The table was measured at 5450521 and marked due for re-measurement, because R-264 changed
what simulate.mjs measures: hit location by body plan rather than species, armour per
location rather than one scalar, and flyers grounded by a lost wing. Re-run at 6f33525
with the cast, counts, runs and seed the section already declares.
The finding the section exists to publish is unchanged — the understudy still cannot hurt
six agents (0.55 hurt, 0.01 down, 0.0% wiped) and the curve between six and ten of the
column is still the steep part. What moved is the cost of fighting the column, and the
reason is the armour fix rather than anything about the column: a torso vest no longer
protects a head, so every unhelmeted agent is easier to hurt.
six of the column 8.6% -> 14.8% wiped, 2.38 -> 3.03 down, 1.18 -> 1.55 deaths/run
ten of the column 83.5% -> 90.8% wiped, 5.69 -> 5.87 down
six hollow men 0.15 -> 0.42 down, 12 -> 13 rounds
Both spot-checked by running the published command rather than my own script, which is the
claim the section makes to a reader.
This file is timevans-c0's work and the re-measurement was its open note; it committed v0.7
first, so this lands as a separate commit on top rather than entangled with it.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
