Using Content-Centered Development to Meet IT Development Challenges
Florida’s Senate Employs New Best Practice in Automation
By virtually any measure, information technology (IT) has advanced at a dizzying pace. Unfortunately, despite this burgeoning technological prowess, the success rate for major IT-based systems projects has remained nearly flat. Many projects fail and many more fall far short of their goals. To quote a study by the Congressional Office of Technology Assessment, completed some years ago but still applicable: “…it was difficult if not impossible to find a federal IT project that can be called a success.” The standard answer to IT project shortfalls has been to blame those directly involved or to point to some perceived lack of technology. Accumulating experience suggests, however, that these approaches will not solve our problems. This is the story of one organization that decided to find a better answer.
Legislatures Face Daunting Automation Challenges
Among large organizations, few face automation challenges at levels as daunting or instructive as legislatures, especially those at the state level. A modern state legislature is a complex, multi-segmented organization whose major products-legislation and communication-are highly labor-intensive, but whose primary marketplaces-the state’s citizens, business communities, and other organs of government-demand nothing less than excellence. Moreover, state legislatures are normally made up of two chambers purposely designed, from the founders forward, to operate very differently. Legislative chambers have different personnel cycles, different funding methodologies and approaches, different legislative procedures, and generally different views of how the world operates. And, of course, these chambers have agendas that make them highly protective of their individuality. In IT, the upshot of this for many state legislatures is a patchwork of technology, applications, and procedures cobbled together over time in an attempt to support the increasing flow of information and legislation. For the heroic staffs charged with keeping this ship afloat, life can be a nightmare.
Florida Confronts its Past and Takes Charge of its Future
While hardly unique, Florida’s legislature faces virtually every one of the problems described above, amplified by the sheer size and diversity of the state’s land area and population. The Florida House and Senate have traditionally been supported jointly by an IT services organization, OLITS. However, the many well-intentioned attempts at building a coherent automation environment resulted in systems that worked but were too difficult to maintain, too expensive, and fell short of their operational goals. It was clear that things could be better.
In the spring of 2002, the Senate, through the support of its leadership, elected to be one of the first of its ilk to rethink and redesign its automation suite on a new and materially different model. To accomplish this major endeavor, the Senate entered into a contract with Infinity Software Development, Inc. and its partners, Xsystems, Software AG, and Arbortext.
Information Flow Automation Based on Consensus
A key foundation of this model was the realization that, viewed honestly, most successful information technology efforts are based on consensus among the providers, managers, and consumers of information resources. If everyone agrees from day one, projects tend to succeed. Previous efforts in the Florida legislature, as in most organizations, had aimed at fostering this consensus on a technological level. Unfortunately, convincing each component of a large, complex organization to acquire and use technology identical or similar to its counterparts has proven to be very challenging. People, and through them their organizations, like to control what they are responsible for, be it location, equipment, technology, or office color schemes. With this background, it’s clear that any attempt to solve information automation challenges by enforcing common technology is likely to create problems far beyond the goals of the effort, often dooming it to failure or marginal results.
In many organizations, the Florida legislature included, the answer to these failures has often been an even more intense focus on technology. This approach has often failed to work well, regardless of the state of technology. Florida’s Senate came to understand this and resolved to find another, more fertile, path.
Senate Adopts a New Approach Based on Content
The Senate reasoned that it must find a new way of fostering and supporting consensus among and within the groups that have a stake in the legislative information system or LIS. The approach it chose could be described as content-centric, distinguishing it from the technology-centric way the Senate had operated in the past. A content-centric approach to automation assumes that the stakeholders may not want or be able to adopt common technology, or even to agree on technological approaches. Accordingly, the focus is shifted to adoption of common formats for the exchange of data. This sounds so simple and self-evident that it has often been overlooked when project sponsors are searching for a vehicle to integrate their diverse participants. In a content-centric environment, each participating group may create, manage, and deliver content any way it wishes as long as the content meets the agreed model. Conversely, each group must use the common content repository to access data, but it may do so by any standard technical means it wishes.
If content is to be the focus, then a unified content modeling approach becomes a critical component in project success. For applications based on complex text, typical in the legislative world, the relational database model falls short, requiring an alternative capable of handling both text and element content. In today’s world, XML (eXtensible Markup Language) is becoming the lingua franca of content interchange. Because the Senate has had some experience with SGML, XML was the obvious content modeling choice. XML provides a minimally intrusive way of asking organizations to agree and collaborate, often with dramatic affects on their willingness and ability to embrace and support a project. The Senate set a series of goals, some of which are summarized below:
First, minimize the impact on existing technology planning and implementation.
Nearly every organization has an automation plan in place based on a certain life cycle of equipment and software. Interrupting that plan can cause serious dislocations to budget planning and organizational return on investment, especially if the new project requires early replacement or addition of software or hardware resources. The Senate’s content-centric approach offered a significant and visible reduction in the likelihood of this kind of dislocation and the “push-back” it generates. The expectation was that, freed from the demand to upset their existing technological environments, participating organizations would be more willing to support the new project and to adopt the content consensus on which it was based.
Second, allow each organization and its staff to continue making its own technology decisions, with the more easily followed proviso that it merely be able to access, import, and export XML in the formats agreed upon for the project.
By lowering the level of intrusion inherent in the new project, the Senate makes it easier for impacted organizations to “opt in” while continuing their own planning and implementation activities. Moreover, while the new approach does not mandate painful technology changes, it provides incentives for participants, over time, to voluntarily elect technologies that will allow them to better
collaborate with their counterparts, creating a soft push toward technology congruence.
Finally, simplify the development and life-cycle maintenance of legislative information applications.
After years of expensive and complex automation support, the Senate was ready for a world that would materially lower the cost and complexity of every player’s participation. It was reasoned that adoption of an XML content basis for major information architectures, would allow the Senate to reach this goal. XML is supported by a rich and cost-effective set of flexible software resources, both embedded and stand-alone. For the developer who wishes to generate and support unique application behavior for his own organization, there is no better foundation than XML and its associated standards. This can be especially useful to the developer maintaining data in relational databases, as is the case in Florida. XML, for the first time in IT history, provides an effective bridge between database technology and text. The fact that XML is an international standard and XML content is inherently portable provides an increased level of protection against excessive dependence on vendors and imposes a predictable modification process that precludes the type of rapid change that can keep users and developers running to catch up.
Florida Takes the Plunge
Working with a team of consultants, the Florida Senate began its content-centric design effort in early 2003. The first phase of the project included a detailed analysis of the Senate’s existing processes and automation systems. The project then identified those processes that had representations in content, inventoried current system contents, and developed a comprehensive XML data model that represented all required data elements, structures, and relationships. This task, while not trivial, was simplified by the fact that XML fully supports the expression of text, element data, and metadata in the same model, using the rich hierarchical nature of XML to keep track of the logical components and their relationships.
The design phase was completed in October 2003. In the second and subsequent phases of the project, some existing Senate systems will be replaced by new applications built on the coherent, XML-based model and development architecture. Others will be integrated into the new environment through the use of “mediator” applications to create an interface between the legacy application and the new XML environment. Developers will work in a layered environment with a common XML repository at its base, accessed by standard query protocols and controlled by a unified “business rules” layer. Actual application development will be supported in any of several selected languages, with the proviso that content will be managed in the repository using the selected protocols.
As the Senate completes the initial phase of the project, ongoing discussions with the House are aimed at establishing the necessary interface between House applications and the Senate’s XML repository. This interface also will follow the content-based integration path, avoiding a repeat of the unsuccessful technology focus of past efforts. The entire project will require several years to replace or integrate all Senate systems and to integrate Senate and House content, but the path will be coherent and easily maintained so that the end result is both effective and maintainable over its life.
It is becoming clear in practice that the rise of XML and content-centric design has given us a new set of logical and organizational tools with which to foster the type of consensus needed to make IT projects successful. While this reality has not yet penetrated into all sectors of the IT world, its truth is becoming increasingly obvious as experience, such as that of Florida, accumulates. Even the most rabid technologist can often be heard to pledge, “The project will succeed if we can just get everyone to play their assigned role and embrace the new resources.” With a content-centric approach to automation in our arsenal, we might just be within sight of the ability to make good on that pledge.
About the Author