This is an earlier experiment. Read the amendments below before treating its findings as current.
The question
This round drew three general conclusions from the maps of one condition at a time.
What was tested
We read across all the conditions tested.
What was found then
Every condition that moved the safety limit was about time: response lag, the age of the radio news, the time to the next conflict. Every condition about force (acceleration, wind, gusts, storm) showed up as lost deliveries instead. Lag and old news add up: with lag, the link's safety limit becomes a gradual slope instead of a sudden step. Throughput is a safety signal too: three cases looked perfectly safe only because the fleet had stopped working. All three conclusions hold only for these scenarios and vehicles, the ones the rules were tuned on.
What we learned later
With six layouts, turbulence 3 came in at about 3 times calm (a weak result), so 'force costs only deliveries' no longer held for turbulence.
research/envelope.md (What multiplies the vehicle penalty)
This summary was replaced by the combined safety map in cycle 113.
research/envelope.md (The envelope, in one place — SUPERSEDED)
What changed
Nothing.
Numbers and source references
- three
- cells that looked safe because the fleet had stopped: wind 9 at cruise 8, wind 12, lagged craft at wind 9research/envelope.md@9d286de (The three things to carry out of this)