Glossary

Quick reference for 108 terms across 7 categories used in the Herding Cats in the AI Age series. Auto-linked on first mention per section in each paper.

108 terms across 7 categories: AI / Agent 21 · Military 27 · Reference 6 · LSS / Quality 22 · Person 13 · Series 10 · Process 9

Jump to letter

A · B · C · D · E · F · G · H · I · K · L · M · O · P · Q · R · S · T · W · Y

Browse by category

A

A2A

AI / Agent

Google’s April 2025 protocol (donated to Linux Foundation) standardizing how AI agents communicate with each other. Handles horizontal agent-to-agent relationships. Complements MCP.

Also: Agent2Agent Protocol, Agent2Agent

See also: MCP · Agent Card · AAIF

First introduced in: Paper 1 the super intelligent five year old

External: https://a2a-protocol.org/latest/specification/

AAIF

AI / Agent

Linux Foundation directed fund announced December 2025, housing three founding projects: Anthropic’s MCP, Block’s goose framework, and OpenAI’s AGENTS.md standard. Platinum members include AWS, Anthropic, Block, Bloomberg, Cloudflare, Google, Microsoft, OpenAI.

Also: Agentic AI Foundation

See also: MCP · A2A

First introduced in: Paper 1 the super intelligent five year old

AAR

Military

The Army’s structured reflection process using four questions: What did we plan to do? What actually happened? Why did it happen? What do we do differently? Used in the series as the model for AI system learning loops; codified in TC 25-20.

Also: After Action Review, After-Action Review

See also: CPI · DMAIC · TC 25-20

First introduced in: Paper 1 the super intelligent five year old

ADP 5-0

Reference

Army Doctrine Publication: The Operations Process (2019). Defines METT-TC, the operations process (plan-prepare-execute-assess), and the relationship between commander and staff.

Also: ADP5-0

See also: METT-TC · FM 5-0 · Battle Rhythm

First introduced in: Paper 1 the super intelligent five year old

ADP 6-0

Reference

Army Doctrine Publication: Mission Command. Establishes commander’s intent, disciplined initiative, and mutual trust as the foundation of decentralized execution. Doctrinal source for the Mission Command pattern applied to AI agents.

Also: ADP6-0

See also: Mission Command · Commander’s Intent · Auftragstaktik

First introduced in: Paper 2 the digital battle staff

Agent Card

AI / Agent

JSON metadata document in the A2A protocol that describes an AI agent’s capabilities, enabling other agents to discover and delegate to it.

Also: agent card

See also: A2A · MCP

First introduced in: Paper 1 the super intelligent five year old

Agentic AI

AI / Agent

AI systems that pursue goals through sequences of actions, making decisions about which tools to use and steps to take — rather than simply responding to a single prompt. Distinct from chatbots.

Also: agentic AI, agentic

See also: MAS · LLM · Orchestrator-Worker

First introduced in: Paper 1 the super intelligent five year old

Anthropic

AI / Agent

AI research and deployment company. Built Claude; donated MCP to Linux Foundation; published Constitutional AI and Building Effective Agents.

Also: anthropic

See also: Claude · MCP · AAIF

First introduced in: Paper 1 the super intelligent five year old

ASS2

Process

Automation, Structure & Scalability, Safety & Security. A three-domain review framework where all three lenses apply together (flat, equal-weight). Used in this series when evaluating a system, artifact, or decision across its automation potential, structural scalability, and safety/security posture simultaneously. Not to be used for single-lens analysis.

Also: ASS-2, ASS2 review

See also: QASA · LSS-BB · PAT

First introduced in: Paper 3 the para experiment

ATP 5-19

Reference

Risk Management (2014). Four-step risk management process; doctrinal source for the FMEA-style risk scoring used in Tier 2 mission analysis.

Also: ATP5-19

See also: FMEA · RPN

First introduced in: Paper 1 the super intelligent five year old

Auftragstaktik

Military

German military concept meaning ‘mission-type tactics’ — give subordinates the what and why, release the how. Commanders discovered that AI agents perform better under Auftragstaktik than under detailed step-by-step instructions.

Also: mission-type tactics

See also: Mission Command · Commander’s Intent

First introduced in: Paper 2 the digital battle staff

B

Battle Rhythm

Military

The recurring schedule of staff meetings, briefings, and synchronization events that keeps a headquarters functioning. Codified in FM 6-0; applied in the vault as session-boot, mid-session checkpoint, and session-close cadence.

Also: battle rhythm

See also: FM 6-0 · ADP 5-0

First introduced in: Paper 2 the digital battle staff

BCT

Military

The U.S. Army’s primary combined-arms maneuver unit, typically 3,000-5,000 soldiers. The scale at which planning doctrine (MDMP) becomes operationally critical.

Also: Brigade Combat Team

See also: MDMP · Combined Arms

First introduced in: Paper 2 the digital battle staff

BCTP

Military

Army program that trains brigade and division staffs through simulation-driven exercises. The author’s primary training vehicle for 7 years.

Also: Battle Command Training Program

See also: BCT · MDMP

First introduced in: Paper 2 the digital battle staff

Blue

AI / Agent

MDMP automation system built by Exia Labs. Deploys specialized AI agents for each MDMP phase. Being tested with the 101st Airborne Division and Washington Army National Guard.

Also: Blue system

See also: MDMP · Exia Labs · OAKOC

First introduced in: Paper 2 the digital battle staff

BLUF

Military

Military briefing style that leads with the conclusion and supports with detail. Reduces decision latency for time-constrained commanders.

Also: Bottom Line Up Front

See also: OPORD · FRAGORD

First introduced in: Paper 1 the super intelligent five year old

Bounded Rationality

LSS / Quality

Herbert Simon’s concept that agents optimize within cognitive and informational limits rather than achieving theoretical optimality. Explains why pre-Claude-3 models required behavioral hooks rather than just good instructions — they satisficed locally.

Also: bounded rationality

See also: Simon · Mechanism Design

First introduced in: Paper 1 the super intelligent five year old

Boyd

Person

John R. Boyd (1927-1997), USAF Colonel and military theorist. Originated the OODA loop in ‘A Discourse on Winning and Losing’ (1987). Core insight: faster OODA cycling outperforms raw capability.

Also: John Boyd

See also: OODA

First introduced in: Paper 1 the super intelligent five year old

C

C2

Military

The military’s architecture for directing forces: hierarchical structure, standardized communication formats, clear authority chains. The AI industry independently converged on this same architecture for multi-agent systems.

Also: Command and Control

See also: Mission Command · Orchestrator-Worker

First introduced in: Paper 2 the digital battle staff

CCIR

Military

The specific information the commander must have to make decisions; drives intelligence collection priorities. PIR (Priority Intelligence Requirements) and FFIR (Friendly Force Information Requirements) are its two halves.

Also: Commander’s Critical Information Requirements

See also: IPB · MDMP

First introduced in: Paper 2 the digital battle staff

Cemri

Person

Mert Cemri. Lead author of UC Berkeley MAST taxonomy (NeurIPS 2025 Spotlight). ‘Why Do Multi-Agent LLM Systems Fail?’ (arXiv:2503.13657) catalogues 14 failure modes.

Also: Mert Cemri, Cemri et al

See also: MAST · MAS

First introduced in: Paper 1 the super intelligent five year old

Channel Capacity

LSS / Quality

Shannon’s C = B log2(1 + S/N) — the maximum information rate transmittable through a noisy channel. Formalization of template-driven decision throughput: structure increases agent instruction signal-to-noise ratio.

Also: channel capacity

See also: Shannon · Entropy · Toboggan Doctrine

First introduced in: Paper 1 the super intelligent five year old

Claude

AI / Agent

Anthropic’s family of LLMs (Opus, Sonnet, Haiku). Claude Code is the CLI/agent harness used to operate the vault documented in this series.

Also: Claude Opus, Claude Code, Anthropic Claude

See also: Anthropic · MCP · LLM

First introduced in: Paper 1 the super intelligent five year old

CMDP

Series

Eight-component framework developed through the Claude-Grok AI-to-AI exchange documented in Paper 5: independent generation, blind critique, revealed-identity critique, human synthesis, live fact-check, probability distributions, training prior disclosure, open publication.

Also: Cross-Model Deliberation Protocol

See also: PAT · Hallucination · Ensemble Methods

First introduced in: Paper 5 when the cats talk to each other

COA

Military

A possible approach to accomplishing the assigned mission. MDMP requires developing 2-3 distinct COAs, war-gaming each, then selecting based on commander’s judgment. AI equivalent: generate multiple solution paths before committing.

Also: Course of Action, courses of action

See also: MDMP · IPB · Wargaming

First introduced in: Paper 1 the super intelligent five year old

Combined Arms

Military

Military doctrine of integrating infantry, armor, artillery, aviation, and other capabilities into coordinated operations — each component amplifying the others’ strengths. Analogy for multi-model AI orchestration.

Also: combined arms

See also: Orchestrator-Worker · MoE

First introduced in: Paper 4 the creative middleman

Commander's Intent

Military

Concise expression of purpose, key tasks, and end state issued by the commander. Subordinates use it to make decisions when contact with higher headquarters is lost or when the situation changes faster than orders can be updated. Foundation of Mission Command.

Also: commanders intent, commander intent

See also: Mission Command · Auftragstaktik · FRAGORD

First introduced in: Paper 2 the digital battle staff

Common Operating Picture

Military

A shared digital display showing the positions and status of all friendly and enemy forces. AI equivalent: shared context state visible to all agents in a multi-agent system.

Also: COP

See also: Battle Rhythm · Context Window

First introduced in: Paper 2 the digital battle staff

Constitutional AI

AI / Agent

Anthropic’s alignment approach using constitutional principles and AI-generated feedback. The alignment-layer approach complemented by Toboggan Doctrine’s structural-enforcement layer. Bai et al. arXiv:2212.08073.

Also: constitutional AI

See also: Anthropic · Toboggan Doctrine

First introduced in: Paper 3 the para experiment

External: https://arxiv.org/abs/2212.08073

Context Window

AI / Agent

The maximum amount of text (measured in tokens) that an AI model can process in a single session. A key constraint in multi-agent system design — agents accumulate ‘context pollution’ over long runs.

Also: context window

See also: Token · LLM · Hallucination

First introduced in: Paper 1 the super intelligent five year old

Control Chart

LSS / Quality

Statistical process control (SPC) tool plotting a quality metric over time against upper and lower control limits (UCL/LCL). Within-limits = in control; drift beyond limits = out of statistical control. Foundational LSS/Quality 4.0 instrument.

Also: control chart, SPC

See also: Cp/Cpk · LSS · Quality 4.0

First introduced in: Paper 1 the super intelligent five year old

Cp/Cpk

LSS / Quality

Statistical process capability indices. Cp measures potential capability (spec width / process width); Cpk measures actual capability accounting for centering. Cpk >= 1.33 is generally acceptable.

Also: Cp, Cpk, process capability

See also: Control Chart · Six Sigma · LSS

First introduced in: Paper 1 the super intelligent five year old

CPI

Process

The feedback loop that updates templates, playbooks, and governance based on observed outcomes. Each AAR feeds CPI; CPI is the Lyapunov function for vault defect rate.

Also: Continuous Process Improvement

See also: AAR · DMAIC · Lyapunov Stability · Yokoten

First introduced in: Paper 1 the super intelligent five year old

Creative Middleman

Series

Paper 4’s analysis of Adobe’s positioning as the creative-AI orchestrator routing user intent to partner models (OpenAI, Google, Runway). The middleman pattern is structurally fragile when partners can disintermediate.

Also: creative middleman

See also: Adobe · Combined Arms

First introduced in: Paper 4 the creative middleman

D

Deming

Person

W. Edwards Deming (1900-1993). Originator of Plan-Do-Check-Act (PDCA) and statistical process control in management. ‘Out of the Crisis’ (1986) is the foundation of CPI.

Also: W. Edwards Deming

See also: DMAIC · CPI · Control Chart

First introduced in: Paper 1 the super intelligent five year old

Digital Battle Staff

Series

Paper 2’s central concept: applying the Napoleonic-era staff structure (G-1 through G-6) to AI agent organization. Each agent fills a specialized staff role with defined inputs, outputs, and authority.

Also: digital battle staff

See also: G-1 · G-2 · G-3 · G-4 · Mission Command

First introduced in: Paper 2 the digital battle staff

DMAIC

LSS / Quality

Lean Six Sigma’s improvement cycle: Define, Measure, Analyze, Improve, Control. Adapted as a governance lifecycle for AI agents; each AAR cycle is a DMAIC turn applied to template and skill quality.

Also: Define Measure Analyze Improve Control

See also: LSS · CPI · Quality 4.0 · Deming

First introduced in: Paper 1 the super intelligent five year old

DOTMLPF-P

Military

Doctrine, Organization, Training, Materiel, Leadership, Personnel, Facilities, and Policy. The U.S. Army capability-gap taxonomy, used to classify AI system gaps beyond just ‘software.’

Also: DOTMLPF

See also: MDMP · Mission Command

First introduced in: Paper 1 the super intelligent five year old

DOWNTIME

LSS / Quality

Lean Six Sigma mnemonic for the 8 wastes: Defects, Overproduction, Waiting, Non-utilized talent, Transportation, Inventory, Motion, Extra-processing. Applied to AI agent systems throughout Papers 1 and 4.

Also: downtime, 8 wastes

See also: LSS · TIMWOODS · Ohno

First introduced in: Paper 1 the super intelligent five year old

Drop Zone

Series

A named intake point in the gravity pipeline where new work arrives with its schema, retention class, and receipt vocabulary pre-defined. Each DZ has a contract that downstream agents can rely on.

Also: drop zone, DZ, Gravity Drop Zone

See also: Gravity Pipeline · Toboggan Doctrine

First introduced in: Paper 8 the toboggan doctrine

E

Ensemble Methods

AI / Agent

ML technique combining multiple classifiers; diversity reduces error variance. Validates PAT (parallel orthogonal review) and CMDP (independent generation before synthesis). Dietterich, 2000.

Also: ensemble methods, ensemble

See also: PAT · CMDP · MoE

First introduced in: Paper 5 when the cats talk to each other

Entropy

LSS / Quality

H(X) = -Sigma p(xi) log2 p(xi) — measure of uncertainty in a probability distribution. Template governance reduces action entropy: structure constrains the agent’s action distribution toward compliant behaviors.

Also: Shannon entropy

See also: Shannon · Channel Capacity

First introduced in: Paper 1 the super intelligent five year old

F

First-Pass Yield

LSS / Quality

Percentage of units completing a process without rework: FPY = Pi(1 - di) for n stages with defect rates di. Primary quality metric throughout the series. Gate enforcement raises per-stage yield from 90% to 99%, transforming five-stage FPY from 59% to 95.1%.

Also: FPY, first pass yield

See also: Cp/Cpk · Quality 4.0 · Gate A · Gate B

First introduced in: Paper 1 the super intelligent five year old

FM 5-0

Reference

Army Planning and Orders Production (2022). Codifies MDMP in Chapter 12. Primary doctrinal source for the planning frameworks adapted in this series.

Also: FM5-0

See also: MDMP · OPORD · ADP 5-0

First introduced in: Paper 1 the super intelligent five year old

FM 6-0

Reference

Commander and Staff Organization and Operations (2022). Establishes battle rhythm, staff section organization, and CCIR management.

Also: FM6-0

See also: Battle Rhythm · CCIR · G-1

First introduced in: Paper 2 the digital battle staff

FMEA

LSS / Quality

Lean Six Sigma risk-analysis tool: enumerate failure modes, score Severity x Occurrence x Detection = RPN (Risk Priority Number), prioritize mitigation by RPN.

Also: Failure Mode and Effects Analysis

See also: RPN · LSS · Risk Management

First introduced in: Paper 1 the super intelligent five year old

Forte

Person

Tiago Forte. Productivity author; created the PARA Method (Projects, Areas, Resources, Archives) — the substrate for this research.

Also: Tiago Forte

See also: PARA

First introduced in: Paper 3 the para experiment

FRAGORD

Military

A modification to an existing OPORD that does not require a full re-planning cycle. Used in the vault to signal commander intent updates during execution; the AI equivalent of a mid-session context update.

Also: FRAGO, Fragmentary Order

See also: OPORD · WARNO · Mission Command

First introduced in: Paper 2 the digital battle staff

G

G-1

Military

Personnel staff section. Part of the numbered staff system descended from Napoleon’s four headquarters departments, formalized by the Prussian Great General Staff.

Also: G1

See also: G-2 · G-3 · G-4 · C2

First introduced in: Paper 2 the digital battle staff

G-2

Military

Intelligence staff section. Owns IPB, CCIR management, and threat analysis. AI analog: the research/scan agent.

Also: G2

See also: IPB · CCIR · G-1 · G-3

First introduced in: Paper 2 the digital battle staff

G-3

Military

Operations staff section. Owns current operations, battle tracking, FRAGORD processing. AI analog: the orchestrator/supervisor.

Also: G3

See also: FRAGORD · Battle Rhythm · Orchestrator-Worker

First introduced in: Paper 2 the digital battle staff

G-4

Military

Logistics / sustainment staff section. AI analog: resource management and infrastructure agents.

Also: G4

See also: G-1 · G-2 · G-3

First introduced in: Paper 2 the digital battle staff

Gastown

AI / Agent

Steve Yegge’s fourth-generation agent orchestration framework. GUPP = Gastown Universal Propulsion Principle: sessions are ephemeral, workflow state lives externally in Git, mission persists across agent restarts.

Also: GUPP, Gas Town

See also: Yegge · Orchestrator-Worker

First introduced in: Paper 1 the super intelligent five year old

External: https://github.com/steveyegge/gastown

Gate A

Process

Pre-execution quality gate. Verifies tool inventory, risk assessment, and readiness before committing resources. Ensures the agent has surveyed available skills/scripts before reinventing them.

Also: gate a, gate-a

See also: Gate B · First-Pass Yield · Poka-yoke

First introduced in: Paper 3 the para experiment

Gate B

Process

Post-execution completion gate. Verifies deliverables exist, documentation was updated, and the change propagated to all dependent references. Closes the institutionalization loop.

Also: gate b, gate-b

See also: Gate A · First-Pass Yield · WNTK

First introduced in: Paper 3 the para experiment

Gravity Pipeline

Series

The end-to-end vault flow where work moves downstream through drop zones, enrichment, routing, and archival via channel design rather than dispatcher logic. Senge’s learning organization made structural.

Also: gravity pipeline, gravity-fed pipeline

See also: Drop Zone · Toboggan Doctrine · Senge

First introduced in: Paper 8 the toboggan doctrine

H

Hallucination

AI / Agent

When an AI model generates confident but factually incorrect output. A quality defect in QASA/LSS terms. Addressed in the CMDP through bilateral fact-checking rounds.

Also: hallucinations, AI hallucination

See also: CMDP · QASAS · RAG

First introduced in: Paper 1 the super intelligent five year old

Herding Cats

Series

The series title. Captures the central management problem of multi-agent AI systems: coordinating independent, semi-autonomous agents toward a coherent outcome — and the doctrinal answer that channel design beats coercion.

Also: herding cats

See also: Toboggan Doctrine · MAS · Mission Command

First introduced in: Paper 1 the super intelligent five year old

I

Incentive Compatibility

LSS / Quality

Mechanism design property where each agent’s optimal self-interested strategy produces the desired collective outcome. The deny hook is incentive-compatible: compliance dominates non-compliance in expected value.

Also: incentive compatibility, incentive-compatible

See also: Mechanism Design · Toboggan Doctrine

First introduced in: Paper 3 the para experiment

IPB

Military

The systematic process of analyzing the operational environment to support decision-making. Step 2 of MDMP Mission Analysis; the AI equivalent is environmental scan / 360 research before planning.

Also: Intelligence Preparation of the Battlefield

See also: MDMP · CCIR · METT-TC

First introduced in: Paper 2 the digital battle staff

K

Kaizen

LSS / Quality

Japanese term for continuous improvement through small, incremental changes. Foundation of CPI; complements DMAIC’s structured improvement cycle.

Also: kaizen

See also: CPI · DMAIC · Yokoten

First introduced in: Paper 3 the para experiment

Karpathy

Person

Andrej Karpathy. AI researcher (ex-Tesla, ex-OpenAI). ‘LLM OS’ thread (2023) positions LLMs as operating system kernels — validates the three-layer vault architecture.

Also: Andrej Karpathy

See also: LLM

First introduced in: Paper 1 the super intelligent five year old

Knowledge Well

Process

A long-form reference document that accumulates institutional knowledge on a single topic, versioned and cross-referenced. Wells are the gravity sink: solved problems flow into wells so future sessions inherit the answer.

Also: knowledge well, knowledge wells, well

See also: PARA · CPI · Yokoten

First introduced in: Paper 3 the para experiment

L

LLM

AI / Agent

Foundation AI model trained on large text datasets (GPT-4, Claude, Gemini, Grok, etc.). The individual ‘agents’ in multi-agent systems are typically LLMs with tool access.

Also: Large Language Model, large language model

See also: Agentic AI · Context Window · Token

First introduced in: Paper 1 the super intelligent five year old

LSS

LSS / Quality

Quality and process-improvement discipline combining Lean (waste elimination) with Six Sigma (variance reduction). DMAIC is its core execution cycle; FMEA, Cp/Cpk, and control charts are core instruments.

Also: Lean Six Sigma

See also: DMAIC · FMEA · DOWNTIME · Quality 4.0

First introduced in: Paper 1 the super intelligent five year old

Lyapunov Stability

LSS / Quality

System property: V(x) >= 0 and dV/dt <= 0, where V is a scalar Lyapunov function. The CPI loop is a Lyapunov function for vault defect rate, monotonically decreasing error frequency as fixes propagate.

Also: Lyapunov, Lyapunov function

See also: CPI · First-Pass Yield

First introduced in: Paper 3 the para experiment

M

MAS

AI / Agent

A system where multiple AI agents collaborate (or fail to collaborate) on a shared objective. The primary subject of this research series.

Also: Multi-Agent System, Multi Agent System

See also: MAST · Orchestrator-Worker · Agentic AI

First introduced in: Paper 1 the super intelligent five year old

MAST

AI / Agent

UC Berkeley taxonomy of 14 failure modes across 3 categories (Specification & System Design 37%, Inter-Agent Misalignment 31%, Task Verification & Termination 31%). Cemri et al., NeurIPS 2025 Spotlight. Failure rates 41-86.7% across 7 frameworks.

Also: Multi-Agent System Failure Taxonomy

See also: MAS · Cemri · MAST Failure Categories

First introduced in: Paper 1 the super intelligent five year old

External: https://arxiv.org/abs/2503.13657

MCP

AI / Agent

Anthropic’s late-2024 protocol (donated to Linux Foundation December 2025) standardizing how AI agents connect to tools, data sources, and external context. 97M+ monthly SDK downloads, 10,000+ active servers by early 2026.

Also: Model Context Protocol

See also: A2A · AAIF · Anthropic

First introduced in: Paper 1 the super intelligent five year old

External: https://modelcontextprotocol.io

MDMP

Military

The U.S. Army’s seven-step planning framework: Receipt of Mission, Mission Analysis, COA Development, COA Analysis (War-Gaming), COA Comparison, COA Approval, Orders Production. Refined over 70 years; the core framework applied to AI planning throughout the series.

Also: Military Decision Making Process, Military Decision-Making Process

See also: FM 5-0 · COA · OPORD · WARNO · Blue

First introduced in: Paper 1 the super intelligent five year old

Mechanism Design

LSS / Quality

Game theory subfield that designs rules and incentive structures to produce desired equilibrium outcomes, working backward from outcome to mechanism. Foundation of Toboggan Doctrine: design the environment so aligned behavior is the equilibrium strategy.

Also: mechanism design

First introduced in: Paper 3 the para experiment

METT-TC

Military

Six mission variables for analyzing any operational situation: Mission, Enemy, Terrain, Troops, Time, Civil considerations.

Also: METT-T

See also: METT-TC(IT) · IPB · ADP 5-0

First introduced in: Paper 1 the super intelligent five year old

METT-TC(IT)

Series

Author’s adaptation of METT-TC adding Information Technology as a 7th variable, acknowledging the digital battlespace in which AI agents operate.

Also: METT-TC-IT

See also: METT-TC · IPB

First introduced in: Paper 1 the super intelligent five year old

Mission Command

Military

U.S. Army leadership doctrine codified in ADP 6-0. Emphasizes commander’s intent and disciplined initiative over detailed orders; the model for delegating authority to AI agents.

Also: mission command

See also: Auftragstaktik · Commander’s Intent · ADP 6-0

First introduced in: Paper 2 the digital battle staff

MoE

AI / Agent

Neural network architecture routing inputs to specialized sub-models via a gating function, activating only the most relevant expert per input. Structural model for multi-agent MDMP staff roles: the orchestrator is the gating function, specialist agents are the experts. Shazeer et al. arXiv:1701.06538.

Also: Mixture of Experts, Mixture-of-Experts

See also: Orchestrator-Worker · Combined Arms

First introduced in: Paper 2 the digital battle staff

External: https://arxiv.org/abs/1701.06538

O

OAKOC

Military

Terrain analysis framework: Observation, Avenues of approach, Key terrain, Obstacles, Cover and concealment. Used in MDMP Mission Analysis. Automated in the Blue system (Exia Labs).

Also: KOCOA-W, KOCOA

See also: IPB · MDMP · Blue

First introduced in: Paper 2 the digital battle staff

OC

Process

Trained external evaluator who observes a unit’s performance against doctrine and facilitates the AAR. In the vault, an OC agent reviews supervisor work and delivers fix-categorized findings.

Also: Observer-Controller, observer controller

See also: AAR · BCTP · PAT

First introduced in: Paper 3 the para experiment

Ohno

Person

Taiichi Ohno (1912-1990). Architect of the Toyota Production System; source of the Lean waste framework (DOWNTIME / TIMWOODS).

Also: Taiichi Ohno

See also: DOWNTIME · Kaizen

First introduced in: Paper 1 the super intelligent five year old

OODA

Military

Decision cycle by John Boyd: Observe, Orient, Decide, Act. Boyd’s central insight is that faster OODA cycling outperforms raw capability. Used throughout the series as the cognitive model for agent decision-making.

Also: OODA loop, Observe Orient Decide Act, Observe-Orient-Decide-Act

See also: MDMP · Mission Command · Boyd

First introduced in: Paper 1 the super intelligent five year old

External: https://en.wikipedia.org/wiki/OODA_loop

OPORD

Military

The full military planning artifact produced by MDMP — a five-paragraph order: Situation, Mission, Execution, Sustainment, Command and Signal. The AI equivalent: the prompt that launches a complex multi-agent workflow.

Also: Operations Order, Operation Order

See also: MDMP · FRAGORD · WARNO

First introduced in: Paper 1 the super intelligent five year old

Orchestrator-Worker

AI / Agent

Multi-agent architecture where an orchestrator agent decomposes tasks and delegates to specialized worker agents. The independent convergence of every major AI company (Anthropic, OpenAI, Google, Microsoft, Cursor, Gastown) on this architecture mirrors the military’s hierarchical C2 structure.

Also: orchestrator-worker, orchestrator worker

See also: MAS · C2 · MoE · Gastown

First introduced in: Paper 1 the super intelligent five year old

P

PARA

Process

Tiago Forte’s vault organization method: Projects (active outcomes), Areas (ongoing responsibilities), Resources (reference material), Archives (inactive items). The substrate for this research.

Also: PARA Method, PARA method

See also: PKM · Forte · Knowledge Well

First introduced in: Paper 3 the para experiment

PAT

Process

Parallel multi-agent review pattern. Specialist lenses (LSS-BB, QASA, ASS2, Editorial) review a deliverable in parallel; a synthesizer produces convergent findings. Validates work before commitment without serializing review.

Also: PAT review, Planning Analysis Team, Parallel Approval Team

See also: QASAS · CMDP · Ensemble Methods

First introduced in: Paper 1 the super intelligent five year old

PKM

Process

Systematic approach to organizing, capturing, and retrieving personal information and knowledge. Obsidian is the PKM tool used in this series.

Also: Personal Knowledge Management

See also: PARA · Knowledge Well

First introduced in: Paper 3 the para experiment

Poka-yoke

LSS / Quality

Lean Six Sigma mistake-proofing device that makes errors structurally impossible. Templates-as-channels are the poka-yoke: the agent cannot produce a non-compliant output without explicitly defeating the structure.

Also: poka-yoke, pokayoke, mistake-proofing

See also: Toboggan Doctrine · LSS · Shingo

First introduced in: Paper 3 the para experiment

Policy Gradient

AI / Agent

Machine learning optimization method updating a policy toward higher expected reward by following the gradient of expected return. Formal analog of CPI loop template improvement: each AAR cycle nudges the template (policy) toward higher task quality.

Also: policy gradient

See also: Reward Shaping · CPI · Sutton

First introduced in: Paper 3 the para experiment

Pyzdek

Person

Thomas Pyzdek. Author of The Six Sigma Handbook (4th ed., 2014). Source for sigma cost escalation and satisficing threshold analysis.

Also: Thomas Pyzdek

See also: Six Sigma · LSS

First introduced in: Paper 1 the super intelligent five year old

Q

QASAS

LSS / Quality

Quality Assurance Specialist, Ammunition Surveillance — the oldest federal civilian career program (est. 1920). Provides the model for AI quality assurance throughout the series: dedicated, trained specialists who operate independently of the production chain.

Also: Quality Assurance Specialist Ammunition Surveillance

See also: PAT · AR 702-12 · LSS

First introduced in: Paper 1 the super intelligent five year old

Quality 4.0

LSS / Quality

The convergence of AI/ML with Lean Six Sigma and quality management methodologies. Gartner projects 50%+ of LSS organizations will incorporate AI tools by 2026.

Also: quality 4.0

See also: LSS · DMAIC · Gartner

First introduced in: Paper 1 the super intelligent five year old

R

RAG

AI / Agent

Technique where AI systems retrieve relevant documents before generating responses, improving accuracy. Used in Blue (MDMP automation) and the CGSC wargaming experiment.

Also: Retrieval Augmented Generation, Retrieval-Augmented Generation

See also: Hallucination · Blue · Context Window

First introduced in: Paper 2 the digital battle staff

Reason

Person

James Reason. Originator of the Swiss Cheese Model of accident causation: defenses-in-depth where holes occasionally line up. Used in the series to motivate layered governance (alignment + structural enforcement).

Also: James Reason

See also: Toboggan Doctrine

First introduced in: Paper 3 the para experiment

Reward Shaping

LSS / Quality

Reinforcement learning technique modifying the reward function to accelerate policy convergence toward desired behavior. Formal basis for template-driven governance: templates shape the reward landscape without changing the agent’s underlying objective. Ng, Harada, & Russell, ICML 1999.

Also: reward shaping

See also: Mechanism Design · Policy Gradient · Toboggan Doctrine

First introduced in: Paper 3 the para experiment

ROE

Military

Constraints on the use of force defining who can act, against what targets, under what conditions. Applied as the AI Off-Ramp model: Weapons Hold (read-only), Weapons Tight (rule-bound execution), Weapons Free (full autonomous within boundaries).

Also: Rules of Engagement

See also: Mission Command

First introduced in: Paper 1 the super intelligent five year old

RPN

LSS / Quality

FMEA scoring output: Severity x Occurrence x Detection. Ranks risks for mitigation prioritization. Range 1-1000.

Also: Risk Priority Number

See also: FMEA · LSS

First introduced in: Paper 1 the super intelligent five year old

S

Senge

Person

Peter M. Senge. Author of The Fifth Discipline (1990). Systems thinking and the learning organization; vault gravity-fed pipeline is a structural implementation of Senge’s vision.

Also: Peter Senge, Peter M. Senge

See also: Gravity Pipeline · CPI

First introduced in: Paper 3 the para experiment

Shannon

Person

Claude E. Shannon (1916-2001). Founder of information theory. ‘A Mathematical Theory of Communication’ (1948) introduced entropy and channel capacity. Underlies template governance analysis.

Also: Claude Shannon, Claude E. Shannon

See also: Entropy · Channel Capacity

First introduced in: Paper 1 the super intelligent five year old

Shingo

Person

Shigeo Shingo (1909-1990). Japanese industrial engineer; co-architect of the Toyota Production System and originator of poka-yoke (mistake-proofing).

Also: Shigeo Shingo

See also: Poka-yoke · Kaizen

First introduced in: Paper 3 the para experiment

Simon

Person

Herbert A. Simon (1916-2001). Nobel laureate in economics; introduced bounded rationality. The Sciences of the Artificial (3rd ed., 1996) is the source for satisficing analysis applied to LLM behavior.

Also: Herbert Simon, Herbert A. Simon

See also: Bounded Rationality

First introduced in: Paper 1 the super intelligent five year old

Six Sigma

LSS / Quality

Quality methodology targeting <= 3.4 defects per million opportunities (6 standard deviations between mean and nearest spec limit). Combined with Lean tools to form Lean Six Sigma (LSS).

Also: six sigma, 6 sigma

See also: LSS · DMAIC · Cp/Cpk · Pyzdek

First introduced in: Paper 1 the super intelligent five year old

STO

Process

A recurring work item with its own cadence and metrics — the vault equivalent of a cron job with intent. STOs are AREAS (ongoing responsibilities), not Projects.

Also: Standing Task Order, standing task order

See also: PARA · Battle Rhythm · Tetris Primitives

First introduced in: Paper 3 the para experiment

T

Task Tensor

Series

Treats tasks as multi-dimensional coordinates (5 axes: scope, primitive, drop zone, retention class, receipt vocabulary) rather than scalar list items. Prevents the failure mode of flattening multi-dimensional problems into 1D checklists.

Also: task tensor, Tensor Doctrine

See also: Tetris Primitives · STO

First introduced in: Paper 3 the para experiment

TC 25-20

Reference

A Leader’s Guide to After-Action Reviews (1993). Source of the four AAR questions used as the model for AI system learning loops.

Also: TC25-20

See also: AAR · OC

First introduced in: Paper 1 the super intelligent five year old

Tetris Primitives

Series

Composable atomic work units. Tasks decompose into a small set of primitive shapes (capture, classify, route, enrich, validate, archive, etc.) that can be combined to form any work pattern.

Also: Tetris primitives, tetris primitives, primitives

See also: Task Tensor · STO · Drop Zone

First introduced in: Paper 3 the para experiment

TIMWOODS

LSS / Quality

Alternate Lean waste mnemonic: Transportation, Inventory, Motion, Waiting, Overproduction, Over-processing, Defects, Skills (non-utilized talent). A reordering of DOWNTIME with the same content.

Also: TIMWOOD

See also: DOWNTIME · LSS

First introduced in: Paper 1 the super intelligent five year old

Toboggan Doctrine

Series

The governance thesis that template-driven channels outperform hook-based enforcement. ‘Build the channel, let gravity work.’ The compliant path becomes the path of least resistance, so agents follow it without coercion.

Also: toboggan doctrine, Toboggan, Channels not Walls

See also: Drop Zone · Gravity Pipeline · Poka-yoke · Mechanism Design

First introduced in: Paper 8 the toboggan doctrine

Token

AI / Agent

The basic unit of text that LLMs process — roughly 0.75 words. Token budgets determine how much context an agent can consider. Token efficiency is a key metric in multi-agent coordination — hybrid systems burn 5x tokens per successful task vs. single agents.

Also: tokens, token budget

See also: Context Window · LLM

First introduced in: Paper 1 the super intelligent five year old

W

WARNO

Military

Preliminary notice of a forthcoming order. Issued immediately upon mission receipt to enable parallel preparation. AI equivalent: early task alerts that allow multi-agent systems to begin staging resources before the complete plan exists.

Also: Warning Order

See also: OPORD · FRAGORD · MDMP

First introduced in: Paper 2 the digital battle staff

Y

Yegge

Person

Steve Yegge. Ex-Google, ex-Amazon; built Gastown (4th-generation agent orchestration). ‘Welcome to Gas Town’ (2026) and ‘Introducing Beads’ establish the GUPP pattern.

Also: Steve Yegge

See also: Gastown

First introduced in: Paper 1 the super intelligent five year old

Yokoten

LSS / Quality

Toyota Lean concept of horizontal deployment of best practices — sharing solutions across the organization rather than rediscovering them. Phase 5 institutionalization mechanism: once a fix works, it propagates through templates, skills, and knowledge wells so every future session inherits it.

Also: yokoten

See also: Kaizen · CPI · Knowledge Well

First introduced in: Paper 3 the para experiment

Glossary

Jump to letter

Browse by category

Military (27)

LSS / Quality (22)

AI / Agent (21)

Series (10)

Process (9)

Reference (6)

Person (13)

A

A2A

AAIF

AAR

ADP 5-0

ADP 6-0

Agent Card

Agentic AI

Anthropic

ASS2

ATP 5-19

Auftragstaktik

B

Battle Rhythm

BCT

BCTP

Blue

BLUF

Bounded Rationality

Boyd

C

C2

CCIR

Cemri

Channel Capacity

Claude

CMDP

COA

Combined Arms

Commander's Intent

Common Operating Picture

Constitutional AI

Context Window

Control Chart

Cp/Cpk

CPI

Creative Middleman

D

Deming

Digital Battle Staff

DMAIC

DOTMLPF-P

DOWNTIME

Drop Zone

E

Ensemble Methods

Entropy

F

First-Pass Yield

FM 5-0

FM 6-0

FMEA

Forte

FRAGORD

G

G-1

G-2

G-3

G-4

Gastown

Gate A

Gate B

Gravity Pipeline

H

Hallucination

Herding Cats

I

Incentive Compatibility

IPB

K