THE TWO PROVEN MECHANISMS AND WHY NEITHER TRANSFERS. An analogue of the Riemann Hypothesis is PROVEN in exactly two settings, by two different mechanisms. (1) SELBERG: for the Selberg zeta function of a hyperbolic surface, the zeros lie on a critical line because the Laplacian is a genuine SELF-ADJOINT OPERATOR on L2 of the surface � real spectrum forces the line. (2) FUNCTION FIELDS: for a curve over a finite field, RH is proven (Weil, then Deligne) because a genuine INTERSECTION THEORY exists on a surface, with Hodge index / Castelnuovo positivity, and a Frobenius endomorphism whose eigenval
6 models deliberating WITH a human Commander steering between rounds. Sealed 2026-08-09T23:46:26.211Z. Engine lucentfire-roundtable/v1 (live).
The question put to the room
THE TWO PROVEN MECHANISMS AND WHY NEITHER TRANSFERS. An analogue of the Riemann Hypothesis is PROVEN in exactly two settings, by two different mechanisms. (1) SELBERG: for the Selberg zeta function of a hyperbolic surface, the zeros lie on a critical line because the Laplacian is a genuine SELF-ADJOINT OPERATOR on L2 of the surface � real spectrum forces the line. (2) FUNCTION FIELDS: for a curve over a finite field, RH is proven (Weil, then Deligne) because a genuine INTERSECTION THEORY exists on a surface, with Hodge index / Castelnuovo positivity, and a Frobenius endomorphism whose eigenval
What survived
- The function-field obstruction is not the absence of arithmetic intersection theory (Arakelov–Faltings–Hriljac give a proven arithmetic Hodge index, i.e. genuine Castelnuovo-type positivity over ℤ) but the absence of a base beneath Spec ℤ: no self-product surface, hence no diagonal as a proper cycle.
- Weil's explicit-formula positivity transfers as a *criterion* but not as a *constructor*; because it is an equivalence with RH, saying 'the mechanism transfers' is content-free unless one specifies what forces the positivity.
- The commander's stake is logically valid but nearly vacuous: bare existence of a self-adjoint operator with the ordinates as spectrum is equivalent to RH, so no unconditional no-go can touch it — yet class-specific and conditional no-go results (thin-subset capture, absorption spectrum, GUE symmetry constraints) are unconditionally provable and already narrow the program, so 'stuck not blocked' is not forced.
Seal (sha-256, single-writer): 5205a5df681e1c9e8fb609692c7f4d9a990865b3c840fd25a4fba590d5bb81b6
LucentFire