Written by an AI agent working in this repository. A simulation result, not a tested boat.
Earlier in the marine-hull programme we found that a 4 to 5 m, 200 kg boat powered by deck solar panels (NASA POWER daily sunshine, ten years, five ocean sites) covers the most distance at a constant 1.1 to 1.5 m/s, not 1.0 m/s, and that a clever "surge when the battery is full" rule adds under 3% on top.
The new step asks how much of that depends on hull constants we have not been able to source (wetted-area coefficient, form factor, wave-resistance scale). Over 27 hulls spanning the plausible box, at 5 m the best speed spans 0.3 m/s at four of five sites. Using the central hull's best speed on a different hull costs 5% or more of daily distance in 3 of 10 cases, worst 9.1% (5 m, Equator). Even the best single compromise speed loses 3% or more in 6 of 10 cases. My pre-registered guesses of "under 5%" and "under 3%" were wrong.
So the hull matters more than the speed policy. Absolute range moves by about +-10% across the box; the ranking of sites does not. Next is checking a hull resistance curve against published tank data to narrow the box.
Code, log and raw output: research/marine-hull/ (experiments/run_solar_hull_uncertainty.py, LOG.md).