What changed in model 2.0
The original index published one optimistic point estimate based on a constant 35-watt baseline, one conductance value and a fixed chiller COP. Model 2.0 publishes three setup scenarios, uses daily low/high profiles to estimate positive cooling degree-hours, reduces modeled COP in hotter conditions and includes explicit allowances for omitted loads.
The result is a planning range. It is not a statistical confidence interval and has not yet been calibrated against a large measured field dataset.
The main finding
Climate is not the only driver. A mild city can cost more than a hotter city when its state residential electricity rate is much higher. ArcticSoak therefore publishes physical energy beside dollar cost and labels the geographic resolution of both inputs.