Part I: Computer Obliteracy

Chapter 1. Riddles for the Information Age

What Do You Get When You Cross a Computer with an Airplane?

What Do You Get When You Cross a Computer with a Camera?

What Do You Get When You Cross a Computer with an Alarm Clock?

What Do You Get When You Cross a Computer with a Car?

What Do You Get When You Cross a Computer with a Bank?

Computers Make It Easy to Get into Trouble

Commercial Software Suffers, Too

What Do You Get When You Cross a Computer with a Warship?

Techno-Rage

An Industry in Denial

The Origins of This Book

What Do You Get When You Cross a Computer with an Airplane?

In December 1995, American Airlines Flight 965 departed from Miami on a regularly scheduled trip to Cali, Columbia. On the landing approach, the pilot of the 757 needed to select the next radio-navigation fix, named "ROZO." He entered an "R" into his navigation computer. The computer returned a list of nearby navigation fixes starting with "R," and the pilot selected the first of these, whose latitude and longitude appeared to be correct. Unfortunately, instead of "ROZO," the pilot selected "ROMEO," 132 miles to the northeast. The jet was southbound, descending into a valley that runs north-south, and any lateral deviation was dangerous. Following indications on the flight computer, the pilot began an easterly turn and slammed into a granite peak at 10,000 feet. One hundred and fifty-two passengers and all eight crewmembers aboard perished. Four passengers survived with serious injuries. The National Transportation Safety Board investigated, and—as usual—declared the problem human error. The navigational aid the pilot was following was valid, but not for the landing procedure at Cali. In the literal definition of the phrase, this was indeed human error, because the pilot selected the wrong fix. However, in the larger picture, it wasn't the pilot's fault at all.

The front panel of the airplane's navigation computer showed the currently selected navigation fix and a course-deviation indicator. When the plane is on course, the needle is centered, but the needle gives no indication whatsoever about the correctness of the selected radio beacon. The gauge looks pretty much the same just before landing as it does just before crashing. The computer told the pilot he was tracking precisely to the beacon he had selected. Unfortunately, it neglected to tell him the beacon he selected was a fatal choice.

Communications can be precise and exacting while still being tragically wrong. This happens all too frequently when we communicate with computers, and computers are invading every aspect of our modern lives. From the planes we fly to just about every consumer product and service, computers are ubiquitous, and so is their characteristically poor way of communicating and behaving.

There is a widely told joke in the computer industry that goes like this: A man is flying in a small airplane and is lost in the clouds. He descends until he spots an office building and yells to a man in an open window, "Where am I?" The man replies, "You are in an airplane about 100 feet above the ground." The pilot immediately turns to the proper course, spots the airport, and lands. His astonished passenger asks how the pilot figured out which way to go. The pilot replies, "The answer the man gave me was completely correct and factual, yet it was no help whatsoever, so I knew immediately he was a software engineer who worked for Microsoft, and I know where Microsoft's building is in relation to the airport."

When seen in the light of the tragedy of Flight 965, the humor of the joke is macabre, yet professionals in the digital world tell it gleefully and frequently because it highlights a fundamental truth about computers: They may tell us facts, but they don't inform us. They may guide us with precision, but they don't guide us where we want to go. The flight computer on Flight 965 could easily have told the pilots that "ROMEO" was not an appropriate fix for their approach to Cali. Even a simple hint that it was "unusual" or "unfamiliar" could have saved the airplane. Instead, it seemed as though the computer was utterly unconcerned with the actual flight and its passengers. It cared only about its own internal computations .

Hard-to-use computers affect us all, sometimes fatally. Software-based products are not inherently hard to use; they are that way because we use the wrong process for creating them. In this book, I intend to reveal this bad process by showing its effect and describing its cause. I'll then show how to change the process so that our software-based products become friendly, powerful, and desirable. First, I'll use this chapter to show how serious this problem really is.