This is an earlier experiment. Read the amendments below before treating its findings as current.

The question

Does straggling grow with heading-copying strength, and does it start at lane restarts (craft reaching a lane's end restart at its start)?

What was tested

Runs on disk: six copying strengths, ideal craft (instant response) and wing, 2 layouts each; excursion starts matched to restarts and to the 3 crossings per lane.

What was found then

For the ideal craft, heading spread grows steadily with copying strength. The distance holding 90 % of craft stays near 83–92 m to quarter strength, then climbs to 226 m, read then as copying doing two things. The wing's spread is flat to half strength, then doubles, and it loses more with copying off (154 against 100). 55–60 % of ideal-craft excursions began within 2 minutes of a restart, a third in mid-lane. The first crossing after a restart starts excursions several times as often as the middle one does. Restarted craft held 38–53 % of far-off-lane time, against 39–40 % of flying time.

What we learned later

LATER / CYCLE 103

Narrowed: two curve shapes describe the data but do not prove two mechanisms. Leaving out excursions carried over from before a restart, about half (not 55–60 %) begin after one.

research/envelope.md:4991

LATER / CYCLE 104

Withdrawn: the extra starts at the first crossing do not split stragglers into restart-caused and genuine ones.

research/envelope.md:4940

LATER / CYCLE 105

Counted by episode, restarted craft hold 41–46 % of far-off-lane time against 39–40 % of flying time: about their fair share.

research/envelope.md:4791

What changed

Nothing.

Numbers and source references
0 / 0.125 / 0.25 / 0.5 / 0.75 / 1
copying strengths testedresearch/envelope.md:4991
2
random starts per strengthresearch/envelope.md:4991
3
crossings along each laneresearch/envelope.md:4940
3.8° → 24.3°
ideal craft heading scatter from copying off to fullresearch/envelope.md:4991
83 / 87 / 92 / 110 / 153 / 226 m
90th percentile off-lane distance across the six strengthsresearch/envelope.md:4991
154 vs 100
wing craft lost per run, copying off vs fullresearch/envelope.md:4991
55–60 %
ideal-craft excursions starting within 2 min of re-placement (as registered)research/envelope.md:4991
a third (31–36 %)
excursions starting in mid-lane flightresearch/envelope.md:4991
1.7–4.1×
excursion-start rate at the first crossing relative to the middle oneresearch/envelope.md:4882
38–53 % vs 39–40 %
re-placed craft's share of far-off-lane time vs of flying timeresearch/envelope.md:4882
Original structured record ↗