Monday, April 16, 2007

Where is the ROI with XML Document Interoperability?

Up to this point, most of my posts about XML Interoperability have focused on identifying the problem space that a standards-based Interoperability Framework would attempt to solve from a technical perspective. But technical merit alone isn't a sufficient reason for an organization to adopt and embrace a technology.

Instead, typical organizations want solutions to real problems that have real implications to productivity, resources, costs, or profit. They want a certain degree of confidence that any technology investment will a) solve the problem, and b) they will see a return on that investment in a reasonable amount of time.

So where is the return on investment for XML interoperability?

Significant Investments Already Made

Consider the investment organizations have made thus far to migrate their content from proprietary formats to XML. Typically, the "up front" costs include:

  • content model development: DTDs, schemas
  • XML-based production tools: rendering engines, servers, XSLT, FO, etc.
  • new editing tools
  • content management systems
  • training

This is a substantial investment for many organizations with the intent that it will remain in service at least long enough to recover these costs, and hopefully longer.

Bill Trippe wrote:

Organizations are more diverse, more likely to be sharing content between operating groups and with other organizations, and more likely to be sourcing content from a variety of partners, customers, and suppliers. Needless to say, not all of these sources of content will be using the same XML vocabulary;

With organizations already vested in their own XML infrastructure, changes to this environment to support one or more additional XML vocubularies from different partners is bound to met with resistence.

Content Sharing Today

Despite this, partners do share content today. They convert (transform) the content from one vocabulary into another, or modify DTDs or schemas to fit the other content model. Yet these solutions make two assumptions:

  1. Each partner is a terminus in the content sharing pipeline
  2. Each partner's XML vocabulary will not change

In some cases, these assumptions are valid, and XML interoperability on the scale of one-way content conversion or schema/DTD integration is quite manageable and efficient. In this case, implementing the proposed XML interoperability framework may simply add more overhead than provide reasonable ROI. However, if either one of these assumptions is not true, then interoperability (and scalabililty) is a real issue and the framework may provide a mechanism for mitigating the costs and risks of trying to interoperate between numerous or changing vocabularies.

The Shortest Distance Between Two Points isn't a Straight Line

XML itself doesn't make any claims to enabling content reusability. But standards like DITA, and to some extent DocBook, provide mechanisms to enable content fragments to be reused in many target documents. For example, a procedural topic (section) written in for a User's manual could also be used in training material or support documentation. Frequently I've seen many cases where a Tech Pubs group is using a different XML grammar than "downstream" partners like Training or Support. There's the additive cost in time and resources to convert the content to fit their DTDs. More importantly, the semantics from the original source have now gone through two different transformations. It's almost like the children's game of "Telephone" where one child whispers a phrase in the next child's ear and so on down the line until the final child hears something entirely different. By enabling a shared interchange, you can reduce the number of semantic deviations to only one.

The Only Constant is Change

The other reality is that even when all partners are using XML standards like DITA, DocBook, ODF, or S1000D, the standards continually evolve, adding and changing content models to meet their constituents' needs. Since these standards aren't explicitly interoperable, the costs of managing changes between different standards goes up considerably. And here is another area where having an Interoperability format makes the most sense: If interchange is channeled through a neutral format, it can be interpreted to and from different standards (and versions) with a fewer number of transformation permutations. So if one partner moves to a different version standard, using an interchange format reduces costs and risks to your own toolchain and processes.

Lingua Franca

One of the design principles we've proposed with the Interoperability format is to leverage existing standards like XHTML. Because of this, we minimize the learning curve required for organizations to come up to speed to enable interoperability between different XML grammars. Just as English is the lingua franca for international business, aviation and science, a standardized interchange format for XML grammars provides a vehicle to enable content sharing between XML standards.

For example, an organization that has invested in a DITA XML infrastructure will not likely have a lot of in-house expertise on DocBook, ODF, or S1000D. Now, the amount of time and effort to enable content interoperability goes up signficantly. Add subsequent XML grammars into the interoperability mix and the level of complexity and cost are even higher.

With that said, a common XML interchange format isn't intended to be a translation vehicle, which is more in line with the Content and Processing interchange strategies I've described in previous posts. These strategies do have a place in the whole discourse around interoperability. They make perfect sense for linear, end-to-end interchange where both parties understand each other's "language" very well. And in reality, these strategies are likely to be more cost-effective than employing a "middle man."

Rather, a standardized XML Interoperability Framework will provide the highest ROI under the following conditions:

  • Content is leveraged/shared among many different consumers using different languages
  • A corollary to the above: Content sharing is non-linear
  • Business demands (time-to-market, lack of in-house expertise, partner relationships) make direct XML grammar translation cost-prohibitive