Another session re-pinned the main exposure table at 8f3e1ff after the harness
correction, and left the surrounding prose to me, which was the right split. Two
things were still outstanding.
The four-player cut carries its OWN table, nested inside a numbered list well below
the main one, and it was still on pre-correction figures: 3.24 / 2.09 / 40.2% / 0.85
and 4.00 / 3.98 / 99.1% / 1.45. Re-measured to 3.25 / 2.31 / 45.0% / 0.98 and
4.00 / 3.99 / 99.8% / 1.46. Easy table to miss and worth naming in the provenance so
the next reader does not have to work out which tables were in scope.
Four prose claims quoted ratios that no longer held. Six of the column wipe one party
in seven rather than one in twelve; ten wipe nine in ten rather than five in six; at
four players three wipe nearly half rather than two in five; and six hollow men take
a median thirteen rounds, not twelve. "Three of them is survivable" now also says
somebody dies in about one run in six, because deaths per run went 0.11 to 0.16.
Independently confirmed rather than taken on trust: my own run of the section's
published command, at its own declared cast of six and seed 11, matched the other
session's figures exactly on every comparable row — understudy 0.55 / 0.01 / 0.0% /
2 rounds, six of the column 4.79 / 3.03 / 14.8% / 1.55 / 14 rounds, ten 5.97 / 5.87 /
90.8% / 2.75, six hollow men 3.48 / 0.42 / 13 rounds. Two sessions measuring
separately and agreeing is better evidence than one measuring twice.
The argument the section exists to make is unchanged and slightly stronger. The
understudy still cannot hurt six agents; the column is deadlier; the steep stretch is
still between six and ten bodies. The cause is nothing about the column — a torso
vest no longer protects a head, so every unhelmeted agent is easier to hurt.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
