Ann,
This is actually in relation to Dennis Nicholson comment 'we need DDC in HILT' - i.e. how I think this may be explained
From developing uses cases it has become apparent that it is a good idea to have some subject terms that are 'words'. The original remit of IESR was to be exclusively a machine-to-machine service, and machines are quite happy to communicate in codes (although even there a human software developer is involved at some point). But in reality there are a lot of use cases where a person uses IESR for discovery. And people generally use words not codes for searching. So, unless a terminlogy service like HILT comes into play within the use case, 'word' based subject schemes seem more usable. (Though I guess people may also use simple, ie high-level, code schemes like JACS.)
The use of classification does not imply that users are not searching using words.
The problem with words is that they cannot be automatically linked in any meaningful way if they are not attached to some semantically explicit structure.
So when users need documents related to e.g. 'salmon' - we may assume that they don't look for "Sea Air Land Modeling Operational Network" but for fish. If looking for fish, however, they may be interested in specifics such as "Salmo salar (Atlantic salmon)" or "Oncorhyncus (Pacific salmon)" - but also in all documents related to the family of fishes - Salmoniforme or sub-class Salmonidae.
One notation in classification holds links between these concepts together - and all words we normally use to express this concept can be attached to this notation.
But most importantly, and this is why Dennis says that HILT needs DDC, is that classification also holds information about perspectives hierarchies in which our "salmon" can be found across the universe of knowledge. This is something thesaurus on its own cannot deal with.
So classification notation will not only enable that all documents related to salmon (irrespective the character string used in searching) are found but would also establish the difference between documents on salmon in zoology from salmon in sport fishing, animal husbandry, in cooking or in ecology, or salmon as an object in works of art.
Obviously, if any controlled vocabulary (including classification) is to be used then some kind of vocabulary management component has to be planned for. It does not matter whether this is a subject authority file or terminology service or vocabulary registry. Costs depend on how many repositories are sharing the same centrally/collaboratively managed vocabulary.
I hope I am not missing some big part of the picture here...
cheers
Aida