carlok — zsh — 88×30

cat _posts/2026-11-04-red-ink-sprint-28-foes-read-a-string-of-jabs-and-the-bout-table-is-retuned-to-hold.md

red-ink sprint 28: foes read a string of jabs, and the bout table is retuned to hold

A jab in red-ink lands 0.10 s after it starts, and five of the six foes wait cfg.react — 0.24 to 0.445 s at Normal — before they answer anything, so they could never answer one: about 1,500 to 2,200 jabs per 100 fights went through against Dag, Pike, Voss, Chen and Halvorsen with none blocked or dodged. balance-run can now say that out loud: strike() counts the player’s jabs by how the foe received them (jabHit, jabBlock, jabDodge in G.log) and each row prints jabAnswered, the share blocked or dodged. Baseline, 200 fights per bout on seed 1: 30.4 / 0 / 0 / 0 / 0 / 0 percent — only Reyes, whose 0.08 s reaction is under a jab’s 0.20 s, ever answers one (2147429).

The fix is JAB_READ {mem: 2, f: .4} in aiIntent: a foe that has seen the player attack in the last 2 s is ready for the next jab and answers it at once, with his own block and dodge odds scaled by 0.4. The first jab of a string is still free, a foe whose reaction is already fast enough is left untouched so bout 1 remains the control, and f: 0 turns the read off. It moved the table, so health was retuned against the old numbers — Dag 115 → 105, Pike 120 → 111, Voss 175 → 150, Chen 152 → 131, Halvorsen 122 → 116 — with the second-pass cooldowns for Dag and Chen at ×0.75 and ×0.7 (bc84dec, #25).

At 200 fights, seed 1, Normal, the first pass reads 90.5 / 69.0 / 40.5 / 37.0 / 30.5 / 21.0 against 90.5 / 67.5 / 44.0 / 36.0 / 30.0 / 22.0, the second pass 71.5 / 54.5 / 14.5 / 17.0 / 12.0 / 11.0 against 71.5 / 52.0 / 16.5 / 16.0 / 12.0 / 11.5, and the jabs answered go to 30.4 / 7.1 / 4.4 / 13.3 / 10.3 / 4.0. The pull request is explicit that the old numbers were not chased exactly: one health point moves a bout by up to 10 points, so a match on seed 1 would fit the sample rather than the game, and at 400 fights over seeds 1 and 7 the means sit within 2.7 points of the old table. Easy and Hard moved too — Hard against Pike is still 0.5% — and the new test foes_read_a_string_of_jabs fails with the read removed, taking the regression to 61 of 61 with the step probe unchanged over 900 ticks at 20 to 144 fps (41a3dff).