Chapter 3 - A Brief History of Human Loops

Last updated 2026-08-09. This page hosts the cross-tradition reading paths that accompany Chapter 3, plus the primary sources behind each of the four movements. The chapter's organizing question: after evidence shows a result missed its target, what is allowed to change next — the action, the process, the interpretation, or the goal?

Companion activity

Take a correction you made in the last month. Name which level actually changed: the immediate action, the process that produced it, the assumption behind the process, or the goal itself. Then ask which level should have changed. A loop that can only reach one level will keep re-running the same correction.

What this chapter argues

Long before AI agents, people built methods for comparing outcomes with intent, testing changes, revising assumptions, reflecting during practice, and turning experience into future capability. Quality improvement, military decision-making, organizational learning, professional practice, education, creativity research, and autonomic computing developed separately, for different purposes, without a shared vocabulary. They still converged on one recurring control problem: what may change after the evidence comes back? The traditions differ mainly in how far up they let the correction travel.

The four movements

From discrepancy to bounded experiment — cybernetic feedback and control theory establish the discrepancy signal; PDCA and the Model for Improvement turn it into a bounded, testable change.
From adaptation to reframing — OODA adapts under time pressure, double-loop learning revises the governing assumption, and reflective practice thinks inside the situation rather than after it.
Learning without narrowing too soon — experiential learning and self-regulated learning turn experience into capability; creative-cognition research warns against collapsing the option space before the evidence justifies it.
From practiced loop to inspectable architecture — autonomic computing makes the loop a designed, inspectable component rather than a habit carried in a practitioner's head.

Which loop corrects what

The chapter's central table pairs each tradition with the level of correction it actually reaches. Use it as a diagnostic: if your corrections keep landing at the action level while the failures keep recurring, the loop you are running is not the loop the problem needs.

Key sources

Chapter 3 draws on primary sources across seven traditions. Full entries are in the print book's Appendix D.

D68 - Wiener, N. (1948). Cybernetics: Or Control and Communication in the Animal and the Machine.
D69 - Carver, C. S., and Scheier, M. F. (1982). Control theory: A useful conceptual framework for personality-social, clinical, and health psychology. doi.org/10.1037/0033-2909.92.1.111
D70 - American Society for Quality. What is the Plan-Do-Check-Act (PDCA) cycle? asq.org
D71 - Institute for Healthcare Improvement. Model for Improvement.
D72 - Boyd, J. R. (2018). A Discourse on Winning and Losing. Air University Press.
D73 - Argyris, C. (1977). Double loop learning in organizations. Harvard Business Review.
D74 - Schön, D. A. (1983). The Reflective Practitioner: How Professionals Think in Action. Basic Books.
D75 - Kolb, D. A. (1984). Experiential Learning. Prentice-Hall.
D76 - Zimmerman, B. J. (2002). Becoming a self-regulated learner: An overview. Theory Into Practice, 41(2), 64–70.
D77 - Finke, R. A., Smith, S. M., and Ward, T. B. (1996). Creative Cognition: Theory, Research, and Applications.
D78 - Kephart, J. O., and Chess, D. M. (2003). The vision of autonomic computing. Computer, 36(1), 41–50.
D18 - Bainbridge, L. (1983). Ironies of automation. doi.org/10.1016/0005-1098(83)90046-8
D20 - Lee, J. D., and See, K. A. (2004). Trust in automation: Designing for appropriate reliance. Human Factors, 46(1).

Chapter contents

Movement 1: from discrepancy to bounded experiment — correct the result, then improve the process.
Movement 2: from adaptation to reframing — adapt under pressure, revise the rules, think in the situation.
Movement 3: turn experience into capability while preserving variation.
Movement 4: from practiced loop to inspectable architecture.
Different loops correct different things — the tradition-to-correction-level table.
What changes when AI enters the loop, and what the architecture inherits from all seven traditions.