A public shelf of finished long-form work and systems proof by Roger Crosby.
This repository contains completed documents in artificial intelligence, cognition, operational systems, recipe-production methodology, systems design, and experimental literature. They are published here as inspectable evidence of finished work rather than as unsupported claims on a résumé.
The documents are not all the same kind of object. Some are Field Manuals, some are production systems, some are Systems Papers, some are speculative analysis, and some are literary experiments. What connects them is the working method:
- large or messy source material is given a defined container;
- claims are separated from metaphor and speculation;
- structure is used to make the material navigable;
- failure modes and control boundaries are made explicit when the subject requires them;
- the document is completed as an artifact rather than left as an idea;
- public work is separated from the private production archive.
Available for remote contract work and project-based collaboration.
Contact: info@paranoidpeoplelivelonger.com
| Document | Primary lane | What it demonstrates |
|---|---|---|
| Systems Papers | AI systems and documentation architecture | Public proof-of-work papers showing real systems, their architecture, failure modes, boundaries, and lessons without publishing the private operating machinery |
| The Dark Cognitive Manual | Cognition and interpretation | Field Manual structure, behavioural observation, constraint-based writing, and practical synthesis |
| Sixty AI Futures | Artificial-intelligence futures | Sustained scenario production, range, disciplined speculation, and systems thinking |
| The Recipe Writing Production Guide | Operational food writing | Translation of kitchen experience into a repeatable production and quality-control system |
| A Zoo at the End of Winter | Experimental literature | Long-form architecture, controlled voice, recurring symbolic systems, and completion at book scale |
| Character Study | Earlier creative archive | Character observation, voice, and public evidence from an earlier stage of the work |
The Systems Papers are a separate public document family. They explain systems that were actually built, used, tested, revised, or recovered without publishing the full internal machinery that operates them.
The purpose is proof, not a consulting giveaway. A Systems Paper should make the architecture legible enough that another person can understand the problem, the design decisions, the control logic, the failure modes, and the limits. It should not function as a complete reproduction kit for the private operating system behind the paper.
A public proof-of-work paper about recovering preferred large-language-model interaction behaviour through external operating controls rather than model-weight modification.
The paper covers the problem created when model behaviour changes across versions, the distinction between model capability and operating behaviour, the role of state discipline and external control documents, drift and failure categories, calibration, human authority, and the boundary between public architectural proof and private implementation detail.
Demonstrates: AI workflow architecture, behavioural continuity, external control systems, model-agnostic documentation, failure analysis, calibration thinking, public/private boundary design, and technical systems writing.
A public proof-of-work paper about converting a noisy AI job search into a constraint-first employment-intelligence system.
The underlying private system was developed through sequential prototypes, beginning with a deliberately mixed batch of 30 live job listings. The public edition shows the architecture and development logic without publishing the private scorecard, candidate constraints, detailed search configuration, test records, or operating Bible.
The paper covers feasibility before relevance, the separation of content fit from employment fit, title-vocabulary discovery, evidence matching, truth-preserving translation, human-gated consequential action, failure-to-infrastructure learning, and the rule that a system should earn complexity through low-cost testing before formalization.
Demonstrates: constraint-first decision architecture, AI-assisted evaluation, human oversight, evidence-backed capability translation, sequential prototyping, failure analysis, knowledge and vocabulary learning, claims discipline, and public/private boundary design.
A public proof-of-work paper about auditing the process behind AI-supported claims rather than treating the visible output as self-explanatory.
The public framework separates claim scope, evidence provenance, AI context, institutional power and incentives, measurement integrity, human oversight and contestability, auditability, and conclusion discipline. It also distinguishes evidence access from evidence quality and keeps material unknowns visible rather than filling them with assumptions.
The public edition deliberately withholds internal agent specifications, detailed prompts, recursive routing logic, branch mechanics, field cards, ledger design, validation records, and the complete private operating package.
Demonstrates: AI evaluation and claims auditing, provenance architecture, human oversight, contestability analysis, quantitative-claim discipline, lawful evidence-access design, bounded findings, institutional power analysis, and public/private system-boundary design.
A completed public Field Manual about cognition, interpretation, pressure, uncertainty, attention, motivation, and the stories people use to explain their own behaviour.
It is not presented as clinical psychology, diagnosis, therapy, or a scientific textbook. It is an operational document: a structured set of observations and working distinctions intended to help a reader notice how thought and behaviour fail under real conditions.
The document demonstrates:
- turning broad behavioural material into a navigable manual;
- maintaining a consistent field-reference voice;
- separating observation from instruction;
- using headings, rules, and repeated forms to control a large subject;
- completing a substantial practical nonfiction artifact.
Sixty compressed future scenarios examining artificial intelligence as infrastructure, institution, collaborator, labour system, commercial force, threat surface, cultural pressure, and ordinary tool.
The collection does not claim that all sixty futures will occur. Its value is range and pressure testing: each scenario changes the conditions and asks what becomes possible, likely, profitable, dangerous, invisible, or normal.
The document demonstrates:
- sustained high-volume conceptual production without collapsing into one repeated prediction;
- the ability to vary scale, tone, stakeholders, and consequences;
- systems thinking across technology, institutions, culture, work, and human behaviour;
- disciplined speculative writing that preserves uncertainty;
- conversion of long-form discussion into a finished public artifact.
A practical production system for writing recipes that survive contact with real kitchens.
The guide treats a recipe as an operating document rather than decorative food copy. It addresses sequencing, ingredient identity, substitutions, timing, equipment, yields, ambiguity, human variation, quality checks, and the kinds of omissions that make a recipe technically attractive but operationally useless.
The document grows out of more than twenty years of food-service and hospitality experience and demonstrates:
- converting tacit professional judgment into explicit procedure;
- writing for operators under time and information constraints;
- anticipating failure points before publication;
- documentation architecture, production standards, and quality assurance;
- translation between subject-matter expertise and usable instructions.
A completed long-form experimental literary work built through a cold institutional and ledger-like voice.
The work uses recurring animals, structures, rituals, and winter systems to examine behaviour before explanation. It is not a conventional novel, devotional work, or simple satire. It uses literary form as an observing instrument.
Available formats:
The document demonstrates:
- maintaining a distinct voice across a substantial manuscript;
- building a symbolic system without explaining every symbol;
- organizing recurring patterns across a long work;
- carrying an unconventional idea through to a finished document;
- literary range alongside operational and technical writing.
A public creative document from the earlier archive.
It is included as evidence of development and range rather than presented as the current flagship. The document shows an earlier approach to character observation, voice, and construction and provides useful contrast with the more controlled systems and long-form work produced later.
This shelf provides direct evidence that the builder can:
- complete long documents rather than only generate fragments;
- impose useful structure on large and ambiguous subjects;
- explain system architecture without overstating implementation evidence;
- maintain distinct voices across different types of work;
- move between operational, analytical, speculative, technical, and literary forms;
- translate practical experience into reusable documentation;
- identify failure modes, boundaries, and authority relationships;
- state limits and avoid inflating the type of claim a document can support;
- package finished work so another person can inspect it directly.
The documents do not prove enterprise deployment, professional certification, academic peer review, clinical authority, or guaranteed commercial outcomes. They prove the existence, range, architecture, and completion of the work that is actually here.
For a fast review:
- Open Systems Paper 001 - Rebuilding the Room for AI systems architecture, behavioural control, failure analysis, and public-safe technical explanation.
- Open Systems Paper 002 - From Job Search to Employment Intelligence for constraint-first decision architecture, evidence matching, human-gated evaluation, and sequential prototype learning.
- Open Systems Paper 003 - AI Claims Audit Stack for AI-claim auditing, provenance, auditability, human oversight and contestability, quantitative claims, institutional power, and bounded conclusions.
- Open The Recipe Writing Production Guide for operational documentation and subject-matter translation.
- Open The Dark Cognitive Manual for Field Manual structure and practical synthesis.
- Open Sixty AI Futures for range, scenario thinking, and sustained production.
- Open A Zoo at the End of Winter for long-form voice and literary architecture.
- Use Character Study as an earlier comparison point rather than the main sample.
For hiring or collaboration discussions, do not treat the repository as a request to read everything. Select the artifact closest to the problem being discussed.
- Human-Gated Agent System - a twelve-role, document-first architecture with structured handoffs, evidence gates, approval levels, tests, and explicit human authority.
- Kaomoji Emotional Confusion Engine - a small tested Python artifact built around bounded symbolic mutation and honest technical description.
- PPLL Signal Archive - the main public index connecting released systems, documents, creative work, and the commercial PPLL door.
- Roger Crosby GitHub Profile - professional orientation and selected proof of work.
This repository contains finished public artifacts. It does not contain:
- the complete private source archive;
- raw conversations and personal records;
- internal operating Bibles;
- complete private implementation recipes behind the Systems Papers;
- unreleased manuscripts;
- private strategy or commercial records;
- credentials, platform configuration, or customer data.
A public Systems Paper may describe a private system's architecture, purpose, failure modes, and lessons. It does not imply that the complete operating documents, control prompts, calibration material, client method, or source chain are public.
The presence of a finished public document does not imply that all source material or production methods are public.
No open-content license is included in this repository. Unless a specific document states otherwise, default copyright applies. Public access permits reading and inspection; it does not automatically grant permission to reproduce, modify, redistribute, train on, package, or sell the documents.
For permissions or professional use, contact the author directly.
Roger Crosby is an Ottawa-based systems builder and operational writer working across:
- human-gated artificial intelligence;
- workflow governance and evaluation;
- documentation architecture;
- operational and plain-language writing;
- creative systems and experimental literature;
- the conversion of practical experience into reusable structures.
His background includes more than twenty years in food service and hospitality and three Algonquin College diplomas in Culinary Management, Business Management and Entrepreneurship, and Hotel and Restaurant Operations Management.
The work is developed through repeated testing, explicit constraints, preserved evidence, version control, failure analysis, and human review. The objective is not to make systems appear intelligent. It is to make the work understandable, inspectable, recoverable, and useful.
Contact: info@paranoidpeoplelivelonger.com
Public portfolio: PPLL Signal Archive
Website: Paranoid People Live Longer