Indexing dynamic text?

["Geoff Hart" <geoff-h -at- mtl -dot- feric -dot- ca>]

I rarely disagree with Mark Baker, and generally learn something 
valuable from his posts (particularly when we disagree), so perhaps 
I'm about to learn something valuable again. In responding to the 
question about indexing a dynamic document (one in which certain 
fields or names or other word labels are filled in only when the 
document is compiled or customized), Mark commented:

<<You can't index a document until its content is determined. It is 
no use indexing the data used to generate the content. Indexing 
only makes sense once the document has been generated.  You 
can do one of 2 things: 1. Don't index. Indexing is just a primitive 
means of providing custom navigation of information. Generating 
information on the fly is a more sophisticated approach, which, if 
done properly, removes the need for indexing. (But, of course, just 
substituting terms isn't going to achieve this.)>>

Would I be correct in assuming that you mean "_manual_ 
indexing" doesn't make sense because of the volume of the work 
and the difficulty of indexing something "on demand, in real time"? 
If not, I think you've missed the point of indexing: to provide 
contextual clues that let readers find information. Concordances 
and full-text searches simply can't meet this need yet, and it'll be 
some time yet before they can. Although generating information on 
the fly is definitely more technologically sophisticated, it's not even 
in the same league as the sophistication a skilled human indexer 
brings to the job. Moreover, I've yet to see a text parser that has 
even an idiot's grasp of context; the only indexes I've ever found 
useful were human-generated, not mechanically assembled.

--Geoff Hart @8^{)}  geoff-h -at- mtl -dot- feric -dot- ca  (Pointe-Claire, Quebec)
"If you can't explain it to an 8-year-old, you don't understand it"--Albert Einstein