Event-Driven Craft Processes as State-Estimation Problems

Traditional crafts often organize continuous or hybrid developmental processes around discrete landmarks: crack in coffee roasting, spring in bread baking, foam transitions in fermentation, gelation in cheesemaking, cone behavior in ceramics, snap in curing, and color changes in heat treatment. This paper proposes that such landmarks should be analyzed as observations of latent process trajectories before they are assumed to be developmental causes. The distinction requires care. A physical transition is not the same object as a sensor signal; a signal is not the same object as a culturally named landmark; and none of these is itself a state estimator. The estimator is the practitioner or algorithm that maps observation and action history into a state estimate. Some physical events may also change the process, so causal mechanism and informational proxy are not mutually exclusive. The framework treats craft as partially observed, viability-constrained, hybrid control. It asks whether a traditional landmark carries replaceable information, marks an unavoidable causal transition, identifies one of several viable routes, or accompanies outputs that mask other valued characteristics. Those questions are domain-specific and require controlled intervention, matched outcomes, practitioner validation, and explicit uncertainty. This revision repairs the event/estimator category error; separates state, control, disturbance, measurement, and landmark classification; distinguishes a safe constraint set from its viability kernel; makes instrumentation and linguistic-transmission claims empirical rather than automatic; integrates human validation, Semantic Manifold Topology, GMB/GSM, and the realization funnel; and supplies a comparative research protocol. No domain experiment or external literature verification was performed.