Emergent Gravity · The Ledger · Foreign Accounts · Annex B

ONE SPHERE,
THREE LIVES

A side expedition builds matter out of holes in a graph, watches every classical road
to the fermion die — and finds the minus sign living where it always had to live: in ℏ, behind a sphere.

✦ ✦ ✦

WRITTEN 10–11 JUNE 2026 · 1 NO-GO, 3 PRONGS · 1 CENTRAL THEOREM · 7/7 ON THE EXCHANGE · 4 KILLS RECORDED

I · Matter From Holes

Mass is the resistance of a loop.

This annex began over coffee with a slogan — matter is absolute, distance is relative — and ended with a theorem about why quantum mechanics is complex. The premise is older than this house: Wheeler's 1957 charge without charge. Take a graph — nodes, edges, weights — and let matter be the harmonic sector of its discrete Hodge theory: the cohomology classes threaded through handles (wormholes) in the graph. A particle's existence is then purely topological — dim ker Δ₁ = b₁ for any weights — while its mass and position are geometric. The harmonic representative is, by Thomson's principle, a Kirchhoff current, and its energy is exact and almost embarrassingly concrete:

mass = Φ² × (effective resistance of the loop)   ··   gravity = the weights relaxing around the topology that cannot bend

Three pre-registered kill-tests, run in ten seconds of Rust, all survived. The direct force between two handles flips sign with flux orientation and decays as a dipole — the electromagnetic sector. The induced force from weight relaxation is universally attractive: the source is quadratic in the field, the response linear in the stress, so the flux sign cancels out of the second-order energy shift, which is negative-definite — Sakharov's 1967 induced gravity, rediscovered on a lattice, with its scaling measured: G ∝ 1/κ, the inverse stiffness of the vacuum.

The gravity sector · pre-registered predictions vs results
MeasurementRegistered predictionResult
Direct (EM) force between handlessign flips with flux; dipole d⁻³flips; d⁻²·⁸
Induced force, gradient elasticityattractive, both orientationsattractive at every separation, monotone
Sakharov scaling, κ = 0.25–1.0E×κ constant−5.8, −6.2, −6.4 ×10⁻⁴ (~10%)
δc(r) around a handle vs point massuniversal far fieldratio → 0.999 by r ≈ 11
On-site elasticity controlshort rangecollapses, d⁻⁵·⁷

The highlighted row deserves its sentence: the deformation field around a handle is identical, to a tenth of a percent, to the field of a structureless point mass of the same energy. The far field forgets the source — an equivalence principle, emergent and unasked-for. And the last row is a condition, not a detail: only gradient elasticity (the vacuum penalising curvature of the weights, not their deviation) produces long-range gravity. We later found that exact condition in the literature: it is Kleinert's "floppy world crystal" requirement, derived in the continuum twenty years ago. A ten-second simulation rediscovered the entry condition of a theory it had never heard of — convergences of that kind are validations, not threats.

A prediction killed the same day — and rehabilitated If a fermion's mass were the half-quantum of its own handle (m ~ ℏ/L), the electron's handle would be Compton-sized — 4×10⁻¹³ m, five orders of magnitude above the 10⁻¹⁸ m pointlikeness bound. Killed in one line of arithmetic, recorded — and later rehabilitated: the lattice itself showed (§III) that the model never made that prediction. Its topological cores are heavy and small; light masses must come from elsewhere. 't Hooft–Polyakov monopoles would nod.
II · Three Roads to the Fermion

Three clean deaths, one no-go.

A theory of matter that cannot tell a boson from a fermion is a theory of soap bubbles. So the expedition went hunting for the −1 — the exchange sign — inside the classical framework, three different ways, each with registered predictions. All three died, and the pattern of the deaths is the result.

✗1

The twisted mode dissolves. KILLED

A Z₂ twist (antiperiodic boundary condition) on a handle destroys its zero mode — the count drops 4 → 3, exactly as cohomology demands. But the would-be fermion mode is pinned at the bulk scale (0.2679, the lattice floor) for every handle length tried: forcing a unit of twisted cycle current costs O(1) regardless of geometry. Div-free rigidity of a one-edge-wide throat: no room for the antiperiodic node. The light twisted quasiparticle does not exist classically. (On global torus cycles the soft mode does exist: the discrete Scherk–Schwarz mass came out as 2(1−cos π/N) to five digits — ratio 1.000.)

✗2

The harmonic mode is gauge-blind. KILLED

Transport a handle's zero mode around a Z₂ vortex ring — a discrete Aharonov–Bohm braid, the mode computed exactly at each step, Wilson-loop holonomy ±1 as readout. Result: +1 on the linking path, +1 on all three controls. The harmonic mode carries emergent flux, not Z₂ charge. Nothing to braid.

✗3

Z₂ cannot orient a braid. KILLED

The last hope was a topological spin: sweep a vortex twice around a handle, read the sign of the double loop. But in a Z₂ configuration space, flipping twice is unflipping: "sweep onward" and "sweep back" are the same sequence of configurations. Every closed classical loop is a retracing. The orientation a braiding phase needs simply does not exist in real, classical variables.

Classical, real, pre-geometric matter is bosonic.
The fermionic minus sign does not live in any configuration space.
It lives in the ordering of operations — and ordering is what ℏ is.

One positive result fell out of Kill 3's setup, and it became the hinge of everything after: sweeping a sphere of flipped edges across one mouth of a handle destroys the zero mode exactly at the closure of the surface — the norm collapses from 0.633 to 1.4×10⁻¹⁴ at the sixth and final edge. The sweep is an all-or-nothing, topological operation: it toggles the handle between its bosonic and twisted sectors. Hold that sphere. It has two more lives to live.

III · The Quantum

One rotor per hole in the universe.

If the sign lives in operator ordering, quantize. The statics of the graph theory turn out to be the Villain limit of compact U(1) lattice gauge theory, and the kinetic term is circuit quantization — Devoret's machinery, edge weights as inverse inductances (so this gravity is, literally, a relaxation of inductances), node capacitances as the kinetic metric. Compactness is not free: it follows from one declared axiom upgrade — the matter field is a U(1) connection, not a real 1-form — which the framework was already whispering, since its spin structures were always Z₂ ⊂ U(1) edge holonomies.

The zero mode of each handle then becomes a free quantum rotor: H = Q²/2M, integer charge, and an identity that is pure Thomson duality, verified to 1.00000 across six geometries:

M × (loop resistance) = 1   ··   the rotor's moment of inertia is the inverse of the classical mass

Time-reversal invariance selects exactly two representations of this rotor — the two T-invariant points α ∈ {0, π} of its theta-circle — and they are precisely the two spin structures: integer charge (bosonic) and half-integer charge (fermionic). The boson/fermion dichotomy is representation theory. And the half-charge state's energy is P/8 — geometric, light for short conductive throats (0.06–0.18, below the bulk floor of 0.268): the light fermion, killed as a classical field mode in Kill 1, resurrects one floor up, as a charge state.

Literature check, run before believing any of it Compact U(1) confines in 2+1D (Polyakov), and a confined photon would kill the electromagnetic sector of §I. A from-scratch 4D Wilson-action Monte Carlo (8⁴, Metropolis, Creutz ratios) localised the deconfinement transition between β = 1.00 and 1.05 — the literature value is 1.0111 — and exhibited the free-photon phase. The framework collects its third structural constraint: vacuum stiffness above βc, after gradient elasticity and matter-as-connection.
IV · The Theorem

The sweep is the fermion parity.

Now the sphere's second life. Quantize the classical sweep of Kill 3: flipping σe = −1 is shifting θe by π, and the unitary that does this is eiπEe — local, one edge, no freedom in the choice. The product over the six edges of a sphere around a wormhole mouth then collapses, by Gauss's law alone, to the one edge the sphere does not contain:

V_sweep = Π_(e∈sphere) e^(iπE_e) = e^(iπG_a) · e^(∓iπE_w) = e^(iπE_w) = e^(iπQ) = (−1)^F # on the Gauss sector V W = − W V # from [Φ, Q] = iℏ and the integer spectrum — nowhere inserted

The vortex sweep is the fermion parity operator. Its anticommutation with the Wilson loop — the fermionic minus sign — emerges from exactly the two ingredients the no-go demanded: ℏ (the canonical commutator; set it to zero and V commutes with everything, recovering the no-go as the classical limit) and compactness (the integer spectrum that makes orientation signs irrelevant mod 2π). No string operators postulated, no stabilizer algebra imported: Kitaev's toric-code operators are derived here from a classical geometric operation plus Gauss plus canonical quantization. The whole chain was verified on an exact model — Z₄ gauge theory on a seven-edge dumbbell graph, the full 16,384-dimensional Hilbert space, Gauss constraint diagonal in the electric basis — five identities, five exact zeros.

V · The Exchange

7/7 — and the sphere's third life.

Statistics is an exchange, not a parity. The exchange phase was computed with a Levin–Wen protocol — three hopping operators around a triangle, the two temporal orderings of the same swap, their ratio a pure scalar — implemented in an exact symplectic Weyl engine (sixty lines of integer arithmetic, no Hilbert space needed). Calibration first, on known ground: the 2D toric code, using only the operators derived in §IV. It returned (e, m, ε) = (+1, +1, −1) with zero tuning — the famous fermion, emerging from our operators. In two dimensions, the chain is closed end to end.

Three dimensions resisted, instructively. A wormhole-mouth dyon — charge plus π-twist — is a strange object: its flux is swallowed by the topology (no real-space Dirac string; a half-monopole forbidden by Dirac's condition in ordinary space, allowed here only because the wormhole hides the evidence). Teleporting the twist from site to site gave +1: a registered hypothesis — teleportation carries no framing — confirmed. The faithful transport membrane could not be guessed; it had to be built from the only object available. It is the sphere's third life: the framing membrane is the world-tube of the §IV sweep sphere, dragged along the path, under one coherence rule — on the closed exchange circuit, every traversed node is swept exactly once. Two consistent tilings exist; a prescription is only real if they agree.

The 3+1D exchange · all predictions registered before the runs
ConfigurationRegisteredR
dyon, teleported twist+1 (no framing)+1
dyon, framed — departure tiling−1−1
dyon, framed — arrival tiling−1−1
membrane without endpoint spheres+1+1
charge alone / twist alone+1 / +1+1 / +1
framed + off-path sphere (robustness)−1 unchanged−1

Seven registered predictions, seven confirmed, in exact integer arithmetic. The fourth row localises the mechanism: the sign lives at the membrane's attachment to its endpoints — the corner spheres crossing the neighbouring hops' charge strings. The wormhole-mouth dyon is a fermion in 3+1 dimensions under faithful transport. And the 2π-rotation half of spin-statistics costs nothing extra: the rotation's operator content is the closed sweep, which is (−1)F by §IV — minus one on odd charge.

VI · Audit

In honesty.

  • The framing prescription — sweep every traversed node once, corners assigned by the circuit tiling — is fixed by a consistency argument (both tilings agree; teleportation and unattached membranes fail), not yet derived from the adiabatic limit of continuous graph morphing. That derivation is the open rigor gap of the expedition, named here rather than hidden.
  • The gravity sector's measured exponents carry finite-size systematics (reference subtraction, periodic images, N ≤ 32); the sign, universality and range-ordering results are robust to all of them; the clean exponents are not yet precision measurements. The point-mass comparator (0.999) is the trustworthy statement.
  • Almost every brick exists elsewhere: Wheeler, Kleinert, Sakharov, Kitaev, Devoret, Goldhaber. What this expedition claims is the assembly, the resistance dictionary (mass = loop resistance; M×P = 1), the three-pronged classical no-go with its measurements, and the derivation chain sweep → (−1)F → framing. A serious literature pass before any external write-up is owed and planned.
  • Prior-art addendum, same day: the headline of §V has forty-five years of ancestry. Friedman & Sorkin, 1980 — Spin ½ from gravity — obtained half-integer spin from spatial handles (geons); the same authors proved a spin-statistics theorem for electric–magnetic composites the same year; Dowker & Sorkin extended it to geon exchange; and the charge-flux exotica of wormhole mouths live in the Alice-string / Cheshire-charge literature (Bucher–Lo–Preskill and successors, including 3+1D lattice phases). What may remain ours: the exact Weyl-engine computation on graphs, the framing membrane as the world-tube of Gauss spheres, and the no-go-to-fermion arc as a single constructive chain. The claim is hereby downgraded from result to mechanism realised in a new setting — pending the comparison this house owes Friedman and Sorkin.
  • The bridge — DELIVERED at the structural level (Phase 4, same day): under two pre-geometric gauge postulates (G1: node-displacement modes of the weight field are gauge — discrete diffeomorphisms; G2: the local trace is constrained, non-radiative), the physical bath is exactly the discrete "+"-polarized gravitational plane waves — dim TT = 3(N−1)+2 in closed form (the run-log's 36 was a one-shot stochastic estimate of 47, 1.3σ low; corrected in completion) — and the selection rule emerges exactly: a sphere in position superposition is invisible to the bath at machine zero (TT fraction 1.1×10⁻³⁴ vs 8.9×10⁻⁶ for a transverse rod, ratio 7×10⁻³⁰), by symmetry, not suppression. New sub-prediction (P4-4): a rod displaced along its own axis is also invisible (3×10⁻³³) — a shape×direction control knob for BMV-class experiments the original framework never stated. Controls: without G1-G2 the raw rod/sphere ratio is ≈2 (the selection comes from the postulates, not the profiles); rod-x/rod-z isotropy = 1.0000. Bonus, completed: the same constraint structure suppresses the gravitating zero-point energy of the weight field by Λ_TT/Λ_naive ∝ N⁻² exactly (measured exponent −1.99 across N = 8…64) — the toy's cosmological constant is parametrically small for the same reason its sphere is silent, rhyming with the locked sector of the main programme. Still owed: the quantitative rate Γ = GM²Q²/ℏR with its exponents (requires the bath's measurement model, not just vacuum dressing) — Phase 5.
  • No laboratory prediction is claimed beyond those this archive already owns. This annex's product is structural: a no-go, a theorem, and an exchange computation, all reproducible from the code below in under a minute.
VII · Reproduce It

The code, so you can break it yourself.

# two Rust crates, no dependencies, every number in this entry: # topo-gravity (classical): cargo run --release # gravity kill-tests: EM + Sakharov + ranges cargo run --release profile # δc(r) vs point mass: 0.999 (Kleinert test) cargo run --release twist # Scherk-Schwarz 1.000; zero-mode counts cargo run --release chain # Kill 1: the twisted mode never softens cargo run --release berry # Kill 2: AB holonomy +1, all controls cargo run --release alg # Kill 3 + the toggle: 0.633 → 1.4e-14 # topo-fermion (quantum): cargo run --release rotor # M×P = 1.00000; fermion = P/8 cargo run --release polyakov # deconfinement at 1.00 < βc < 1.05 cargo run --release toggle # V = (−1)^F: five exact zeros (4⁷ states) cargo run --release exchange # 2D calibration: (e,m,ε) = (+1,+1,−1) cargo run --release framed # 3D: 7/7, the dyon is a fermion