The SECI Spiral

In the AI Story

The SECI model emerged from Nonaka's systematic observation of Japanese companies that outperformed Western competitors in the 1970s and 1980s — Honda, Canon, Matsushita, Sharp. Western management theory, still dominated by Herbert Simon's computational model of organizations as information-processing systems, could not explain why these firms consistently produced breakthrough innovations. Nonaka's insight was that the relevant activity was not information processing but knowledge creation, and that knowledge creation required the conversion between forms of knowing that Western theory had collapsed into a single category.

The four modes correspond to distinct organizational practices and physical settings. Socialization occurs in what Nonaka later called originating ba — spaces of shared physical presence where tacit knowledge flows through observation and imitation. Externalization occurs in dialoguing ba, where peer conversation crystallizes tacit insight into communicable form. Combination occurs in systemizing ba, where explicit knowledge is reconfigured through documentation, databases, and now AI tools. Internalization occurs in exercising ba, where practitioners convert explicit knowledge into embodied skill through deliberate practice.

The spiral's most important feature is its interconnection. The quality of Combination depends on the quality of Externalized inputs, which depends on the depth of tacit knowledge built through Socialization and Internalization. Break the cycle at any point and the remaining modes produce diminishing returns. This structural interdependence is what makes the framework diagnostic rather than merely descriptive — it specifies where in a knowledge-creation process a distortion is occurring and what its downstream consequences will be. Segal's Orange Pill documents a twenty-fold productivity multiplier at Napster's Trivandrum training, which SECI analysis reveals as overwhelmingly a Combination gain.

The framework was built for human organizations populated by human knowers. AI's arrival in 2022–2025 represents its most consequential test. Large language models perform Combination at unprecedented scale, contribute meaningfully to Externalization as conversational partners, and are structurally absent from Socialization and Internalization. The spiral that AI produces is one mode spinning rapidly while the others lag or atrophy — a deformed wheel rather than an accelerated one.

Origin

Nonaka first articulated the framework in a 1991 Harvard Business Review article, 'The Knowledge-Creating Company,' and developed it fully in the 1995 book of the same name co-authored with Hirotaka Takeuchi. The intellectual lineage runs through Michael Polanyi's The Tacit Dimension (1966), which provided the distinction between tacit and explicit knowledge, and Kitarō Nishida's concept of basho, which Nonaka adapted as ba.

The empirical foundation was decades of fieldwork in Japanese companies — particularly the bread-making machine project at Matsushita, where software developer Ikuko Tanaka apprenticed with a master baker to absorb through her body the kneading technique that no specification could capture. The case became canonical because it demonstrated all four modes in sequence, producing a commercial product through knowledge conversion that Western information-processing theory could not explain.

Key Ideas

Tacit and explicit are categorically different. Not two points on a spectrum but two kinds of knowing that interact through conversion but never fully reduce to each other.

The spiral ascends through cycles. Each complete rotation — Socialization, Externalization, Combination, Internalization — raises the organization's collective understanding to a higher level of sophistication.

Each mode depends on the others. Combination without Socialization produces recombination without grounding; Externalization without deep tacit knowledge produces plausible emptiness; Internalization without explicit inputs has nothing to embody.

The framework is diagnostic. It does not prescribe behavior but reveals where in a knowledge-creation process distortions are occurring and what their consequences will be.

Acceleration of one mode without balance deforms the spiral. The wheel that is round on one side and flat on the other rotates without rolling smoothly — producing output without understanding, speed without depth.

Debates & Critiques

Contemporary scholars have attempted to extend SECI for the AI era through frameworks like GRAI (Böhm and Durst) and the GenAI SECI model (Ogawa et al.), which split each mode into human and machine perspectives. Critics argue that SECI's Japanese cultural origins make it difficult to apply in Western contexts where shared physical space and long employment tenure cannot be assumed. Nonaka's response was that the principles are universal even if the specific implementations must be culturally adapted.