Time a tap from touch-down, and give the opponents' legs a knee

Oliver after step 1: much harder, stuck at the Steward, "trotz gutem
Timing nicht mehr bekommen". A tap only proves it is a tap when the
finger lifts, so the swing started 100-150 ms after he meant it; the old
volley window hid that, step 1's groundstrokes did not. A tap made on the
gold ring fell from q 0.82 to 0.09 (and a third were missed at 150 ms),
murky balls landed short and the opponent came in. The swing clock now
starts at the touch's age (rally modes only); flicks keep their crossing.

The opponents' legs had no ceiling: 7.3 m/s at the Runner, 15.7 at the
Champion against the player's ~6.2. Past legKnee (6 m/s) only 15% of the
growth survives, monotone, so the ladder order holds; champion ~7.5. v362.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

PROJECT.md

@@ -5262,6 +5262,30 @@ late opponent at the net covers every pass, so the bot loses ~95% of the points
 comes in. That is step 4 (and the passing/lob game of steps 2-3); it is deliberately not mixed into
 step 1, so its phone verdict reads one change.
 
+### Step 1's phone verdict, and the two fixes it surfaced (2026-10-14, v362)
+Oliver: *"schritt 1 ist schon gut ... Das Spiel ist momentan wesentlich schwerer geworden (not a bad
+thing) — ich komme als geübter Spieler über the Climb vielleicht höchstens bis zum Steward. das liegt vor
+allem gefühlt an der Geschwindigkeit und manchmal scheine ich den Ball trotz gutem Timing nicht mehr zu
+bekommen."* Two real defects, measured, and neither of them was a balance value:
+- **A TAP SWUNG WHEN THE FINGER LIFTED.** A touch only proves it is a tap at touchend (it could still
+  become a flick), so the swing started 100-150 ms after the player meant it. The volley window used to
+  absorb that (x1.8 window, x1.4 reach). Step 1 made the rally groundstrokes, and the lag became visible.
+  Ring-tapping bot at opponent 6, finger down exactly on gold:
+  0 ms of tap = 100% returned at q 0.82, 100 ms = 95% at **q 0.09**, 150 ms = **40%** returned.
+  Murky returns land short, a short ball pulls the opponent in, and his legs then made him unpassable: a
+  cascade, felt as "speed". **Fix:** input.js hands the touch's age to `keyPressed('Space', lead)`, and
+  the rally swing clock starts at `lead` (rally modes only). Timing is judged from touch-down; the ball
+  still has to arrive inside the window. Flicks keep their crossing-fired swing (the left-flick saga).
+  After the fix: 100 ms taps 99-100% at q 0.96. Verified with real touch events in the pane (a 120 ms
+  tap starts the swing at 0.122 s). Test: a gold-ring tap with 120 ms on top returns >= 90% at q >= 0.6
+  (it fails at q 0.08 without the fix).
+- **THE LEGS GET A KNEE** (step 4, first cut). `gameplay.knee`: up to `legKnee` (6.0 m/s) nothing
+  changes; past it only `legSoft` (15%) of the growth survives. Runner 7.3 -> 6.2, Umpire 8.3 -> 6.3,
+  Coach 10.6 -> 6.7, Champion 15.7 -> ~7.5. The player is ~6.2 at level 1 and gets more with level and
+  peer gifts. The mapping is monotone, so the ladder order and the like-for-like speed tests hold. Only
+  the rig guard (`> 7 m/s`) moved, to `> legKnee*0.8`. Rungs 1-7 are unchanged: the Steward (6) already
+  ran 5.6. Ball PACE is untouched; it is at full pace times the league tier from rung 5.
+
 # OPEN — leftovers, in rough priority order
 
 Nothing here is broken; these are known gaps and unverified risks.

index.html

@@ -1140,6 +1140,6 @@ if (location.hostname !== 'splatter-tennis.wuwi.one' && !window.Capacitor?.isNat
   </div>
 </div>
 <span id="recLamp" hidden>REC</span>
-<script type="module" src="src/main.js?v=361"></script>
+<script type="module" src="src/main.js?v=362"></script>
 </body>
 </html>

src/config.js

@@ -578,6 +578,9 @@ export const TUNE = {
                         // the peer gift; reaching parity earns all of it. Never a penalty.
   peerRun: 0.08,        // parity adds 8% legs on top of the permanent level bonus
   peerWindow: 0.12,     // ...and 12% contact window. Skill still wins; grinding opens the margin
+  legKnee: 6.0,     // m/s past which an opponent's legs stop growing at full rate (gameplay.knee) —
+                    // the player's auto-run is ~6.2. Was uncapped: the champion ran 15.7 m/s
+  legSoft: 0.15,    // ...and the share of the growth past the knee that survives (champion ~7.5)
   matchSpeedBase: 1.25, // racketAI speed multiplier for opponent ONE. 1.4 was calibrated against
                         // the broken recovery (the player never walked back to centre, so every
                         // wide return looked like opponent strength) — with real footwork restored

src/gameplay.js

@@ -1198,7 +1198,15 @@ const matchStep = () => {
   const n = G.match.opp - 1;
   return n + (!G.net && !G.match.cup && !G.match.train ? 4 * Math.max(0, 1 - n / 7) : 0);
 };
-const rzChaseSpeed = rz => TUNE.racketSpeed * leaguePace()
+// THE LEGS HAVE A KNEE (2026-10-14). The ladder multiplied speed rung on rung with nothing on top:
+// 7.3 m/s at the RUNNER, 10.6 at the HEAD COACH, 15.7 at the CHAMPION — a sprinter's top speed and
+// 2.5x the player's ~6.2. Under step 1's geometry a late opponent at the net then covered every pass.
+// Up to legKnee nothing changes; past it only legSoft of the growth survives, so the order of the
+// ladder (and every like-for-like speed test) holds while the champion lands near ~7.5. Wrong-footing,
+// the read, placement and consistency are what climb past it now.
+const knee = v => v <= TUNE.legKnee ? v : TUNE.legKnee + (v - TUNE.legKnee) * TUNE.legSoft;
+const rzChaseSpeed = rz => knee(rzRawSpeed(rz));
+const rzRawSpeed = rz => TUNE.racketSpeed * leaguePace()
   * (G.match ? (TUNE.matchSpeedBase + matchStep() * TUNE.matchSpeedPer) * ((rz.foe && rz.foe.speed) || 1)
              : 1 + TUNE.duelSpeedMax * duelSkill())
   * (G.wasServe && ball.lastHit === 'P' ? TUNE.aiReturnDash : 1)
@@ -1750,7 +1758,7 @@ function meleeCooldown() {
                   TUNE.meleeCool / (1 + Math.max(0, near - 1) * TUNE.meleeCrowd));
 }
 
-export function keyPressed(code) {
+export function keyPressed(code, lead = 0) {
   // THE SPRITE BENCH takes the key before anything else reads it: it is a back-room art test with no
   // ball and no scoring, so every other meaning a key has here is meaningless inside it.
   if (G.bench) return benchKey(code);
@@ -1840,8 +1848,15 @@ export function keyPressed(code) {
     return;
   }
   if (G.phase === 'rally' && P.swingT < 0) {
-    recordEvent('swing', { window: TUNE.swingWindow * winLevel(), assist: assistK() });
-    P.swingT = 0; P.pose = 'swing'; P.poseT = 0; G.swung = true;
+    recordEvent('swing', { window: TUNE.swingWindow * winLevel(), assist: assistK(), lead });
+    // A TAP COUNTS FROM THE MOMENT THE FINGER LANDS (rally modes). A tap is only known to BE a tap
+    // when the finger lifts — it could still turn into a flick — so it used to swing ~100-150 ms
+    // after the player meant it. Inside step 1's groundstroke window that collapsed the contact
+    // quality of a tap made exactly on the gold ring from 0.82 to 0.09 (Oliver, 2026-10-14: "trotz
+    // gutem Timing nicht mehr bekommen"). input.js hands the touch's age in as `lead`, and the swing
+    // clock starts that far along: the timing is judged from touch-down, the ball still has to be
+    // there for the contact.
+    P.swingT = rallyFeel() ? lead : 0; P.pose = 'swing'; P.poseT = 0; G.swung = true;
   }
 }
 

src/input.js

@@ -223,7 +223,7 @@ const endTouch = e => {
       // (a bare hold used to fire at a free auto-target; that was the whole exploit)
       if (aim && dist >= TUNE.pullMin) dragServe(aim);
     }
-    else if (!stick.moved && dt < 250) { if (!stick.nearRack) keyPressed('Space'); } // tap — dead if it was a near-miss at the rack
+    else if (!stick.moved && dt < 250) { if (!stick.nearRack) keyPressed('Space', dt / 1000); } // tap — dead if it was a near-miss at the rack; timed from touch-down (keyPressed)
     else if (!stick.adopted && dist > TUNE.flickMin
              && (stick.swung || dt < FLICK_MS || (rallyFeel() && G.phase === 'rally'))) {
       // flick = aimed swing. The FLICK_MS deadline only still applies where the swing could NOT

test.js

@@ -920,7 +920,8 @@ test('league 5 has no speed dip below league 4, including serve and wrong-footed
     const fourth = cfg.leagueOf(4).band, fifth = cfg.leagueOf(5).band;
     let floor = 0;
     for (let n = fourth[0]; n <= fourth[1]; n++) floor = Math.max(floor, measure(n, serve, surprised));
-    assert.ok(floor > 7, 'the rig measures a real full-speed chase');
+    // since the legs have a knee (gameplay.knee) the late ladder runs at ~6-7.5 m/s, not 7-15
+    assert.ok(floor > TUNE.legKnee * 0.8, 'the rig measures a real full-speed chase: ' + floor.toFixed(2));
     for (let n = fifth[0]; n <= fifth[1]; n++) {
       const speed = measure(n, serve, surprised);
       assert.ok(speed >= floor, `${cfg.matchFoe(n).short}: ${speed} must clear ${floor} (serve ${serve}, surprised ${surprised})`);
@@ -1136,6 +1137,38 @@ test('a SHORT ball is an approach, a deep one keeps the baseliner back, and nobo
   } finally { TUNE.matchFlub = flub; }
 });
 
+test('a TAP is timed from touch-down: a normal 120 ms tap on the gold ring still returns cleanly', () => {
+  // Oliver, 2026-10-14: "manchmal scheine ich den Ball trotz gutem Timing nicht mehr zu bekommen".
+  // A tap only proves it is a tap when the finger LIFTS, so it swung ~100-150 ms late — inside the
+  // groundstroke window that took a tap made exactly on the gold ring from q 0.82 to 0.09, and at
+  // 150 ms a third of the balls were simply missed. keyPressed's `lead` starts the swing clock at
+  // touch-down; this is the ring-tapping player, with a real finger's 120 ms of tap on top.
+  const assist = TUNE.autoAssist; TUNE.autoAssist = 1.15;
+  try {
+    startMatch(); G.interlude.introT = 0; G.match.opp = 6;
+    for (let i = 0; i < 2.2 * 60; i++) update(DT);
+    const LAG = 7; // frames: ~120 ms between the finger landing and it lifting
+    let fire = -1, armed = false, incoming = 0, returned = 0, sumQ = 0;
+    for (let i = 0; i < 90 * 60; i++) {
+      G.modal = null; G.modalQ.length = 0; G.hp = G.hpMax; G.match.opp = 6;
+      if (G.phase === 'pretoss' && !(G.interlude && G.interlude.introT > 0)) { if (i % 40 === 0) keyPressed('Space'); }
+      else if (G.phase === 'toss') { if (ball.vy < 0.4 && ball.y > 2.4) keyPressed('Space'); }
+      else if (G.phase === 'rally' && ball.active && ball.lastHit === 'O' && ball.vz > 0 && !armed) {
+        const d = Math.hypot(ball.x - P.x, (ball.z - P.z) * 0.8);
+        if (Math.abs(d - TUNE.matchRingSweet) < TUNE.matchRingBand) { armed = true; incoming++; fire = i + LAG; }
+      }
+      if (i === fire) keyPressed('Space', LAG / 60);   // the lift: the touch's age rides along
+      const before = ball.lastHit;
+      update(DT);
+      if (armed && before === 'O' && ball.lastHit === 'P') { returned++; sumQ += ball.pow; armed = false; }
+      else if (armed && (!ball.active || G.phase !== 'rally')) armed = false;
+    }
+    assert.ok(incoming >= 30, 'a real sample: ' + incoming);
+    assert.ok(returned / incoming >= 0.9, `a gold-ring tap returns the ball: ${returned}/${incoming}`);
+    assert.ok(sumQ / returned >= 0.6, 'and returns it cleanly: mean q ' + (sumQ / returned).toFixed(2));
+  } finally { TUNE.autoAssist = assist; }
+});
+
 test('rally depth: deep by default, mistiming lands it SHORTER, never longer', () => {
   // the only out a rally ball may find is wide — depth must never put one past the baseline
   startMatch(); G.interlude.introT = 0;