Messaging standards are like telephone systems: they can function sensibly only if there is a large network of willing users. This means that to succeed such artifacts must be marked by clear use of language and clear documentation. The RIM documentation, as we have seen, is systematically ambiguous. What is needed, if HL7v3 is to satisfy its need for a uniform information representation that is coherent, clear and implementable, are two separate, though of course related, artifacts, which might be called 'Reference Ontology of the Healthcare Domain# and 'Model of Healthcare Information', respectively. The former would include those categories, such as thing, process, anatomical structure, disease, infection, procedure, etc., needed to provide a compact and coherent high-level framework in terms of which the lower-level types captured in vocabularies like SNOMED CT could be coherently organized. The latter would include those categories, such as message, document, record, observation, etc., needed to specify how information about the entities that instantiate the mentioned types can be combined into meaningful units and used for further processing. HL7's Clinical Document Architecture could then be related in an appropriate way to this Model of Healthcare Information, thereby avoiding the current counterintuitive stopgap, which forces a document to be an Act. And HL7's Clinical Genomics Standard Specifications could similarly be related in an appropriate way to the Reference Ontology of the Healthcare Domain, and it could thereby avoid the problematic identification of an individual allele as a special kind of Observation.