Real value (was implementing single-source) - demonstrated!

[HALL Bill <bill -dot- hall -at- tenix -dot- com>]

Even though the weekend has passed, I can't resist adding my $0.02 to the
debate between Andrew Plato and Dan Emory about the potential $M value of
implementing XML content management systems in a technical documentation
environment.

As I remember the thread, Dan presented a number of clear benefits to be
gained from implementing XML systems, with which I am in 100% agreement.
Andrew argued that such systems were massive overkill, suggested that the
solution was to train authors to write better, and doubted that anyone could
actually demonstrate the achievement of real value from implementing such
systems.

Based on long experience with structured information management (starting in
1982/83 with a system based on WordPerfect merge tables) and recently the
very happy implementor of RMIT University's native XML database and content
manager - Structured Information Manager (SIM - http://www.simdb.com) I have
some real numbers to offer.

The scenario is authoring, management and delivery of maintenance routines
and associated logistics technical data for a project to build and support
10 frigates for the Australian (8 ships) and New Zealand (2 ships) Navies.
Each ship requires approximately 2,000 equipment-related maintenance
routines - with individual routines ranging from 1-2 printed pages of text
to more than 70 pages filled with complex tables and figures.

Because some of the data was already structured in the WordPerfect merge
tables, we were able to 'single-source' more than 20 different deliverables
and extracts from the one data set, where many of the output requirements
(including complete replacement of our paper text and data deliverables with
electronic ones) were completely unanticipated when the document structure
was determined. Alone, this version of 'single sourcing' probably saved the
company several $M in documentation costs.

However, because the WordPerfect system could not cope with the need to
maintain ship-specific differences between the documents as a consequent of
language differences between the two customers and engineering changes from
one ship to the next, we found it necessary to independently maintain each
ship-set of documents. This year, as we are completing documentation for the
delivery of our 5th ship we were faced with the need to manage 10,000 and
deliver separate maintenance routines.

By designing an SGML DTD to meet our content management requirements (it
would have been XML had the original specs been developed a year later),
migrating the WordPerfect data to the SIM XML environment and implementing
document applicability to specific equipment configurations, effectivity to
specific engineering changes, and dual language elements, we reduced our
requirement to manage and deliver 10,000 separate ship-specific documents
(increasing each year by another 2,000) to manage ~1,800 class documents and
deliver only about 50 routines (net changes against the class set). The 80%
reduction in management and 98% reduction in delivery for Ship 5 were
achieved through phase 1 of our project which is based single-sourcing from
SGML managed at the file level.

We had a major authoring task to convert 4000+ Australian and New Zealand
texts to ~1,800 single class documents - particularly since we had also
accepted the obligation to review and substantially beef up the use of
warnings and cautions at the element level in the documents.

Andrew argued that the cost of training authors to work productively in such
a system would be prohibitive. The fact is that it took us one day to teach
authors (who had no prior contact with either SGML or FrameMaker) the
mechanics of using the SIM-based work flow environment and the basics of
structured authoring in FrameMaker+SGML's authoring interface, and a week or
less of ad-hoc support before the authors were exceeding their productivity
in the WordPerfect environment.

Early in the new year we will be implementing content management and re-use
at the level of single SGML/XML elements. This will reduce the volume of
text under management by another 50-70% and give us much tighter control
over content.

However, even though we are still managing documents at the file level in
our Phase 1 system, the structured authoring environment and the ability to
easily query the system for similar texts has led to a very substantial
improvement in the quality of our delivables in the 31/2 months it took 4-5
authors to transform 4000+ ship specific routines into 1,800 class routines.
In addition to beefing up the warnings and cautions (a simple select
point-and-click operation for standard texts) we were also able to
substantially improve the quality to ensure better consistency between text
references and metadata (e.g., labour resource requirements, spares tools
and materials lists, etc.). 

We have also found that we can generate ad-hoc extracts from the SGML data
for a small fraction of what it cost to produce similar extracts from the
WordPerfect environment (e.g., an hour or so of programming using SIM's ACE
language vs several days or even weeks using WordPerfect merge/macro code).

Did we get real value? Andrew is correct in one area. Implementing such a
system is expensive. However, even though we were very late in the lifecycle
of the ANZAC Ship Project we anticipated breaking even on the remaining
IDENTIFIED work against the data. Since beginning the implementation we have
already successfully completed major unanticipated work, such as the upgrade
to warnings and cautions not included in our original costing model.

Also, now that we have the XML content management technology in-house, the
savings from implementing it for new projects will significantly impact the
overall project cost. Probably 5-10% of the cost of major defense projects
is documentation related. By implementing truly effective management and
reuse of our corporate knowledge as enabled by structured authoring and
content management across the whole project life cycle, we anticipate we can
reduce documentation related costs by up to 50%. For a project the size of
the ANZAC Ship Project ($A 5 BN in 1989 $) that would equate to savings on
the order of $A 100's of millions.

Obviously, some content management applications are better than others.
Because I mistakenly passed the product off as an academic research project
rather than commercial software, SIM wasn't even on our horizon when we
started our system selection process. 

Only after deciding that none of the well known international products would
give us what we wanted within our risk and cost recovery constraints did we
look thoroughly at the local product to discover that in many areas it was
(and still is) a generation ahead of all the other products. SIM was
developed from first principles to manage and process structured text.
Essentially the entire system (including the repository itself) was
programmed in RMIT's own high level object oriented language (Ace) which is
also used as its integral scripting language. No third party applications
are used anywhere in the architecture - ensuring that everything has been
optimised to work together and greatly simplifying implementation and
maintenance labour and costs. I have given more details SIM architecture and
capabilities in earlier postings to Techwrl, and these can be found on the
archive by searching on SIM or my name.

In any event, our implementation is still on budget and schedule, and
although we are still waiting on some of our desired functionality, what we
already have more than justifies the expense of the project.

If anyone is seriously interested in the benefits from XML/SGML content
management, I can post copies of the case study presentation I gave last
week at the XML Asia Pacific 2000 Conference in Sydney.


Bill Hall
Documentation Systems Specialist
Integrated Logistic Support
ANZAC Ship Project
Tenix Defence Systems Pty Ltd
Williamstown, Vic. 3016 AUSTRALIA
Email: bill -dot- hall -at- tenix -dot- com
URL: http://www.tenix.com

^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
Develop HTML-based Help with Macromedia Dreamweaver! (STC Discount.)
**NEW DATE/LOCATION!** January 16-17, 2001, New York, NY.
http://www.weisner.com/training/dreamweaver_help.htm or 800-646-9989.

Sponsored by SOLUTIONS, Conferences and Seminars for Communicators 
Publications Management Clinic, TECH*COMM 2001 Conference, and more 
http://www.SolutionsEvents.com or 800-448-4230

---
You are currently subscribed to techwr-l as: archive -at- raycomm -dot- com
To unsubscribe send a blank email to leave-techwr-l-obscured -at- lists -dot- raycomm -dot- com
Send administrative questions to ejray -at- raycomm -dot- com -dot-  Visit 
http://www.raycomm.com/techwhirl/ for more resources and info.