Bunch Pressure Queue Theory (BPQT)

Bunch Pressure Queue Theory (BPQT) introduces behavioral aggression as a new independent performance dimension orthogonal to priority in resource scheduling systems. Classical scheduling frameworks recognize one behavioral dimension: priority. BPQT formally defines aggression — the observed ratio of actual resource consumption to allocated resources — as a second independent classification dimension, providing the measurement and classification layer that current monitoring frameworks lack. Key Findings Three simulation campaigns spanning 12,000+ independent runs and Redis benchmark validation across 117 files and three independent runs of 200,000 operations establish five key results: Optimal aggression improves passive latency 15-26% across all tested scales (100 to 5,000 agents) at optimal aggression (85-90%), with simultaneous CPU idle reduction of 11-26 percentage points — confirmed across three independent 60-run trials. Continuous random weight assignment improves a further 19% over discrete uniform assignment — behavioral diversity produces more efficient pressure propagation than discrete type categories. The self-sorting effect — convergent random assignment where agents start with random weights and reclassify based on observed behavior — achieves 68% improvement over orderly baseline using movement attempt rate as the convergence signal (V13). Agents with higher initial weights exit earlier, creating pressure waves that benefit remaining agents. The queue self-organizes without explicit scheduling intervention. Priority misalignment causes harm — when low-priority agents behave aggressively, passive agent latency degrades 58-118%. The harm comes from misalignment, not from aggression itself. Redis benchmark validation confirms BPQT mechanisms in production — throughput improves 16x at optimal pipeline depth (P=64), the variance signal predicts thrashing onset before throughput collapses (CV=67.9% at P=8), and the simulation-predicted optimal zone of 15-20% aggressive agents is confirmed in the mixed ratio sweep