Throughout the 1980s, requirements rose while competition intensified in our segment of the engineering field, which can broadly be called scientific data processing. As hardware got faster, generic workstations became better at meeting customer requirements without need for the clever tricks we used at Edom Engineering to achieve maximum performance. Meanwhile, the standard libraries and advanced features of Unix operating systems grew to the point where even our crack programming staff couldn't reproduce everything customers expected.

Our core scientific and engineering market was also shrinking because of an unrelated external factor: the U.S. military in the 1980s reduced the research funding upon which many of our customers depended.

It was in the late 1980s when Edom Engineering managers decided on a leap that they hoped would establish a new beachhead in advanced computing. Throughout its existence, we had been happy basing our systems on Motorola chips. But the new wave of Reduced Instruction Set Computing (RISC) processors promised much better performance. The sages among our hardware engineering staff checked out these processors and selected one that they said could deliver the performance we needed.

The impetus behind RISC processors was the increasingly multilayered decision-making required within conventional chips from Intel and Motorola, which computer scientists now categorized with the demeaning term Complex Instruction Set Computing (CISC). Conventional chips got that name because they contained large numbers of instructions, some narrowly tailored to particular software tasks. Different instructions sometimes require different numbers of cycles, forcing the chip's engineers on the hardware side, as well as compiler developers on the software side, to build in more complex scheduling. The weight of all these components led to a need for more wires connecting the processing unit to memory.

By the 1980s, computer scientists determined that increased overhead from these trends was eating up processor power, and concluded that they could create a competitive new generation of chips using a stripped-down set of very short, simple instructions that behaved in rigorously similar ways in a fixed number of cycles. Implementations from various companies hit the market in that decade. RISC became the major story in the trade press, and benchmarks bore out the inventors' predictions.

Thus began the Longjump project at Edom Engineering. It was fraught with uncertainty from the start. We were basing a system on a new product based on a new computing paradigm, and depending on a small firm with an unknown track record. At the same time as the hardware team wrapped the chip into an audacious new system trying to take advantage of every possible feature for speed, the software teams had to port our unique, finely tuned operating system, compiler, and libraries. Because RISC was so different an architecture from the Motorola chips we had always used, we had to base our compiler and a lot of our system utilities on those provided by the RISC vendor. The porting effort steered through uncharted waters.

An engineer I'll call George was appointed project leader for Longjump. George was a relatively young project leader, an earnest, restless fellow with an unusual personality for an engineer in those days. He bore a trim mustache, kept his shoulders slightly hunched as he sat in meetings, and looked about with tight lips that got tighter when he listened to news he didn't want to hear. He had a foot in marketing as well as engineering, and tended to wear suit jackets in an age when traditional computer nerd attire was even sloppier than it is today. He must have known that a more professional look would bolster his credibility when meeting with top management. And because this was his first attempt as project leader, he felt the need to convey credibility.

At the company meeting where Longjump was presented, George stood before the assembled management and staff to make a brief speech that included a seemingly formulaic phrase: "I'd like to thank the company's managers for entrusting me with the leadership of this project."

I don't think he anticipated how the rest of us would react to this gracious statement. We all knew Longjump was a hazardous undertaking, and this was no moment for even a hint of timidity. By thanking management for their trust, he planted in our minds the seed of doubt that he could live up to it.

But ultimately, if George played any counterproductive role, it was by showing too much deference to his managers as they directed the project into stagnant channels.