As with most W3C efforts, you can track the progress of work within any one activity by the state and version of the documents released. The first document released by the WebOnt group on OWL listed a set of requirements for an ontology language, followed by documents for test cases, abstract syntax and semantics, and, finally, a language reference and user guide.
The roots for OWL exist in the OWL Use Cases and Requirements document, released in July 2002 and recently updated. According to this document, we've been working with ontologies all along by using vocabularies such as ones I've used in the book like Dublin Core and PostCon. These are ontologies because they define the data for a specific knowledge domain, which is what the Use Case and Requirements document defines as ontology.
Specifically, ontology encompasses four concepts:
Relationships between classes
Properties of classes
Constraints on relationships between the classes and properties of the classes
When you consider that these concepts can be used, equally, with RDF and RDFS, you can see why there is some confusion about where RDFS ends and OWL begins. The Use Cases document, while demonstrative of applications facilitated by the use of an ontology, didn't exactly help with clarifying when to use OWL and when to use RDFS, other than suggesting use of RDFS for defining OWL and then using OWL for everything else.
In addition to use cases, design goals given in the document were:
Ontologies must be sharable, so that more than one business within a particular business domain could use the same ontology defined for that domain.
Evolving ontologies should be given version numbers and the schema defining the ontology given a separate URI for each new version (such as PostCon with its http://burningbird.net/postcon/elements/1.0/version). Ontologies would then be related through the use of rdfs:subClassOf.
Ontologies must be interoperable.
Inconsistencies in ontologies must be detected automatically to prevent them from occurring.
Ontologies must balance expressivity and scalability.
Ontologies must be consistent with other standards.
Internationalization must be supported.
None of the use cases or design goals is overwhelmingly complex, except possibly testing for inconsistencies. The next document in the series released by the working group then contained test cases to see if the OWL met the various design goals.