Interchange vs Analysis

[mcarr at allette.com.au]

Bernard wrote:

> I generally have to agree with what is written below, but I have to take a
> minor exception on (a) regarding 'the line that you can do it with DITA or
> DocBook'. Out of the box does a good job, but unfortunately the 'good job'
> is at scaring people away.

I consider DITA to be an interchange format. If two organisations can
figure out how to convert their own structure to and from DITA, they can
freely exchange data. Add five more organisations and impose the same
requirement on them and everyone can exchange data, whereas in the past,
each organisation would have to code the conversion for all of their data
partners. This is very powerful and very useful, but it doesn't replace
the structure that the organisations use on their own side.

The long and short of it is that saying "our data can be characterised by
nested blocks" is not a replacement for analysis of a dataset. The more
generic the structure you use, the less representative it is of your
particular dataset. The more you customise the generic structure, the less
interoperable it is with tools and the less you're able to interchange
meaningfully with partners who require your data anyway.

Proper analysis and structure design should be a selfish exercise. You can
always develop an interchange strategy down the track - all you're ever
going to have to do is map your structure into something that someone else
can use. First priority is to develop a structure that satisfies your own
needs, not those with whom you intend to trade data. DITA is fine, but it
frequently gets mispositioned, IMHO. I do try to scare people away from
misusing it - that's true... ;-)


Marcus