Who Can Hold 2 Billion Transistors in His Head at Once?

It’s impossible to do engineering anymore without using mostly other people’s knowledge


I was visiting a high-tech company whose principal business was advanced chip design. A young engineer showed me his latest prototype. It was a circuit board dominated by a single large integrated circuit. It contained, he told me, more than 2 billion transistors.

I’d never done anything remotely similar myself, and I wondered how I’d feel as a new engineer in some company being given a “Mission: Impossible” assignment like that. “How can you possibly design something so complex?” I asked.

The young engineer just shrugged his shoulders. No big deal, he appeared to be saying. After a little further prodding, he confessed that, after all, there was a large collection of previously designed cells, and the computer design tools were pretty awesome. Not much to it, really.

Technology has gotten exponentially more complex with the passing years, and yet engineers are turned out of universities in the same four-year cycle that they used decades ago. How is this possible?

I remember dimly that when I got out of school, I thought I knew all there was to know about electrical engineering and that, to use the term of the old trade unions, I was a “master of engineering.” Of course, that wasn’t right, even back then, but now there is no pretense: No one knows all of electrical engineering. Our profession has splintered into many specialties, and they, in turn, into subspecialties.

As with any deeply complex problem, the approach in engineering design is to partition the task into noninteracting subsystems and attack each independently. Thus our world has become compartmentalized and layered, with well-defined interfaces between the layers. It is as if each had a warning label that said, “No user-serviceable parts inside.”

The young engineer working on chip design depends on the work of other people, in the adjoining layers, who have already designed the cells he uses and who worried about the electrical characteristics at the base circuit level. Still others will, in turn, take his chip for granted and write software to produce functionality, implementing algorithms derived by theoreticians, while still others will worry about the user interface.

The young engineer will work shoulder to shoulder with a computer whose software tools codify design rules that have been learned through the years of experience by many other engineers—intricacies none of them know except collectively. Where necessary the software tools will do the math for him—math he may have forgotten or never learned. In every direction around him, he will rely on other people’s knowledge. The bad news is that this engineer will never know all of engineering; the good news is that he doesn’t have to. The world doesn’t work that way anymore.

What is wonderful today is how easy it is to tap into other people’s knowledge. In the olden days—that is, before the Internet—we had to find the right person to ask for help, or we had to wade through libraries for stale, encrusted information. Now we start by asking the world at large: “How do I do or fix this?” Chances are, we’re not the first to ask, and some stranger has already provided the answer. We can get continuously updated tutorials from Wikipedia, and there is an ever-growing collection of online lectures on technical subjects. There may even be a YouTube video titled “How to Design a 2-Billion-Transistor Circuit.”

In such a world, what does the engineer need to know? First, how to skillfully acquire information—but today, everyone is fairly good at that. Next, he or she should know something about all the layers, a good deal about his or her own layer, and even more about some niche within it. But most engineers are good at turning the crank once a problem is defined; the talent that characterizes the best engineers is realizing what problems need to be solved.

After a few years out of school, much of the knowledge an engineer applies will have been learned on the job, through continuous education and just-in-time knowledge acquisition. His or her university education will have provided the fundamentals and core principles of engineering—whatever they are. I’m not so sure anymore.

This article originally appeared in print as “Other People’s Knowledge.”

#ieee #spectrum  JANUARY 2013