What longer course measurements keep—and can hide

The bounded controls are complete. They support keeping the current acceptance gate while investigating a more carefully defined motion check.

The same six files chosen for the IMU6 unit study contain five substantial recorded ground-track changes: at least 3 degrees over 30 seconds, with complete receiver timing. All five were detected at 2-, 6- and 12-second resolution in both receiver track and navigation course. These are route changes, not independent measurements of aircraft yaw. Only two of the six files supplied these controls, so this is limited evidence about larger maneuvers rather than a complete flight-domain validation.

Short real corrections can disappear under averaging. Thirty-six deterministic controls with known course changes tested durations from 0.2 to 40 seconds and rates from 0.06 to 0.5 degrees/second. The course-rate limit was held at 0.05 degrees/second.

For example, a 0.1-degree correction over one second has a true rate of 0.1 degree/second. The 12-second measurement reports a peak of only about 0.0125 degree/second and misses the limit crossing. The 2-second measurement still detects this example. These controls are deliberately noiseless course signals, not Earth-model simulations or an estimate of how often actual aircraft make each maneuver.

This explains why the 53 resolution-sensitive files cannot automatically become accepted scientific windows. A longer timescale preserves the tested substantial turns but can suppress real short corrections, just as it suppresses measurement noise. It does not prove the original excursions were artifacts.

The implementation is analysis/lll/ilvis0_motion_controls.py. The worker completed after the unit worker released its lock, keeping total numerical threads at two. Source hashes, strict packet checks and GPS timing passed again in all six. Sources, inputs and numerical environment are frozen separately in data/ilvis0-course-controls-20261007/; this directory preserves the manifest, source snapshots and complete report. The scientific gate, stored originals and frozen Earth-model studies remain unchanged.

A revised screening rule needs an explicit treatment of short aircraft motion and evidence that the downstream analysis tolerates it. Selecting 12 seconds solely because it recovers more data would not supply that evidence. No new motion interval has been adopted here.