diff --git a/poc-memory/src/neuro/scoring.rs b/poc-memory/src/neuro/scoring.rs
index 3a74e74..39d26fd 100644
--- a/poc-memory/src/neuro/scoring.rs
+++ b/poc-memory/src/neuro/scoring.rs
@@ -181,18 +181,36 @@ impl ConsolidationPlan {
         if self.run_health {
             runs.push(("health", 0));
         }
-        for (name, count) in [
-            ("replay", self.replay_count),
+        // Build per-type batch lists, then interleave so different agent
+        // types alternate rather than running all-replay-then-all-linker.
+        let types: [(&str, usize); 4] = [
             ("linker", self.linker_count),
+            ("replay", self.replay_count),
             ("separator", self.separator_count),
             ("transfer", self.transfer_count),
-        ] {
-            let mut remaining = count;
+        ];
+        let mut queues: Vec<Vec<(&str, usize)>> = types.iter().map(|(name, count)| {
+            let mut q = Vec::new();
+            let mut remaining = *count;
             while remaining > 0 {
                 let batch = remaining.min(batch_size);
-                runs.push((name, batch));
+                q.push((*name, batch));
                 remaining -= batch;
             }
+            q
+        }).collect();
+
+        // Round-robin interleave
+        loop {
+            let mut added = false;
+            for q in &mut queues {
+                if let Some(run) = q.first() {
+                    runs.push(*run);
+                    q.remove(0);
+                    added = true;
+                }
+            }
+            if !added { break; }
         }
         runs
     }
@@ -239,28 +257,32 @@ fn consolidation_plan_inner(store: &Store, detect_interf: bool) -> Consolidation
 
     // Target: α ≥ 2.5 (healthy scale-free)
     if alpha < 2.0 {
-        plan.replay_count += 10;
-        plan.linker_count += 5;
+        plan.replay_count += 50;
+        plan.linker_count += 100;
         plan.rationale.push(format!(
-            "α={:.2} (target ≥2.5): extreme hub dominance → 10 replay + 5 linker",
+            "α={:.2} (target ≥2.5): extreme hub dominance → 50 replay + 100 linker",
             alpha));
     } else if alpha < 2.5 {
-        plan.replay_count += 5;
-        plan.linker_count += 3;
+        plan.replay_count += 25;
+        plan.linker_count += 50;
         plan.rationale.push(format!(
-            "α={:.2} (target ≥2.5): moderate hub dominance → 5 replay + 3 linker",
+            "α={:.2} (target ≥2.5): moderate hub dominance → 25 replay + 50 linker",
             alpha));
     } else {
-        plan.replay_count += 3;
+        plan.replay_count += 10;
+        plan.linker_count += 20;
         plan.rationale.push(format!(
-            "α={:.2}: healthy — 3 replay for maintenance", alpha));
+            "α={:.2}: healthy — 10 replay + 20 linker for maintenance", alpha));
     }
 
     // Target: Gini ≤ 0.4
+    // High Gini means degree inequality — most nodes under-connected.
+    // Linker fixes this by adding edges to low-degree nodes.
     if gini > 0.5 {
-        plan.replay_count += 3;
+        plan.replay_count += 10;
+        plan.linker_count += 50;
         plan.rationale.push(format!(
-            "Gini={:.3} (target ≤0.4): high inequality → +3 replay",
+            "Gini={:.3} (target ≤0.4): high inequality → +10 replay + 50 linker",
             gini));
     }