A new president had just taken over the Bison Valley Ax Works. He was selected by the Board of Directors because he had "vision." Ogg, the Cro-Magnon design engineer, was skeptical. He asked his colleague Grnk, "Did they say he had vision, or visions?"
This was all clarified when Mr. Big addressed his collected employees in his first of many pep-talks. "I have a vision of a mega-ax—an ax so advanced, our competitors will be struck down in awe. We will use materials that are lighter and tougher than wood and more durable than stone. We will make it user-programmable, and we will combine advanced intelligence," Mr. Big said. "A mega-ax by 10,000 B.C. That is our slogan!"
Well, 10,000 B.C. came and went, and the project was late. An organization that had once proudly executed its projects on time and on budget was hopelessly overspent and delayed. This is a familiar story to any of you who have ever worked on a megaproject. Why do these megaprojects so frequently self-destruct? There are several important reasons.
First, there are marketing challenges. Megaprojects try to forecast long-term customer requirements. The difficulty of doing this rises exponentially with the distance to the forecasting horizon. Forecasting a market four years in advance isn't twice as hard as forecasting a market that's two years away—it's at least 10 times harder.
Second, there are technical challenges. The more elements of a system we change simultaneously, the more complex our system-integration problems become. The number of interactions between elements rise roughly as 2n, where n is the number of elements. This means that we have more potential failure modes.
Third, it's hard to motivate teams working on megaprojects. The first year of a five-year program has no urgency. The fifth and sixth years of the same project are complete panic. The payoff on a megaproject is simply too far off to be an effective motivator.
Fourth, long projects acquire a life of their own. The more money that has been spent on a bad idea, the more likely it is that managers have gone on the record assuring the boss that everything is under control. It becomes very hard for the managers to tell the boss that everything he's been told for the last four years was wrong.
Fifth, megaprojects tend to pick up excess baggage along the way. They are so big that they require senior management support. This creates an interesting effect. The megaproject begins to operate under different rules than other programs. It has a large checkbook, and the company depends on its success. It can't not be supported. This makes it an attractive target for anybody who needs resources to accomplish a pet activity.
How can we avoid the megaproject death spiral? First, recognize that costs of large projects rise non-linearly. When in doubt, go small. Second, set a number of intermediate deadlines for shippable versions of the project. You don't need all the the capability in the first product that you introduce. Third, if your megaproject is going down the tubes, don't be afraid to put new people in charge. Since new leaders have no vested interest in what has already gone by, they are more likely to look at the project objectively.
With a little care, we can take the mega out of megaprojects.