Lists Home |
Date Index |
On Tuesday 18 February 2003 12:01 pm, Jonathan Robie wrote:
> >The main thing is that if people use fonts (and hopefully styles)
> >consistently, you can work with the data. There are reasonable, and
> > improving conversion tools that can work with that as a basis.
> When I write a document from scratch, I do.
. . .
> For me, it's easier to just write it using text in the first place,
> but we're all angle-bracket geeks here. Most authors don't think
> like we do.
Quite right. I learned that lesson the hard way ;-)
> FWIW, after insisting that I use Microsoft Word, a publisher is also
> telling me I have to make sure, by hand, that I am following all the
> guidelines that an XML editor would have enforced for me....
That happens. It's worse in some cases because they also have poor styles
(everything is 10 or 12pt courier) so visual people (me included) get totally
lost. In a situation like that, I would(did), of course, write a few macros
to convert from *my* styles to *their* styles, so I could write comfortably.
Here's two observations though:
1) Most publishers have to deal with far more variety than you might
expect.... many people I've seen have trouble using styles *at all*, and find
even font consistency hard. For many people, they necessarily escape the
bounds placed upon them, because they cannot work any other way.
2) Relatively few people have the skill level that xml-dev folk have... so
they can't even write macros, or conversion tools to make their life easier.
They almost certainly can't appreciate the value in conformance enough that
they might even desire to do so.
Both (1) and (2) lead me to believe that even if really cool XML authoring
tools existed, it would still be very hard to count on the data being
correct. You will *have* to, at some point, do further manipulation.
Another data point: tagging accuracy for data-entry folk (double keying) is
typically around 90%, and these people are *paid* to care. Even in the world
of relational databases, I read that the overall data accuracy in most
corporate databases is something like 95-98%.