First, The Solution

With all the talk in Washington about reducing government bureaucracy, it's refreshing to see someone actually doing something about it. It started when IBM pulled support for ATMS, a mainframe-based document entry and archival system. This action prompted the Wisconsin Legislature to look for a new system. Andersen Consulting got their attention with a Sun-based client-server publishing system.

The new system, called Text 2000, based on Documentum (for document management) and Interleaf (for document composition) has improved and simplified the publishing of laws and statutes. Composing documents is now done in-house instead of being sent to an external agency, saving both time and money. The system is such an improvement, in fact, that representatives from other state legislatures and even private industry, such as automobile companies and pharmaceuticals, have visited the Wisconsin capital to learn from this example, which has been written up in several magazines, such as Computer World and Governing.

Now, the Problem

But planning a new system, even installing a new system, is not enough. The ATMS legacy data was completely incomprehensible to the new software, and the Wisconsin Legislature which is government, after all had a multitude of documentation in the old format. For this data to be usable, it would have to be converted to Interleaf. The job was given to DCL, Andersen Consulting's Business Partner. (Effective January 1, 2001, Andersen Consulting changed its corporate name to Accenture.)

"There is an obvious difficulty to a conversion like this," explains Mike Gross, head of development for DCL. "Interleaf has a more structured format than the source material. In other words, it contains more information. After all, they're using Interleaf to compose documents, whereas before they sent these documents out to be composed."

DCL was able to adapt modules from other projects to aid in the programming of this conversion. Mark Gross, president of DCL (and no relation to Mike), expands on this:

"We've converted ATMS to Interleaf before, but no two conversions are ever alike. Still, our general experience converting from word processor formats to highly structured formats has made us practiced at writing software to infer information from a document's appearance, so now it just becomes a matter of tuning. Like, what does a list look like in this document? What about a section heading? Let's revise the code accordingly. This is how we make programmatic solutions a viable alternative to doing everything manually. Projects with a certain minimum volume and complexity can be automated or partially automated at a significant savings."

Project Management

It would be wrong, however, to think that the work is done once the programs have been written. Janet Feld, DCL's Project Manager for the Andersen conversion, told me about her role.

"If a program could do the entire conversion perfectly, then there would be virtually nothing for me to manage because that's a one-step process. But complex conversions, particularly conversions where we add information to documents, are never consistent enough for a 100% automated solution. In some cases, it is theoretically possible to automate everything, but this isn't a puzzle, this is business. We only automate when it's cost-effective. That's why I used our editorial staff on this project. After the programmatic conversion, they went through the documents to check for unusual cases the software might have missed. Not only that, they performed complicated tasks that did not occur frequently enough to warrant programming. They did some table editing, for example. When they're done, then the files have to parse. These additional steps of editing and review, parsing and correction, complicate the process. After all, I have to track every file and what step it's in. Fortunately for me, DCL has had the foresight to develop tracking systems that can be adapted to each project."

Samples

Also important was the sample stage. Todd Harter of Andersen Consulting told me that "the samples we received helped set expectations for the entire project." This is one reason why DCL stresses samples and pilots. Usually, clients have rarely or never converted data before (in this case, that would be the Wisconsin Legislature, rather than Andersen), and therefore cannot predict all the issues or be sure of the results they need and the results that cannot be realistically or affordably achieved. Because of this, and because pilots reveal potential problems early on, DCL recommends a pilot stage to virtually every customer.

Thanks to the efforts of Andersen Consulting and DCL (and their own good judgment), the Wisconsin Government was able to implement a new system in a timely and cost-effective manner. A government project on schedule and within the budget ... impossible? Not in Wisconsin.

Want more information on this topic? Click here!