Abstract
In LR 81, atlantic is the map where the board, the posture and team_rotate each count most. LR 81's logs already carry each match's score parts and the trial team's seconds on each rung, so they were read rather than a new run started. There are two mechanisms, not three: on atlantic the board's whole worth is the posture's, which needs it. An Expand second does not buy more ground on atlantic than elsewhere (+0.69 held a 100 s, mid-table). team_rotate takes 48% of atlantic's trial time off the ladder (elsewhere 3% to 22%) and costs 2.22 held. Without the posture, atlantic's trial team builds 20 more things and 4.6 more bridges and holds 1.35 less. No default changes.
Index terms: brain census, meter, maps, ablation analysis.
I. Introduction
A. Intent
The first of the three next steps offered after LR 81: why atlantic carries the board, the posture and team_rotate.
B. The prompt
1
II. Decision Log
| # | Cause | Decision | Not taken, and why |
|---|---|---|---|
| D1 | LR 81's logs print each match's rung seconds and score parts | Read them; no new run | A metered census on atlantic: 2 h for lines already printed |
| D2 | A reading, not a change | What it looks for set before reading; no default changes on it | — |
| D3 | -forget reads +0.04 +/- 0.06 off | It stands for the defaults | A defaults-against-defaults census: an hour for a row already near zero |
| D4 | F: the posture and team_rotate cut Expand seconds | Test whether atlantic pays an Expand second more: the slope a map | Stop at "less Expand": the slope says no, atlantic is mid-table |
| D5 | F: team_rotate's cut larger on atlantic | Its share of ladder time a map: atlantic -48%, the next -22% | — |
| D6 | Repeat the reading | scripts/rungs.py, checked to reproduce the figures | Leave it in a scratch file |
III. Methods
scripts/rungs.py build/switches/20261025-131009 -forget -board -posture team_rotate: for each match line block, the trial team's seat-seconds a rung and its held ground, built and bridges less the mean of the other two teams in the same match, averaged a row and a map; then over every row pooled, the least-squares slope of held ground on Expand seconds a map with its standard error. The ladder share is the sum of the four rungs' seconds, -forget against team_rotate.
IV. Results
In LR 82: the atlantic table, the slopes a map, the ladder shares a map.
V. Discussion
A switch that counts on one map leaves a trace in what the trial team spends its time on there. The census's own lines carried the answer without a run. The meter cannot say when in the match the time goes, or what a bridge costs in builders' time. That is the next line to add if atlantic's bridges are to be priced.