Essentials of a Publishing System
In part 1 of this series, we discussed some of the key problems of capturing and sharing information - problems like delivering information to multiple types of media, making updates faster and easier, and reducing the cost and time to translate and publish.

Given those challenges, what should we do?
In this part, we'll describe the essentials for solving these problems, which include building a "single source" of information that eliminates redundancy, creating information in reusable modules, and automatically assembling and publishing information for multiple audiences and multiple media.

Let's take a trip through the Wayback Machine to 1960.
You find yourself responsible for publishing a technical manual. You begin the project by figuring out what information the book will contain and assigning various sections to subject matter experts (SMEs). The SMEs are mostly engineers, and they write out their content in longhand on foolscap. (If you don't know that "foolscap" is writing paper, congratulate yourself on your youth and vitality.)

The secretarial pool types out the handwritten notes from the engineers. An editor - that's your job too - reviews and edits those typewritten notes and then sends them back to the secretarial pool to re-type.

You assemble all of the sections in order and proofread them again to make sure everything flows and that you mark all the typing errors. Then you add typesetting instructions (also known as "markup") in the margins to indicate the formatting of each section of the document. The markup specifies font, line measure, leading and other formatting instructions that you copy from a set of corporate guidelines.

You then send the typewritten documents with your markup to a printer. Their typesetter works on a Linotype machine to create typeset versions of your documents. The printer returns your copy in "galleys," which are just long rolls of paper with your document printed out in a single column.

Then a designer or layout artist cuts up the galleys and lays them out onto pages. An editor - you again - reviews the layouts both for typesetting accuracy and design. Then someone - poor you - scans through the layouts and builds a table of contents and an index, which you then send to the typing pool, review for accuracy, send to the typesetter, give to the designer to lay out on pages and then review for accuracy again.

After that, you send the entire book to the printer who makes all of the corrections, lays out the type to match your page layouts, and produces a copy of the book for someone - you know who - to check again. You send the book back, the printer makes any further corrections, and finally - finally! - he prints it.

Mercifully, your book contains no illustrations.

Back to the Future
Get back into the Wayback Machine and return to the present. Same job. Different millennium. Now you have desktop publishing tools that let you do everything yourself. The specialists are all gone except for the SMEs. No secretaries, no typesetters, no layout designers. There are other benefits: you no longer have to create a table of contents and index manually and you have much less proofreading to do.

What work remains? You create content, apply formatting and lay out pages. But there's more.

In the last 45 years, the demands on you have changed. It's no longer enough to produce a revision of the book every year or two; update cycles are now measured in months instead of years. You have to produce the book both in print and online. The book is translated into three additional languages and some day you'll have to produce it in eight. You're under pressure to produce customized versions for various types of readers, and some day you know that you will have to let your customers build their own versions. And you have to do all this while facing escalating pressure to deliver the book faster at lower cost with fewer people.

You need to find quicker, better ways to get your job done. So you take a page from the office automation projects of the past - the projects that eliminated adding machines, pencils and handwritten account ledgers. You need to eliminate the redundancy and rework that comes from copying and pasting the same content. You need to stop copying data from its original source into your documents. You need automation - lots of it, everywhere that it makes sense.

And you need to achieve all of that while gaining the flexibility to produce an ever-wider variety of information products.

So just as we are inspired by the benefits of office automation, we are also inspired by the approach. The key to successful automation, regardless of what process you're automating, is an absolutely consistent structure and data format. While the formatting of an individual business document suggests a structure that's obvious to anyone looking at it, authors freely alter structure and formatting to suit their circumstances and tastes. The result is that documents of the same type will have similar but not identical formatting and similar but not identical structures. These inconsistencies, however small, make automation impossible.

Another key to successful office automation is modularization. By splitting up data into pieces, it's not only easier to manage but also easier to incorporate into various reports. Translating that to documents, it means splitting up them into reusable components big enough to be worth managing separately but small enough to reuse in multiple instances.

By reusing information instead of re-creating it, you can create a single source of information so that making just one change can update multiple documents. A single source also leads to the reduction or elimination of redundancy, which not only lets you reduce translation costs to only those parts that have changed, but also ensures the integrity and accuracy of your information. Can you imagine the problems your colleagues in accounting would face if data such as net income appeared separately in multiple locations, each with no connection to the other? How much work would it be to track them all down for updating? What are the risks of missing one?

Modularization is also critical for personalization. Creating information out of smaller components lets you set up a system to assemble those components dynamically to suit the needs of various audiences.

For modularization to work the components must be interchangeable and they must fit correctly into their "parent" document. For example, you may choose to create two different sizes of modules such as warnings and topics, where warnings fit into topics and topics fit into books.

Also key to automation is separating the information from its presentation. In office automation, that means storing data in databases and extracting it for various purposes. Accountants can produce tabular reports, charts and graphs in virtually endless combinations, even though the data in its raw form would be unintelligible to them. Applying this principle to documents means that you can use the same information in different kinds of documents with different formatting, all without touching the information itself.

Let's review the essentials of your automated publishing system:

• Modularization - enables reuse, single-sourcing and personalization
• Structure and consistency - essential for automation
• Separation of content from presentation - required to deliver the same information to multiple different media types of media
• Automation - helps you assemble information for multiple audiences and publish to multiple types of media without human intervention
• Single source - helps you eliminate the ongoing time and expense of maintaining redundant information while ensuring its integrity

XML also enables the separation of content from its presentation. XML represents information in a "media neutral" form that's not constrained by the limitations and capabilities of any particular medium, so you can create information in its "pure" form and process it separately to produce information products.

One of the advantages of this separate process is that through automation, the information can be presented with absolutely consistent formatting regardless of author, and it can take full advantage of the capabilities of each medium.

Through XML, you can also specify the size of your reusable information modules. And you can do this in a consistent way so that you can easily interchange one module for another.

Another key ingredient to an automated publishing system is the software that makes it all work - software that lets you create information in reusable modules of XML, assemble those modules for specific audiences and publish the result for multiple types of media.

The final piece of the puzzle involves changing the roles and responsibilities of the people involved in creating and improving your information. This article went into great detail about how roles changed between 1960 and now, and in a future installment you will see that roles will change yet again..

In part three, we'll look at the process of developing a solution: developing data models (DTDs, schemas), designing stylesheets and integrating various tools and technologies.