Overview
Topoethics is a conditional theorem. It does not claim that human society is
a chaotic dynamical system—it proves that if a system satisfies specific
mathematical axioms, then the removal of any single agent is
structurally irreversible. The strength of the result is entirely determined by the
strength of these axioms.
This note provides an accessible but mathematically rigorous introduction to the core
axiomatic pillars of the framework. We assume familiarity with
metric spaces and
continuity, but no prior background in
dynamical systems theory is required.
Why Axioms, Not Assumptions?
In mathematics, an axiom is a statement accepted as a starting point for deduction.
By calling these conditions axioms rather than assumptions, we
emphasize that Topoethics is a theorem of pure mathematics: it follows deductively
from the axioms, with no empirical guesswork involved. Whether these axioms hold for
any particular real-world society is a separate empirical question—one we deliberately
leave open.
Pillar 1: The Impulsive State-Space Model
Axiom 1 & 8 (State-Space Model & Macroscopic Agency)
The aggregate state of the world at time \( t \) is represented by a point \( X(t) \) on a smooth, finite-dimensional
manifold \( \mathcal{M} \subseteq \mathbb{R}^n \). Between impulse times, the dynamics are governed by a
locally Lipschitz, autonomous vector field \( F \):
\[
\frac{dX}{dt} = F(X), \quad t \in (\tau_i, \tau_{i+1})
\]
The locally Lipschitz condition guarantees, via the Picard–Lindelöf theorem, that solutions exist and are
unique for every initial condition—without it, the very notion of "two distinct trajectories diverging"
would be ill-defined.
Each human agent can, at decision epochs \( \tau_i \), inject a non-zero impulse \( I_i \in \mathbb{R}^n \)
(\( I_i \neq 0 \)) that resets the state discontinuously:
\[
X(\tau_i^+) = X(\tau_i^-) + I_i
\]
We assume all impulses are admissible: the post-impulse state
\( X(\tau_i^+) = X(\tau_i^-) + I_i \) remains on the manifold \( \mathcal{M} \).
Physically, this means human actions, however large, do not eject the system from its state space.
Crucially, these impulses manifest as classical, macroscopic events—a word spoken, a line of code committed,
a physical movement—operating at scales of \( \sim 10^{-2} \) to \( 10^{0} \) meters, exceeding the Planck
length (\( \sim 10^{-35} \) m) by over thirty orders of magnitude. The universe possesses no rounding
operator capable of truncating such massive structural changes to a mathematical zero.
Note: In the paper, these appear as Axiom 1 (State-Space Model) and
Axiom 8 (Macroscopic Agency), separated by several intermediate definitions and remarks.
We present them together here as a single "pillar" because the impulse equation is meaningless without the
state-space it acts upon.
The Anatomy of Intervention
Classical dynamical systems are continuous. However, human actions—a word spoken, a line of code committed, a door held open—are
discrete. By formalizing these interventions as impulses, we define the global system as piecewise-autonomous.
The system follows its natural mathematical flow \( F(X) \) until a human agent injects a macroscopic displacement, after which the
exact same flow resumes from the new coordinate.
Connection to Computer Science: Event-Driven Architecture
In the language of software engineering, human history is not a deterministic batch job running from fixed initial conditions.
It is an event-driven process whose state is mutated by hardware interrupts (impulses) arriving at runtime.
The universe's kernel (the vector field \( F \)) simply resumes execution from the newly mutated state register after each interrupt.
Pillar 2: The Bounded Attractor
Remark 2 (Bounded Attractor)
The flow of \( F \) admits a compact, forward-invariant set \( \mathcal{A} \subset \mathcal{M} \) (a global attractor)
to which all trajectories of interest converge after a finite transient. The diameter \( D := \sup_{x,y \in \mathcal{A}} \|x - y\| \) is finite.
Note: This condition appears as Remark 2 in the paper. It is not a formal
axiom but a standing assumption that constrains all subsequent analysis to trajectories confined
within the attractor.
Why Dissipation and Confinement Matter
Without bounded dissipation, a nonlinear system can simply explode—trajectories escape to infinity. Dissipation ensures that the system is confined:
all trajectories are eventually trapped in a bounded region of phase space. This is a necessary prerequisite for the existence of a strange attractor.
Because the state space is highly nonlinear, a trajectory can be attracted in one direction while being repelled in another, folding back onto itself
infinitely many times while remaining strictly within the finite bounds of the attractor diameter \( D \).
Connection to Deep Learning: Nonlinearity & Capacity
Without nonlinear activation functions (ReLU, sigmoid, tanh), a neural network—no matter how many layers—collapses to a single linear
transformation \( \mathbf{y} = W\mathbf{x} + \mathbf{b} \). It is nonlinearity that grants a network the capacity to fold coordinate spaces
to separate complex data (Universal Approximation Theorem). The exact same principle is at work here: nonlinearity combined with bounded
confinement is what gives a dynamical system the capacity for infinite structural complexity (chaos).
Pillar 3: Topological Mixing
Axiom 6 (Topological Mixing)
Let \( \phi^t \) denote the continuous-time flow generated by \( F \), and let \( f := \phi^1 \) be
the time-one map (the map that advances every point by one unit of time under the flow).
The autonomous dynamics on \( \mathcal{M} \) are topologically mixing:
for any two non-empty open sets \( U, V \subseteq \mathcal{M} \), there
exists an integer \( N \) such that:
\[
f^n(U) \cap V \neq \varnothing \quad \text{for all } n > N.
\]
Here \( f^n = \underbrace{f \circ f \circ \cdots \circ f}_{n} \) denotes the \( n \)-fold composition of
the time-one map—equivalently, the state of the system after \( n \) units of autonomous evolution.
Note: This appears as Axiom 6 in the paper (numbered sequentially with
intermediate Definitions and Remarks sharing the same counter).
Intuition: Kneading Dough
Imagine kneading bread dough. Each fold and stretch is a "flow" that maps one region of the dough into another. After sufficient kneading,
any microscopic drop of food coloring injected into the dough will eventually be smeared throughout the entire mass—every region of the dough
will contain traces of the coloring.
Topological mixing is the precise mathematical statement of this process. It dictates that the flow eventually spreads every open region
across the entire attractor. No subset of the state space can be "quarantined" from the cascading effects of a past impulse.
The Coffee Analogy
Consider milk poured into coffee. Over time, the white streak becomes visually indistinguishable—yet the molecular configuration of the
liquid with milk is permanently and globally distinct from coffee without milk. The milk's information is not destroyed; it is folded
into the entire volume. By Axiom 6, an individual's intervention is similarly distributed across the full state space but never annihilated.
Mixing Time and Impulse Frequency
Topological mixing is an asymptotic property (\( n > N \)). If impulses arrive so frequently
that no single inter-impulse interval achieves full mixing, this does not invalidate the result. What
matters is the cumulative effect: each autonomous segment partially stretches and folds open sets,
and the composition of many such partial stretches is asymptotically mixing—much as repeated short strokes
of stirring eventually homogenize a fluid even if no single stroke completes the job.
Pillar 4: The Chaotic Regime
Axiom 3 (Chaotic Regime)
The system possesses a positive maximal Lyapunov exponent:
\[
\lambda_{\max} := \sup_{\delta X(0) \neq 0} \limsup_{t \to \infty} \frac{1}{t} \ln \frac{\|\delta X(t)\|}{\|\delta X(0)\|} > 0
\]
where \( \delta X(t) \) denotes the evolved tangent vector of an infinitesimal perturbation \( \delta X(0) \).
The supremum is taken over all initial perturbation directions, ensuring that \( \lambda_{\max} \) captures the fastest possible
rate of exponential separation.
Note: This appears as Axiom 3 in the paper (the shared counter includes
Axiom 1, Remark 2, and then this Axiom as the third numbered statement).
The Quantitative Engine of Chaos
Topological mixing (Axiom 6) tells us that perturbations spread. The Lyapunov exponent tells us how fast.
A positive \( \lambda_{\max} \) means that nearby trajectories separate exponentially over time:
\[
\|\delta X(t)\| \sim \|\delta X(0)\| \, e^{\lambda_{\max} t}
\]
For the Lorenz system with standard parameters (\( \sigma = 10, \rho = 28, \beta = 8/3 \)), the maximal Lyapunov exponent is
approximately \( \lambda_{\max} \approx 0.906 \). This means the separation between initially nearby trajectories doubles rapidly,
destroying any possibility of long-term predictive control.
Deterministic Chaos Is Not Randomness
A critical distinction: chaotic systems are deterministic, not random. Given identical initial
conditions, the evolution is uniquely determined. What makes the system "chaotic" is that initial conditions
can never be specified with infinite precision, and any finite imprecision is amplified without bound. The
unpredictability is epistemic (we cannot measure precisely enough), not ontological (the
universe does not "roll dice" at the classical scale). This means every intervention has a definite
causal consequence—we simply cannot compute what that consequence will be beyond the Lyapunov horizon.
From Impulses to Irreversibility
The gap between an infinitesimal tangent perturbation and a macroscopic human intervention is bridged by the
interplay of these axioms. Let us trace the logical chain precisely:
-
Instant separation. The impulse (Axiom 8) displaces the trajectory by
\( \delta(\tau_i^+) = \|I_i\| > 0 \), creating a nonzero initial separation.
-
Exponential amplification. The positive Lyapunov exponent (Axiom 3) guarantees
that this separation grows exponentially on average:
\( \|\delta X(t)\| \sim \|\delta X(0)\| \, e^{\lambda_{\max} t} \).
Crucially, if \( \delta(t) \) were to converge to zero, the limsup growth rate would be
non-positive—contradicting \( \lambda_{\max} > 0 \). This proof by contradiction is what
establishes irreversibility: \( \delta(t) \not\to 0 \).
-
Bounded saturation. Since the attractor \( \mathcal{A} \) is compact with finite diameter
\( D \) (Remark 2), the separation cannot grow without bound. Instead, it saturates at the attractor
scale, after which the two trajectories behave as uncorrelated points—macroscopically distinct
states bearing no resemblance to each other.
-
Global distribution. Topological mixing (Axiom 6) ensures this divergence is not confined
to a small corner of the state space but is eventually distributed across the entire system.
Note the distinct roles: Axiom 3 alone proves that the separation is permanent (the two trajectories
never reconverge). Remark 2 clarifies the fate of the separation (saturation, not unbounded growth).
Axiom 6 establishes that the influence is global, not merely local. Together, the original
trajectory is permanently, structurally lost.
The Conditional Nature of the Theorem
It cannot be stressed enough: Topoethics is a conditional theorem. It asserts that if the axioms are
satisfied, then the irreversibility of individual removal follows as a mathematical necessity—not as a moral opinion, not
as an emotional plea, but as a rigid deductive consequence of the geometry of the state space.
The cascading nature of historical events—the way macroscopic perturbations produce massive, uncomputable global effects—strongly
suggests that complex socio-technical systems exhibit sensitive dependence. But whether they rigorously satisfy these abstract
axioms is a question for empirical science, not mathematics. The framework intentionally stays on the mathematical side of this boundary.