Introduction. A fitness landscape maps each possible configuration of a system to a performance value. Kauffman’s NK model makes the landscape’s ruggedness tunable: N components each contribute to fitness, and each contribution depends on the component’s own state and on K others. With K = 0 the landscape has a single peak and local search finds it; as K rises the landscape becomes rugged, local search stalls on one of many local peaks, and the founding configuration determines the outcome. Levinthal brought the model into organisation theory, and Rivkin, Siggelkow and Levinthal used agent-based simulation to study how decision structure shapes search. Three results matter. A hierarchical organisation with delegated decision rights, interdependent domains and divergent local incentives can come to rest at a sticking point that is not even a local peak of the organisation’s own landscape. Temporary decentralisation followed by reintegration outperforms both permanent centralisation and permanent decentralisation, and the advantage grows with environmental change. And because different structures have different sticking points, a sequence of structures outperforms any fixed structure even in a stable environment: changing structure dislodges the organisation and restarts search. The models are abstract, and their parameters (N, K, incentive divergence, shift size) are choices that a reader should check before transferring a result.
Important authors. Stuart Kauffman, then at the University of Pennsylvania and the Santa Fe Institute, introduced NK landscapes; Sewall Wright’s adaptive landscape of 1932 is the ancestor. Daniel Levinthal (Wharton) brought the model into management research and, with Nicolaj Siggelkow (Wharton) and Jan Rivkin (Harvard Business School), produced the simulation results that bear on the VSM. Giovanni Gavetti, Sendil Ethiraj and Felipe Csaszar are among the later contributors; Oliver Baumann’s group has reviewed the literature.
Importance for cybernetics and the VSM. The VSM’s central tension—System One autonomy against System Three cohesion—has been modelled explicitly for twenty-five years, and the results contradict its structural answer. The VSM has no representation of a performance landscape, so it cannot express a sticking point, and its design manufactures them by construction: bounded autonomy under central constraint with no mechanism to detect that search has stopped short of a local optimum. More fundamentally, the VSM prescribes a fixed architecture and treats deviation as pathology; this literature’s finding is that the architecture should vary in time and that fixity is the pathology. Cybernetics generally treats structure as the variable to be designed once, not sequenced.
Importance for the article. §3.3 (rival mechanisms) states the sticking-point and structure-sequence results as rivals to the VSM’s fixed architecture and notes that alternation also answers Perrow’s claim that decentralised-then-centralised operation is infeasible—a three-way disagreement the paper stages rather than settles. §10.5’s candidate register lists “a fixed five-function architecture is the design for adaptation” against “sequences of structures outperform any fixed structure”, with replication of the NK simulations including a VSM-specified fixed structure among the alternatives as the cheapest discriminator. The import is from VSM vs Complexity Science §2 (C6). A reviewer will press on whether the VSM really prescribes a fixed structure or only a fixed set of functions realisable by changing structures (the degeneracy defence of T7.3), on the sensitivity of NK results to parameter choices, and on the fact that “structure” in these models means decision-rights allocation, not the VSM’s channel architecture; the replication receipt has to specify how a VSM structure is encoded.
Sources in the reading list.
- the NK model and the ruggedness argument; Tier C, read the NK chapters only.
- the paper that brought NK landscapes into organisation theory; adaptation versus selection, tight versus loose coupling.
- the sticking-point result and the three conditions that produce it.
- the reintegration result and its growth with environmental change; the answer to Perrow.
- sequences of structures beat fixed structures even in stable environments.
Other important sources and authors.
- Wright, S. (1932). The roles of mutation, inbreeding, crossbreeding and selection in evolution. Proceedings of the Sixth International Congress of Genetics, 1, 356–366 — the original adaptive landscape.
- Kauffman, S. A., & Levin, S. (1987). Towards a general theory of adaptive walks on rugged landscapes. Journal of Theoretical Biology, 128(1), 11–45 — the NK model’s first full statement.
- Gavetti, G., & Levinthal, D. (2000). Looking forward and looking backward: cognitive and experiential search. Administrative Science Quarterly, 45(1), 113–137 — search with a cognitive model of the landscape; the nearest thing to a System Four in this literature.
- Ethiraj, S. K., & Levinthal, D. (2004). Modularity and innovation in complex systems. Management Science, 50(2), 159–173 — how a mismatch between the designed decomposition and the true interdependence structure degrades search; relevant to recursion boundaries.
- Baumann, O., Schmidt, J., & Stieglitz, N. (2019). Effective search in rugged performance landscapes: a review and outlook. Journal of Management, 45(1), 285–318 — the review of the organisational NK literature.
- Csaszar, F. A. (2018). A note on how NK landscapes work. Journal of Organization Design, 7, 15 — a short technical primer for anyone who must replicate a simulation.