21st Century Information Architectures

After almost 20 years of information management theory, movements and gurus, why isn't anything any better? Critical success factors continue to be elusive; perhaps the most frustrating has been meta data and information management. Corporations recognize information as an asset, but still can't treat information as an asset. The vendor community continues to provide lip service to standards while producing targeted, product-specific solutions. At the root of all of this seeming failure of information management is that many of our previous and treasured concepts of information management may not address the current reality of information technology. No one has done anything wrong. Technology, application taxonomies and the explosive increase in business user reliance on information and collaborative intelligence have modified the rules.¹ In this environment, the current mind-sets of information management will never work. Corporations cannot exploit the information asset without new processes and perspectives on information.

There is tremendous emphasis placed on meta data standards, repositories, stewardship and alignment. Some of these are still good ideas. Others will be placed into a new perspective. The bottom line is simple - information as it is known is evolving outside of its traditional role, and the inadequate tools and philosophies of the past are even less adequate now. IT departments must overcome several barriers to effective use of information.

Mind-Sets

The first barrier to change is the mind-set of information managers toward meta data and information management (IM). The good news is that meta data management has proven to be vital to project success.² Use of information models and repository layers increases utility and relevance of applications. SAP's architecture is a splendid example of this, but there is a trap lurking under these supposed examples of current IM thought.

The first mind-set to toss aside is the relevance of a comprehensive global repository for an enterprise. Meta data has created successful data warehouses, data marts, etc., but less than a handful of organizations have implemented an enterprise repository product with any hint of success. Few, if any, of these implementations are based on a global repository. Those companies using repositories for impact analysis and management of operational applications rarely tie use of the repository into decision support. This is not saying there haven't been some successes. The track record is too disappointing to call the current instantiations of repositories a success.

History aside, the real reason to avoid pursuit of the classic repository is simple - the idea they represent is irrelevant to most businesses. Corporations will acquire their operational application "meta layers" via an integrated package such as SAP, PeopleSoft or Oracle. Or they will choose to purchase disparate applications that are functionally adequate. Within the confines of a product such as SAP, all is semantically consistent. Outside the tool, traditional information chaos reigns. The reason for this is few organizations have 100 percent implementations of Oracle, SAP or PeopleSoft. Many have some of all three, plus a few supply chain products thrown in. This means, by default, the organizations must use different definitions and business rules within subject areas. The relevance of a repository is diminished in the face of practicality.

The next mind-set to scrutinize is "information as an asset." Information (and knowledge) is an asset, but the current treatment is out of date. Currently, the repository should function as the overseer, with stewards and owners acting to implement and enforce rules and standards. Through this process, great universal changes can be made to applications, and the organization knows at a glance the relevance of a piece of data.

Unfortunately, this view ignores reality. The successful uses of meta data tools do not track this path. Successful uses of meta data are usually navigational and knowledge-based, i.e., the user knows where to get information, and why and how it got there.³ Modern repositories barely have any capability in navigation. Current repositories are acceptable at the administration of data and contain some level of definitions (but poor semantics and synonym management).

Another mind-set is that information has to be in rows and columns to be useful to be real business intelligence. However, we all realize that a great deal of the actions and decisions in a business are based on unstructured information as well. In fact, the types of content available in the 21^st century up the ante on any meta data strategy.

There are different categories of information, knowledge and data. Therefore, the blanket stewardship/ownership of a piece of information is subject to the context within which the information exists. Stewardship and ownership are dynamic within the enterprise. Rules may be put in place to make stewards and owners accountable, but where is the business measure of accountability? An asset has value. Many authors (e.g., Karl Sveiby) are working on valuation of information and knowledge. As such, tools and process will have to treat information with the rigor assigned to money and labor. Again, the concept of the universal semantic repository cannot do this.

The usage and context of a particular piece of information determine who is to be the caretaker and definer of that information, not a global concept of steward and owner. This simplified example shows how one simple element can be viewed and treated across an organization.

Without a business goal and business context, stewards and owners are not accountable and interest wanes rapidly. In addition, structured information has received all of the attention of the tool and rules-makers. However, most of the powerful content in an organization is unstructured. No tools to date deal effectively with structured and unstructured information (unstructured data includes digital media, text and documents).