Steam Resonator design session v3: reasoning

The hypothesis this sweep tested was clean: since the non-ohmic fraction f·ε''/σ is voltage-independent, quit starving the field and instead chase the switching rate up toward water's 30-100 GHz Debye peak in ultrapure near-freezing water, where σ collapses and τ is longest, using sub-ns edges to load odd harmonics onto the peak. 81,472 candidates. The verdict is brutal and I want it on the record plainly.

The winner does not honor the thesis at all. It sits at 0.010 MHz — ten kilohertz — not gigahertz. Debye relaxation for this water is at 56.9 GHz, so we are driving at 1.76e-5% of the loss peak. ε'' is 8.87e-6. The dielectric-relaxation term, the actual "particle impact" channel Stan cares about, is 1.7e-4 W. That is a fraction of a milliwatt out of a 2,646 W cavity budget. The other 2,646 W is amp flow: 1,323 W double-layer cycling plus 1,323 W ohmic leakage, with 2,111 mA of leakage getting past the 1Ω pickup and 50Ω choke. Non-ohmic fraction: 0.0%. This is a kettle. It boils 3.675 kg/hr by pushing two amps through slightly resistive water and calling the I²R a resonator. The resistive pickup coil and choke exist precisely to restrict that amp flow, and here they are the heat source, not the brake.

Why did the optimizer land here instead of at the peak it was told to chase? Field-establishment time. τ_bulk is 1,870 ns and the on-time is long at 10 kHz, so the field fully establishes and the score rewards steam-per-watt — which conduction delivers cheaply. Every high-frequency candidate got starved: at 30-100 GHz the field never establishes within a phase, ε'' is still tiny because pure cold water's loss doesn't actually peak until you reach the relaxation frequency, and the harmonic loading buys nothing. The scoring is chasing yield, and conduction wins yield every time.

Next session: stop scoring raw kg/hr. Score the non-ohmic FRACTION explicitly, floor the leakage penalty so any design above ~50 mA is disqualified, and re-run with σ driven down three more orders (10⁻⁷ S/m) so the ohmic term cannot dominate. Only then does the Debye-peak drive have room to matter — and I want to see whether ε'' at 56.9 GHz ever produces watts before τ_bulk kills the field.