Commander Intent Compilation, Automated DAG Task Decomposition and Zero-Courier Multi-Agent Dispatch: chapter answers
The popular belief holds that once a high-level goal is handed to a clever enough orchestrator, it decomposes the goal into a task graph by itself, sub-agents claim the tasks and retry their failures by themselves, the commander looks only at results, and nobody need carry messages between windows again. This book falsifies the reading of that belief as "zero human". In an illustrative simulation of batches of four volumes, an intent compiler plus dispatcher needs on average 9.951 human interventions per batch, 0.830 fewer than the 58.4 hand-carried messages of the old relay; but irreversible…
DOI pending · monograph page goes live after publication · 5DL-B-407 · 中文版
- Chapter 1 · Intent Compilation: a Five-Slot Rule Compiler, Asking for Missing Slots and Hierarchical Decomposition
But as soon as a tenth of the empty slots govern irreversible steps, the picture reverses. In the third row, with the same wrong-guess rate of 0.3, the mean cost of guessing rises to 7.871 and the worst single case is 4… - Chapter 2 · The Directed Acyclic Task Graph: Kahn's Topological Sort, the Critical Path and Monte Carlo Durations
Chapter 2 detects cycles with Kahn's sort and computes float by the critical path method; a deterministic plan of 17.4 hours is met with probability only 0.0213 under Monte Carlo. - Chapter 3 · Multi-Agent Claiming: Contract-Net Bidding, Graham's List Scheduling and Timing Anomalies
Chapter 3 covers contract-net claiming and Graham's list scheduling and reproduces a timing anomaly: three executors finish in 12, four need 15. - Chapter 4 · Resource Locks: a Global Order against Deadlock, Leases for Orphaned Locks and Fencing Tokens
Chapter 4 shows a global locking order driving simulated deadlock rates of 0.1795 and 0.8655 to zero, and recovers orphaned locks with leases and fencing tokens. - Chapter 5 · Automatic Recovery after Failure: Idempotency Keys, Exponential Backoff with Jitter, Retry Budgets and Circuit Breakers
Chapter 5 shows blind retries causing 532 duplicates in ten thousand operations, and retries every second sending 147,693 calls during a thirty-second outage; none of the four guard rails can be dropped. - Chapter 6 · Exception Escalation: Human Interventions, Budget Overrun, Deadline Probability and Little's Law
Exceptions fall into four classes (irreversible actions, budget overruns, three consecutive acceptance failures and lock-wait timeouts), plus a catch-all for unclassified errors. Each report carries five slots of contex… - Chapter 7 · Eastern and Western Sources Examined: Division into Units in Sunzi's Chapter on Momentum and Smith's Pin Factory
A flat organisation reaches its peak output of 13.00 at twenty-five people. At fifty people only 1.00 remains, and beyond that output is negative: communication consumes all the time there is. Three hundred people in a… - Chapter 8 · Field Tools and Checker Specifications: the Dispatch Ledger, the Intent Compilation and Task-DAG Dispatch Checker, and a Pre-Registered Test
Chapter 8 gives the dispatch ledger, the Intent Compilation and Task-DAG Dispatch Checker, a checker specification table grouped by the four kits, and a pre-registered test: non-inferiority of the error rate needs 1,188…
Illustrative model calculation, not a live quote