Re: Exceptable Errror Wrates

[Robert Plamondon <robert -at- PLAMONDON -dot- COM>]

Nobody should walk into the Quality fire without clutching Deming's
OUT OF THE CRISIS to their bosom.  The task of learning Quality
Assurance and applying it to those parts of Tech Pubs where it
makes sense to do so is no idle task.  You either need to pony
up the appropriate level of expertise and effort, or punt.  It's
far more scientific to sprinkle chicken blood on one's monitors
in the hope of generating good juju than it is to make up Quality
metrics without a solid grounding in the field.

The Quality biz is aimed squarely at manufacturing and other fields
with very repetitive tasks -- Customer Support is one example, perhaps.
But the essence of manual creation is not a set of repeated assembly-
line tasks; one manual is not like another in the way that one
camshaft is like another, or one day running the same electric power
plant is like another.  The nature of the product and the methods used
in creating it vary from title to title.  This frustrates the
more simplistic approaches to Quality Assurance.

Sure, you can apply the usual stuff to compositing, but this is hardly
where the value in Tech Pubs lies.  If the company has an editorial
staff, sure, why not -- but when manuals are created by writers more
or less working in self-editing mode, the mechanisms for assuring
quality at the copy-editing and compositing level are missing, so
what's the point of measurement?

Anyway, a statement that "the only acceptable error level is zero" is
fatuous.  It's the same as stating that "we will spend whatever it takes,
accept whatever delays are necessary, and test the manual until we
are certain that no errors remain."  A decision to stop looking for
errors at a certain point and ship the document is the same thing as
declaring the probable level of remaining errors to be acceptable.

        -- Robert
--