This is an earlier experiment. Read the amendments below before treating its findings as current.
The question
The design predicted that matching neighbours' speed and direction would be the main brake on a spreading disturbance. We measure the share of a disturbance each craft passes on to the next; below 1 it dies out.
What was tested
On the same shoved grid of hovering craft, we turned the strength of speed-matching from 0 up to 8.
What was found then
The share passed on fell steadily, from 0.85 with no matching to 0.55 at full strength. It always stayed below 1, the line where a disturbance would start to grow. The last layout's summary showed no close calls at any strength. So speed-matching is the brake, as predicted.
What we learned later
The close-call counts came from the last layout only, a reporting bug found by the 2026-09-08 outside review and then fixed; these runs were not repeated.
README.md:1244; git a17e672
What changed
Nothing.
Corrections retained in the record
Earlier wording: α falls monotonically from 0.85 to 0.56
Record correction: The recorded value at alignment 8 is 0.5547, i.e. 0.55 (the README of the same commit also printed 0.56).
results/results.json e2_align_sweep
Numbers and source references
- 0 to 8
- alignment strength sweptresults/results.json e2_align_sweep
- 0.85
- α at alignment 0results/results.json e2_align_sweep
- 0.55
- α at alignment 8 (0.5547; the README of the time rounded it to 0.56)results/results.json e2_align_sweep
- 1
- the stability line for α (definition)research/flight_plan.html (Rev B); definition of string stability
- 0
- breaches at every alignment value (last-seed summary)results/results.json e2_align_sweep