A Computational Taxonomy for Distributed Environmental Governance: Object Classes, Required Operations, and the Representation of Absence - A Preliminary Discussion

Abstract

The preceding papers of this series develop an architecture for governing interdependent Earth–human systems: conditions as the object, three distinct structures of the field, coordination through reported change, assembly of heterogeneous knowledge, an event layer, criticality detection with graded duties, and observational capacity as a governed condition. Each of these commits the architecture to representing something, and the commitments are not arbitrary. This paper collects them. The method is analytical: from each preceding position the paper derives what an implementation would have to hold and what operations it would have to support, states the result as eight object classes with their required operations and their soundness constraints, and assesses existing standard families against them. The paper's central finding concerns what the architecture requires that data standards do not supply. Standards for environmental information are built to represent what has been observed; the architecture requires, at parity with observations, the representation of absences: couplings that obtain and are unwitnessed, ignorance a party acknowledges, claims that could not be established and by whom, transition classes an indicator suite does not precede, the silence regime in force on a channel, and the difference between a low reading and an unread instrument. A schema that cannot express these cannot carry the architecture's findings, since several of its central determinations are determinations about absence. Three further constraints are developed. Event streams admit only a causal partial order, so a representation timestamping to a global clock encodes a fact the field does not supply. Heterogeneous holdings must be carried without forced aggregation, so a schema permitting one value per field cannot hold the disagreement the architecture preserves as content. And assessments of change are horizon-relative, so an unqualified magnitude is not a well-formed claim. The paper surveys observation and sensor standards, alerting formats, provenance vocabularies, argument interchange, and graph and ontology languages, and maps their coverage against the eight classes. It closes with numbered requirements for a reference architecture. It reports no implementation and specifies no encoding.