Knowledge Creation

The book began with a practical puzzle: why sophisticated hotel companies, despite rich data and expensive analytical tools, so often failed to convert them into durable profitability. Yin built models, tested them against real hotel portfolios, sat in operating meetings, and watched general managers ignore dashboards that had cost millions. The models could be technically correct and operationally inert at the same time. He then found the same structural gap in financial services, industrial manufacturing, higher education, and public administration: elaborate theory and analysis on one side, complex reality on the other, joined by a thin and often broken bridge. What started as a book about research methodology became a theory of what knowledge is, why some knowledge grows while other knowledge decays, and why civilizations progress.

Reality-Driven Knowledge Evolution is the master claim: knowledge advances only when every phase of its creation, evaluation, and revision remains answerable to reality — not to publication incentives, citation networks, methodological conformity, or revenue. The book contrasts RDKE with two incomplete modes. Literature-driven production starts from gaps in the literature, tests theories against other theories, and counts publication and citation as success; the book grants it archival value but denies it can alone produce living knowledge. Market-driven production lets paying clients define the problem and revenue validate the theory; it creates real value but underproduces knowledge that serves flourishing without serving profit.

The operational core is a recurring cycle in which every phase carries a discipline and a characteristic failure. Reality demands disciplined perception before a familiar theory prematurely names the phenomenon — the failure is premature framing. Question demands a specific, investigable puzzle with subject, scope, and criterion of success — the failure is over-generalization. Thinking demands genuinely different, falsifiable candidate explanations — the failure is premature convergence. Theory demands a determinate account that can be criticized — the failure is elaboration for its own sake. Practice demands using the theory where its implications are genuinely at risk — the failure is decoupling. Evidence demands reading outcomes against the theory, especially anomalies — the failure is confirmation bias. And the new reality is not the original starting point but a richer one. The strategic rule: reality is the first and last phase of every cycle.

One of the book's sharpest diagnostics is that high output, rigor, and consensus can be symptoms of knowledge stagnation rather than health. The tell is simple: nobody is being surprised any more. This sits inside a full life cycle — Birth → Growth → Maturity → Stagnation → Decay → Regeneration → Evolution — with the caution that some decayed knowledge should be replaced rather than rescued.

On AI, the formulation is that AI has industrialized skill; capability remains scarce. The book distinguishes an augmentative pattern, in which AI expands what humans can attempt while they retain and build underlying capability, from a substitutive pattern, in which near-term productivity gains hollow out the very ability to recognize AI error.

What's inside

The four original models: the Knowledge Value Pyramid (Information → Knowledge → Understanding → Wisdom → Civilizational Contribution, which the book argues modern institutions have inverted by treating countable outputs as the summit); the Knowledge Evolution Spiral (a loop is stationary; a spiral moves); the Knowledge Ecosystem of seven interacting actors — enterprises, universities and research institutes, AI systems, governments, markets, communities and civil society, and individual practitioners and researchers — bound by six couplings; and the Knowledge Life Cycle.

The book is in explicit dialogue with Popper, Kuhn, Polanyi, Nonaka and Takeuchi, Simon, Argyris and Schön, Senge, and Peirce, and does not claim to start from nothing. Because strict falsification is often unavailable in complex organizations and at civilizational scale, RDKE broadens the test to disciplined revisability through patterned encounters with reality over time.