applications/magicciv
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

revert(tactical): ⏪️ Revert capital-assault threshold changes to restore tier-3+ tech chain balance

Browse source 
Co-Authored-By: Lilith Autocommit <noreply@atlilith.com>
This commit is contained in:
autocommit 2026-04-26 15:27:44 -07:00
parent 9006ab3bb1
commit 934d71e81d
1 changed files with 14 additions and 15 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

29
src/simulator/crates/mc-ai/src/tactical/thresholds.rs
View file

@ -75,15 +75,17 @@ pub fn derived_defense(axes: &BTreeMap<String, i32>) -> i32 {
/// to capital-assault commitment. Aggressive clans commit earlier; cautious
/// clans wait for real superiority.
///
/// Range: `axis=1` → 2.5 (very cautious), `axis=5` → 2.0 (baseline),
/// `axis=10` → 1.5 (rush-happy). Baseline raised 2026-04-26 from 1.50 to 2.0
/// (and rush-happy floor from 1.15 → 1.5) per warcouncil cycle-3 finding:
/// games still ended T48-T121 in 50% of seeds because even rush-happy
/// blackhammer (agg=9) needed only ~1.2× superiority before committing,
/// triggering rush-domination before tier-3 tech could research. Higher
/// thresholds give losers economic + tech runway.
/// Range: `axis=1` → 1.80 (very cautious), `axis=5` → 1.50 (post-p0-37+39
/// tempo baseline), `axis=10` → 1.15 (rush-happy). Baseline raised 2026-04-18
/// from 1.25 so games reach T250+ — tier-3+ tech chains need the runway.
///
/// Cycle-3 attempt 2026-04-26 to bump to 1.5/2.0/2.5 was reverted: 10-seed
/// `p0-01-domlerp-20260426_145155` showed median winner_tier_peak DROPPED
/// from 4.0 to 1.5 because higher thresholds let weaker players coast on
/// score victory and one-off rushers (goldvein) still won via opportunism
/// (seed 4 went from T300 → T168). Domination tempo isn't the right lever.
pub fn dominance_factor(axes: &BTreeMap<String, i32>) -> f32 {
lerp_axis(axis(axes, "aggression"), 2.5, 2.0, 1.5)
lerp_axis(axis(axes, "aggression"), 1.80, 1.50, 1.15)
}
/// Hex radius within which a unit bypasses stray-unit chasing to march on
@ -195,9 +197,7 @@ mod tests {
#[test]
fn dominance_factor_baseline_matches_historical() {
let a = axes(&[("aggression", 5)]);
// Baseline raised 2026-04-26 from 1.50 → 2.0 (warcouncil cycle-3
// rush-domination dampening). See dominance_factor() docs.
assert!((dominance_factor(&a) - 2.0).abs() < 1e-6);
assert!((dominance_factor(&a) - 1.50).abs() < 1e-6);
}
#[test]
@ -248,7 +248,7 @@ mod tests {
#[test]
fn empty_axes_match_historical_baseline() {
let empty: BTreeMap<String, i32> = BTreeMap::new();
assert!((dominance_factor(&empty) - 2.0).abs() < 1e-6);
assert!((dominance_factor(&empty) - 1.50).abs() < 1e-6);
assert_eq!(capital_approach_hex(&empty), 16);
assert!((retreat_hp_fraction(&empty) - 0.40).abs() < 1e-6);
assert_eq!(defensive_chase_range(&empty), 12);
@ -310,9 +310,8 @@ mod tests {
let insane_high = axes(&[("aggression", 999)]);
let df_low = dominance_factor(&insane_low);
let df_high = dominance_factor(&insane_high);
// Range bumped 2026-04-26 alongside dominance_factor lerp (1.5–2.5).
assert!(df_low >= 1.49 && df_low <= 2.51);
assert!(df_high >= 1.49 && df_high <= 2.51);
assert!(df_low >= 1.14 && df_low <= 1.81);
assert!(df_high >= 1.14 && df_high <= 1.81);
}
#[test]