The Intelligence Behind Successful Software Management: “I Want Predictability”
- “What do you want for Christmas?” I asked a neighbor of mine in northern Virginia, at a little pre-Christmas good-cheer gathering. He is the chief executive officer of a mid-sized company.
- “Well,” he responded with a smile, “I’ve got a couple hundred neckties, and I don’t even ‘suit-up’ every day anymore.” Then his smile faded. “You know what I really want? I want software development to be predictable.”
- He knew I had a software company, but we had never talked about it before.
- “I want my software people to be able to predict the cost and schedule of every job they undertake,” he went on. “I want them to be able to predict the reliability of the product. I guess I would even like to have some glimpse beforehand of its quality.”
- “Funny you should say that,” I said. “That’s the business I’m in.”
- “You know, in every other part of my business, the executives in charge are able to do those things. They forecast their budgets, and they come pretty close to them. If they are bidding on an outside job, they come up with a price and a delivery date, and I can count on their coming close to those numbers. Not so, my CIO. He seems to be operating in the dark.”
- “Yes, we’ve found that about one third of the software projects on which we have collected data exceeded their budgets or their schedules by more than thirty percent.” I started to pull up more statistics from memory.
- Just then, the pretty lady from down the street shoved an empty glass between us. “Would one of you gentlemen be kind enough to push through that crowd at the bar?” she asked. That was code for “Don’t be so serious at our holiday party.”
- “Delighted,” I said, forcing a smile as I took the glass.
- “Stop by my office,” the CEO called after me as I disappeared into the crowd.
The Software Situation Is Serious
“Software development projects are in chaos,” The Standish Group International reported.1 Only 16 percent of projects are “completed on time and on budget, with all features and functions as initially specified.” Thirty-one percent are cancelled during development. Fifty-three percent are completed short of the original goals for schedule, budget, or functionality.
In other words, if software projects were bridges, about one third would fall down during construction; about one half wouldn’t get all the way across the river; and only one sixth would carry traffic. The ferrymen would not have to worry about planning a career change.
Unfortunately, as software takes over more and more business and governmental tasks, we do have to worry. Some of us even retreated into the woods as the dreaded changeover to the Year 2000 approached, fearful that the software on which modern life depends would fail. Fortunately, it didn’t, though its successful operation came at a tremendous fixing cost.
The Standish Group estimates that there are 175,000 projects each year in the United States alone, costing $250 billion. The cost of failures and overruns eats up a large fraction of that huge sum. No one can put a price tag on the cost in agony of the people involved, but we suspect that young people are avoiding the software field and older ones are leaving it. American companies increasingly depend on developers from overseas who may be more willing to endure the agony.