Another Information System Fails--Why?

It seems as though we hear every month of yet another multimillion-dollar information technology failure in the California state government. In the latest case, a system to link welfare networks was abandoned after the expenditure of $18 million. It is time we learned the lessons of why these systems fail. To date, many managerial, oversight and technical problems have been identified, but two very important elements are underplayed or missing altogether in the discussion.

The first missing element is a deep understanding of the human factor in technology implementation and use. Information technology often is looked upon as something you go out and buy and install, like bringing home a new lamp and plugging it in. But, in fact, to function effectively, information technology has to be fully and successfully integrated into the activities of the people who use it.

A new automated information system, especially one built around the core activities of an agency, will change almost every activity of the agency and will inevitably alter power positions and job descriptions of everyone along the way. A manager’s power may be increased because the new information system enables the integration of two departments under one command, or it may be weakened because fewer workers are now needed. That manager will not be passive about these changes.

Even the lowliest clerk may dig in and sabotage the project because it necessitates a move from an office with a window to one without a window. High-level managers commissioning new systems ignore these factors at their peril--yet they often do ignore them.

The other big piece missing from the discussion is information expertise--an understanding of how people search for and use information, and how best to organize information within the computer to facilitate retrieval. For example, a system that holds the names of 50,000 people will not be too problematic when retrieving duplicate names; with a system of 5 million names, the duplication problem explodes and can halt the entire system unless sophisticated solutions are employed. More generally, information retrieval systems of all kinds are notoriously size-sensitive and do not scale up well without significant design changes.

In most real-world information systems, simple, conventional database structures are not adequate for handling the messiness of the information.

Imagine a child welfare agency’s information system: At any given moment, a child may be under several governmental jurisdictions, may be at one stage of a lengthy, multistage process determining the disposition of the child’s case and may be under a complicated arrangement set up by the court for visitation by a parent or guardian.

The design of a system to organize such information so that it is retrievable and useful for the various agencies and individuals who need to access that file is a major task in and of itself. It requires expertise not only in computer systems, but also in information organization and retrieval techniques. Each type of information in that child’s file may need to be indexed according to different principles of categorization, with different search capabilities made available for the file users for each type of information.

For example, one type of information may lend itself best to a classification using just a few categories; another type may best be searched using a thesaurus of technical vocabulary in use in that particular agency. Information-related design skills do not replace computer systems analysis; rather, people with these skills should be working with the systems analysts--and their expertise should be given equal weight with that of the analysts.

The technology is highly visible; the information, and the social meaning of the information technology, are not so visible. Usually, in the design of such systems, great attention is paid to buying the computers and programming them, while the human and information-related factors are shortchanged or ignored altogether.