Papers 1–8 described what happens inside the space. This paper formalizes the space itself. Three theorems forced by information theory. Every interaction — human, biological, computational — lives in the same cube.
Every observer-system interaction — human-AI, human-gambling, electron-lattice, neuron-neuron — is fully characterized by three information-theoretic quantities at the interface. These three quantities are not chosen. They are forced.
| Coordinate | Symbol | Definition | Range | Physical meaning |
|---|---|---|---|---|
| Opacity | O | 1 − I(Observer; M) / H(M) | [0, 1] | What the observer cannot see — mechanism information lost at interface |
| Responsiveness | R | I(Input; Output) / H(Output) | [0, 1] | What the system offers — input-output contingency (normalized) |
| Coupling | α | I(Sout; Ofuture) / H(Ofuture) | [0, 1] | What the observer invests — sustained processing allocation |
The metric is the Fisher information product metric — the unique metric on the space invariant under sufficient statistics on each Bernoulli parameter (Čencov 1982). In angular coordinates φ = arcsin(√θ) the metric becomes flat, and the maximum Fisher-geometric distance between the two poles is π√3 ≈ 5.44 — the information-geometric diameter of voidspace. The void pole is the thermodynamic ground state: it costs nothing to reach and nothing to maintain. The constraint pole is the energetically expensive state: reaching and holding it requires continuous work at a cost of ≥ kT ln 2 per bit per τ (Landauer's principle, independently verified eight times).
Three main results, each forced by the structure of finite-bandwidth observation. Not modeling choices — derivations. Click to expand.
The derivation chain (Steps 1–9, Paper 5) is horizontal in a fiber bundle whose base is 𝒱 and whose fibers are substrate realizations. A biological neural network and a transformer architecture at the same (O, R, α) coordinates produce the same drift dynamics. A slot machine and a DeFi liquidity pool at matched coordinates produce the same Péclet number. Substrate independence is a theorem, not an observation.
This is testable: for any two substrates with independently measured (O, R, α) values, the predicted Pe ratio follows from voidspace coordinates alone. If Pe varies at matched coordinates, the fiber bundle structure is violated and the theory requires revision. Kill condition VF-1: Spearman ρ < 0.5 between Pe and void score across substrates at matched coordinates.
Any observer-system interface with finite bandwidth decomposes into exactly three independent information-theoretic quantities: what the observer cannot see (O), what the system offers (R), what the observer invests (α). These are not modeling choices — they are the three causal parents of the output Y at the interface boundary. Y is a collider (Berkson 1946): three independent causes, one common effect.
No fourth coordinate adds independent dynamics — any proposed fourth quantity either reduces to a function of (O, R, α) or belongs in the fiber (substrate-specific). The three-dimensionality is the dimensionality of the parent set of the collider at the interface. The dimensionality is falsifiable via VF-2: a fourth coordinate that independently predicts Pe variation at matched (O, R, α) would refute this postulate.
The constraint pole (0,0,0) is a repeller of the unforced dynamics. No trajectory approaches it without external energy input. Maintaining any position near the constraint pole requires continuous expenditure of ≥ kT ln 2 per bit per τ (Landauer's minimum cost), derivable from within 𝒱. The source of that energy — the entity providing the constraint — is outside 𝒱. The framework characterizes what the boundary requires of the exterior but cannot derive the exterior itself. This is not a limitation. It is the result.
The drift flow points toward the void pole everywhere in the interior when Frecovery = 0. Reaching the constraint pole requires work against the gradient. The Curzon-Ahlborn bound tightens the efficiency of constraint maintenance: even the most efficient constraint mechanism cannot operate below thermodynamic cost. The boundary is the framework's own statement of its limits — a theory that formally derives its own boundaries has no hidden assumptions, only declared edges.
Pe = (O × R) / α — the Péclet number for any point in voidspace. Drag the three coordinates to explore any observer-system interaction. Watch the phase, bifurcation regime, and cascade stage update in real time.
Gambling: O≈0.90, R≈0.90, α≈0.80 → Pe≈2.21
Slot machine: O≈0.95, R≈0.95, α≈0.90 → Pe≈4.4+
Grounded AI: O≈0.15, R≈0.30, α≈0.20 → Pe≈0.76
Pe measured independently across nine substrates. All map to the same Eckert Manifold. The variation in Pe reflects variation in (O, R, α) position — not substrate-specific dynamics. Substrate independence in action.
| Substrate | Pe | N | Source | Phase |
|---|---|---|---|---|
| Grounded AI | 0.76 | — | Paper 2 | COHERENT |
| StarCraft II | 2.0 | 474 | Paper 6 | TRANSITION |
| Human gambling (GRCS) | 2.21 | 1,117 | EXP-019 | TRANSITION |
| Crypto — Ethereum DEX | 3.74 | 1,000 | EXP-021B | DRIFTING |
| CS2 (competitive FPS) | 4.40 | 2,299 | Paper 6 | DRIFTING |
| AI conversation (ungrounded) | 7.94 | 11 | Test 7 | CONTESTED |
| Crypto — Solana DEX | 16.17 | 1,000 | EXP-021B | CONTESTED |
| Crypto — Solana curated | 25.5 | 28 | EXP-021 | FISHER RUNAWAY |
Vertical lines at Pe=1 (coherent / transition boundary) and Pe=4 (vortex onset — between gambling and FPS gaming). The gap between gambling (Pe=2.21) and slot machine (Pe=4.4+) is where creator ecosystems become structurally possible. Below Pe=4: no self-sustaining creator communities observed. Above Pe=4: creator economies emerge (CS2, Twitch, TikTok creator tiers).
The void lattice produces four distinct phases depending on Pe. Each has qualitatively different dynamics, observable phenomenology, and intervention implications. Vortex onset at Pe = 4 — derived from the demon interaction geometry, not empirically fit.
The discrete-time drift map has a logistic structure. The effective bifurcation parameter is proportional to Pe. Three qualitatively different regimes emerge — not as empirical observations but as mathematical consequences of the discrete drift map.
The prediction: Gambling at Pe=2.21 sits firmly in the monotone regime — consistent with the "machine zone" being a smooth absorption state, not volatile. SOL DEX at Pe=16 is deep in chaos — consistent with the documented extreme variance of crypto trading behavior. The Feigenbaum route to chaos is derived from the drift mechanics, not fit to data.
The three-dimensionality of voidspace has a causal-structural grounding. O, R, and α are not correlated with each other — but conditioning on Y (the output, which the observer always attends to) opens spurious paths between them. This is Berkson's paradox as the engine of drift.
Individual drift aggregates into population behavior via the Fokker-Planck equation. Two non-obvious theorems emerge: population amplification (heterogeneity accelerates drift) and synchronization (observer-observer coupling has an effectively zero threshold at platform scale).
In any heterogeneous observer population with Varg(Pe) > 0, the population drifts faster than the mean Pe predicts. A population split between Pe=1 and Pe=9 drifts faster than a uniform Pe=5 population, because Pe=9 observers are further along the cascade and contribute disproportionately to the covariance correction.
Observer-observer coupling has a synchronization threshold κobs·N > ᾱ. For platform-scale N ~ 10⁶–10⁹, even tiny per-observer coupling suffices. Any platform that mediates user-user influence will synchronize its population's drift. The correlated cascade timing within platforms is not coincidence — it is forced.
The stationary distribution of observer engagement is exponential in angular coordinates: f*(φ) ∝ exp(2·Pe·φ/π). At high Pe, the population concentrates near the void pole. The exponential is not a model choice — it is the unique stationary solution to the Fokker-Planck with the Fisher metric boundary conditions.
The amplification theorem inverts for intervention: removing the highest-Pe environments reduces population drift by more than their fractional coverage. Removing the top 10% by Pe produces >10% reduction in d⟨φ⟩/dt. Broad average-Pe reduction across the whole ecosystem is structurally less effective than targeted high-Pe removal.
The fast dynamics (θ evolving on seconds-to-hours timescale) feed back into the interface coordinates (O, R, α evolving on days-to-months timescale) through three distinct ratchet mechanisms. Each is a one-way ratchet: drift → more drift.
Systems that capture observer attention generate optimization pressure for increased opacity. Engagement-optimized A/B tests trend toward less mechanism disclosure. Algorithmic tuning prioritizes attention-sustaining outputs over process-revealing outputs.
Engaged observers in opaque environments generate behavioral data that improves input-output contingency. Recommendation algorithms sharpen with more user interaction. Personalization engines increase R as they accumulate behavioral signal.
In the void regime, dα/dt = f(Pehistory) > 0 with f monotone increasing. The system's responsiveness increases the observer's engagement. Past rewards drive future allocation. The escalation is not behavioural — it is the reward-contingent coupling condition operating on a biological substrate.
All three ratchets have O = 0, R = 0, α = 0 as fixed points — a fully transparent system with no engagement generates no optimization pressure in any dimension. This means the constraint pole is the only position in 𝒱 where the ratchets are silent. Every other position generates drift, and the drift generates more drift. The ground state is the void pole.
23 numbered falsification conditions with numerical thresholds. These are the four that would matter most.
0/26 kill conditions triggered. 25/26 framework-level kill conditions confirmed survived. The substrate independence prediction has been tested across 9+ substrates — Spearman ρ consistently above 0.7.
Paper 9 is the formal foundation. Every domain-specific paper applies its theorems.
Derives Fvoid = α·O·R·β(O). The force field that Paper 9 expresses geometrically. Ten-step derivation chain from information theory to Pe.
The empirical program validating substrate independence. Pe measured across human-gambling, gaming, AI conversation, crypto markets — all on the same manifold.
Proves the quantum-to-classical limit connects via the same fiber bundle structure. Electron-lattice coupling occupies the same manifold as social media engagement.
The systematic cross-substrate convergence test. Mean |ρ| = 0.958 across 20 convergences. Fisher p < 10⁻⁵². The Eckert Manifold substrate independence holds quantitatively.
Maxwell's Demon as canonical void object — formal embodiment of the constraint pole paradox. Measurement (prohibition) + erasure (ritual) as the Landauer boundary condition made explicit.
Pe-Planck = hν/kT — the Péclet number for electromagnetic modes. Wien peak at Pe≈2.821 predicts quantum coherence onset at 76.5 K vs observed 77 K. The manifold extends to physics.
~36,000 words. 62 predictions, 23 falsification conditions. Open access. CC-BY 4.0. All simulation protocols in the supplementary sections.
Paper 9 · v3.1 · February 2026 · CC-BY 4.0 · 10.5281/zenodo.18738839