Our Computational Heritage

In the sixties, there was a generation of engineers trained in both slide rules and computers. Those who went before never really adopted computers, and those who went after never really learned how to compute by hand. But this group of young engineers received their most intensive engineering education working with both the tools of the previous generation and the tools of the next at a time when the tools couldn't have been more different. This, in turn, cultivated an entire behavioral culture toward engineering results. Among the tenets were the following:

• Accept no answer just because you had computed it. Question everything. By the time we had gotten to computers, this principle had morphed its way to "trust no input data."

• Break complex computations into reasonable-sized chunks, and check every intermediate result to avoid error propagation.

• For both manual and computer calculations, figure out a way to debug your computation so that when you do generate a patently ridiculous result, you have a built-in way to sort it out.

These ideas were not abstractions; they were keys to survival. And from a pragmatic point of view, they became instincts that enabled survival in a cold, ruthless world of problem-solving.

My theory is that this golden generation of engineers became the leaders of today's Silicon Valley colossus, and that we are now at risk as this generation approaches retirement age. We have bigger, faster, more powerful computers, and magnificent software running on them. But who is going to tell us when the answers are wrong?