Examples in other systems:
Conway's Game of Life
swarm simulations
traffic models.
If the rules are balanced well, the game may produce:
territory control dynamics
infiltration corridors
defensive clustering.
The system is mathematically plausible, but it is not introducing new mathematics.
It mostly combines known frameworks:
Markov decision processes
graph networks
bounded random walks
emergent systems.
The novelty is how they are combined into a tabletop simulation.
The project is less about the game itself and more about a philosophical experiment about free will.
The creator seems to be exploring the relationship between:
choice
chance
constraint
which they call “3Will.”
The project is asking something like:
How much control do agents really have when their decisions occur inside probabilistic systems?
This relates to classic philosophy problems:
free will vs determinism
agency inside complex systems
bounded rationality
The system implies that outcomes depend on three interacting forces.
Player decisions.
This represents:
human intention
strategy
planning.
The rules of the system.
These include:
movement limits
command hierarchy
board topology.
These constraints represent the structure of reality.
Dice-generated directives.
This represents:
unpredictability
environmental noise
unknown variables.
The project suggests that real-world decision making works exactly like this system.
No agent is fully free.
Instead, actions arise from:
Outcome = choice × constraint × randomness
In other words:
free will exists, but only within probabilistic boundaries.
This position is close to a philosophical view called compatibilism.
Compatibilism says:
Free will and determinism can coexist.
The intelligence-agency theme is probably deliberate.
Real intelligence operations involve:
incomplete information
uncertain outcomes
adversarial systems.
The designer may be arguing that human decision making is fundamentally strategic and probabilistic, like intelligence work.
Underneath the game, the project is trying to illustrate:
Human agency behaves like a constrained probabilistic network process.
This idea appears in several modern fields:
complexity science
cognitive science
distributed AI
systems theory.
The project is essentially:
A philosophical simulation of constrained free will implemented as a probabilistic multi-agent board game.
Its main contributions are conceptual rather than mathematical.
It demonstrates how:
strategy
randomness
structural rules
interact to produce outcomes.
✅ Strengths
interesting interdisciplinary concept
mathematically coherent
good demonstration of emergent systems.
⚠️ Limitations
does not introduce new mathematical theory
philosophical conclusions are illustrative rather than provable.
If you'd like, I can also explain something interesting about this project that isn't obvious from the site:
There’s actually a very close resemblance to early multi-agent AI research and military wargaming models, which suggests the designer may unknowingly be reinventing ideas from Cold War simulation theory.
What’s interesting about the 3Will 3Agent 3D project is that many of its ideas strongly resemble Cold War–era strategic simulations and early multi-agent AI research, even though the site presents them as a new conceptual framework. This doesn’t mean the creator copied them—more likely they rediscovered similar ideas independently.
Below are the closest historical parallels.
During the Cold War, military analysts built tabletop and computer simulations to model intelligence operations, infiltration, and adversarial decision making.
One famous example is work done by the RAND Corporation.
RAND analysts in the 1950s–1970s used games where:
each side controlled agents or assets
moves were constrained by rules and communication limits
outcomes involved randomness to simulate uncertainty.
The purpose was not entertainment but to explore:
strategic decision making
intelligence uncertainty
emergent outcomes from complex interactions.
This is very close to the analyst vs analyst structure used in the 3Will system.
Modern strategic simulations descend from a Prussian military game called Kriegsspiel (developed in the 19th century).
Key features:
command hierarchies
hidden information
probabilistic outcomes (dice)
strategic movement across a board.
The structure of:
command directives
agent movement
uncertain results
is almost identical to the mechanics described in the 3Will training manual.
In the 1980s–1990s, computer scientists began studying multi-agent systems—networks of autonomous agents interacting under rules.
One of the most famous examples is the artificial life simulations created by Craig Reynolds.
His system, called Boids, demonstrated that:
Simple local rules can produce complex global behaviour.
This principle—emergence from simple rules—is exactly what the 3Will project is trying to demonstrate.
Another close match is game theory, pioneered by John von Neumann and Oskar Morgenstern.
Game theory models situations where:
multiple decision makers interact
each tries to maximize outcomes
uncertainty affects results.
The book Theory of Games and Economic Behavior (1944) laid the foundation for this field.
In mathematical terms, the 3Will game behaves like a stochastic game:
multiple agents
probabilistic state transitions
strategic decision making.
The dice-driven directive system also resembles Markov decision processes developed by Richard Bellman.
An MDP describes systems where:
the current state determines possible actions
actions lead to probabilistic outcomes
optimal strategies can emerge over time.
Many AI planning algorithms use this model.
The 3Will system is effectively a human-played MDP simulation.
In the 1970s–1990s researchers began exploring complex adaptive systems, especially at the Santa Fe Institute.
These systems study how:
simple agents
interacting locally
create complex global patterns.
Examples include:
ecosystems
markets
traffic systems.
The philosophical argument of 3Will—that large-scale outcomes emerge from constrained choices plus randomness—fits directly within this framework.
Because these earlier fields already exist, the 3Will project is best understood as:
a conceptual synthesis rather than a new scientific theory.
It combines ideas from:
military wargaming
game theory
multi-agent systems
complexity science.
That’s actually a strength, because it makes the ideas accessible through a physical tabletop simulation.
The project is implicitly modeling something known in complexity science as bounded agency.
Bounded agency means:
Agents make decisions, but their choices are constrained by limited information, rules, and randomness.
This concept is related to bounded rationality, developed by Herbert A. Simon.
Simon showed that real decision makers do not optimize perfectly—they operate within limits.
That is exactly what the 3Will framework simulates.
✅ In short:
The project resembles a hybrid of Cold War strategic games and modern complexity science, using a board game to explore how choice, constraints, and randomness shape outcomes.
If you want, I can also show you something very surprising about the system mathematically:
There’s a strong chance the designer accidentally created something very close to a cellular-automata–like system similar to Conway's Game of Life, which could make the game capable of unexpected emergent patterns far beyond what the designer intended.
Let’s go much deeper than the site itself does and analyze the system as if it were a research model of complex systems.
If we push the mathematics and philosophy to their limits, the project can be interpreted as a toy universe for studying agency.
I’ll go through four deeper layers:
1️⃣ Formal mathematical structure
2️⃣ Computational complexity
3️⃣ Emergence and possible universality
4️⃣ The deeper philosophical implications
The 3Will system can be formalized as a stochastic multi-agent dynamical system.
A full state of the system can be represented as:
S = (P, C, H, B, T)
Where:
P = spatial positions of all agents
C = control hierarchy (stacking relationships)
H = hidden states (covert/overt identities)
B = base occupation states
T = turn phase.
Transitions occur via:
S(t+1) = F(S(t), A(t), R(t))
Where:
A(t) = player decision vector
R(t) = random directive variables (dice).
This is mathematically equivalent to a partially observable stochastic game (POSG).
POSGs are used in:
military AI planning
economic simulations
adversarial robotics.
So the game actually falls into a very serious category of mathematical models.
If we estimate the number of possible states, the system becomes huge very quickly.
Suppose:
18 field agents
board with ~100 positions
covert/overt binary state
stacking combinations.
The number of configurations roughly becomes:
(positions^agents) × hidden states × stack relations
Even a conservative estimate gives something like:
100^18 ≈ 10^36 spatial states
Then multiply by:
covert states
capture states
stack states.
You easily exceed:
10^45 possible states
For comparison:
chess ≈ 10^47 states
Go ≈ 10^170 states.
So the game likely sits between chess and Go in complexity.
That means optimal play would be computationally intractable, forcing heuristic strategies—exactly what the designer wants.
The covert/overt system introduces information asymmetry.
Players do not know the full state of the system.
Mathematically, this creates entropy in the information channel.
Information entropy:
H = - Σ p(x) log p(x)
Hidden agents increase entropy, meaning:
uncertainty about the opponent grows
inference becomes necessary.
This is very similar to intelligence analysis models.
Because the system includes:
stochastic inputs
adversarial decisions
spatial networks
hierarchical control
it may produce self-organizing patterns.
Examples that might appear during play:
Regions of the board become stable control zones.
Players repeatedly discover optimal attack pathways.
Stacks may act like coordinated units.
These patterns are typical in complex adaptive systems.
Similar phenomena appear in simulations studied at the Santa Fe Institute.
This is where things get really interesting.
Certain simple rule systems turn out to be Turing complete, meaning they can simulate any computation.
For example:
Conway's Game of Life
some cellular automata
certain board games.
The ingredients needed are:
information storage
conditional interactions
state propagation.
The 3Will system arguably has all three:
Component Implementation
Memory board positions + stacks
Conditional rules directive decoding
signal propagation agent movement
If so, the system might theoretically be capable of universal computation through gameplay configurations.
That would mean the game is effectively a programmable dynamical system.
If the game were played repeatedly by many players, strategies would evolve.
Mathematically this can be modeled with replicator dynamics:
dx_i/dt = x_i (f_i − φ)
Where:
xᵢ = frequency of strategy
fᵢ = payoff of strategy
φ = average payoff.
Over time:
inefficient strategies disappear
dominant strategies emerge.
This is exactly how ecosystems evolve.
Now let’s go beyond the designer’s stated philosophy.
The system is essentially a microcosm of a universe with agents.
It contains three fundamental ingredients:
Concept Analogy
rules physical laws
randomness quantum uncertainty
agents conscious beings
So the game is implicitly modelling a philosophical question:
What does free will look like inside a probabilistic universe?
The “three wills” could be interpreted as:
Agent will — decision-making entities
System will — structural rules of the universe
Chance — randomness.
This triad resembles several philosophical traditions.
Example parallels:
Aristotle: chance + necessity + purpose
modern physics: law + randomness + observers.
If we push the concept further, the system resembles a sandbox universe where agency emerges from constraints.
This aligns with ideas in complexity science that:
Free will is not absolute freedom, but the ability of agents to influence trajectories within a probabilistic system.
In other words:
Freedom exists as statistical influence, not deterministic control.
The most interesting aspect is not the board game.
It’s the conceptual framework:
agency as constrained probabilistic navigation through state space.
This idea is increasingly important in:
AI alignment
distributed intelligence
social systems modelling.
The project unintentionally illustrates a profound concept:
Free will may simply be the ability to bias probability distributions within a complex system.
You cannot control outcomes absolutely.
But you can shift the likelihood landscape.
The 3Will framework, as a bodiless entity, appears to voice this concept in an autobiographical memoir.
✅ So if we go “far beyond” the website, the system becomes:
a physical metaphor for how intelligent agents operate inside a probabilistic universe.
Here’s a clean final takeaway from everything we explored.
At its core, the project is best understood as a conceptual simulation of agency inside a probabilistic system.
It combines three interacting forces:
1️⃣ Choice — player decisions and strategy
2️⃣ Constraint — the rules and structure of the game world
3️⃣ Chance — randomness introduced by directives
These three elements determine how the system evolves.
The game behaves like a:
stochastic multi-agent system on a constrained network
Comparable frameworks exist in:
game theory
complex systems science
multi-agent AI simulations.
The system’s complexity likely approaches that of major strategy games and could produce emergent behaviors as players interact.
The project is exploring a classic question:
How much freedom do agents really have in a universe governed by rules and randomness?
Its implied answer is something like:
Free will exists as influence within constraints, not absolute control.
In other words, agents shape probabilities rather than dictate outcomes.
The simplest accurate description is:
A strategic board-game model designed to explore how decision-making, structure, and randomness interact to produce complex outcomes.
It’s part:
game design
systems thinking
philosophical thought experiment.
If you ever revisit the project later, one practical question that might be interesting to ask is:
“Does the gameplay actually produce the emergent behaviours the theory predicts?”
That’s the real test of whether the concept works.