Leviathan

By Douglas A. Samuelson

Long ago, in a land far, far away, the king's ministers needed to understand the workings of a large, complex system. They hired a company to build them a model that would help them understand: "What if...?"

The company proceeded to build a large, detailed model. It was, in fact, so large that it required massive amounts of data to feed it and keep it functioning. So Harry and Tom, the two analysts who programmed most of it, were joined, in time, by two almost equally senior people whose job was to work with the clients to keep the model well-fed with data. These two data analysts and the project manager spent more and more of their time at the ministers' palace, cultivating the company's good relationship with the clients and trying to acquire still more data.

In time Tom moved on to other work. This was fine with the project manager, as the real needs were for client liaison and data management, not more coding and testing and certainly not more design. He hired a couple of junior modelers to help with documentation, which would also help these analysts learn how to modify the model to do new analyses if anyone asked. This arrangement lowered the average cost per hour of the project, which also helped the company keep the ongoing business.

Eventually, senior management pushed to have another more senior person work on the model, so that there would not be a catastrophe if anything happened to Harry. This did not work out well, however, as the new senior modeler asked, "Why is this model so large, so detailed in so many ways, but also so cumbersome, inflexible, and hard to understand or modify?" The program manager pointed out sternly, "We keep this line of work because our model predicts so accurately and in such detail. Look at these printouts! Look at these charts! How can you criticize such a successful effort?"

So the model grew and grew. The new senior analyst called it Leviathan: "It's a monster," he explained, "and it will be harder and harder to keep feeding it, or to make it move in any new direction. You'll see."

This offended the data-gathering senior analysts and displeased the managers. After all, Leviathan was a big successãso much so that they were trying to sell more work with it to other organizations within the government. The last thing they needed was someone on their own team claiming Leviathan wouldn't even meet the needs of the first client much longer. The senior modeler was assigned to documentation. Not surprisingly, he was not much faster at it than the junior modelers were. "You are not cost-effective," the project manager told him. So he was gone.

They brought in another senior modeler. Leviathan continued to grow. They brought in another data analyst to work with the client. The program manager spent even more time—nearly full time, in fact—at the client's site. The company that made the modeling software got bought out, the next "upgrade" made the software unstable, and major modifications were needed to keep Leviathan running.

"This would be a good time to switch software packages," the new senior modeler recommended.

"Nonsense!" the project manager barked. "We told the client this was the right package. We'll look bad if we tell them it isn't the right package now." And Leviathan continued to grow.

The junior modelers kept quiet about the whole thing and plugged away diligently making the changes. The new senior modeler was assigned to documentation. Like his predecessor, he was found to be not cost-effective and removed from the project. Besides, the junior modelers were obviously doing quite well enough, as far as the project manager could tell. "The younger people seem to be quicker to learn, and they're less expensive anyway," he beamed. "And they don't ask obstructive, argumentative questions."

About this time, the clients decided to try a whole new technological approach to running their big, complex system. "Run the model for us," they said, "and tell us what to do."

The project manager consulted with the senior data analysts and announced that the company would be happy to do this, for a large sum of money. This would be needed because a vast amount of new data would have to be collected, and a great deal of code modification would be required to make sure the model tracked the new data.

"We already paid you a great deal of money to develop this model," the ministers replied, "and you have convinced us it predicts nearly perfectly. Show us how the new approach will work, NOW, or we will find someone who will."

At this point, Harry decided to retire.

Because of a strange reluctance by other senior modelers to take on the task, the managers suddenly found the estimated cost of the new task skyrocketing. For some reason, none of the other offices and agencies wanted to be first to sponsor new development work. "If the office that paid for it can't use it now," they said, "how can we?"

So Leviathan sits, unused and unsupported, still magnificent in its ability to track perfectly a set of conditions that no longer exist in real life. The program manager and the senior data analysts, meanwhile, have been promoted and assigned to new projects within the company. After all, their previous project was a shining success—wasn't it?



Douglas A. Samuelson is president of InfoLogix, Inc., a consulting company in Annandale, Va. He is also an adjunct professor at The George Washington University and at the University of Pennsylvania, and an external research professor at the Krasnow Institute, George Mason University.